Diversity of Chiasmocleis in the Amazon and Guiana Shield:

The type species of the genus is C. albopunctata (Boettger, 1885), originally described from Paraguay, but with several records in Brazil (see review in Forlani et al., 2011). The genus was proposed by Méhelÿ (1904) to accommodate the species Engystoma albopunctatum, based on putative morphological differences of the species and the remaining Engystoma. Hamptophryne boliviana (Parker, 1927) was originally described as Chiasmocleis but was later assigned to the monotypic genus Hamptophryne (Carvalho, 1954). Several empirical phylogenetic analyses support the distinctiveness of Hamptophryne in relation to Chiasmocleis (Frost et al., 2006; van der Meijden et al., 2007; Pyron and Wiens, 2011; Trueb et al., 2011) and the species is, therefore, not treated further here. Only a few species of Chiasmocleis were described until very recently: C. hudsoni Parker, 1940; C. ventrimaculata (Andersson, 1945); C. bassleri Dunn, 1949; C. shudikarensis Dunn, 1949; and C. anatipes Walker and Duellman, 1974. No additional species were described for the Amazon basin until early in the 21st century. From 2001 to the present, five new species were named in the region of study: C. jimi Caramaschi and Cruz, 2001; C. magnova Moravec and Köhler, 2007; C. avilapiresae Peloso and Sturaro, 2008; C. devriesi Funk and Cannatella, 2009; and C. supercilialba Morales and McDiarmid, 2009. Recently, de Sá et al. (2012) proposed an alternative taxonomy of microhylids, which affected the content of Chiasmocleis, and Syncope.

A Brief Taxonomic History of Syncope:

Syncope was proposed by Walker (1973) to include a diminutive microhylid species with fused presacral vertebrae I and II and reduced digits. On that occasion, the genus was proposed as monotypic, and the sole species Syncope antenori was named and described. Soon after that, Nelson (1975) described another species, Syncope carvalhoi, from the Loreto region, sharing the fused presacrals I and II and reduced digits. Duellman and Mendelson III (1995) described Adelophryne tridactyla from the Loreto region, Peru. Duellman and Mendelson's specimens also had reduced digits, but were assigned to Adelophryne without justification for its generic placement. Silva and Meinhardt (1999) noted several shared osteological characters between Adelophryne tridactyla and the two known species of Syncope, thus transferring A. tridactyla to the genus Syncope.

Phylogenetic Relationships and the Current Status of Chiasmocleis and Syncope:

Most of the characters used for definition of gastrophrynine genera lie on pectoral girdle elements, and this is no different for Chiasmocleis and Syncope. Some authors have commented on the overall similarity in osteology of the two genera (Walker, 1973; Duellman and Menselson III, 1995; Moravec and Köhler, 2007) and phylogenetic analyses based on morphology alone had found the two genera to be closely related (Zweifel, 1986; Wild, 1995).

Prior to de Sá et al. (2012), few species of Chiasmocleis (and none of Syncope) were included in any phylogenetic studies using molecular characters (but see Trueb et al., 2011). The only works to include more than one species were those of van der Meijden et al. (2007), Pyron and Wiens (2011), and Trueb et al. (2011),2 which included two species of Chiasmocleis (C. hudsoni and C. shudikarensis) in their analysis and found them to form a monophyletic clade, either sister to the remaining Gastrophryninae or sister to all gastrophrynines except Ctenophryne and Nelsonophryne (currently synonyms, de Sá et al., 2012). De Sá et al. (2012) included a much larger set of Chiasmocleis and Syncope species, and also included most New World genera of microhylids (missing only Adelastes), thus providing rigorous tests of the monophyly of Chiasmocleis and Syncope and their relationships with other microhylid taxa. Chiasmocleis was found to be polyphyletic, and to remedy this situation two changes were made to the classification (de Sá et al., 2012). Chiasmocleis panamensis was removed from the genus and transferred to Elachistocleis—several phenotypic characters corroborate this arrangement (i.e., osteology, advertisement call, our personal obs.). The second change is larger and more important to this work; de Sá et al. (2012) found C. magnova, C. bassleri, and C. hudsoni to be nested within Syncope Walker, 1973 (represented in their analysis by three putative species, one of them labeled as Syncope, sp.). They then decided to transfer C. bassleri, C. hudsoni, C. jimi, C. magnova, and C. supercilialba to Syncope, hence maintaining both genera as valid. The authors based their decision on the argument that they “want to recognize the separate evolutionary trajectory of this lineage based on shared morphological and life history traits” (de Sá et al., 2012). Although the taxonomic decision is in accordance with their preferred tree (maximum likelihood; see fig. 2)3 and with the International Code of Zoological Nomenclature (ICZN, 1999), their arguments are not convincing.

Fig. 2. 

Part of the phylogenetic tree of microhylids from de Sá et al. (2012), highlighting the relationship between “Chiasmocleis” and “Syncope.” Maximum likelihood (ML) tree of molecular data (16S: 673 aligned sites; BDNF: 711 aligned sites; tyrosinase: 551 aligned sites; 28S: 738 aligned sites), using GTR + I + G as the model of nucleotide substitution. Alignments performed using MAFFT (parameters not given, presumably default) and further modified by eye (hypervariable regions excluded). Gaps were presumably treated as missing data. Numbers on nodes refer to ML bootstrap nodal support values. Additional support values (i.e., Bayesian posterior probabilities, minimum evolution “distance” bootstrap) are available in the original. Nodes with no values received less than 70% bootstrap values or were not given in the original.

The authors discuss digit and body size reduction and the shape of the clavicles as morphological support for their arrangement.4 De Sá et al. (2012) mention “Syncope species have lost two vertebrae and have reduced and/or lost fingers I and IV,” and add that “a similar pattern of small adult body size and digit reduction is present in the species of Chiasmocleis that we found to share phylogenetic affinities with Syncope: C. bassleri, C. hudsoni, and C. magnova.” De Sá et al. (2012) further mention that the new arrangement “is also consistent with morphological variation in the pectoral girdle where there has been a complete loss of the connection between coracoids and epicoracoid in S. antenori and S. magnova (and a reduced connection in S. jimi and S. hudsoni) whereas the connection is present in Chiasmocleis.” The authors do not mention the state of the clavicles of either C. bassleri or C. supercilialbus, in which a connection of the coracoids with the epicoracoids is present.

De Sá et al. (2012) also suggested that reproductive mode may unite Syncope and the few species of Chiasmocleis they transferred into Syncope. Data on reproduction is scarce for many species of Chiasmocleis and Syncope, but is sufficiently well known to reject de Sá et al.'s (2012) claim. Syncope antenori has free-swimming nonfeeding tadpoles while direct development was suspected to occur in Syncope carvalhoi (Krügel and Richter, 1995) and C. magnova (Moravec and Köhler, 2007)—this was never proved by empirical evidence, only inferred from the relative large size and pigmentation of the eggs. On the other hand, most members of Chiasmocleis lay hundreds of small pigmented eggs and have exotrophic tadpoles (Duellman, 1978; Wogel et al., 2004; Lima et al., 2005; Oliveira-Filho and Giaretta, 2006; Rodrigues et al., 2008; Menin et al., 2011; Santana et al., 2012; P.L.V.P., personal obs.), and species with aquatic free-swimming exotrophic larvae include some of those of which de Sá et al. (2012) claim to share reproductive ecology traits with “Syncope” (i.e., C. bassleri, C. hudosni). The heterogeneity of reproductive modes within “Syncope” therefore does not support de Sá et al.'s (2012) hypothesis.

Nonetheless, because de Sá et al. (2012) provide the largest test of phylogenetic relationships within microhylids, we adopt the general microhylid classification proposed by them (but see Results and Discussion sections for a novel classification proposal to the Chiasmocleis/Syncope section of the tree). We took the opportunity provided by de Sá et al. (2012: i.e., the publication of novel DNA sequence data for Chiasmocleis and hypothesized evolution of phenotypic characters) to include a phylogenetic framework in our own work. By including a larger taxon sampling and several representatives for each species, we were able to further test the monophyly Chiasmocleis and Syncope as proposed by de Sá et al. (2012), as well as to infer species limits within both genera.

Institutional Abbreviations:

Effort was made to examine only specimens deposited in scientific collections, but this was not always possible. When no collection number is available, specimens are referred to by their field number.5 Some specimens illustrated (especially live specimens) do not have any number associated with them and this is noted. A complete list of the material examined is available in appendix 1 (except for type material of newly described species, which are given in the text).

Field Series Numbers:

Exact meaning of the acronym is given in “quotes” and is followed by additional details on locality (if acronym referes to any specific place and not to a collector's name).

Morphology:

The following measurements (adapted from Peloso and Sturaro, 2008) were taken with an electronic caliper under a stereomicroscope (to the nearest 0.1 mm). SVL (snout-vent length); 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 (interorbital distance; distance between anterior corner of the eyes); IND (internarial distance); END (eye-nostril distance; from the anterior corner of the eye to the posterior margin of nostril); THL (thigh length; from the middle of the cloacal opening to the outer edge of the flexed knee); TBL (tibia length; from the outer edge of the flexed knee to the heel); FL (foot length; from tibio-tarsal articulation to tip of fourth toe); 4TD (diameter of fourth toe disc). Fingers and toes are numbered and abbreviated as follows: Fingers I–IV  =  FI–IV, Toes I–V  =  TI–V. Summary of measurements and landmarks are illustrated in figure 3.

Fig. 3. 

Outlines of stereotyped gastrophrynine frog in (A) dorsal, (B) ventral, and (C) lateral views, with standardized morphometric variables used in this study: SVL, snout-vent length; HL, head length; HW, head width; ED, eye diameter; IOD, interorbital distance; IND, internarial distance; END, eye-nostril distance; THL, thigh length; TBL, tibia length; FL, foot length; 4TD, diameter of fourth toe disc. Additional measurement details are given in Material and Methods.

Discrete morphological characters were evaluated from direct observation of specimens (see fig. 4 for clarification). Relative finger and toe lengths were addressed with the following point marks: point where tip of adpressed FI touch subarticular tubercle on FII; point where tip of FIV touch FIII (in relation to the subarticular tubercles on FIII); point where adpressed TI touch TII; point where adpressed TV touch TIV. Presence and distribution of spines: we looked for dermal spines in the chin, cloacal region, dorsum, fingers, and toes. Presence and development of subarticular tubercles, especially in FI and TI. Presence and development of palmar tubercles. Presence and amount of webbing.

Fig. 4. 

Ventral views of (A) right hand and (B) right feet of Chiasmocleis, evidencing characters frequently mentioned in the diagnosis and comparisons throughout the text. Abbreviations: F, finger; T, toe; d., distal; in., inner; m., metatarsal; pal., palmar; pr., proximal; t., terminal; s., subarticular; tb., tubercle.

Several color parameters were used to discriminate species. We used mainly ventral color pattern, presence or absence of inguinal blotches, and presence or absence of light stripes on the posterior surface of thigh (femoral lines). Coloration in life for all species is based on field notes and photos take by ourselves and colleagues or are based on descriptions from the literature. Due to time restrictions while visiting some of the collections, and in favor of analyzing a larger volume of material, measurements were not taken for every specimen examined. Sample sizes for each measurement are given.

Some specimens, including type material, were examined solely from photographs (see appendix 1). We used those specimens in the diagnosis of the species only if the images were of high resolution, were taken from different views/angles, and were focused on the diagnostic characters.

Advertisement Calls:

The advertisement calls of several Amazonian species of Chiasmocleis have been described at least once. Santana et al. (2009) presented a discussion of known calls and summarized the main call parameters in a table. Data summarized by Santana et al. (2009) revealed that calls were described idiosyncratically and consequently descriptions are not, in many cases, directly comparable. We attempted to conduct a standardized comparison among available calls of species of Chiasmocleis treated here. Availability and quality of recordings was, however, a limiting factor for a thorough review. Advertisement calls were obtained from colleagues or from the acoustic collection of the USNM.

Calls were analyzed using the software Raven, version 3.1 for Mac (Charif et al., 2008), and figures drawn in Praat, version 5.3.23 for Mac (Boersma and Weenick, 2006). For call terminology, we used the following scheme: vocalization is any sound emitted by the vocal apparatus of the animal, while calls and notes are subcategories of vocalizations. Advertisement calls were treated as vocalizations that were discrete temporal units composed of the call emitted by males assumed to be vocalizing to attract females for mating. The calls of all species analyzed here are composed of a series of multipulsed notes. Notes were distinguished from each other when the interval between them was greater than the interval between individual pulses.

Parameters measured were: CR, call rate (notes/min); ND, note duration (milliseconds, ms; measured from the audiospectrogram, with doubtful limits checked on the waveform); IBN, interval between notes (ms); DFH, dominant frequency of first harmonic (Hz); DFC dominant frequency of the call (Hz); HRN, number of harmonics by note; PPN, number of pulses per note. The dominant frequency (“Max Frequency” in Raven 1.3) as used here is the frequency at which maximum power occurred (Charif et al., 2008).

Phylogenetic Analyses:

To help in the evaluation of species limits in Amazonian species of Chiasmocleis we conducted a phylogenetic analysis including up to three genes for 96 specimens of Chiasmocleis or Syncope. All currently recognized species known to occur within the geographical limits of this study were included with the regrettable exception of C. supercilialba. Specimens from the putative new species identified a priori based on the analyses of morphological and acoustic characters were also included. For phylogenetic inference, we used fragments of two mitochondrial (the large ribosomal subunit 16S, up to 557 bp; and cytochrome oxidase I [COI], 658 bp), and one nuclear genes (Tyrosinase, 532 bp).

Kalophrynus internilineatus (Kalophryninae) was used to root the tree. In addition to that, another nongastrophrynine microhylid, Microhyla heymonsi (Microhylinae), was included as an additional nongastrophrynine outgroup. Seven gastrophrynines were also included as outgroups, including two former members of Chiasmocleis (Hamptophryne boliviana and Elachistocleis panamensis). Hamptophryne boliviana was removed from Chiasmocleis by Carvalho (1954) on the basis of morphology, whereas E. panamensis was moved to Elachistocleis by de Sá et al. (2012) based on a phylogenetic analysis of DNA sequences. Additional gastrophrynines included were Ctenophryne barbatula, Ct. geayi, Dermatonotus muelleri, Gastrophryne carolinensis and Elachistocleis helianneae.

Although inferring relationships outside the Chiasmocleis/Syncope clade was not our intention, our outgroup sampling provided the first test of the new positioning of E. panamensis proposed by de Sá et al. (2012), and also a further test of the monophyly of the Chiasmocleis/Syncope clade.

Laboratory protocols. Total genomic DNA was extracted and isolated from frozen and ethanol-preserved tissues using the Qiagen DNeasy kit following the manufacturer's guidelines. Polymerase chain reactions (PCR) for amplification of DNA fragments were carried out in 25 µl reactions using either Fisher Taq together with Buffer A (catalog number FB600030) or Illustra PuRe Taq Ready-To-Go PCR Beads (GE Healthcare Life Sciences). Primers used in amplification and sequencing of the 16S and tyrosinase fragments are commonly used in amphibian systematics while a new pair of primers was designed for sequencing the Folmer et al. (1994) fragment of COI. For the 16S fragment, we used 16SAR (forward, 5′-CGCCTGTTTATCAAAAACAT-3′) and 16SBR (reverse, 5′-CCGGTCTGAACTCAGATCACGT-3′) (Palumbi et al., 1991). For tyrosinase, we used TyrC (forward, 5′-GGCAGAGGAWCRTGCCAAGATGT-3′) and TyrG (reverse, 5′-TGCTGGCRTCTCTCCARTCCCA-3′) (Bossuyt and Milinkovitch, 2000). A primer pair was designed de novo for targeting a 658 bp fragment of COI; COI.PF-A (forward, 5′-TTTCAACHAAYCAYAAAGAYATYGG-3′) and COI.PR-A (reverse, 5′TANACTTCNGGGTGDCCAAARAATCA-3′).

PCR setups were those of Faivovich et al. (2005), with standard annealing temperature of 48° C for 16S and COI; 56° C for tyrosinase. PCR products were cleaned and desalted using AMPure (Agencourt Biosciences Corporation) in a Beckman Coulter Biomek 2000 robot (“Becky”) or by hand. Cycle sequencing using BigDye Terminators, v. 3.0 (Applied Biosystems), was run in 10 µL reactions following the protocol of Platt et al. (2007), and products were cleaned and desalted using cleanSEQ (Agencourt Biosciences Corporation) in “Becky.” Sequencing was done with an ABI 3730XL automated sequencer. Samples were sequenced in both directions to check for sequencing errors and ambiguities. Sequence contigs were assembled and edited in Geneious 6.0.3 (Biomatters,  www.geneious.com).

Genbank sequences. To complement our dataset, several sequences were obtained from previously published works (Genbank; see appendix 2). Only terminals for which at least the 16S was available were included in the final dataset.6 One sequence labeled as C. hudsoni from Fouquet et al. (2007: accession number, EU201100) was not included. This represents C. haddadi, sp. nov. (described below), and derives from the same population (French Guiana, Mont Kotika) for which we sequenced four other individuals.

