The diversity of evolutionary lineages with little phenotypic differences (i.e., cryptic species) might be better understood in the light of genetic delimitation of evolutionary units (Thomé et al. 2012, Hambäck et al. 2013). Recent molecular phylogenies of anuran, including work on species from the Brazilian Atlantic Forest, did not recovered as monophyletic the species groups clustered based mostly on morphology (Amaro et al. 2009, Canedo and Haddad 2012, Fouquet et al. 2012, Thomé et al. 2012).

Species are segments of population level evolutionary lineages and do not necessarily need to be phenetically distinguishable, diagnosable, monophyletic, intrinsically reproductively isolated, ecologically divergent, or anything else to be considered species, but they only have to be evolving separately from other lineages (de Queiroz 1998, de Queiroz 2007).

A recent molecular phylogeny (de Sá et al. 2012) recovered a polyphyletic Chiasmocleis and, to render the genus monophyletic, transferred one species to Elachistocleis and three species to Syncope. Recently, Peloso et al. (2014) placed Syncope in the synonymy of Chiasmocleis. Chiasmocleis is the most diverse genus of Neotropical microhylids, with 29 species distributed throughout Amazonia, Atlantic Forest, and open areas in South America, such as the Brazilian Cerrado and the Chaco of Bolivia and Paraguay (Cruz et al. 1997, de Sá et al. 2012, Peloso et al. 2014).

Tonini et al. (2013) in a phylogeographic analysis estimated high genetic divergence and low gene flow among populations of Chiasmocleis lacrimae (described as Chiasmocleis carvalhoi Cruz et al. 1997) in the Brazilian Atlantic Forest. Samples of two potential new species and of “Chiasmocleis capixaba” with less feet webbing were included as populations of Chiasmocleis lacrimae. Moreover, the study suggested that populations isolated-by-distance could represent recently diversified species, estimated to Miocene and Pliocene. Increasing the sampling along the distribution of Chiasmocleis lacrimae and Chiasmocleis capixaba and using additional molecular and morphological data, we were able to differentiate the phylogenetic structure and morphological differences associate to intraspecific and interspecific variation. We found that Chiasmocleis lacrimae and Chiasmocleis capixaba were not recovered as monophyletic, in fact populations corresponding to undescribed distinct evolutionary lineages. Although these undescribed lineages have similar body size and shape, and low nuclear divergence, they are exceptionally divergence in mitochondrial markers and are geographically structured.

Herein, we describe a new species of Chiasmocleis from the Atlantic Forest of southeastern Brazil and present a phylogenetic hypothesis for the species group.

Specimens and tissues used herein and comparative material are deposited in the following collections: 1) CFBH: Coleção de Anfíbios Célio Fernando Baptista Haddad, Departamento de Zoologia, Universidade Estadual Paulista Rio Claro, Rio Claro, São Paulo State, Brazil; 2) MNRJ: Museu Nacional do Rio de Janeiro, Rio de Janeiro, Rio de Janeiro State, Brazil; 3) Museu de Zoologia, Universidade de São Paulo, São Paulo, São Paulo State, Brazil; 4) MBML: Museu de Biologia Mello Leitão, Santa Teresa, Espírito Santo State, Brazil; 5) CTA: Coleção de Tecidos e DNA da Universidade Federal do Espírito Santo (UFES) and LGA: Laboratório de Genética Animal, Vitória, Espírito Santo State, Brazil; 6) RN and CTRN: Universidade Federal Rural do Rio de Janeiro, Seropédica, Rio de Janeiro State, Brazil. Field numbers correspond to M. T. Rodrigues (MTR), Universidade de São Paulo, São Paulo, São Paulo State, Brazil; P. Rocha (PEU), Universidade Federal da Bahia, Salvador, Bahia State, Brazil; and J. F. R. Tonini (JFRT), vouchers are at UFES. Specimens examined and tissues samples are listed in Appendix 1 and 2, respectively, and sample localities are shown in Figure 1.