De Sá et al. (2012) included 23 sequences of Chiasmocleis and Syncope representing 14 species in their analysis of microhylid relationships. We have included 22 of them in our analyses.7 Genbank accession numbers for novel and previously published sequences, together with their source, are given in appendix 2.

Phylogenetic analyses. Phylogenetic analyses were performed under direct optimization (Wheeler, 1996; Varón and Wheeler, 2012) in POY 5.0 (Varón et al., 2013) using equal weights for all transformations (substitutions and insertion/deletion events) and the parsimony optimality criterion. Justifications for the use of dynamic homologies in parsimony, which we endorse herein, can be found elsewhere (Kluge and Grant, 2006; Wheeler et al., 2006; Grant and Kluge, 2009). Contiguous sequences were preliminarily delimited in fragments of putative homology to allow incorporation of partial sequences and accelerate dynamic homology cost calculations (Wheeler et al., 2006). External gaps due to missing data (and not due to natural length variation) were coded as “N”s.

Analyses were performed using the command “search,” which implements a driven search composed of random addition sequence Wagner builds (RAS), subtree pruning and regrafting (SPR) and tree bisection and reconnection (TBR) branch swapping, parsimony ratcheting (Nixon, 1999), and tree fusing (Goloboff, 1999), storing the shortest trees of each independent run and performing a final round of tree fusing on the pooled trees. When running in parallel, trees are exchanged among processors at the end of each search and the best unique trees are then selected and added to the pool of trees to be fused following the next round of search. Ten consecutive searches runs consisting of 12 hr driven searches were conducted in parallel on 16 Intel Xeon 3.0 GHz processors at the American Museum of Natural History computer cluster. Best trees from each run were saved and pooled together for additional rounds of tree fusing and TBR. For a more thorough search of optimal topologies, trees from the initial searches were submitted to additional rounds of TBR under an iterative pass optimization (Wheeler, 2003).

We report Goodman-Bremer, GB (Goodman et al., 1982; Bremer, 1988; Grant and Kluge, 2008b), for each node on the strict consensus tree of the most parsimonious trees. GB was chosen over other nodal support measurements (e.g., bootstrap, jackknife) because of its objectivity and direct relatedness to optimality criteria (Grant and Kluge, 2008a; Wheeler, 2010). GB was calculated by comparing the strict consensus tree with a pool of suboptimal trees, assembled by saving all trees visited during TBR searches performed on the optimal trees.

Uncorrected pairwise genetic distances for 16S sequences were calculated using MEGA 5.2 (Tamura et al., 2011) based on multiple sequence alignment performed in MUSCLE (Edgar, 2004) using the default parameters.

Character Evolution:

During evaluation of phenotypic characters for diagnosis of the species treated here a few were considered consistent enough (fixed within species) to be diagnostic and useful for evaluating their evolution (transformation series). We have selected three morphological characters (femoral line, dermal spines, and amount of webbing in males), and one reproductive character (egg clutch size) for evaluation of their evolution in the study group. Characters were optimized in a summary tree, derived from the strict consensus of the most parsimonious trees, with all terminals representing the same taxon collapsed to a single terminal. Morphological characters were scored as absent or present, while the amount of webbing in males was simplified to constitute only two states, absent (if absent or only basal webbing observed), and present (if extensive webbing observed) (see fig. 5). Because of limited taxon sampling, optimization of characters on the outgroup, which can influence optimization on the ingroup, was a matter of uncertainty. We opted to collapse all outgroup terminals into a single leaf and coded all characters as ambiguous observations (equal probability of the character being present, or absent). Character optimization was accomplished using the parsimony algorithm in Mesquite 2.74 (Maddison and Maddison, 2007).

Fig. 5. 

Variation in amount of toe webbing in Chiasmocleis. (A) Webbing absent in a male C. haddadi, sp. nov. (FL 454). (B) Basal webbing in a female C. shudikarensis (FL 410). (C) Extensive webbing in a male C. shudikarensis (FL 487). Note sexual dimorphism in webbing (B–C, syntopic specimens). Not to scale.

Coding of phenotypic characters was done from direct observation of specimens (listed in appendix 1) or from the literature. In the later case, the following sources were consulted: Nelson, 1975; Duellman and Menselson III, 1995; Krügel and Richter, 1995; Cruz et al., 1997; Cruz et al., 1999; Nascimento and Skuk, 2006; Moravec and Köhler, 2007; Funk and Cannatella, 2009.

Presentation of Data:

First, we present the results of the phylogenetic analysis. This is followed by a review of relevant characters deemed of importance in Chiasmocleis (and Syncope) taxonomy. We then present species accounts, with data for all known species and the description of new taxa. For known species, data are presented in the following order: Holotype (collection number; whether the specimen was examined; notes on preservation state; photos of most holotypes are included); Type locality (usually copied from the original publication; when vague, additional details are given, if available); Diagnosis (based on our observation of specimens; when literature data are used in the diagnosis, that is mentioned); Variation (same as for Diagnosis); Calls and Tadpoles (data on calls are given for newly analyzed calls and reanalyzes of previous descriptions and from the literature); Remarks (if needed: any important remarks relevant to the taxonomy or systematics of the taxa are provided); and Distribution (a review of available specimen records as well as literature records of populations not examined by us where pertinent; distribution map for the species, except C. albopunctata, which was given by Forlani et al., 2011; maps include only records for which we examined at least one specimen, by direct examination or solely from photographs, except for type localities, which are sometimes based solely on the literature, e.g., C. devriesi). For new taxa, we provide data as follows: Holotype, Type Locality, Paratypes, Diagnosis, Comparisons (with all species treated in this work), Description of the Holotype, Measurements of the Holotype, Variation, Etymology, Calls and Tadpoles, Remarks (if needed), and Distribution.

A compilation of data on natural history, both from the literature and novel data (based on field notes and photographs), is given at the end of the manuscript, as is a brief comment on conservation. For these final sections, information on all species is compiled together.

The Chiasmocleis bassleri Clade:

This clade contains C. bassleri and C. superciliaba. Although well supported in our analysis, the position of C. bassleri differs markedly from that reported by de Sá et al. (2012), whose sole specimen of C. bassleri was found nested within what we recovered as members of the C. hudsoni clade. Morales and McDiarmid (2009) suggested that C. supercilialba “is related to C. bassleri and could be [from] a supra-species group” and proposed the C. bassleri species group. We agree that there are striking morphological similarities between the two taxa (e.g., body shape, inguinal spot, digit morphology), and therefore place C. supercilialba, although tentatively, as a member of the C. bassleri clade. This arrangement is pending further test, preferably in light of a phylogenetic analysis that actually includes specimens of C. supercilialba.

The Chiasmocleis hudsoni Clade:

Chiasmocleis haddadi, sp. nov., is sister to a cluster of specimens from several populations identified (original labels) as either C. hudsoni and C. jimi. Our topology shows that both C. hudsoni and C. jimi are not monophyletic (C. jimi is considered here as a synonym of C. hudsoni, see below). Chiasmocleis magnova is sister of a clade containing C. antenori, C. tridactyla, and C. carvalhoi. One sequence of C. carvalhoi from San Martin, Peru (KU 215720; from de Sá et al., 2012) groups with our sequences of C. antenori. Since we have not examined this specimen, we do not take any nomenclatural action regarding C. antenori or C. carvalhoi. Both sequences of Syncope sp. (de Sá et al., 2012) are nested within S. carvalhoi, and are tentatively assigned to this taxon, pending examination of the voucher specimens.

The Chiasmocleis shudikarensis Clade:

Peloso and Sturaro (2008) described C. avilapiresae, differentiating it from C. shudikarensis mostly on color pattern, presence/absence of the inguinal blotch, and development of fringes. Our phylogenetic analysis recovered two well-supported monophyletic lineages, one with all specimens of C. avilapiresae and another with all specimens of C. shudikarensis (fig. 8). There are considerable differences in the advertisement calls among populations of C. shudikarensis as well as a certain degree of morphological variation, which led Peloso and Sturaro (2008) to consider that additional species may be hidden in the name C. shudikarensis. Present evidence is insufficient to either support or reject Peloso and Sturaro's (2008) prediction. This is discussed further in the Variation and Remarks section in the C. shudikarensis species account.

The Chiasmocleis ventrimaculata Clade:

The overall morphological similarity of C. anatipes, C. devriesi, C. papachibe, sp. nov., C. royi, sp. nov., and C. ventrimaculata is explained by their close relationship, with the five forming a well supported monophyletic clade. The sole specimen of C. devriesi is clustered with some of our samples of C. anatipes, and the two recovered as sister to a clade containing the other three species in the C. ventrimaculata clade: C. royi, sp. nov., C. papachibe, sp. nov., and C. ventrimaculata.

The clade as a whole has a curiously disjoint distribution, with most species being found in the western Amazon (Colombia, Ecuador, Peru, Bolivia, and western Brazilian states of Acre and Rondônia), while C. papachibe is found in eastern Amazon (vicinities of Belém, Pará). The absence of species of this clade in the majority of the Amazon basin in Brazilian territory (especially in the Tapajós and Xingu river basins) deserves further scrutiny, and the negative effects of inadequate sampling in those areas cannot be disregarded.

The Chiasmocleis albopuncata Clade:

A monophyletic “Atlantic Rainforest clade” was not recovered, as C. albopunctata is nested within the Atlantic Forest species. Chiasmocleis leucosticta is sister to C. albopunctata, and the two form a clade that is sister to the Atlantic Forest endemics C. alagoanus, C. capixaba, C. lacrimae, and C. schubarti. The arrangement of this clade is similar to that reported by de Sá et al. (2012; most of the sequences used derive from that work), except for the position of C. schubarti. Within samples of C. albopunctata we found three well-supported clades, one with specimens from northeastern Brazil (mostly Caatinga/Cerrado habitat transition), one with specimens from southeastern Brazil (Cerrado habitats in the states of Minas Gerais and São Paulo), and a third with specimens from Bolivia.

External Morphology:

The morphological analyses revealed a great degree of variation in some characters commonly used to diagnose species in the genus Chiasmocleis (e.g., inguinal blotches, body stripes, presence and distribution of dermal spines, toe webbing, and ventral color patterns), which we feel deserves further scrutiny.

Morphometrics:

Despite the fact that not all specimens were measured (see Material and Methods), we have compiled a large amount of data on morphometrics. These data are summarized in table 1.

Table 1 

Summary of morphometric data for the Chiasmocleis species studied ^{a, b} Abbreviations are listed in the Material and Methods

Inguinal Blotch (fig. 9):

Fig. 9. 

Variation in size and shape of the inguinal blotch in species of Chiasmocleis. (A–C) C. bassleri; (A) MPEG 27766, (B) UFAC-RB 3872, and (C) UFAC-RB 1251. (D) C. shudikarensis, INPA 278.</emph>

This character was used for the first time in Chiasmocleis taxonomy by Dunn (1949) to describe a “rather vague and irregular” mark in the inguinal region of the holotype of C. bassleri. These black or brownish marks can vary widely in form and shape, from a small round spot on the dorsolateral region to a larger and more irregularly shaped blotch that extends from the groin to the dorsum (fig. 9). Although often referred to as an inguinal spot, we feel that this character is better designated as inguinal blotch.

The presence or absence of the inguinal blotch proved to be a taxonomically informative character. Three species present conspicuous inguinal blotches: Chiasmocleis bassleri, C. shudikarensis, and C. supercilialba. Some specimens of C. bassleri and C. supercilialba also have additional blotches/spots on the lateral region (fig. 9B). A very faint blotch in the inguinal region of C. avilapiresae is present in a more limited number of specimens, all from a single locality (Maués, Amazonas, Brazil). The blotch in C. avilapiresae is, however, limited to the inguinal region and does not extend dorsolaterally.

Light Stripes on Middorsum and Posterior Surface of Thigh (femoral line):

The presence and shape of a light stripe on the dorsum varies within populations and proved to be of no taxonomic utility in Chiasmocleis (fig. 10 illustrates this character's variation).

Fig. 10. 

Schematic drawing of variation in presence and extent of dorsal white lines in Chiasmocleis.

On the other hand, a lightly colored (white or cream) line on the posterior surface of the thigh, extending to the knee and over the cloacal opening, is consistently either present or absent within populations, proving to be a useful diagnostic character. We label this character as femoral line. All examined specimens of C. avilapiresae, C. shudikarensis, and C. royi, sp. nov. (described below), present femoral lines, whereas the line is absent in all other Amazonian/Guiana Shield species of Chiasmocleis. The femoral line can sometimes be connected, and perpendicular, to the dorsal line. The femoral line is present in non-Amazonian Chiasmocleis such as C. capixaba and C. leucosticta.

A different stripe on the posterior surface of thigh and flanks is present in Chiasmocleis bassleri and C. supercilialba. These lines, called here split stripes, are distinct from the femoral line described above—the limits of the split stripe are not so well defined, but they usually mark a separation (or split) between the dorsal and ventral color patterns. Split stripes are not necessarily a line, but can be formed by a series of irregular light-colored stains (fig. 9A, C). Split stripes may also be evident on the lateral region (fig. 9A, C), and extend throughout the hind limbs. This is treated in more detail ahead (see variation under C. bassleri). Split stripes are present in other microhylids (e.g., Ctenophryne, Hamptophryne, and Hypopachus) and can be white as in Chiasmocleis and Ctenophryne or black as in some Hamptophryne and Hypopachus. Split stripes are present in most of the specimens of C. bassleri and C. supercilialba, whereas a few specimens do not show a clear differentiation between dorsal and ventral patterns.

Dermal Spines (fig. 11):

Fig. 11. 

Example of distribution of dermal spines on the head of Chiasmocleis.

Several authors had acknowledged the importance of dermal spines in microhylid systematics (see Nelson, 1972; Zweifel and Myers, 1989; Cruz et al., 1997; Lehr and Trueb, 2007). They have also, however, indicated that the distribution and development of spines may vary sexually and taxonomically. Spines, when present, are usually more evident in reproducing specimens; therefore, the character may also vary seasonally. At this point it is not clear whether even the spines are homologous structures among all gastrophrynines, or whether different lineages evolved spines independently (e.g., the spines in the chin of Chiasmocleis hudsoni are much larger and less numerous than in C. shudikarensis). The function and morphological nuances of the spines were never studied in detail, but some authors have related the presence and development of dermal spines to reproductive activity or as a defensive strategy against predators (Lehr and Trueb, 2007).

For the reasons above, although we use the presence/absence and distribution of spines to diagnose species, this character should be evaluated with care. The presence of dermal spines in Microhylidae is well illustrated in Lehr and Trueb (2007: figs. 5–6), Peloso and Sturaro (2008: fig. 6), and Cruz et al. (2007a); spines are also illustrated here, in figure 11.

In Chiasmocleis, the distribution, size, and density of spines vary both taxonomically and sexually. Spines are usually more common in males than in females. At least five species have males with many spines spread over the dorsum, chin, cloacal region, limbs, and digits: C. anatipes, C. avilapiresae, C. royi, sp. nov. (described ahead), C. shudikarensis, and C. ventrimaculata. Spines are rare, but sometimes present, in C. albopunctata, C. bassleri, and, although slightly more common, in C. supercilialba. Chiasmocleis hudsoni also shows some spines on the chin (in males) and in the cloacal region (more commonly in males, but also present in females). The low number of specimens available for C. magnova makes it difficult to define the distribution pattern of spines in males; females have many spines in the dorsolateral region and scattered spines on the tarsus and TV (Moravec and Köhler, 2007). Spines are unknown in C. devriesi (known only from the female holotype, but see remarks in species accounts), C. haddadi, sp. nov. (described below), and C. papachibe, sp. nov. (described below). Several Chiasmocleis from the Atlantic Rainforest also show a conspicuous presence of dermal spines (Cruz et al., 2007a; Cruz et al., 2007b).

Toe Webbing:

This character varies widely between sexes. When present, webbing is much more common and extensive in males than in females, and thus comparisons should be limited to specimens of the same sex. Marked sexual dimorphism is present in Chiasmocleis avilapiresae and C. shudikarensis, e.g., males generally have fully webbed toes, whereas those in females are only basally webbed or show no web at all (fig. 5). Toes of females Chiasmocleis anatipes are reported to be widely webbed (Rodriguez and Duellman, 1994), but examination of a few female specimens refutes this observation (see Remarks under C. anatipes). In some taxa, both sexes lack webbing completely (e.g., C. antenori and C. hudsoni).

There have been several attempts to standardize the description of variation in webbing among frogs (Savage and Heyer, 1967; Myers and Duellman, 1982), but we decided not to apply any notation to the present work for the following reasons: (1) some species present digit reduction, especially in TI and TV (e.g., Chiasmocleis antenori, C. carvalhoi, C. haddadi, sp. nov., C. hudsoni, C. magnova, and C. tridactyla), so they are not directly comparable to species with more developed digits (given that the formula correlates the extent of webbing to digit phalanges); (2) time is constrained when visiting collections—even with experience, taking webbing notation is very time consuming and in many cases would come at the expense of the amount of material examined, since time in several collections was limited. Given the complexity of the group, we decided to prioritize volume of material examined over describing in detail variation in webbing. Instead, variation in webbing was noted solely as absent, basal, or extensive (fig. 5).

Our approach will certainly draw criticism and we are aware that some degree of subjectivity is inserted when not following “standardized recipes” for describing variation. However, we feel that the approach used is of sufficient reliability for diagnosing the currently known diversity in Chiasmocleis.

Ventral Color Pattern (fig. 12):

Fig. 12. 

Terminology used in this paper to describe ventral color pattern: (A) small light spots and stains over a dark background; (B) large dark spots over a light background; (C) dark blotches and stains over a light background; (D) large dark blotches over light background; or alternatively light vermiculations over a dark background; (E) dark vermiculations or reticulation over a light background; (F) intense dark vermiculations over a light background.

Color pattern of the ventral region (throat, chest, belly, and undersurfaces of limbs) were highly variable both within and among species. Nonetheless, it is a useful diagnostic character to differentiate species. As with dermal spines, although used in all diagnoses, this character should be examined with care, bearing in mind that variation does exist. Whenever a good series of specimens was available variation of the ventral and dorsal color patterns are provided in the variation section. Some clarifications to terminology are, however, needed. We define the ventral patterns observed in Chiasmocleis as follows (adapted from Kok and Kalamandeen, 2008):

Uniform. We refer to a uniform pattern when no spots, blotches, or vermiculations are present. Most specimens examined, however, do show some pattern of markings on the venter.

Spots or Stains. Small to medium-sized roughly round markings (herein called spots) or irregularly shaped ones (stains) may be light or dark in contrast to with the background color.

Blotches. Medium to large-sized irregular, light or dark markings contrasting with the background color. A blotch may show more than one shade of color (e.g., darker outline with a lighter center).

Vermiculations, Reticulations, or Anastomoses. Dark or light network of lines or elongated stains contrasting with the background color.

It should be noted that some specimens might show a combination of different patterns, for example, a uniform dark throat followed by a reticulation of dark stains on the chest and small spots on the belly. Again, a small level of subjectivity occurs with this character as, for example, different authors may interpret dark vermiculations against a light background as a dark background with many small irregular stains, and vice versa.

Advertisement Calls:

The calls of all species examined consist of repetitive series of relatively short, multipulsed, notes. This call structure has been argued to be a putative synapomorphy uniting members of Chiasmocleis (Hartmann et al., 2002), but Santana et al. (2009) found the call structure to be very different in C. mantiqueira (unfortunately C. mantiqueira was never included in a phylogeny of the genus). Up to this point, the calls of several species remain unknown and further investigation is needed to assess whether call structure is a synapomorphy of Chiasmocleis or the call found in C. mantiqueira is a single exception (autapomorphy).

The duration of notes, intervals between them, as well as number of pulses vary between species and among populations of the same species. A summary of advertisement call parameters found by our analysis of recordings is given in table 2.

Table 2 

Summary of acoustic data from advertisement calls of Chiasmocleis examined for this study A review of data from literature is available in Santana et al. (2009)

Holotype:

ZMB 10588 (fig. 13) according to Bauer et al. (1966). An adult specimen, examined from photographs; in very good state of preservation.

Fig. 13. 

Holotype of Chiasmocleis albopunctata from Paraguay (ZMB 10588) in (A) dorsal and (B) ventral views. SVL  =  25 mm.

Type Locality:

Paraguay, Amer[ica]. Merid[ional].

Diagnosis. A large species for the genus; SVL in males 23.2–32.2 mm, in females 28.2–38.0 mm (SVL data from Caramaschi and Cruz, 1997). Body ovoid and robust; snout rounded in dorsal and lateral views. Four distinctive fingers, all but FI slightly fringed in both sexes; fingers not webbed; FI well developed with a distinct, well-developed subarticular tubercle between the proximal phalanges; subarticular tubercles present on all fingers; adpressed FI does not reach the subarticular tubercle of FII; adpressed FIV reach but do not pass distal margin of proximal tubercle of FIII; palmar tubercles protuberant, divided; relative finger lengths I<IV<II<III. Five distinctive, well-developed toes present; toes fringed in both sexes; toes free of webbing; TI with a distinct, well-developed subarticular tubercle; adpressed TI touch or barely touch subarticular tubercle of TII; adpressed TV does not touch or reaches the middle of the middle subarticular tubercle of TIV; toes lack terminal discs; relative toe lengths I< II<V<III<IV. Males may present few, small dermal spines on hands. Femoral line absent. Dorsum usually grayish or dark brown with white spots and stains, especially on snout and dorsolateral region; venter (belly and undersurfaces of thigh) usually brownish or grayish with many white spots.

Variation:

Number of specimens examined in detail is insufficient to present conclusive data on geographic variation. Forlani et al. (2011: fig. 1) reported significant variation in dorsal color pattern between distinct populations along the species' range, but did not comment further on it. Dorsal white spots vary in number, but are always present.

Call and Tadpole:

The call of Chiasmocleis albopunctata was described twice, first from a population in Bolivia (De la Riva et al., 1996) and then from a population in southeastern Brazil (Oliveira-Filho and Giaretta, 2006). De la Riva et al. (1996) reported a call with a fast series (584.4–907.7 notes/min) of multipulsed notes (5–8 pulses/note) with a mean dominant frequency of 4311.0–4664.4 Hz. Oliveira-Filho and Giaretta (2006) reported a slower call rate (423–501 notes/min), with more pulses per call (9) and a very similar dominant frequency (4306 Hz). Unlike Oliveira-Filho and Giaretta (2006: 67–68), we consider these differences trivial. A larger series of calls from distinct populations is needed to further assess variation in the species.

The tadpole of Chiasmocleis albopunctata was described and illustrated by Oliveira-Filho and Giaretta (2006). External morphology is similar to other species of Chiasmocleis (McDiarmid and Altig, 1999), but differs from known larvae in having wartlike ornamentations on the oral flaps.

Remarks. Genetic distances between all specimens of Chiasmocleis albopunctata included in the phylogenetic analysis are given in table 3.

Table 3 

Uncorrected pairwise distances between 16S sequences of Chiasmocleis albopunctata

Distribution:

This species is common in open habitats and its distribution in the Amazon is restricted to transitional Cerrado/Amazonia. The species is widely distributed in open areas in Paraguay, Bolivia, and central and northern Brazil. Forlani et al. (2011) recently discussed the range of the species, provided an updated map, and commented on a dubious record of a specimen from Pará, Brazil. We have no additional information on the distribution of this species and refer to Forlani et al. (2011) for data on this matter. The species is common in the Rio Tocantins basin with several records in the state of Tocantins, Brazil (Forlani et al., 2011).

Holotype:

KU 146035 (fig. 14); adult male, examined from photographs; in very good state of preservation.

Fig. 14. 

Holotype of Chiasmocleis anatipes (KU 146035) in (A) dorsal and (B) ventral views. SVL  =  19.0 mm. Photo M. Bustamante- AmphibiaWebEcuador (see Ron et al., 2014).

Type Locality:

Santa Cecília, Província Napo, Ecuador. Coordinates not given in the original. We used the following coordinates, from Paynter (1993): 00°03′N / 76°58′W.

Diagnosis:

A medium-sized species for the genus; SVL in males 18.0–24.2 mm (Walker and Duellman, 1974; our data), sole female examined 24.9 mm. Body moderately slender; snout rounded in dorsal and lateral views. Four distinctive fingers, all but FI fringed; fingers not webbed; FI well developed with a subarticular tubercle present between the proximal phalanges; tubercles present on all fingers; adpressed FI surpasses the subarticular tubercle of FII; adpressed FIV does not touch distal tubercle of FIII; palmar tubercles visible, divided; relative finger lengths I<IV<II<III. Five distinctive, well-developed toes present; toes fringed and extensively webbed in males (see remarks for comments on females); TI with a distinct well-developed subarticular tubercle; adpressed TI does not touch or barely reaches subarticular tubercle of TII; adpressed TV reaches the middle of the middle subarticular tubercle of TIV; TII–V with terminal discs; relative toe lengths I<II<V<III<IV. Many dermal spines on dorsum and cloacal region in both sexes; spines on chin of males present, absent in females; spines always more abundant and developed in males. Inguinal blotch and femoral line absent. Venter, belly, and undersurfaces of thigh cream colored with large dark brown irregular spots or blotches.

Variation:

The number of specimens analyzed do not allow for a detailed analysis of variation. Intensity of marbling on dorsum may vary and a dorsal midline may be absent or present (pl. 1C–H). Number and shape of ventral dark spots vary among specimens (figs. 14–15, pl. 1F, H), being almost entirely absent in the specimen depicted in plate 1F. Detailed description of color in life was given by Walker and Duellman (1974).

Call and Tadpole:

The call of Chiasmocleis anatipes was never described; reported to consist of a “short buzz” (Rodriguez and Duellman, 1994). A recording and a brief sonogram are available at from Ron et al. (2014), but the recording seems to be unsuitable for a detailed bioacoustic analysis. From the sonogram published by Ron et al. (2014) it is clear that the call consists of a short series of multipulsed notes.

The tadpole of Chiasmocleis anatipes was first described by Walker and Duellman (1974) and briefly redescribed by Duellman (1978) and Rodriguez and Duellman (1994). According to those descriptions the tadpole of C. anatipes presents paired spiracles located ventrolaterally and a dextral vent tube, a character state present only in pipoids (Orton, 1953; Haas, 2003). Duellman apparently noticed the mistake, but the correction never appeared in print; the following is taken from Donnelly et al. (1990): “Duellman (1978) described paired ventral spiracles in C. anatipes but the spiracle is single and midventral (W.E. Duellman, personal commun. to R. Altig).” Note, however, that Ronn Altig was not an author of the Donnelly et al. paper, but likely acted as a reviewer for the manuscript. We have not examined the mentioned specimens and cannot provide further clarification on the issue. Nonetheless, it is clear that the tadpoles of C. anatipes need to be accurately redescribed.

Remarks:

Walker and Duellman (1974) described Chiasmocleis anatipes based on a series of seven specimens, all males, collected at the type locality. The species was diagnosed based mostly on the amount of toe webbing, extending “to the bases of the terminal phalanges of each digit” (Walker and Duellman, 1974). Females were unknown at that time. Rodriguez and Duellman (1994) reported a population of C. anatipes from northern Peru (Iquitos region), including males and females, and reinforced in their diagnosis that “the toes are fully webbed” (Rodriguez and Duellman, 1994: 73). We examined two of the female specimens from Iquitos (MHNSM 15697, 15700, from photographs only: fig. 15A, B) and they clearly do not show full webbing on toes. From the photos it is impossible to discern, however, whether webbing is basal or even totally absent. One of those specimens (MHNSM 15700) shows a ventral pattern very similar to that of the holotype of C. ventrimaculata (fig. 38) and agrees with the description given by Andersson (1945). The other specimen (MHNSM 15979), however, shows a ventral pattern intermediate between the holotype of C. devriesi (MHNSM 21540, not examined, but see figs. 15D, pl. 5G) and that of a specimen of C. ventrimaculata (MHNSM 21539: fig. 15C) from the same region. Both the holotypes of C. ventrimaculata and C. devriesi are females. We had access to a series of specimens from two localities in Napo, Ecuador, that we refer to C. anatipes. All have ventral color patterns very similar to examined types (holotype examined only from photographs; see fig. 14). Most of specimens of this series from Napo (LACM specimens) are juveniles and cannot be unambiguously identified, but two are adults, one male (QCAZ 51042), and one female (QCAZ 51041). The toes of the male specimen are fully webbed and it is unambiguously identified as C. anatipes. The female specimen is not fully webbed, but webbing can be easily seen between TII–TIII, TIII–TIV, and TIV–TV (see also pl. 1G). The two adult specimens (QCAZ 51041–51042) were also included in the genetic analyses and they have identical sequences for 16S and only two SNPs in the COI sequence. These two specimens are sister terminals in the phylogenetic analysis, and therefore we confirm that the female specimen is unambiguously assigned to C. anatipes, clearly showing that full webbing is not a diagnosis of the species as a whole, but solely of its males (contra Rodriguez and Duellman, 1994). In addition to the specimens from Napo, two specimens from Orellana, Ecuador, were also examined for morphology and included in the phylogenetic analyses (QCAZ 44341–44342). These are juvenile specimens (sex could not be determined), but basal webbing is present in both. They would, therefore, be assignable to C. anatipes. In the phylogeny, these two specimens cluster with the single representative of C. devriesi (MHNSM 21540, from Genbank). Genetic distances in the 16S sequences between all specimens of C. anatipes, as well as C. devriesi, is given in table 4.

Fig. 15. 

Ventral views of Chiasmocleis specimens from the Iquitos region, Loreto, Peru. (A) Female, SVL 20.0 mm (MHNSM 157000: L. Rodriguez); (B) Female, SVL 28.8 mm (MHNSM 15697: L. Rodriguez); (A–B) Identified as C. anatipes by Rodriguez and Duellman (1994). (C) Female, SVL 34.4 mm of C. ventrimaculata. (D) Female holotype of C. devriesi (MHNSM 21540: C. Funk). Not to scale.

Table 4 

Uncorrected pairwise distances between 16S sequences of Chiasmocleis anatipes and C. devriesi

Distribution (fig. 16):

Fig. 16. 

Distribution of Chiasmocleis anatipes. Star  =  type locality; circle  =  examined specimens.

Known from northeastern Ecuador, at Santa Cecilia, Provincia Sucumbios (type locality), and at the neighboring provinces of Orellana and Napo. Rodriguez and Duellman (1994) reported the species from the Iquitos region, Peru, but we consider the record questionable (see Remarks). Reports of this species for the Manaus region are based on misidentified specimens (P.L.V.P., personal obs.).

Syncope antenori: (Walker, 1973).

Holotype:

KU 124009; adult female, examined from photographs (fig. 17). Specimen well preserved, although a very large incision was made on the belly region. Color pattern only slightly discernible on the venter; cream venter with small whiteish spots.

Fig. 17. 

Holotype of Chiasmocleis antenori (KU 23299) in (A) dorsal, and (B) ventral views. SVL  =  12.3 mm. Photo M. Bustamante- AmphibiaWebEcuador (see Ron et al. 2013).

Type Locality:

Puerto Libre, Río Aguarico, 570 m, Provincia Napo, Ecuador.

Diagnosis:

A small species for the genus; SVL in females 12.4–13.6 mm, no data collected for males. Body slender to slightly robust, snout rounded in dorsal and lateral views. FI very much reduced without subarticular tubercle, tip slightly pointed; FII and FIII well developed with hardly visible subarticular tubercles on FIII, one on FII and two on FIII; FIV reduced but clearly visible, without subarticular tubercles; palmar tubercles not present; relative finger lengths I<IV<II<III. Four distinctive toes present; TI lost, TV reduced but clearly distinguishable; toes not webbed; tip of toes III–IV rounded with terminal discs present. Relative toe lengths II<V<III<IV. No dermal spines. No femoral stripe present. Venter (belly and undersurfaces of thighs) beige with small light spots.

Variation:

Despite adding several specimens of Chiasmocleis antenori into the genetic analysis, the number of specimens analyzed for morphological characters is fairly limited, restricted to three paratypes and a few specimens collected by P.L.V.P. in Serra do Divisor, Acre, Brazil. The specimens from Acre show variable amount of white spots on dorsum. Color of the iris color varies from golden to orange and red (see pl. 2A–D).

Call and Tadpole:

Call is unknown. Chiasmocleis antenori has endotrophic larvae developing inside terrestrial bromeliads (Krügel and Richter, 1995). The tadpole was described in detail, including several developmental stages, by Krügel and Ritchter (1995).

Remarks:

This species is similar in color pattern and external morphology to Chiasmocleis magnova, from which it differs in having an external tympanum visible (not visible in C. magnova: Moravec and Kohler, 2007), by its smaller size (maximum recorded SVL 13.6 mm, versus 18.3 mm in C. magnova), and in having seven presacral vertebrae (eight in C. magnova).

Phylogenetic analysis found two clades within Chiasmocleis antenori, one with three specimens from Ecuador (Orellana and Pastaza) and the other with all of the syntopic samples from Brazil and a sample labeled as C. carvalhoi (as Syncope carvalhoi in de Sá et al., 2012) from San Martin, Peru. Genetic distances between specimens of C. antenori vary from 0%–11.4%, with as much as 9.1% distance between syntopic specimens from Acre, Brazil. Genetic distances between all specimens of C. antenori included in the phylogenetic analysis are given in table 5.