The following measurements were adapted from Duellman (2001) and Peloso and Sturaro (2008); measurements were taken for 56 individuals with a digital caliper under a stereomicroscope to the nearest 0.1 mm: SVL (snout-vent length); HDL (hand length; from the base of the thenar tubercle to the tip of the third finger); HDL4 (hand length from the base of the thenar tubercle to the tip of the fourth finger); HL (head length; from snout to angle of the jaw); HW (head width; between the angle of jaws); ED (eye diameter; between anterior and posterior corner of the eye); IOD (inter-orbital distance; distance between anterior corner of the eyes); IND (inter-nostril distance); END (eye-nostril distance; from the anterior corner of the eye to the posterior margin of nostril); THL (thigh length; from the center of the cloaca opening to the outer edge of the flexed knee); TBL (tibia length; from the outer edge of the flexed knee to the heel); FAL (forearm length); FL (foot length; from tibia-tarsal articulation to tip of fourth toe); 3FD (diameter of third finger disk); 4TD (diameter of fourth toe disk). Fingers and toes are numbered and abbreviated as follows: Fingers I–IV = FI–IV, Toes I–V = TI–V.

Molecular Analyses: Total genomic DNA was extracted from ethanol-preserved liver or muscle tissues using Qiagen DNeasy kit (Valencia, California, USA). We used four molecular markers (mtDNA: 12S, 16S, and NADH dehydrogenase subunit 2 [ND2]; nucDNA: brain-derived neurotrophic factor [BDNF]), amplified using previously published primer sets and PCR profiles (de Sá et al. 2012, Tonini et al. 2013). We performed a multiple loci alignment using an iterative procedure to compute a series of alignment/tree pairs in SATé-II (Liu et al. 2012), using default settings. GenBank accession numbers are given in Appendix 2.

The following outgroup were chosen based on published phylogenies including species of Chiasmocleis (de Sá et al. 2012): Chiasmocleis leucosticta, Chiasmocleis mantiqueira, Chiasmocleis crucis, Chiasmocleis schubarti, and Chiasmocleis cordeiroi. We selected a total of 100 samples (ingroup includes 69 samples) for a data set consisting of 2, 473 base pairs. The best partition schemes and substitution models (Table 1) were chosen using PARTITION FINDER v1.1.1 (Lanfear et al. 2012) and used in phylogenetic analysis downstream.

We applied two approaches of phylogenetic estimation: 1) Maximum Likelihood (ML) and 2) Bayesian inference (BI) using the dataset containing the markers 12S, 16S, ND2, and BDNF. Maximum Likelihood in RAXML v7.2.8 (Stamatakis 2006) used a rapid-bootstrap with 1000 replications. Bayesian Inference in BEAST v1.7.4 (Drummond et al. 2012) used birth-death process as tree prior, linked tree models across partition, relaxed clock model with linked mitochondrial markers, but not the nuclear gene. The BI analysis ran for 50 million generations and the parameters were sampled every 5, 000 generations producing a total of 10, 000 trees. We discarded the first 1, 000 trees as burnin in TREEANOTATOR. The output file was checked using TRACER v1.5 and values of Effective Sample Size >200 were considered suitable. Nodes having bootstrap values >70 in ML and posterior probabilities >0.95 in BI were considered as well supported. Analyses were performed through Cipres (Miller et al. 2010) and trees were visualized and edited using FIGTREE. Data available from the Dryad Digital Repository: http:// 10.5061/dryad.gm41t. Genetic distance (p-uncorrected) was calculated in MEGA5.0 (Tamura et al. 2011). A second species awaits description (Forlani et al. submitted) and it is referred throughout this manuscript as Chiasmocleis sp.

The phylogenetic hypotheses generated through ML (Figure 2) and the BI (Figure 3) resulted in similar topology. The ML tree (Figure 2) supported two new species as sister group of Chiasmocleis capixaba, Chiasmocleis lacrimae correspond to a basal node; whereas in the BI tree (Figure 3) Chiasmocleis capixaba was estimated as sister to Chiasmocleis lacrimae, but the posterior probability of this node was lower than 0.95. Both the ML and BI trees showed clades of Chiasmocleis leucosticta, Chiasmocleis mantiqueira, Chiasmocleis crucis, Chiasmocleis schubarti, and Chiasmocleis cordeiroi, but not Chiasmocleis lacrimae and Chiasmocleis capixaba (Figure 2, 3). Populations of Chiasmocleis capixaba and Chiasmocleis lacrimae from the north of the Espírito Santo State, as well as populations of Chiasmocleis lacrimae from southern areas of the Bahia State, would represent two new cryptic lineages closely related to Chiasmocleis lacrimae and Chiasmocleis capixaba. In the ML analysis populations from southern Espírito Santo formed a clade that makes Chiasmocleis lacrimae polyphyletic (Figure 2), whereas in the BI these populations formed a clade including also populations of Chiasmocleis lacrimae from the states of São Paulo and Rio de Janeiro (Figure 3). However, given the lack of support for this node in both analyses, basing taxonomic change on the presumed polyphyly is not warranted at present.