Table 5 

Uncorrected pairwise distances between 16S sequences of Chiasmocleis antenori

Unfortunately, we have not examined the Ecuadorian or the Peruvian specimens, making any taxonomic conclusions about the status of these populations unattainable at present.

Distribution (fig. 18):

Fig. 18. 

Distribution of Chiasmocleis antenori and C. carvalhoi. Star  =  type locality of C. antenori. Circles  =  examined specimens of C. antenori; cross  =  type locality of C. carvalhoi (see text for details). triangles  =  examined specimens of C. carvalhoi.

Ecuador (Napo, Orellana, and Pastaza), and western Brazil (Acre). Coloma et al. (2004) gave a wider distribution and included a large portion of northeastern Peru. A specimen identified as Chiasmocleis carvalhoi from San Martin, Peru (de Sá et al., 2012), was found inside our samples of C. antenori and is regarded as a confirmed record of the species for Peru.

Fig. 19. 

Holotype of Chiasmocleis avilapiresae (MPEG 23299) in (A) dorsal, and (B) ventral views. SVL  =  34.9 mm.

Holotype (fig. 19):

MPEG 23299; adult female, in very good state of preservation.

Type Locality:

Estação Científica do Programa Pró-Biodiversidade da Amazônia (PPBio) (aprox. 01°59′S, 51°39′W), Floresta Nacional Caxiuanã, Municipality of Portel, state of Pará, Brazil.

Diagnosis:

A large species for the genus; SVL in males 22.2–28.0 mm (N  =  28), in females 23.6–37.8 mm (N  =  82). Body ovoid and robust; head triangular, snout rounded in dorsal and lateral views. Four distinctive fingers, all but FI fringed in males, less fringed in females; fingers not webbed; FI well developed, with a distinct, well-developed subarticular tubercle present between the proximal phalanges; distinctly visible subarticular tubercles present on all fingers; adpressed FI reaches or surpasses the subarticular tubercle of FII; adpressed FIV does not reach or barely reaches distal tubercle of FIII; palmar tubercles protuberant, divided; relative finger lenghs I<II<IV<III. Five distinctive and well-developed toes present; toes fringed, less distinct in females; toes usually extensively webbed in males, only basally webbed in females; TI with a distinct, well-developed subarticular tubercle; adpressed TI does not touch or barely touches subarticular tubercle of TII; adpressed TV does not touch or reaches only to the middle of subarticular tubercle of TIV; TII–IV with terminal discs, usually more developed in females, but also present in males; relative toe lengths I<II<V<III<IV. Males with many dermal spines on fingers and toes; both sexes may show dermal spines, but they are much more numerous and developed in males; males with many spines on anterior portion of chin, absent in females. Inguinal blotch usually absent. Femoral line present. Color pattern given in the variation section.

Variation:

Ventral (throat and belly) color pattern can vary from almost entirely uniform (cream, white, brown, or black) to densely reticulated (fig. 20; see also Peloso and Sturaro, 2008). A single specimen, out of 206 examined, showed strong dark vermiculations on the venter (fig. 20D). Western populations (west of the state of Amazonas, Mato Grosso and Rondônia) have, in general, fewer spots on the ventral region than those from eastern populations (eastern Amazonas and all Pará). Additionally, the specimens from Aripuanã, Mato Grosso, are slightly smaller than those of remaining populations. A dorsal midline is usually absent, but is present in some specimens, commonly in contact with the femoral line posteriorly (pl. 3D).

Fig. 20. 

Variation of ventral color pattern in Chiasmocleis avilapiresae. (A) Aripuanã, Mato Grosso, Brazil (UFMT 7124). (B–C) Floresta Nacional Caxiuanã, Pará, Brazil (MPEG 23315, and MPEG 23317, respectively, both paratypes). (D) Resex do Alto Juruá, Rio Juruá, Amazonas, Brazil (INPA 17259, paratype).

The dorsum of Chiasmocleis avilapiresae shows a great degree of color variation in life (see pls. 2E–H and 3). In general individuals have a dark, grayish dorsum (pl. 2E, G) while the dorsal surfaces of limbs are usually brightly colored—yellow, orange, or red. In several specimens the flanks are reddish (pls. 2G, 3G) and in a few specimens the reddish pattern extends through most of the dorsum of specimens (pls. 2H, 3H). Some specimens from Mato Grosso (pl. 3E) and Rondônia (pl. 3A, C, F) have golden yellow or orangeish dorsums. The snout may be whitish or follow the general dorsal pattern. For additional notes on variation, see Peloso and Sturaro (2008).

Call and Tadpoles:

The advertisement call of Chiasmocleis avilapiresae was recently described by Barros et al. (2010). The call consists of a fast series (295.4 notes / min) of multipulsed notes (7–18 pulses/note). Mean note duration was 98.0 ± 9.8 ms, and mean interval between notes 111.1 ± 0.1 ms. Mean dominant frequency of was 3368.2 ± 73.2 kHz. Tadpoles are unknown.

Distribution (fig. 21):

Fig. 21. 

Distribution of Chiasmocleis avilapiresae. Star  =  type locality, solid circle  =  examined specimens.

Brazil, widespread in rainforest areas south of the Rio Amazonas and west of the Rio Tocantins. Known in the states of Acre, Amazonas, Rondônia, Mato Grosso, and Pará. We are unaware of records of this species in adjacent Bolivia and Peru.

Remarks:

In the list of paratypes given by Peloso and Sturaro (2008: 42), collection numbers from MNRJ were mistyped. The publication reads MNRJ 14231–80, but the correct collection numbers for that lot are MNRJ 44231–44280. We thank M. Targino for pointing out this mistake to P.L.V.P.

Genetic distances between all specimens of C. avilapiresae included in the phylogenetic analysis are given in table 6.

Table 6 

Uncorrected pairwise distances between 16S sequences of Chiasmocleis avilapiresae

Fig. 22. 

Holotype of Chiasmocleis bassleri (AMNH 42699) in (A) dorsal, and (B) ventral views. SVL  =  18.5 mm.

Syncope bassleri: (de Sá et al., 2012).

Holotype (fig. 22):

AMNH 42699; poorly preserved, portions of limbs destroyed, and color completely faded.

Type Locality:

Type locality given by Dunn (1949) as “Río Utoquinia to Río Tapiche, Peru (near the Brazilian border).”

Peloso and Sturaro (2008) considered the type locality of Chiasmocleis bassleri as “vague.” Despite that, Rodrigues et al. (2011) gave an “approximate” coordinate for the type locality of C. bassleri, based on unclear arguments. In an attempt to solve this issue, we revisited Harvey Bassler's field notes (deposited at the AMNH) and several hydrographic maps of eastern Peru.

From the field notes, it seems that the holotype (AMNH 42699) was collected at the region where the Río Utoquinia would meet the Río Tapiche. However, from the maps we analyzed, it is not clear whether there is a water connection between the rivers. It is possible that minor waterways are present, thus providing a connection point between the rivers, but it is also possible that the only connection is by land. Regardless, this putative connection must be somewhere near the borders of departamentos of Ucayali and Loreto (both in Peru) and the state of Acre (Brazil)—this region being roughly 120 km east of the city of Contamana, Peru, and 140 km west of Cruzeiro do Sul, Brazil. Dunn (1949) likely used the position of the Utoquinia and Tapiche to draw his map (Dunn, 1949: fig. 4). The locality shown in Rodrigues et al.'s map (Rodrigues et al., 2011: fig. 2) is, however, clearly far from the one presented by Dunn (1949: fig. 4) and certainly does not correspond to any point along the course of either the Río Tapiche or the Río Utoquinia. Coordinates given by Rodrigues et al. (2011) are ca. 450 km southeast of the locality where the distance between the Utoquinia and the Tapiche is at its minimum. After examination of maps and satellite images, we use the following coordinates as a proxy for the type locality of C. bassleri (07°32′30″S / 73°59′35″W: see fig. 25).9

Diagnosis:

A medium-sized species for the genus; SVL in males 19.2–22.1 mm (N  =  11), in females 21.2–28.8 mm (N  =  20). Body ovoid to elongate; head triangular; snout rounded in dorsal and lateral views. Four distinctive fingers; FI may be reduced in some specimens, but always clearly visible; tubercle on FI may be absent or present; subarticular tubercles present on all remaining fingers; adpressed FI never extends beyond the distal margin of subarticular tubercle of FII; adpressed FIV does not reach distal tubercle of FIII; palmar tubercles present, not divided; relative finger lengths I<II<IV<III. Five distinctive toes present, first may be much reduced; toes may be slightly fringed; toes not webbed; TI lacks tubercle; adpressed TI does not touch subarticular tubercle of TII; adpressed TV does not touch or reaches only to the middle of the middle subarticular tubercle of TIV (touches in the holotype); TII–IV with terminal discs, usually more developed in females, but also present in males; relative toe lengths I<II<V<III<IV. An inguinal blotch, variable in shape, is always present. There is usually a marked contrast in color pattern between the dorsum and the venter that can be delimited by a split stripe.

Variation:

We have found a high degree of variation in external characters of Chiasmocleis bassleri, especially in color pattern.

The original description reports a “belly white with five large circular black spots” (Dunn 1949: 9). We have examined sympatric specimens that vary in spot counts, from just a couple to several (over five) well-defined circular spots (see, for example, fig. 23A). Some specimens do not have circular spots, but show elongated stains or, sometimes, a reticulated pattern (fig. 23F; this pattern is common in some specimens from Amazonas and Mato Grosso, Brazil, but is also present in some specimens from Ecuador).

Fig. 23. 

Variation of ventral color pattern in Chiasmocleis bassleri. (A) Parque Nacional da Serra do Divisor, Acre, Brazil (UFAC-RB 1611). (B) Via Tarapacá, Letícia, Amazonas, Colombia (ICN 50249). (C) Porto Walter, Acre, Brazil (OMNH 34829). (D) Rio Curanja, Balta, Ucayali, Peru (KU 197036). (E) Floresta Nacional do Pau-Rosa, Rio Paraconi, Maués, Amazonas, Brazil (MPEG 27766). (F) Aripuanã, Mato Grosso, Brazil (UFMT 7136). Not to scale.

Chest and throat usually consist of dark vermiculations against a light background (fig. 23A, B), but in some specimens the throat is uniformly dark colored, especially in males (fig. 23C, E–F). Few specimens show a uniform pattern (whitish with black spots or stains) over the entire ventral surface (fig. 23D); this pattern is common in specimens from Peru. There is also extensive variation in length ratios between fingers and toes, especially length of FI, FIII/FIV ratio, length of TI, and TIV/TV ratio.

Color in life varies strikingly among specimens (pl. 4), but data are limited and prevent a detailed analysis of geographic variation. Dorsum can have a uniform color pattern, varying between tones of black, dark brown, purple, reddish, and pink. Several specimens may show blotches on the dorsum, usually lighter than the dorsum and varying between shades of red, orange, green, yellow, and pinkish. A specimen from Vaupés, Colombia, shows a bizarre dorsal pattern, with the left side dark reddish brown and the right side light cream with dark brown mottling (pl. 4G).

Most specimens show a well-marked differentiation between dorsal and ventral pattern, evidenced by a split stripe. In some specimens, the stripe is wide and black, extending from the posterior corner of the eye to the inguinal region (pl. 4B). The split is, however, absent in a few specimens (pl. 4D), but we found no clear pattern of geographical variation in this character. Sympatric specimens can show the absence or presence of the split (e.g., in Porto Walter, Acre, Brazil).

Call and Tadpoles:

The advertisement call of C. bassleri has been previously described by Santana et al. (2009). The description here is based on the combined calls of three specimens from different populations, including a reanalysis of Santana et al.'s (2009) recording. Acoustic parameters for each individual call are listed in table 2. The call is composed of a fast, repetitive series of multipulsed notes (mean 5.2 ± 1.2 pulses per note, 2–16, N  =  764), emitted at a rate of 568.2 notes/min. Mean note duration 53.3 ± 11.1 ms (15.0–163.0, N  =  786) and mean interval between notes 49.0 ± 26.3 ms (11.0–302.0, N  =  783). Mean dominant frequency was 2747.0 ± 145.8 Hz (2584–3125, N  =  786). Pulse duration was 6.9 ± 1.6 ms (4.0–11.0, N  =  368). A call from Rio Madeira, Rondônia, Brazil, is illustrated in figure 24.

Fig. 24. 

Advertisement call of Chiasmocleis bassleri. (A) Oscillogram and (B) spectrogram of two consecutive multipulsed notes. Recorded at Ilha da Pedra, Rondônia, Brazil; recorded on 12 March 2010 at 26.4° C air temperature (no voucher specimen; recorded by A. Lima, recording number FNJV 30715).

Tadpoles are unknown.

Remarks:

The phylogenetic position of Chiasmocleis bassleri, given our results, is strikingly distinct from that found by de Sá et al. (2012), represented in his analysis by a single specimen from Loreto, Peru, also included here. We have included several samples of C. bassleri from throughout its distribution in the phylogenetic analysis, and although two well-supported clades were found, we did not find any phenotypic evidence to diagnose the clades as separate taxa. The levels of genetic distance within C. bassleri are generally low (0–6.8%). Genetic distances between all specimens of C. bassleri included in the phylogenetic analysis are given in table 7.

Table 7 

Uncorrected pairwise distances between 16S sequences of Chiasmocleis bassleri

Distribution (fig. 25):

Fig. 25. 

Distribution of Chiasmocleis bassleri. Star  =  type locality, solid circle  =  examined specimens.

Distributed in western Brazil (Acre, Amazonas, Mato Grosso, Pará, and Rondônia), eastern and northern Peru, southwest Colombia, and Ecuador. Icochea et al. (2004) presented a much wider distribution for the species in Peru, and mention the occurrence of the species in Bolivia.

Syncope carvalhoi: (Nelson, 1975).

Holotype:

Fig. 26. 

Paratype of Chiasmocleis carvalhoi (AMNH 88065) from Estirón, Rio Ampi-Yacú (type locality) in (A) dorsal, and (B) ventral views. SVL  =  11.1.

MZUSP 36429; an adult male, in very good state of preservation.

Type Locality:

Río Ampi-Yacú ( =  Ampiyacú), Estirón, Loreto, Peru. No coordinates given in the original. We found at least three different localities in Loreto named Estirón. Because the original explicitly states the specimens were collected along in the Rio Ampiyacú river, we assume the type locality refers to Estirón del Cuzco, for which we provide the following approximate coordinates 03°22′S / 72°00′W.10

Diagnosis:

A small species for the genus; SVL of the male holotype 9.4 mm (Nelson, 1975) and 10.9–11.7 mm in the two female paratypes (N  =  2: Nelson, 1975). Body relatively robust, snout round in dorsal and lateral views. Tympanum is usually not visible. FII–IV much reduced; FI not visible; FIII well developed with subarticular tubercle barely visible, tip of FIII pointed; FIV almost not distinguishable; fingers slightly fringed; palmar tubercles not present; relative finger lengths IV<II<III. Four toes visible, TI not visible; toes only slightly fringed, not webbed; tips of toes slightly expanded, especially in TIII–TIV, tips of TIII–TIV pointed. Relative toe lengths II<V<III<IV. Dermal spines absent. No femoral stripe present. In preservative, venter (belly and undersurfaces of thigh) beige with white spots or blotches. In life, dorsum brownish or orangish with darker (brown to dark brown blotches); cream-colored canthal stripe that may end at the shoulder region or extend all the way to the inguinal regions, becoming wider and more irregularly defined posteriorly; belly with large anastomosing cream-colored large blotches (brown area forms an irregularly reticulated pattern).

Variation:

Number of specimens examined in detail (N  =  7) is much reduced to present data on variation, but we provide a few remarks on color in life, based on examination of photographs (pl. 5A–F). Two syntopic species, from Umarital, Rio Ampiyacú region, Loreto, Peru (∼ 30 km from the type locality) vary markedly in dorsal pattern; JMP 1499 (pl. 5C–D) has a more uniform dorsum (brown) and a canthal stripe extending to the shoulder, while JMP 1451 (pl. 5A–B) has dark brown blotches over a brown dorsum and the stripe extends all the way to the inguinal region. Because these two specimens were examined only from photographs, we do not know whether this difference is due to sex or ontogeny or by chance.

Call and Tadpole:

Call is unknown. Nelson (1975) reported large unpigmented ovarian eggs for the species suggesting endotrophic larvae (e.g., direct development or nonfeeding tadpoles that develop in water or moist soil). Nothing more is known about its reproductive behavior and, at this point, there is no evidence that a tadpole phase even exists.

Distribution (fig. 18):

Known from east Loreto, Peru, and west Colombia (see Acosta-Galvis, 2000, for the Colombian record). Ortega-Andrade and Valencia (2010) provided a record of Chiasmocleis tridactyla (as Syncope tridactyla) for Lorocachi, Pastaza, Ecuador. We have not examined these specimens, but from the figures in their publication it seems those specimens should be tentatively identified as C. carvalhoi based on reduced number of fingers (three) and toes (four) and ventral color pattern (large white blotches) versus small spots in C. tridactyla. The only ambiguous character is the presence of a tympanum (barely visible) in both specimens reported by the authors (QCAZ 9577, 9520; tympanum absent in the types of C. carvalhoi).

Remarks:

Two specimens from Rio Nanay, Loreto, Peru, included in de Sá et al. (2012) as Syncope sp. were found nested within our Chiasmocleis carvalhoi clade. These specimens show 0.0% genetic distance from specimen MNCNADN 27392, collected near Iquitos, Loreto, Peru. Pending examination of a larger series of C. carvalhoi and the vouchers of the specimens sequenced by de Sá et al. (2012), we include those specimens in C. carvalhoi. Genetic distances between all specimens of C. carvalhoi included in the phylogenetic analysis are given in table 8.

Table 8 

Uncorrected pairwise distances between 16S sequences of Chiasmocleis carvalhoi

Holotype (figs. 15D, pl. 5G–H):

MHNSM 21540; examined only from the photographs in the original publication (Funk and Cannatella, 2009).

Type Locality:

Amazon Conservatory for Tropical Studies (ACTS) Field Station, 65 km NE of Iquitos, 1 km N of the Río Napo, and 1.3 km SSE of the ACTS Field Station (3°15′34″S, 72°54′10″W), Departamento Loreto, Peru.

Diagnosis:

Since we have not examined the sole specimen of Chiasmocleis devriesi (except from the photographs published in the original description), readers should refer to the original description (Funk and Cannatella, 2009) for a diagnosis of the species.

Variation:

Known only from the female holotype.

Call and Tadpoles:

Call and tadpoles are unknown.

Remarks:

Funk and Cannatella (2009) described the species on the basis of a single female specimen, which renders any evaluation of morphological variation unattainable. Additionally, many diagnostic characters in Chiasmocleis are present only or are more pronounced in males (i.e., webbing, fringes of digits, and dermal spines). The ventral pattern of the holotype (fig. 15D, pl. 5H) is similar to some specimens of C. anatipes, C. avilapiresae, and C. ventrimaculata and cannot be unambiguously set apart from those on the basis of this character. Chiasmocleis devriesi differs from all species by its larger size (SVL 42.4 mm in C. devriesi; rivaled only by C. avilapiresae, with a maximum recorded SVL, according to our sample, of 37.8 mm) and its moderately pointed snout. The absence of a femoral line further differentiates C. devriesi from C. avilapiresae (femoral line present). From our phylogenetic analysis, it seems clear that the relationship of the sole specimen of C. devriesi lies with populations assigned to C. anatipes and not with C. avilapiresae or C. ventrimaculata (fig. 8). The holotype is related to a couple of specimens from Orellana, Ecuador, forming a well-supported sister clade to another three specimens, unambiguously assigned to C. anatipes. Because the specimens from Orellana are juveniles they cannot be unambiguously assigned to C. anatipes nor do they fit precisely the diagnosis of C. devriesi. For the purpose of our phylogenetic analysis, we have labeled the specimens from Orellana as C. anatipes, but they clustered with the holotype of C. devriesi. There is zero genetic distance in the 16S between the sole specimen of C. devriesi and each of the specimens of C. anatipes included in our analysis (see table 4). We have examined only the holotype of C. devriesi from photographs published in the original description (Funk and Cannatella, 2009; also reproduced here as fig. 15D and pl. 5H). Due to the lack of evidence supporting or falsifying the status of C. devriesi as a valid taxon, we chose not to synonymize it with C. anatipes, but highlight that this issue deserves close attention in the future. For the phylogenetic analysis, for example, only the 16S fragment is present for the holotype of C. devriesi (which shows no genetic differentiation among populations; table 4). Given the phylogenetic proximity of these populations, it would be wise to expand character sampling for these taxa in the future.

Distribution (fig. 27):

Fig. 27. 

Distribution of Chiasmocleis devriesi and C. magnova. Star  =  type locality of C. magnova; solid circle  =  examined specimens of C. magnova; cross  =  type locality of C. devriesi.

Known only from the type locality.

Fig. 28. 

Holotypes of (A–B) Chiasmocleis hudsoni (BMNH 1939.1.1.3: photos: G. Bitencourt), in dorsal and ventral views respectively; and (C–D) Chasmocleis jimi (MNRJ 15459: photos I. Nunes), in dorsal and ventral views respectively.

Fig. 29. 

Variation of ventral color pattern in Chiasmocleis hudsoni. (A–C) Parque Nacional da Amazônia, Itaituba, Pará, Brazil (MPEG 18479, MPEG 18511, and MPEG 18554, respectively). (D) Berbice River 18 mi SW Kwakwani, Guyana (AMNH 166437). (E) Estação Ecológica Grão Pará, Alenquer, Pará, Brazil (CN 1881).

Chiasmocleis jimi: (Caramaschi and Cruz, 2001). Holotype, MNRJ 14549, figure 28C–D, new synonymy.

Syncope jimi: (de Sá et al., 2012).

Syncope hudsoni: (de Sá et al., 2012).

Holotype (fig. 28A–B):

BM 1939.1.1.3; adult male (according to the original publication: Parker, 1940) in very good state of preservation (examined from photographs).

Type Locality:

The type locality was given in the original publication as New River, British Guiana—no geographic coordinates given (Parker, 1940). Frost (1985), based on a personal communication from M.S. Hoogmoed, reported that the locality is in SW Suriname. The New River, however, is in an area of territorial dispute (the New River Triangle) between the Co-operative Republic of Guyana (former British Guiana) and the Republic of Suriname (former Dutch Guiana). The issue is unresolved at present, but historical documents indicate the British Guiana should have precedence over the territory (Donovan, 2003). Although irrelevant to the species distribution, based on the interpretation of ancient and current resolutions (see Donovan, 2003, for a historical perspective), we consider the type locality to be New River, Co-operative Republic of Guyana.

Diagnosis:

A small species for the genus; SVL in males 14.1–23.4 mm (N  =  20); in females 17.2–28.4 mm (N  =  40). Body ovoid, robust; head much narrower than body, snout rounded in dorsal and lateral views. Four distinctive fingers; all but FI slightly fringed, less fringed in females; fingers not webbed; FI reduced with a rounded or slightly pointed tip, lacking subarticular tubercle; finger tips of FII–IV rounded, tips of FII and III swollen, may present discs; adpressed FI does not reach subarticular tubercle of FII; adpressed FIV reach middle or distal margin of distal tubercle on FIII; palmar tubercles protuberant, divided; relative finger lengths I<IV<III<II. Five distinctive toes, first slightly reduced; toes slightly fringed, less distinct in females; toes not webbed in either sex; TI usually without subarticular tubercle (limited to a swelling in some specimens but also may be only slightly visible in few others); adpressed TI does not touch subarticular tubercle of TII; adpressed TV does not touch or reaches only up to the middle of the subarticular tubercle of TIV; TII–IV with terminal discs; relative toe lengths I<II<V<III<IV. Some males with few well-developed dermal spines on chin. Color pattern given in under the variation section.

Variation:

Ventral pattern (throat, chest, and belly) varies between almost completely unpigmented (fig. 29A) to having different amounts of mottling that can extend from the anterior portion of the chin to the hind limbs (fig. 29B–E). Throat is usually more heavily pigmented in males, sometimes completely pigmented. Ventral pigmentation varies from brown to grayish. Most of the variation in this species is with coloration. The rostral region may vary from a completely uniform brown (fig. 29E) to showing a conspicuous pigmentation of orange (fig. 29B, D), cream (fig. 29C, F), or whitish (fig. 29A) extending from the snout to above the eye. In some specimens the colored pigmentation can extend posteriorly to the eye to the inguinal region (fig. 29B–D) becoming more (fig. 29C) or less (fig. 29B, D) faint posteriorly. Dorsal surface of the hind limbs and sometimes the posterior surface of the dorsum may show a variable amount of minute spots of mottling that vary in color among white, golden, purple, and bright orange (fig. 29).

Some males possess few, but large, dermal spines on the chin.

Call and Tadpole:

The advertisement call and a sporadic call of Chiasmocleis hudsoni were described by Rodrigues et al. (2008). The advertisement calls described here (fig. 30) are based on the combined calls of three specimens each belonging to a different population, including one specimen also analyzed by Rodrigues et al. (2008). The call is composed of a repetitive series of multipulsed notes. An introductory note was observed in three calls, with more pulses per note (mean 17.3 ± 2.6 pulses per note, 15–20, N  =  3) and longer duration (427.3 ± 81.6 ms, 360.0–518, N  =  3) than subsequent call notes. The introductory note had a mean dominant frequency of 4249.2 ± 263.1 Hz (3962.1–4478.9, N  =  3), and pulse duration was 7.6 ± 1.8 ms (4.0–11.0, N  =  50). The chorus notes are shorter (mean 5.6 ± 0.6 pulses per note, 5–10, N  =  1108), emitted at a rate of 498.0 notes/min. Mean note duration was 95.4 ± 2.2 ms (7.0–19.8, N  =  1108) and mean interval between notes was 24.8 ± 3.9 ms (15.0–72.0, N  =  1106). Mean dominant frequency was 4845.2 ± 379.0 Hz (3962.1–6373.8, N  =  1108), and the pulse duration was 7.2 ± 2.9 ms (2.0–14.0, N  =  357). Harmonics could be observed in one call, one with a lower frequency than the dominant frequency (2584.0 ± 114.3 Hz, 1894.9–2756.2, N  =  51) and two with frequencies higher than the dominant.

Fig. 30. 

Advertisement call of Chiasmocleis hudsoni. (A) Oscillogram and (B) spectrogram of 11 consecutive notes (last one incomplete) emitted by calling male from Reserva Ducke, Amazonas, Brazil (not collected; recorded by M. Menin; recording number FNJV 30713). (C) Oscillogram and (D) spectrogram of seven consecutive notes emitted by a calling male from Barro Vermelho, Rio Juruá, Amazonas, Brazil; recorded on 27 October 1991, at 24.4° C air temperature (INPA 3353; recorded by C. Gascon; recording number USNM tape 254, cut 7).

The sporadic call is also composed of a repetitive series of multipulsed notes (mean 11.8 ± 4.7 pulses per note, 7–30, N  =  107) emitted at a rate of 294.5 notes/minute. Mean note duration was 158.3 ± 27.8 ms (121.0–292.0, N  =  108) and mean interval between notes was 32.3 ± 9.0 ms (22.0–78.0, N  =  105). Mean dominant frequency was 4726.1 ± 928.5 Hz (3789.8–6373.8, N  =  108), and the pulse duration was 6.8 ± 1.5 ms (3.0–12.0, N  =  282).

The tadpole of Chiasmocleis hudsoni was described by Rodrigues et al. (2008) and the external morphology is similar to that of other tadpoles described in the genus. The absence of an arc shape around the interocular region, a light line between the eyes, and a flagellated tail tip differentiate the tadpole of C. hudsoni from those of C. royi (Schlüter and Salas, 1991).

Remarks:

The status of Chiasmocleis jimi Caramaschi and Cruz, 2001. Caramaschi and Cruz (2001) described C. jimi based on a series of preserved specimens from Humaitá, Amazonas, and Parque Nacional da Amazônia, Pará, both in Brazil. Caramaschi and Cruz (2001) likely did not examine any specimens of C. hudsoni—no list of specimens examined is provided—and therefore based their comparisons solely on the description by Parker (1940). The authors considered C. jimi to differ from C. hudsoni by the swollen tips of fingers III and IV and toes II–V and by color pattern (Caramaschi and Cruz, 2001: 6). We find that additional discussion of both of these characters is warranted.

Swollen digital tips. Caramaschi and Cruz (2001) argue that “tips of all digits not expanded in the other [Chiasmocleis anatipes, C. bassleri, C. hudsoni, C. shudikarensis, C. ventrimaculata] species,” but this statement is inaccurate. The original descriptions of C. anatipes, C. bassleri, C. hudsoni, C. shudikarensis all report discs on at least some digits (Parker, 1940; Dunn, 1949; Walker and Duellman, 1974). This was confirmed by our observations of larger series of specimens, including a large series of specimens of both C. hudsoni and C. jimi (including both holotypes). We found no differences in the morphology of hands and feet of the types of C. hudsoni and C. jimi (fig. 28).

Color in life. The description of the color in life of Chiasmocleis jimi was based on field notes by Celso Morato de Carvalho (Caramaschi and Cruz 2001: 7). Considering only the original descriptions of both species (Parker 1940; Caramashi and Cruz, 2001), they can be distinguished only by (1) “dorsum purple brown with some fine lighter stipplings white forming an indefinite light zone from the tip of snout above canthus rostralis, along the edge of upper eyelid…flanks and limbs with large areas of light pink stippling” in C. hudsoni (Parker, 1940), while “dorsum of body, arms and legs uniform reddish brown with minute irregular white dots” in C. jimi (Caramaschi and Cruz, 2001); (2) “under surfaces white with brown stippling on gular region” in C. hudsoni (Parker, 1940), while “ventrolateral region and venter cream, heavily grayish spotted” in C. jimi (Caramaschi and Cruz, 2001). Apparently the specimen described by Carvalho lacks the “light zone on the snout” (hereinafter, white snout). Color pattern is a variable character and examination of recently collected specimens assigned to both C. jimi or C. hudsoni showed a great degree of variation in the color of live specimens and that the snout may or not be white (pl. 6). The ventrolateral and ventral regions of a C. jimi population from Parque Nacional do Amazonas, Pará, Brazil (where some of the paratypes were collected), are brownish and not grayish (M.S. Hoogmoed, field notes), therefore agreeing with the description of C. hudsoni by Parker (1940).

Advertisement calls. The call of Chiasmocleis hudsoni was described by Rodrigues et al. (2008) from Reserva Florestal Adolpho Ducke, in Manaus, state of Amazonas, Brazil. The authors recognized two call types for C. hudsoni, and named them an advertisement call and a sporadic call (Rodrigues et al., 2008). We had access to two calls of Chiasmocleis from Rondônia, Brazil, that are referable to C. jimi (based on the original description of the species). In the recording, we detect both the advertisement and sporadic call types and all acoustic parameters are very similar to those of C. hudsoni from Amazonas (Rodrigues et al., 2008). Unfortunately, no calls of topotypes of C. jimi are available.

Phylogenetic evidence. The phylogenetic analysis did not recover monophyly of specimens identified (by the collectors) as Chiasmocleis jimi or C. hudsoni. Despite uncertainty about the relationships among populations, together all specimens form a well-corroborated clade, sister to C. haddadi, sp. nov. The amount of uncorrected genetic distance between populations is considerably variable (0%–11.4%), with up to 2.3% distance between syntopic species. Genetic distance values between all specimens of C. hudsoni included in the phylogenetic analysis are given in table 9.

Table 9 

Uncorrected pairwise distances between 16S sequences of Chiasmocleis hudsoni

The Status of Chiasmocleis jimi:

Given the above discussion and character analyses, we feel that the present evidence is overwhelming and our conclusion is that C. jimi Caramaschi and Cruz, 2001, must be regarded a junior synonym of C. hudsoni Parker, 1940.

Distribution (fig. 31):

Fig. 31. 

Distribution of Chiasmocleis hudsoni and C. haddadi. Star  =  type locality of C. jimi (considered here as a synonym of C. hudsoni); circles  =  examined specimens of C. hudsoni; cross  =  type locality of C. haddadi; half circles  =  examined specimens of C. haddadi.

Most of the Amazon basin in Brazil, west of the Rio Tocantins. Although we have not examined specimens from several populations in the Guiana Shield countries, there are known records of the species in Colombia, Venezuela, Guyana, Surinam, and French Guiana (Barrio-Amorós and Schargel, 2003; Rodrigues et al., 2004a). Possibly present in Peru, although we are unaware of any records.

Syncope magnova: (de Sá et al., 2012).

Holotype:

Fig. 32. 

Chiasmocleis magnova (AMNH 103550) from Iquitos, Peru, in (A) dorsal, and (B) ventral views. SVL  =  16.7 mm.

MHNSM 19993, not examined.

Type Locality:

The type locality is 31 km on the road from Iquitos to Nauta, ca. 40 km straight SW of Iquitos (04°00′ S, 73°26′ W), Departamento Loreto, Peru.

Diagnosis:

A small species for the genus; SVL in females 16.7–18.3 mm (Moravec and Köhler, 2007). We had no access to male specimens. Body slender, snout rounded in dorsal and lateral views. FI much reduced without subarticular tubercle, tip slightly pointed; FII and FIII well developed with hardly visible subarticular tubercles, one on FII and two on FIII; FIV reduced, without subarticular tubercles; fingers only slightly fringed according to Moravec and Köhler (2007), not fringed in the sole specimen examined (AMNH 103550); palmar tubercles not present; relative finger lengths I<IV<II<III. Five distinctive toes present; TI reduced, tip does not reach tubercle of TII by a wide margin; toes only slightly fringed, not webbed; tips of TI pointed, TII–IV rounded with terminal discs present. Relative toe lengths I<II<V<III<IV. Dermal spines present on dorsum, posterior surface of thighs, toes, and tarsus; absent on fingers. No femoral stripe present. Venter (belly and undersurfaces of thighs) beige with small light spots. Few large unpigmented eggs (Moravec and Köhler, 2007).