Therefore, our results show that the new species clusters within the genus Chiasmocleis.

Chiasmocleis quilombola sp. n. has been misidentified as Chiasmocleis lacrimae and Chiasmocleis capixaba due to an overlap in feet webbing and body size (e.g., Tonini et al. 2013, see below). Chiasmocleis quilombola sp. n. corresponds to clade N2, Chiasmocleis sp. to clade N1, Chiasmocleis capixaba to central clade, and Chiasmocleis lacrimae to southern clades of Tonini et al. (2013). Our morphological observations and comparisons with other species combined with molecular information support Chiasmocleis quilombola sp. n. and Chiasmocleis sp. as separate evolutionary lineages (see below and Figure 1, 2). Low levels of genetic divergence in the BDNF (Table 3) between Chiasmocleis quilombola sp. n. and closely related species is consistent with recent cladogenetic events and supports a previous study that estimated initial speciation within this clade (i.e., Chiasmocleis lacrimae, Chiasmocleis capixaba, Chiasmocleis quilombola sp. n., and Chiasmocleis sp.) during the Miocene/Pliocene (Tonini et al. 2013). Chiasmocleis quilombola sp. n. and Chiasmocleis sp. corresponds to an earlier lineage split dated to approximately the Pliocene/Pleistocene (Tonini et al. 2013).

Chiasmocleis quilombola sp. n. is distinct from Chiasmocleis schubarti (species with which occurs in sympatry) in having smaller snout-vent length, feet slightly webbed, cream ventral surface, and marbled light brown and cream dorsolateral pattern instead of larger snout-vent length, absence of feet webbing and belly pattern roughly marbled in dark brown and light cream in Chiasmocleis schubarti (Cruz et al. 1997). Chiasmocleis quilombola sp. n. is most similar to Chiasmocleis lacrimae and Chiasmocleis capixaba, species with which it has been previously confused (e.g. Tonini et al. 2013). However, the new species is distinguished from closely relatives by the following set of characters: 1) a smaller body size, shorter head length, shorter thigh and tibia compared to Chiasmocleis lacrimae, Chiasmocleis capixaba, and Chiasmocleis sp. (Table 4), 2) smaller eye diameter, inter-orbital distance, inter-nostril distance, diameter of third finger disk, and diameter of fourth toe disk than Chiasmocleis capixaba, 3) smaller eye-nostril distance, feet length, hand length, and hand length to the tip of the fourth finger than Chiasmocleis lacrimae. Moreover, males of Chiasmocleis quilombola sp. n. have less webbing on the foot (more extensive web on the foot in Chiasmocleis capixaba, absent in Chiasmocleis sp., and ranging from little to absent in Chiasmocleis lacrimae; Cruz et al. 1997, Peloso et al. 2014, Forlani et al. submitted). The new species has slender arms, legs, finger, and toes (robust arms and legs in Chiasmocleis lacrimae; thick fingers and toes in Chiasmocleis capixaba), as well as smaller and less abundant dermal spines in males (spines larger and abundant in Chiasmocleis lacrimae; abundant in Chiasmocleis capixaba). Males of Chiasmocleis quilombola sp. n. posses less amount of fringes between fingers II and III and a slender third finger than males Chiasmocleis capixaba.