Call and Tadpoles:

Unknown.

Remarks:

Moravec and Köhler (2007), when describing the species, suggested that this species might be associated with the genus Syncope (all former members of Syncope are now included in our Chiasmocleis hudsoni clade). This was based mostly on the basis of the presence of few large eggs (a condition also present in Syncope: Krügel and Richter, 1995; Silva and Meinhardt, 1999) and the significant digit reduction.

In our analysis, the sole specimen of Chiasmocleis magnova (sequence from de Sá et al., 2012) is well supported as the sister to a clade containing C. antenori, C. carvalhoi, and C. tridactyla while they all form the sister clade to C hudsoni plus C. haddadi, sp. nov. Studies on the ecology of C. magnova can provide useful insights on the evolution of the group as a whole, as it seems C. magnova can be a transitional form between the more general Chiasmocleis body plan, and the miniaturized C. antenori, C. carvalhoi, and C. tridactyla.

Distribution (fig. 27):

Known from the vicinities of the type locality, in Iquitos, Departamento Loreto, Peru.

Holotype (fig. 33):

Fig. 33. 

Holotype of Chiasmocleis shudikarensis Dunn, 1949 (AMNH 43674), in (A) dorsal, and (B) ventral views. SVL  =  23.3 mm.

AMNH 43674; adult, probably female specimen. The chest is cut open. Both hands are partially destroyed and so is the dorsal portion of the left foot. Other than that, the specimen is well preserved and some details of color pattern are still clearly visible.

Type Locality:

Shudikar-Wau, Upper Essequibo River, British Guiana (now the Co-operative Republic of Guyana). No geographic coordinates were given in the original description.

Diagnosis:

A medium-sized species for the genus; SVL in males 18.7–25.5 mm (N  =  22); in females 20.9–29.7 mm (N  =  11). Body ovoid, robust; head triangular, snout rounded in dorsal and lateral views. Four distinctive fingers, all but FI fringed in males, much less fringed in females; fingers not webbed; FI well developed with a distinct well-developed subarticular tubercle present between the proximal phalanges; distinctly visible subarticular tubercles present on all fingers; adpressed FI reaches or extends beyond the subarticular tubercle of FII; adpressed FIV does not reach or barely reaches distal tubercle of FIII; palmar tubercles protuberant, divided; relative finger lengths I<IV<II<III. Five distinctive and well-developed toes present; toes fringed, less distinct in females; toes usually extensively webbed in males (fig. 5C) and only basally webbed in females (fig. 5B); TI with a distinct well-developed subarticular tubercle; adpressed TI does not touch or barely touches subarticular tubercle of TII; adpressed TV does not touch or reaches only the middle of subarticular tubercle of TIV; TII–V with terminal discs; relative toe lengths I<II<V<III<IV. Males with many dermal spines on fingers and toes; both sexes may show dermal spines on dorsum, fingers, and toes, much more numerous and developed in males; males with many spines on anterior portion of chin, lacking in females. Femoral line is always present (see pl. 7E).

Variation:

In almost all populations where large series of both males and females were examined (e.g., FLONA do Amapá, Amapá, and Manaus, Amazonas), marked sexual dimorphism in fringes on hands was observed, with males having conspicuously more developed fringes than females. The fringes in males are usually ornamented with numerous dermal spines. Males from Petit Saut, French Guiana, also show more developed fringes in fingers than females, but the difference is not as conspicuous as in other populations (e.g., Manaus and Amapá).

In life, the dorsum of males is usually reddish or grayish but is commonly marbled in shades of yellow or orange (pl. 7). Dorsum of females is usually more uniformly colored, gray or blackish. Snout follows the same general pattern of the dorsum, or is whitish (pl. 7A, C). A dorsal line may be present, varying in extent (as shown in fig. 10). Hind limbs of males and females usually yellow, orange, or reddish. Forelimbs grayish, yellow, or reddish, being uniform in females and usually marbled (as in the dorsum) in males. Venter color pattern varies in much the same way as it does in C. avilapiresae (illustrated in fig. 20). An inguinal blotch is invariably present, although variable in shape and size.

Call and Tadpole:

The advertisement call of Chiasmocleis shudikarensis was described by Zimmerman and Bogart (1988) from Manaus, Amazonas, Brazil; and by Lescure and Marty (2000) from French Guiana. Both calls presented distinct acoustic parameters: Manaus, Brazil—“frequency range” 5890–7510 Hz; note duration 10–60 ms; interval between notes 10–30 ms (Zimmerman and Bogart, 1988); French Guiana—call rate 360 notes/min; dominant frequency 3380–3750 Hz; note duration 98 ms (Lescure and Marty, 2000). The difference in dominant frequency is striking and deserves further scrutiny as it suggests the possible that at least two distinct taxa are present.

The tadpole of Chiasmocleis shudikarensis was supposedly illustrated but not described by Hero (1990). The illustrations were later reported to belong to another microhylid, Ctenophryne geayi (Menin et al., 2011). Tadpoles of C. shudikarensis were recently described in detail by Menin et al. (2011).

Remarks:

Ávila-Pires et al. (2010) referred to a presumably undescribed species from the portion of the state of Pará north of the Amazon River. P.L.V.P. was present on a fieldtrip where two of the specimens were collected (Estação Ecológica Grão Pará: CN2112, CN2185—MPEG 28327, MPEG 28328, respectively; see pl. 7A) and we have carefully examined some additional specimens from the series mentioned by Ávila-Pires et al. (2010) at the MPEG collection. We disagree with Ávila-Pires et al. (2010) and consider the populations to be unambiguously assignable to either C. shudikarensis or C. hudsoni (see appendix 1 for identifications). The C. shudikarensis specimens from “Calha Norte” do have a smaller SVL than specimens from French Guiana and Amazonas, but they agree with the variation we observed in larger samples of C. shudikarensis, and also agree with the original description of the species (Dunn, 1949) and the general morphology with the holotype (AMNH 43674).

Despite distinct advertisement call parameters, we have decided not to name an additional species of Chiasmocleis allied to C. shudikarensis at this time for the following reasons: we have examined specimens of the population assigned to the “Manaus call type” (INPA 245, 277–78, 289, 320), but have not examined the vouchers for the “French Guiana call type.” The specimens from Manaus agree with the original description and with the holotype C. shudikarensis. We have examined a series of specimens from Petit Saut, French Guiana (MPEG 5106–5107, 5121, 5029–5033, 5060) and the only morphological differences are that male specimens from Sinnamary River show considerably less fringed fingers than those from Manaus, Amazonas, Brazil, and females from French Guiana are slightly more robust. Therefore, the population from French Guiana is also unambiguously assignable to C. shudikarensis. In this particular case, examination of specimens and advertisement calls from the vicinities of the type locality (Essequibo River) is advisable before nomenclatural actions are taken. Genetic distances between 16S sequences included in our analysis very low and are given in table 10.

Table 10 

Uncorrected pairwise distances between 16S sequences of Chiasmocleis shudikarensis

Distribution (fig. 34):

Fig. 34. 

Distribution of Chiasmocleis shudikarensis and C. papachibe. Circles  =  examined specimens of C. shudikarensis; half circles  =  specimens of C. shudikarensis examined solely from live photographs; cross  =  type locality of C. papachibe; diamond  =  examined specimens of C. papachibe.

Restricted to the Guiana Shield in Guyana, Surinam (we have not examined any specimens from Surinam), French Guiana, and Brazil, north of Rio Amazonas (Amapá, Amazonas, Pará, and Roraima). Rodrigues et al. (2004b) provided a map that likely overestimates the distribution of the species, extending south of Rio Amazonas (Acre, Mato Grosso, and Rondônia). All of the specimens previously identified as Chiasmocleis shudikarensis in collections from southern Amazonia were identified by us as either C. avilapiresae or C. bassleri.

Fig. 35. 

Paratype of Chiasmocleis supercilialba Morales and McDiarmid, 2009 (USNM 342862), paratype in (A) dorsal, and (B) ventral views. SVL  =  18.8 mm.

Chiasmocleis supercilialbus (Morales and McDiarmid, 2009). Incorrect spelling in the original description (see Peloso et al., 2013).

Syncope supercilialbus (de Sá et al., 2012).

Holotype:

MHNSM 16174; examined solely from a photograph, in life (not shown).

Type Locality:

Pakitza, Reserve Zone, Manu National Park, ca. 57 km northwestern from the mouth of Rio Manu (11°56′47″S / 71°17′00″W), on Rio Manu, Departamento Madre de Dios, Peru.

Diagnosis:

A medium-sized species for the genus; SVL in males 18.4–18.7 mm (Morales and McDiarmid, 2009). Females unknown. Body ovoid to elongate; head triangular, snout rounded in dorsal and lateral views. Four distinctive fingers; FI well developed, subarticular tubercle may be absent or present; subarticular tubercles present on all remaining fingers, sometimes hardly visible; adpressed FI never extends past the distal margin of subarticular tubercle of FII; adpressed FIV does not reach distal tubercle of FIII; palmar tubercles present, not divided; relative finger lengths I<II<IV<III. Five distinctive toes present, first may be much reduced; toes may be slightly fringed; toes not webbed; TI lacks tubercle; adpressed TI does not touch subarticular tubercle of TII; adpressed TV does not touch middle subarticular tubercle of TIV; TII–IV with terminal discs, usually more developed in females, but also present in males; relative toe lengths I<II<V<III<IV. An inguinal blotch of varied shape is always present. Venter (belly and under surfaces of thighs) usually white or cream with black or dark brown irregular spots.

Variation:

Some variation in color pattern was observed from photos of live specimens (pl. 8) and examination of a paratype (USNM 342862: fig. 35). The dorsum can be dark brown, to almost uniformly whitish (silver) and may show reddish blotches and white spots (when not entirely whitish). A wide white stripe is almost invariably present from the snout to past the eyelids, sometimes extending to the inguinal region (forming the split stripe). A very narrow dorsal white line commonly present, extending from the occipital region to the cloacal region. Dorsal surfaces of forearm yellowish, orange or red, while dorsal surfaces of hind limbs usually follow the general dorsal pattern. As reported by Morales and McDiarmid (2009) the shape and number of inguinal and lateral blotches (or spots) is variable, but commonly dark brown. Throat region with intense brown reticulation over a white (or cream-colored) background. Chest, belly, and undersurfaces of limbs with large, dark brown spots or irregular blotches (fig. 35, pl. 8D, F, H).

Call and Tadpole:

The advertisement call of C. supercilialba was described by Morales and McDiarmid (2009). Call consists of a fast series of multipulsed notes (mean note duration 31.0 ms; mean interval between notes 37.0 ms) with dominant frequency between 2985.8–3205.0 Hz. Number of pulses per note not given. Tadpoles are unknown.

Remarks:

A population from Rio Formoso, state of Rondônia, Brazil (OMNH 37204–37206, 37308–37309, 37319), present dermal spines on dorsum, upper lips, cloacal region, fingers, and toes. The specimens also show well-developed fringes on the fingers and toes, and the toes are basally webbed. These specimens, therefore, agree (in part) with the diagnosis of Chiasmocleis supercilialba (see Morales and McDiarmid 2009, and text above). We tentatively assign the Rio Formoso specimens to C. supercilialba.

Distribution (fig. 36):

Fig. 36. 

Distribution of Chiasmocleis supercilialba. Star  =  type locality; circle  =  additional specimens examined; cross  =  specimens from Rondônia, Brazil, tentatively labeled as C. cf. supercilialba.

Southern Peru (Madre de Dios) and western Brazil (Acre and Rondônia).

Fig. 37. 

Chiasmocleis tridactyla (MPEG 5637) from Tabatinga, Amazonas, Brazil, in (A) dorsal, and (B) ventral views. SVL  =  11.5.

Adelophryne tridactyla: (Duellman and Mendelson III, 1995).

Syncope tridactyla: (Silva and Meinhardt, 1999).

Holotype:

URP-WED 59935. Not examined. Numbers are W.E. Duellman field numbers. According to the original publication, the specimens were to be deposited at the Museo de Historia Natural, Ricardo Palma, Lima, Peru (Duellman and Menselson III, 1995)

Type Locality:

San Jacinto, Departamento Loreto, Peru (02°18′45″S / 75°51′46″W, 180 m).

Diagnosis:

A small species for the genus; SVL in females 11.1–12.4 mm (Duellman and Menselson III, 1995). Body relatively robust, snout round in dorsal and lateral views. A distinct tympanum is visible. FI–IV much reduced; FI almost completely lost; FIV reduced but clearly identifiable, with tip round; FIII well developed with two subarticular tubercles, tip pointed; FIV much reduced, without subarticular tubercles, tip round; fingers slightly fringed; palmar tubercles not present; relative finger lengths I<IV<II<III. Four toes visible, TI not visible; toes only slightly fringed; not webbed; tips toes pointed, discs absent. Relative toe lengths II<V<III<IV. Dermal spines absent. No femoral stripe present. In preservative, venter (throat, chest, belly, and undersurfaces of thigh) beige with small light spots. In life, dorsum brownish with light flecking (very small spots) and a cream-colored canthal stripe. Throat, chest, and belly brown with small spots (larger and more irregular posteriorly. Iris is red.

Variation:

Number of specimens examined in detail (N  =  6) is much reduced to present data on variation, but we provide a few remarks on color in life, based on examination of photographs (pl. 9A–D). Duellman and Mendelson III (1995) reported a dorsolateral cream-colored stripe in specimen KU 221992 (from Teniente Lopez, Loreto, Peru) but the specimens from Leticia, Colombia, lack the dorsolateral portion of the stripe. In those specimens, the stripe ends on the posterior edge of the eyelid. Spots on the dorsum seem to be more numerous and larger in the specimen from Teniente Lopez.

Call and Tadpoles:

Unknown.

Distribution (fig. 38):

Fig. 38. 

Distribution of Chiasmocleis tridactyla. Star  =  type locality; circles  =  examined specimens.

Known from Loreto, Peru (San Jacinto, the type locality, and Teniente López), and from the vicinities of the Brazil-Colombia border near the cities of Leticia (Colombia, Amazonas) and Tabatinga (Brazil, Amazonas).

Remarks:

Silva and Meinhardt (1999), when transferring the species to the genus Syncope, provided a detailed osteological description of the species based on a specimen from the same population examined by us (Tabatinga, Brazil: RMNH 26812). The specimens examined here, as well as the one examined by Silva and Meinhardt (1999) were listed by Duellman and Mendelson III (1995) “as tentatively referred to A.[delophryne] tridactyla.” We have not examined the types of Syncope tridactyla and therefore follow Duellman and Mendelson III (1995) to refer to the Tabatinga specimens as Chiasmocleis tridactyla. The specimens from Leticia, Colombia (JMP 1948, 2057, 2231: examined through photographs, see pl. 9A–D) agree with the original description and resemble the specimens we have examined from Tabatinga. The specimens from Leticia were, however, collected just around 15 km away from the city of Tabatinga. Genetic distances between the three specimens of C. tridactyla included in the phylogenetic analysis are given in table 11.

Table 11 

Uncorrected pairwise distances between 16S sequences of Chiasmocleis tridactyla

Fig. 39. 

Holotype of Chiasmocleis ventrimaculata (Andersson 1945) (NRM 1943, holotype) in (A) dorsal, and (B) ventral views (photos: J.M. Padial).

Engystoma ventrimaculata: (Andersson, 1945).

Chiasmocleis ventrimaculata: (Dunn, 1949).

Holotype:

NRM 1943; adult female, examined from photographs (fig. 39). Specimen desiccated, but still preserving overall morphological features and some of the color pattern.

Type Locality:

Río Pastaza, Ecuador. No geographic coordinates given in the original description (Andersson, 1945). The Pastaza River is one of the major rivers in Ecuador, running through a large portion east of the country in the Provinces of Chimborazo, Morona-Santiago, Pastaza, and Tungurahua. It then runs into Peru, where it merges with the Marañon River. The type locality is, therefore, not precise.

Diagnosis:

A medium-sized species for the genus; SVL in males 20.0–23.0 mm (Rodriguez and Duellman, 1994; Funk and Cannatella, 2009); in females 21.0–33.4 mm (Rodriguez and Duellman, 1994; Funk and Cannatella, 2009). Body slender; snout rounded in dorsal and lateral views. Four distinctive fingers, all but FI slightly fringed in males; fingers not webbed; FI well developed with a subarticular tubercle present between the proximal phalanges; subarticular tubercles usually present on all fingers; adpressed FI may reach, but never reaches beyond, the subarticular tubercle of FII; adpressed FIV may reach distal tubercle of FIII; palmar tubercles visible, divided; relative finger lengths I<IV<II<III. Five distinctive and well-developed toes present; toes fringed without webbing or with only basal webbing present between TIII–TIV and TIV–TV; TI with a distinct well-developed subarticular tubercle; adpressed TI barely touches subarticular tubercle of TII; adpressed TV reaches the tip of subarticular tubercle of TIV; TII–IV with terminal discs; relative toe lengths I<II<V<III<IV. Males with many dermal spines on chin, fingers, toes, and dorsum; spines less abundant in females, more common on cloacal region. Inguinal blotch absent. Femoral stripe on posterior thighs absent. Venter, throat, belly, and undersurface of thighs white or cream colored with large dark brown, irregular blotches; blotching commonly more dense on throat; in preservative, dorsal surface of arms with large white or cream-colored stains.

Variation:

Venter is usually light colored, white, or cream, with brown to black stains and blotches, sometimes forming a slightly reticulated pattern, extending from chin through the throat, chest, belly, and undersurfaces of limbs (fig. 40). In life, the dorsum is almost uniformly dark brown, but may show minute whitish spots (pl. 9E–H). Snout is sometimes golden or whitish, while dorsal surfaces of limbs may show orange or yellow stains of irregular shape (pl. 9E, H).

Fig. 40. 

Variation of ventral color pattern in Chiasmocleis ventrimaculata. (A) Cantadera Tacaná, Letícia, Colombia (ICN 50245). (B) Reserva Extrativista do Alto Juruá, Cruzeiro do Sul, Acre, Brazil (UFAC RB 2252). (C–D) Reserva Extrativista Riozinho da Liberdade, Tarauacá, Acre, Brazil (UFACF 641, and UFACF 910, respectively).

Call and Tadpole:

Calls assigned to Chiasmocleis ventrimaculata were described twice in the past, once by by Nelson (1973) from a population from Colombia, and again by Schlüter (2005) from Panguana, Peru. We consider the call described by Schlüter (2005) to represent a distinct species, C. royi, sp. nov. (described below). The call described by Nelson (1973) has a mean dominant frequency range of 3562 ± 110 Hz (3350–3700 Hz). Attempts to find the recordings analyzed by Nelson (1973), for potential reanalysis, were unfruitful. Tadpole is unknown.

Remarks:

Several populations from southern Peru and Bolivia have been called Chiasmocleis ventrimaculata in the past (De la Riva, 1995; Schlüter, 2005; Peloso and Sturaro, 2008; Von May et al., 2009). After examination of photographs of the holotype of C. ventrimaculata (fig. 39) and a large sample of specimens from throughout the taxon's distribution, we concluded that two species, one of which we describe below (C. royi, sp. nov.), are currently associated with the name. Our phylogenetic analysis corroborates this, and the two species are not even sister to each other. Instead, C. royi, sp. nov., is sister to C. papachibe, sp. nov., + C. ventrimaculata.

Genetic distances between the three specimens of Chiasmocleis ventrimaculata is zero.

Distribution (fig. 41):

Fig. 41. 

Distribution of Chiasmocleis ventrimaculata and C. royi. Crosses  =  examined specimens of C. ventrimaculata. Dashed area refers to the extension of the Río Pastaza ( =  Pastaza River), inside the territory of Ecuador, given as the type locality of C. ventrimaculata by Andersson (1945). Star  =  type locality of C. royi; circles  =  additional examined specimens of C. royi.

Colombia, Ecuador, northern Peru, and western Brazil (Acre and western Amazonas).

Holotype (fig. 42, pl. 10A–B):

MNHNP 2011.0139 (field number PG 403), an adult male, collected by P. Gaucher, on February 21, 2005.

Type Locality:

Montagne Kotika (03°56′00″N / 54°11′05″W, 700 m), French Guiana.

Paratypes:

MNHNP 2011.0140, MNHNP 2011.0143–2011.0145, all collected with the holotype. IEPA (FL 453–455, FL 511–513), Igarapé Santo Antônio, Floresta Nacional do Amapá (1°06′04″ N / 51°53′36″ W), state of Amapá, Brazil, collected by J.D. Lima and A. Ferreira-Sobrinho in February 28 to March 20, 2005. IEPA (TQ 1036), from Rio Anacuí, Parque Nacional Montanhas do Tumucumaque (1°50′41″N / 52°44′28″W), state of Amapá, Brazil, collected by J.D. Lima and J.R.F. Lima from February 22 to March 11, 2006.

Diagnosis:

A small sized species for the genus; SVL in males 13.7–18.2 mm (N  =  12); females unknown. Body ovoid, robust; head much narrower than body, snout rounded in dorsal and lateral views. Four distinctive fingers; all but FI slightly fringed, less fringed in females; fingers not webbed; FI reduced with a rounded or slightly pointed tip, subarticular tubercle usually absent, hardly visible if present; finger tips of FII–IV rounded, tips of FII and III swollen, may present discs; adpressed FI does not reach subarticular tubercle of FII; adpressed FIV reaches middle or distal margin of distal tubercle of FIII; thenar tubercle present in most individuals; relative finger lengths I<II<IV<III. Five distinctive toes, first slightly reduced; toes slightly fringed, less distinct in females; toes not webbed; TI with without subarticular tubercle; adpressed TI does not touch subarticular tubercle of TII; adpressed TV does not reach or reaches only the middle of the middle subarticular tubercle of TIV; TII–IV with terminal discs. Some males with few well-developed dermal spines on chin.

Comparisons Chiasmocleis haddadi differs from C. antenori in having four externally evident fingers (FI not evident externally in C. antenori), five externally evident toes (TI not evident), toe tips rounded (toe tips pointed), and in having a reticulated venter (dark with scattered light spots). It differ from C. albopunctata in having a reduced first finger (well-developed finger in C. albopunctata) and by its smaller size, its snout and dorsum lacking large white spots (snout and dorsum with large white spots in C. albopunctata), its reduced FI (well-developed in C. albopunctata), and its advertisement call. Chiasmocleis haddadi differs from C. anatipes by the absence or much-reduced webbing between toes in males (extensive webbing in males C. anatipes) and its having a reduced first finger (well-developed finger in C. anatipes) and a reticulated venter (light colored venter with large brown spots or blotches in C. anatipes). The new species is distinguished from C. avilapiresae by: its smaller size, the absence of a subarticular tubercle on TI (present in C. avilapiresae), the absence of or much-reduced webbing between toes in males (extensively webbed in C. avilapiresae), its mottled venter (usually blotched in C. avilapiresae), the lack of a femoral line (present in C. avilapiresae), and its advertisement call. The new species differs from C. bassleri in body shape (more robust in C. bassleri), in having a mottled venter (usually large dark spots over a light venter in C. bassleri), by the absence of an inguinal blotch (present in C. bassleri), and by the advertisement call (table 2). The new species differs from C. carvalhoi in having four externally evident fingers (FI not evident externally in C. carvalhoi) and five externally evident toes (TI not evident). Chiasmocleis haddadi differs from C. hudsoni by: its dark brown dorsum (usually light brown or cream dorsum in C. hudsoni), dark brown vermiculations on the throat (sometimes largely brown with small white spots), and its advertisement call (table 2). The new species differs from C. magnova in having finger IV fully developed, although reduced when compared to most other species (finger VI much reduced in C. magnova), and by the dark brown reticulation on throat, chest, and undersurfaces of limbs (belly and undersurfaces of thighs beige with small light spots in C. magnova). From C. royi the new species differs by having a reduced FI (fully developed in C. royi), by the absence of a femoral line (present in C. royi) and an externally protruding black vocal sac in males (protruding in C. royi), and by its advertisement call (table 2). The new taxon differs from C. papachibe, sp. nov., by its smaller size, and in having a reduced first finger (fully developed with a distinct subarticular tubercle in C. papachibe). Chiasmocleis haddadi is distinguished from C. shudikarensis by the absence of or much-reduced webbing between toes in males (males with extensive webbing in C. shudikarensis), by the lack of spines on fingers of males (many spines in males of C. shudikarensis), a femoral line (well-defined white line in C. shudikarensis), and an inguinal blotch (present in C. shudikarensis), and by its advertisement call (see text). From C. supercilialba, the new taxon differs by having a mottled venter (large dark spots or stains over a light venter in C. supercilialba), by the absence of an inguinal blotch (present in C. supercilialba), and by its advertisement call (see text). Finally, it differs from C. tridactyla in having four externally evident fingers (FI not evident externally in C. tridactyla), five externally evident toes (TI not evident), rounded toe tips (toe tips pointed), and a reticulated venter (dark with scattered light spots).

Description of the Holotype (fig. 42, pl. 10A–B):

Body ovoid, relatively robust. Head short, wider than long (HW 1.25 × HL); head much narrower than body trunk; snout rounded in dorsal and lateral views; nostrils not protuberant, positioned anterolaterally, directed laterally. IOD 2.2 ×IND. Canthus rostralis indistinct, round in cross section. Eyes are small. Occipital or supratympanic folds absent; tympanum not visible. Upper jaw projecting beyond lower; lower lip with truncate, trilobed anterior margin; tongue large, elongate, with free lateral and posterior edges, extended beyond the extent of the jaw; vocal slits present, one on each side of the tongue; choanae small, rounded, widely separated, anterior to eye; vomerine teeth absent. Vocal sac not externally evident.

Forelimb slender; FI reduced, FII, FIII, and FIV well developed; tips of FI and FII rounded; discs on tips of FIII and FIV; all fingers nearly round in cross section; no webbing between fingers; relative finger lengths I<II<IV<III; tips rounded, with small discs on FIII and IV. Subarticular tubercles present on all fingers, but inconspicuous on FI; one subarticular tubercle on FII and FIV, two on FIII; all tubercles well developed; no supernumerary tubercles; thenar tubercle present, rounded; palmar tubercle divided, not prominent. No dermal spines visible on arms, hands, or fingers.

Legs short (combined THL, TBL, and FL 1.6 ×the SVL), relatively robust, lacking tubercles, tibial and tarsal ridges, and warts; toes not webbed; TI weakly developed (tip fails to reach subarticular tubercle on TII), without subarticular tubercle; toe tips rounded with small discs on all except TI. Inner metatarsal tubercle present. Subarticular tubercles present on all toes except TI, large but very weakly developed; no outer metatarsal tubercle. TI barely reaches proximal margin of the tubercle of TII; TV reaches distal margin of proximal subarticular tubercle of TIV; relative toe lengths I<II<V<III<IV; toes without dermal spines.

Skin smooth dorsally and ventrally.

Measurements of the Holotype:

SVL 14.4, HL 3.5, HW 4.1, ED 1.7, IOD 3.0, IND 1.3, END 1.2, THL 6.2, TBL 6.9, FL 10.1, 4TD 0.6.

Variation:

Dorsum uniform dark brown or mostly dark brown with scattered small white dots. Throat, chest, and ventral surfaces of limbs dark brown with small cream-colored spots; venter varies from dark with white dots to cream colored with intense dark brown vermiculations (fig. 43). In life, ligher portions of venter are blueish (pl. 10B, F) but may show some yellow regions (pl. 10D).

Color in Preservative:

Dorsum and dorsal surfaces of limbs dark brown with scattered small white spots (usually associated with small warts and more abundant on forelimbs). Throat, chest, and ventral surfaces of limbs dark brown with small cream spots; venter cream colored with intense brown vermiculations.

Etymology:

The specific epithet is used as a noun in the genitive case and honors Célio F.B. Haddad, friend, amphibian enthusiast, and professor at the Universidade Estadual Paulista, Rio Claro, São Paulo. We honor Célio for his invaluable contribution to South American herpetology, as well as for his continuous help (including on several Chiasmocleis-related matters), incentive, and friendship over the years.

Call and Tadpoles:

The advertisement call of Chiasmocleis haddadi (fig. 44) consists of a series (164 notes/minute) of multipulsed notes (mean note duration 330 ± 40 ms; mean interval between notes 60 ± 20 ms) with a mean dominant frequency of 4146.5 ± 60.1 kHz.

Fig. 44. 

Advertisement call of Chiasmocleis haddadi. (A) Oscillogram and (B) spectrogram of two consecutive multipulsed notes. Recorded at the type locality, Montagne Kotika, French Guiana (recording number FNJV 30718).

Remarks:

An individual from Mont Bakra, French Guiana, previously labeled as C. hudsoni (Fouquet et al., 2007), included in the phylogeny (voucher specimen not examined) is sister to four samples of C. haddadi from Mont Kotika, French Guiana (all paratypes) (fig. 6), and is included in this taxon. Genetic distances between all specimens of Chiasmocleis haddadi included in the phylogenetic analysis are given in table 12.

Table 12 

Uncorrected pairwise distances between 16S sequences of Chiasmocleis haddadi

Distribution (fig. 31):

Known from the type locality and from Mont Bakra in French Guiana, and from two localities in the state of Amapá, Brazil.

Holotype (fig. 45, pl. 11A–B):

Fig. 45. 

Holotype of Chiasmocleis papachibe, sp. nov. (MPEG 30683), in (A) dorsal and (B) ventral views. SVL  =  24.8 mm.

MPEG 30683 (field number PRG 043), an adult male, collected by M.J. Sturaro and A. D'Angiolella in April 18, 2010.

Type Locality:

Paragominas (03°06′23″ S / 47°46′55″W), municipality of Paragominas, state of Pará, Brazil.

Paratype:

MPEG 30684 (field number PRG 055), an adult female (pl. 11C–D), from the type locality, collected by M.J. Sturaro and A. D'Angiolella, in April 23, 2010.

Diagnosis:

A medium-sized Chiasmocleis; SVL of adult male 24.8 mm, of female 32.6 mm. Body ovoid and relatively slender; snout rounded in dorsal and lateral views, IOD about 2.1–2.3 ×the IND. Four distinctive fingers present; all but FI fringed in males; FI well developed with a subarticular tubercle usually present between the proximal phalanges; adpressed FI does not touch subarticular tubercle on FII; adpressed FIV touches distal subarticular tubercle of FIII; distinct subarticular tubercles present on all fingers; palmar tubercles protuberant, divided; relative finger lengths I<II<IV<III. Five distinctive and well-developed toes present; toes fringed, less so in female; adpressed TI does not touch or barely touches subarticular tubercle of TII; adpressed TV does not reach or reaches only to the middle of the middle subarticular tubercle on TIV; TII–IV with terminal discs, usually more developed in females; toes basally webbed; relative toe lengths I<II<V<III<IV. Males without dermal spines on fingers and toes; both sexes without dermal spines on dorsum and toes. Vocal sac undeveloped, vermiculated by black and white. Femoral line and inguinal blotch are always absent.

Comparisons:

Chiasmocleis papachibe differs from C. antenori by its much larger SVL (maximum recorded SVL 13.6 mm in C. antenori) and in having four externally evident fingers (FI not evident externally), five externally evident toes (TI not evident), toe tips rounded (toe tips pointed), and a reticulated venter (dark with scattered light spots).

The new species differs from Chiasmocleis anatipes in having basal webbing in toes of males (extensive webbing in males of C. anatipes), and by its vermiculated venter (dark spots in C. anatipes). The new species is distinguished from C. avilapiresae by: its smaller size (SVL of C. avilapiresae up to 37.8 mm in females), its basally webbed feet in males (extensively webbed in male C. avilapiresae), the absence of a femoral line (present in C. avilapiresae), and its venter with intense dark reticulations over a white background (brown stains or blotches in C. avilapiresae). The new species differs from C. bassleri in having a vermiculated venter (large dark spots or stains over a light venter in C. bassleri) and by the absence of an inguinal blotch (present in C. bassleri). The new species differ from C. carvalhoi by its much larger SVL (maximum recorded SVL 13.2 in C. carvalhoi: Nelson, 1975), four externally evident fingers (FI not evident externally), and five externally evident toes (TI not evident). The new species is distinguished from C. haddadi and C. hudsoni by its larger size (maximum SVL 23.4 mm in C. hudsoni and 18.2 mm in C. haddadi), fully developed FI (reduced in C. haddadi and C. hudsoni), and a round finger I tip (tip slightly pointed in C. haddadi and C. hudsoni). Chiasmocleis papachibe differs from C. magnova by having all fingers fully developed (fingers I and IV reduced in C. magnova), and by its larger size (maximum SVL 18.3 mm in C. magnova: Moravec and Köhler, 2007). Chiasmocleis papachibe is distinguished from C. shudikarensis in having basal webbing on toes of males (males with extensive webbing in C. shudikarensis) and in lacking an inguinal blotch (present in C. shudikarensis) and dermal spines on dorsum and digits (spines numerous and conspicuous in C. shudikarensis). Chiasmocleis papachibe differs from C. ventrimaculata in having a venter with a heavy amount of brown vermiculations (venter light colored with brown stains or blotches, sometimes forming a weakly defined vermiculated pattern in C. ventrimaculata). The new taxon differs from C. tridactyla by its much larger SVL (maximum recorded SVL 12.4 mm in C. tridactyla; Duellman and Mendelson III, 1995) in by its four externally evident fingers (FI not evident externally in C. tridactyla), five externally evident toes (TI not evident), rounded toe tips (toe tips pointed), by its reticulated venter (dark with scattered light spots).