Chiasmocleis quilombola sp. n. are distinguished from other Chiasmocleis species by: 1) four externally evident fingers and five toes distinguishes it from Chiasmocleis antenori, Chiasmocleis carvalhoi, and Chiasmocleis tridactyla (digit reduction; Walker 1973, Nelson 1975, Duellman and Mendelson 1995); 2) a shorter snout-vent length differentiate it from Chiasmocleis alagoanus, Chiasmocleis albopunctata, Chiasmocleis anatipes, Chiasmocleis atlantica, Chiasmocleis avilapiresae, Chiasmocleis bassleri, Chiasmocleis centralis, Chiasmocleis cordeiroi, Chiasmocleis crucis, Chiasmocleis devriesi, Chiasmocleis hudsoni, Chiasmocleis leucosticta, Chiasmocleis magnova, Chiasmocleis mehelyi, Chiasmocleis papachibe, Chiasmocleis royi, Chiasmocleis sapiranga, Chiasmocleis shudikarensis, Chiasmocleis supercilialba, and Chiasmocleis ventrimaculata (larger snout-vent length; Dunn 1949, Bokermann 1952, Walker and Duellman 1974, Caramaschi and Pimenta 2003, Cruz et al. 1997, Cruz et al. 1999, Caramaschi and Cruz 1997, Cruz et al. 2007a, Moravec and Köhler 2007, Peloso and Sturaro 2008, Funk and Cannatella 2009, Morales and McDiarmid 2009, Peloso et al. 2014); 3) small feet webbing of males and females distinguish the new species from Chiasmocleis cordeiroi, Chiasmocleis leucosticta, Chiasmocleis mantiqueira, and Chiasmocleis sapiranga (more extensive webbed feet in males and females; Cruz et al. 1997, Cruz et al. 2007a, b); 4) a light cream belly pattern without dark spots distinguished it from Chiasmocleis alagoanus, C. atlantica, Chiasmocleis haddadi, Chiasmocleis leucosticta, and Chiasmocleis mantiqueira (belly pattern roughly marbled in dark brown and pale cream, Cruz et al.1997, Cruz et al.1999, Cruz et al. 2007b, Peloso et al. 2014); and 5) snout rounded and belly light cream colored differentiate it from Chiasmocleis gnoma (snout truncate and belly boldly marbled in brown and pale cream; Canedo et al. 2004).

Chiasmocleis quilombola sp. n. occurs in sympatry with Chiasmocleis schubarti at the Floresta Nacional do Rio Preto, Municipality of Conceição da Barra, and at the Reserva Biológica Córrego Veado, Municipality of Pinheiros; it also occurs with Chiasmocleis capixaba and Chiasmocleis schubarti at the Reserva Natural Vale, Reserva Biológica de Sooretama, and at Cocoa plantations in Povoação, sites in the Municipality of Linhares. The new species is allopatric to Chiasmocleis sp. and Chiasmocleis lacrimae (Figure 1). We did not have access to tissues samples of Chiasmocleis capixaba from Nova Viçosa, Bahia State (Van Sluys 1998), to include in the genetic analysis, thus the phylogenetic relationship of this population remains unclear.

Cryptic species have challenged our ability to assess current levels of biodiversity. Anuran taxonomy has used various data sources to describe the species diversity, e.g., advertisement calls, external morphology, osteology, tadpoles, ecology, molecular data, karyotypes (Duellman and Trueb 1986, Haas 2003). However, Chiasmocleis systematics has been based on external morphology from adults and behavioral information (Wogel et al. 2004, Hartmann et al. 2002, Nascimento and Skuk 2006, Oliveira Filho and Giaretta 2006, Langone et al. 2007, Peloso and Sturaro 2008, Rodrigues et al. 2008, Santana et al. 2012, but see Peloso et al. 2014). Sexual dimorphism in size, amount of webbing, and color pattern have been useful characters to diagnose species (Cruz et al. 1997). Recent molecular studies 1) demonstrated the non-monophyly of traditionally recognized species groups (de Sá et al. 2012, Peloso et al. 2014) and 2) reported high genetic divergences and low gene flow along small geographical scales, suggesting that some populations could represent new species (Tonini et al. 2013). Given the current overall biodiversity crisis and specifically the worldwide threats to amphibian biodiversity, molecular studies should move beyond the identification of genetic clades and should make every effort to formally describe those evolutionary lineages. Herein, we have taken this approach and described a new species based on a combination of morphological characters in a clade of cryptic species with shown high genetic diversity and low gene flow.

The new species occupy coastal areas North of Espírito Santo State, a region that is under strong human pressure. Therefore, marine and coastal communities are susceptible to impacts of proposed modifications in the landscape for the exploitation of mineral resources. In this context, Chiasmocleis quilombola sp. n. may face imminent threat of habitat loss, as consequence of the deforestation and intensive occupation of the space by human activities.