Description of the Holotype:

Body ovoid, relatively slender. Head short, wider than long (HW 1.1 × HL); head slightly narrower than body; snout rounded in dorsal and lateral views; nostrils not protuberant, positioned anterolaterally, directed laterally. IOD 2.3 ×the IND. Canthus rostralis poorly defined, round in cross section; loreal region slightly convex. Eyes are small. No occipital fold present, supratympanic fold present; tympanum not apparent externally. Upper jaw projecting beyond lower; lower lip with truncate, trilobed anterior margin; spines on chin absent; tongue large, elongated, with free lateral and posterior edges; vocal slits not developed; choanae small, rounded, widely separated, just anterior to eye; vomerine teeth absent. Vocal sac is not developed.

Forelimbs slender; all fingers developed, nearly round in cross section; no webbing between fingers; relative finger lengths I<II<IV<III; tips rounded, with little developed fringes, without discs. FI swollen, no distinct subarticular tubercle visible. Subarticular tubercles on remaining fingers present but not very developed and not prominent; one subarticular tubercle on FII and FIV, two on FIII; both tubercles are equal in size; no supernumerary tubercles; palmar tubercle not visible, thenar tubercle present, large, round prominent; outer metacarpal tubercle present. All fingers without dermal spines laterally.

Legs short (combined THL, TBL, and FL 1.36 × SVL); legs relatively robust, lacking tubercles, lacking tibial and tarsal ridges or warts; toes not webbed (a very rudimentary vestige of web between TIII–TIV and TIV–TV is present and may be considered basal webbing by some observers); TI weakly developed (tip fails to reach subarticular tubercle of TII), without a visible subarticular tubercle; toe tips rounded with little developed lateral fringes and small discs on all but TI. Subarticular tubercles present in all toes; no outer metatarsal tubercle. TI barely reaches proximal margin of the tubercle of TII; TV reaches distal margin of proximal subarticular tubercle on TIV; relative toe lengths I<II<V<III<IV. Toes without lateral dermal spines.

Skin smooth dorsally and ventrally without scattered spines.

Measurements of the Holotype:

SVL 24.8, HL 4.7, HW 5.3, ED 1.8, IOD 4.1, IND 1.8, END 2.0, THL 9.0, TBL 9.4, FL 15.4.

Variation:

The species is only known from three specimens (two of which are illustrated in pl. 11) and all have very similar morphology, except for trivial differences in size and body shape between males and females. Additionally, males seem to have more numerous white dots on the dorsum than females.

Etymology:

The specific epithet, a noun in apposition, refers to the vernacular composite word “Papa-Chibé,” meaning “the one who eats chibé.” Chibé is made by soaking manioc (Manihot esculenta Crantz) flour—locally known as farinha de mandioca—in water. The resulting porridge, or soupy liquid, is then commonly seasoned with salt and hot peppers and eaten, sometimes with fish or game, but often as the only source of nutrients (Murrieta, 1998). Farinha de mandioca and its derivatives, including chibé, are an important source of nutrition to several Amazonian populations in the Lower Rio Amazonas region (Murrieta, 1998; Murrieta et al., 2008). The name “Papa-Chibé,” or “Papa-Xibé,” is colloquially used to refer to anyone who is native to the state of Pará, Brazil (Sobral, 2005). Chiasmocleis papachibe is, up until now, known only from a couple of localities in Pará, and the name is given as homage to the inhabitants of the Brazilian state of Pará (“paraenses”).

Call and Tadpoles:

Unknown.

Remarks:

Genetic distances between the three specimens of Chiasmocleis papachibe included in the phylogenetic is zero.

Distribution (fig. 34):

Known from the type locality and from a nearby locality, Barcarena, Pará, Brazil.

Fig. 46. 

Holotype of Chiasmocleis royi, sp. nov. (USNM 343266), in (A) dorsal and (B) ventral views. SVL  =  20.6 mm.

Holotype (fig. 46):

USNM 343266 (field number USNM-FS 153416), an adult male, collected by R.B. Cocroft on December 12 1990.

Type Locality:

Explorer's Inn, 30 km southwest of Tambopata Reserve (12°50′S / 69°17′W; 280 m.a.s.l.), Puerto Maldonado, Departamento Madre de Dios, Peru.

Paratypes:

USNM 269000, adult male, collected by R.B. Cocroft and K. Hambler on 12 July 1986; USNM 269001, adult male, collected by K. Hambler; USNM 346267–346268, two adult males, collected with the holotype; USNM 343033, adult female, collected by R.W. McDiarmid, in January 10, 1989; USNM 247432, juvenile, anonymous native inhabitant (information taken from label), in August 28, 1983; USNM 247433, juveniles, collected by R.B. Cocroft, on September 13, 1984. All paratypes collected at type locality.

Additional referred (nontype) material listed in appendix 1.

Diagnosis:

A medium-sized Chiasmocleis; SVL in males 18.8–23.1 mm (N  =  24); females 20.7–29.1 mm (N  =  16). Body ovoid and relatively slender; snout rounded in dorsal and lateral views. Four distinctive fingers present; all but FI fringed in males, less so in females; fingers not webbed; FI well developed with a subarticular tubercle usually present between the proximal phalanges; distinct subarticular tubercles present on all fingers; palmar tubercles protuberant, divided; relative finger lengths I<II<IV<III. Five distinctive, well-developed toes present; toes fringed, less so in female; toes basally webbed in both sexes; TI with a distinct well-developed subarticular tubercle; adpressed TI does not reach or barely reaches subarticular tubercle of TII; adpressed TV does not reach or reaches only to the middle of the middle subarticular tubercle of TIV; TII–IV with terminal discs, usually more developed in females, but also present in males; relative toe lengths I<II<V<III<IV. Males with dermal spines on fingers and toes; both sexes with dermal spines on dorsum and toes, more numerous and more developed in males. Males with few spines on anterior portion of chin (may be absent in some specimens). Vocal sac well developed, black, directed anteriorly when males are actively calling (pl. 12G). The femoral line is always present.

Comparisons:

Chiasmocleis royi differs from C. antenori by having four externally evident fingers (FI not evident externally in C. antenori), five externally evident toes (TI not evident), rounded toe tips (toe tips pointed), and a reticulated venter (dark with scattered light spots). It differs from C. albopunctata by the absence of white blotches on the snout and dorsum (present in C. albopunctata), its light venter with dark mottling (dark with white dots and blotches in C. albopunctata), and its advertisement call (see text for details). It differs from C. anatipes by having basal webbing on toes of males (extensive webbing in males C. anatipes) and by its mottled venter (dark spots in C. anatipes). The new species is distinguished from C. avilapiresae by its smaller size (mean SVL of C. avilapiresae 32.0 mm in females), its having basally webbed feet in males (extensively webbed in C. avilapiresae), a mottled venter (usually blotched in C. avilapiresae), and a slender body (robust in C. avilapiresae), and by its advertisement call (table 2). The new species differs from C. bassleri in having a mottled venter (large dark spots over a light venter in C. bassleri), by the absence of an inguinal blotch (present in C. bassleri) and the presence of the femoral line (absent in C. bassleri), and by its advertisement call (table 2). The new species differs from C. carvalhoi in having four externally evident fingers (FI not evident externally in C. carvalhoi) and five externally evident toes (TI not evident). Chiasmocleis royi differs from C. devriesi by its smaller size (SVL of the holotype C. devriesi 42.2 mm), its mottled venter (large dark spots over a light venter in C. devriesi), presence of a femoral line (absent in C. devriesi), and its slender body (robust in C. devriesi). It is distinguished from C. hudsoni by a round finger I tip (tip slightly pointed in C. hudsoni), the presence of a femoral line (absent in C. hudsoni), its having a vocal sac protruding forward when calling (vocal sac round, not protruding forward in C. hudsoni), and by its advertisement call (table 2). Chiasmocleis royi differs from C. magnova in having all fingers fully developed (fingers I and IV reduced in C. magnova) and by containing many small, pigmented, eggs in oviduct (few large, unpigmented, eggs in C. magnova; Moravec and Köhler, 2007). Chiasmocleis royi is distinguished from C. shudikarensis by its basal webbing in toes of males (extensive webbing in C. shudikarensis), the absence of an inguinal blotch (present in C. shudikarensis), a vocal sac protruding forward when calling (vocal sac round, not protruding forward in C. shudikarensis: see pl. 14A–B), and its advertisement call (table 2). Chiasmocleis royi differs from C. supercilialba by the absence of inguinal blotches (present in C. supercilialba), by the mottled venter (with large dark spots), and by its advertisement call (table 2). Finally, it differs from C. tridactyla in having four externally evident fingers (FI not evident externally in C. tridactyla), five externally evident toes (TI not evident), rounded toe tips (toe tips pointed), and a reticulated venter (dark with scattered light spots).

The new species has been historically confused with Chiasmocleis ventrimaculata (Cocroft and Hambler, 1989; De la Riva, 1995; Schlüter, 2005; Peloso and Sturaro, 2008), from which it differs by having a smaller, more slender body (usually more robust in C. ventrimaculata), ventral pattern usually reticulated, becoming more densely reticulated on ventral thigh (brown spots in C. ventrimaculata), longer finger IV, with tip reaching distal tubercle of finger III (finger shorter in C. ventrimaculata, finger tip does not pass the distal margin of the proximal tubercle on finger III), and a femoral line present (femoral line not visible in the photos of the holotype and lacking according to M.S. Hoogmoed personal notes: “back of thighs without light line”; also lacking in all other specimens examined).

Description of the Holotype (fig. 46):

Body ovoid, relatively slender. Head short, wider than long (HW 1.2 × HL); head slightly narrower than body; snout rounded in dorsal and lateral profiles; nostrils not protuberant, positioned anterolaterally, directed laterally. IOD 2.2 × IND. Canthus rostralis poorly defined, round in cross section; loreal region slightly convex. Eyes small. No occipital fold present, supratympanic fold visible; tympanum not apparent externally. Upper jaw projecting beyond lower; lower lip with truncate, trilobed anterior margin; few spicules on chin present; tongue large, elongated, with free lateral and posterior edges; vocal slits present, extended through most of the extent of the jaw, choanae small, rounded, widely separated, just anterior to eye; vomerine teeth absent. Vocal sac large and prominent, dark colored, directed forward.

Forelimb slender; all fingers developed, nearly round in cross section; no webbing between fingers; relative finger lengths I<IV<II<III; tips rounded, without discs. Subarticular tubercle usually not visible in FI; subarticular tubercles on remaining fingers present but not very developed and not prominent; one subarticular tubercle on FII and FIV, two on FIII; proximal tubercle more prominent than distal, which is hardly visible; no supranumerary tubercles; palmar tubercle not visible, thenar tubercle present, large, round not very prominent; outer metacarpal tubercle present. All fingers with many dermal spicules laterally.

Legs short (combined THL, TBL and FL 1.57 × SVL); relatively robust, lacking tubercles, tibial and tarsal ridges or warts; toes not webbed (a very rudimentary vestige of web between TIII–TIV and TIV–TV is present and may be considered as basal webbing by some observers); TI weakly developed (tip fails to reach subarticular tubercle of TII), without a visible subarticular tubercle; toe tips rounded with small discs on all but TI. Subarticular tubercles present in all toes except TI; no outer metatarsal tubercle. TI barely reaches proximal margin of the tubercle of TII; TV reaches distal margin of proximal subarticular tubercle of TIV; relative toe lengths I<II<V<III<IV. Toes with few lateral dermal spicules.

Skin mostly smooth dorsally and ventrally, with scattered spines, especially on dorsum, hind limbs, fingers, and toes.

Measurements of the Holotype:

SVL 20.6, HL 4.0, HW 5.3, ED 1.5, IOD 3.2, IND 1.4, END 1.4, THL 9.0, TBL 8.4, FL2 13.5, 4TD 0.7.

Variation:

Throat in males is dark brown or black and a vocal sac can be easily observed in adult male specimens. Ventral patterns usually consist of dense grayish or dark brown vermiculations against a light background (figs. 46, 47, pl. 12B), but in some specimens it consists of numerous small dark spots and stains (fig. 47B).

Fig. 47. 

Variation of ventral color pattern in Chiasmocleis royi, sp. nov. (A) Male, from Parque Ambiental Chico Mendes, Rio Branco, Acre, Brazil (UFAC-RB 3539). (B) Male, paratype, from Puerto Maldonado, Madre de Dios, Peru (USNM 343268, paratype). (C–D) Females, Guajará-Mirim, Rondônia, Brazil (CHUNB 23547, 25555, respectively).

Color in life seems to be variable. Dorsal color varies from brown to blackish or reddish (pl. 12). Dorsum is not uniformly colored and may show white, orange, or black mottling. A middorsal line is present in most specimens examined, but it may be absent, while the femoral line is always present. Dorsal portion of forelimbs either has the same overall pattern as dorsum, or is either orange or reddish.

Etymology:

The specific epithet is used as a noun in the genitive case and honors Roy McDiarmid, friend and researcher at the USGS Patuxent Wildlife Research Center and National Museum of Natural History, Washington, DC. Roy has been actively studying frogs in South America for a long time. He was the first one to call the attention of P.L.V.P. to the possibility of a species complex within Chiasmocleis ventrimaculata and continuously encouraged us to describe this new form, which he first collected, along with other colleagues, over 20 years ago.

Call and Tadpoles:

Schlüter (2005) described a call of Chiasmocleis ventrimaculata from Panguana, Madre de Dios, Peru. The specimens are actually representatives of C. royi. The range of the dominant frequency is 5000–7000 Hz. We had access to additional calls of the species, deposited at the USNM collection, made at Tambopata, Peru, by R. Cocroft (fig. 48). The call is composed of a repetitive series of multipulsed notes (mean 14.6 ± 0.7 pulses per note, 12–16, N  =  109) emitted at a rate of 503.2 notes/minute. Mean note duration is 89.5 ± 5.2 ms (77.0–104.0, N  =  109) and mean interval between notes 21.9 ± 4.4 ms (15.0–41.0, N  =  107). Pulse duration was 4.0 ± 0.8 ms (3.0–7.0, N  =  201). Mean dominant frequency, belonging to the second harmonic, was 7629.1 ± 1204.0 Hz (5512.5–13953.5, N  =  109). The fundamental frequency is 3729.9 ± 526.1 (3273.0–4823.4). Two harmonics with frequencies higher than the dominant frequency could be seen.

Fig. 48. 

Advertisement call of Chiasmocleis royi. (A) Oscillogram and (B) spectrogram of a series of nine consecutive notes emitted by a calling male from the type locality, at Tambopata, Madre de Dios, Peru; recorded on 12 December 1990, at 23.4° C air temperature (USNM 343268, paratype; recorded by R.B. Cocroft; recording number USNM tape 269, cut 13).

Tadpoles are unknown.

Remarks:

Chiasmocleis royi has been historically confounded with C. ventrimaculata (De la Riva, 1995; Morales and McDiarmid, 1996; Peloso and Sturaro, 2008; Von May et al., 2009), and most of the data on ecology published on the latter actually refers to the new taxon named here (Cocroft and Hambler, 1989; Schlüter and Salas, 1991). The two are not, to our knowledge found in sympatry (fig. 41) with C. ventrimaculata showing a more septentrional distribution than C. royi.

Genetic distances between the three specimens of C. royi included in the phylogenetic analysis are given in table 13.

Table 13 

Uncorrected pairwise distances between 16S sequences of Chiasmocleis royi

Distribution (fig. 41):

Known from eastern and southern Peru, northern Bolivia, and western Brazil.

Taxonomy:

The topology of this clade, is substantially different from that of de Sá et al. (2012). Their specimen of C. carvalhoi (as Syncope carvalhoi in their analysis; see fig. 2) was found at a basal position, and sister to all remaining species in the clade (which also included C. bassleri in their results), while we recovered that same specimen nested within our samples of C. antenori. The two specimens labeled as Syncope sp. by de Sá et al. (2012) were found here to be part of C. carvalhoi. We have included three specimens of C. tridactyla, from Leticia, Colombia—this species was not sampled by de Sá et al. (2012). We have also named a new species, which is included in this clade (C. haddadi).

Evolution:

A few members of this clade show an advanced degree of miniaturization, which includes lost of vertebrae and digits (Silva and Meinhardt, 1999). Reproductive biology characters also are of great interest as at least one species has a derived reproductive mode, with tadpoles developing inside bromeliads. Direct development has been postulated in other members of the clade as well, (Krügel and Richter, 1995; Moravec and Köhler, 2007), but to our knowledge never confirmed by empirical observations.

Unfortunately, limited data on osteology, as well as missing data on ecology and behavior precludes any detailed analysis of evolutionary trend in the clade. Fieldwork directed to collect data on reproduction and behavior of species in this clade can prove useful for understanding the evolution of reproductive biology of this very interesting lineage of frogs.