Generic names follow Faivovich et al. (2005) and AmphibiaWeb (2016).

DNA was extracted from muscle or liver tissue preserved in 95% ethanol or tissue storage buffer, using standard phenol–chloroform extraction protocols (Sambrook et al. 1989). We used a polymerase chain reaction (PCR) to amplify DNA fragments for mitochondrial genes 12S rRNA (12S), Cytochrome Oxidase sub-unit I (COI), two overlapping fragments for the last ~320 bp of 16S rRNA (16S), NADH dehydrogenase subunit 1 (ND1) and adjacent tRNAs (tRNA^Leu, tRNA^Ile and tRNA^Gln), and the nuclear gene POMC, using the primers listed in Goebel et al. (1999), Moen and Wiens (2009) and Wiens et al. (2005). PCR amplification was performed under standard protocols and sequenced by the Macrogen Sequencing Team (Macrogen Inc., Seoul, Korea). The combined DNA matrix had up to 3792 bp.

The newly generated DNA sequences are available on GenBank under accession numbers listed on Table 1. To optimize taxon sampling within Hylidae, we blasted our E. tuberculosa 12S and ND1 sequences to the GenBank database (blastn procedure). These searches shown unequivocally that the most similar samples belonged to the genera Osteocephalus and Tepuihyla, tribe Lophiohylini. For example, a blastn for the ND1 sequence of E. tuberculosaQCAZ 53542 retrieved 50 most similar sequences with identity values ranging from 82.7% to 89.1%. All 50 sequences belonged to the Lophiohylini tribe (Corythomantis, Osteocephalus, Osteopilus, Tepuihyla, and Trachycephalus). To confirm the close relationships between E. tuberculosa and Tepuihyla, we carried out an additional exploratory analysis based on Pyron (2014) matrix for the genes 12S and 16S. We included our sequences as well as those from all species of the tribes Hylini (including Ecnomiohyla) and Lophiohylini from Pyron (2014) matrix. We also included several species of Agalychnis, Litoria and Phyllomedusa as outgroups. The final matrix had 169 terminals. A maximum likelihood phylogenetic analysis using GARLI 2.0 (Zwickl 2006) confirmed with strong support that Ecnomiohyla tuberculosa is not closely related to other Ecnomiohyla. Instead, it is a member of the tribe Lophiohylini and is nested within Tepuihyla. The phylogeny based on Pyron (2014) matrix is available as Suppl. material 1.

For final phylogenetic analysis we included GenBank sequences from all available species of Lophiohylini as well as sequences from representative species of all other tribes within Hylinae (Cophomantini, Dendropsophini, and Hylini). GenBank sequences were originally published by Darst and Cannatella (2004), Faivovich et al. (2005), Salducci et al. (2005), Wiens et al. (2005), Moen and Wiens (2009), Moravec et al. (2009), Salerno et al. (2012), Jungfer et al. (2013), and Salerno et al. (2015). Samples of Phyllomedusinae were included as outgroups (Agalychnis spurrelli, Phyllomedusa tomopterna, and Phyllomedusa perinesos). We also added new sequences from 25 individuals of Ecuadorian and Peruvian Trachycephalus to assess their taxonomic status.

Preliminary sequence alignment was done with MAFFT 7.2 software with the L-INS-i algorithm (Katoh and Standley 2013). All sequences in the matrix were visually examined and translated in MESQUITE (version 3.01; Maddison and Maddison 2014). The matrix was partitioned to allow independent inferences of models of evolution by gene and by codon position in coding genes. We used software PARTITIONFINDER v. 1.1.1 (Lanfear et al. 2012) to simultaneously estimate both the best-fit model for each partition and the best partition strategy for our data.

Phylogenetic trees were obtained using maximum likelihood and Bayesian inference. Maximum likelihood searches were carried out with software GARLI 2.0 (Zwickl 2006). We made two independent searches with 10 replicates each. The first search started with random trees and the second with stepwise addition trees. We modified the settings for the number of generations without topology improvement required for termination (genthreshfortopoterm = 200000) and the maximum range for localized SPR topology changes (limsprrange = 10) to increase exhaustiveness of the tree space search. Other settings were set on default values. Node support was assessed with 200 pseudoreplicate non-parametric bootstraps (npb), configured with the same settings of the full search, but with three replicates per run.

Bayesian inference was performed with MARBAYES 3.2.1 (Ronquist et al. 2012). The analysis consisted of four parallel runs of the Metropolis-coupled Monte Carlo Markov chain for 5 × 10⁶ generations; each run had six chains with a temperature of 0.08. We used TRACER 1.6 (Rambaut et al. 2014) to assess convergence and stationarity of the runs and to obtain effective sample sizes (ESS) for all model parameters. The search was considered finished when all ESS were > 200. Each run was sampled every 1000 generations. The first 25% of the samples was discarded as “burn-in”, using the remaining samples to estimate the Bayesian tree, posterior probabilities (pp) and other model parameters.

Diagnostic characters and comparisons are based on specimens from the following institutions. Ecuador: Fundación Herpetológica Gustavo Orcés, Quito (FHGO); Museo de Zoología at Pontificia Universidad Católica del Ecuador, Quito (QCAZ). Peru: Centro de Ornitología y Biodiversidad, Lima (CORBIDI); Colección Referencial de Biodiversidad del Instituto de Investigaciones de la Amazonía Peruana, Iquitos (IIAP). United States of America: American Museum of Natural History, New York, (AMNH); National Museum of Natural History, Washington DC (USNM); Natural History Museum at the University of Kansas, Lawrence (KU). United Kingdom: Natural History Museum, London(BMNH). Venezuela: Museo de Historia Natural La Salle, Caracas (MHNLS). Our comparisons included the revision of the holotype of Hyla tuberculosa (Natural History Museum, London, BMNH 1947.2.13.34) and photographs of the holotype of Hyla macrotis (Swedish Museum of Natural History, Stockholm, NRM 1958). Examined specimens are listed as Suppl. material 2. Character definitions follow Duellman (1970) and Luna et al. (2012) for nuptial excrescences. Notation for hand and foot webbing is based on Myers and Duellman (1982). Sex was determined by presence of nuptial pads or vocal slits, and by gonadal inspection. Descriptions of coloration in life are based on digital photographs. Adults were measured with digital calipers for the following eight morphological variables, following Duellman (1970): (1) SVL; (2) head length; (3) head width; (4) tympanum diameter; (5) femur length; (6) tibia length; (7) foot length; and (8) eye diameter. Additionally, we measured the hand length (HAL) from proximal edge of palmar tubercle to tip of third finger for the new species described herein following Batista et al. (2014).

Morphometric analyses were performed based on measurements of adult males (number of specimens in parenthesis): E. tuberculosa (5), T. shushupe sp. n. (1), T. edelcae (17), T. obscura (17), and T. rodriguezi (14; see Suppl. material 2). We did a log₁₀ transformation of the dataset, a common practice in morphometric analyses due to the multiplicative nature of allometric growth and to make measurement variances comparable. A principal components analysis was later performed on the data. We used Principal Component 1 as a proxy for size and its effects, given that most variation is due to size and size effect differences. Other principal components were interpreted as a proxy for shape (Bookstein 1989). We also performed a Discriminant Analysis with species as group priors, in order to determine if they can be diagnosed from each other with morphometric characters.

Calls were analyzed from two adult males, QCAZ 53699 from Juyuintza, Pastaza province, Ecuador (2.110°S, 76.190°S, 200 m of elevation) and CORBIDI 12513 from Ere River, Putumayo Basin, Loreto Department, Peru (1.67903°S, 73.7197°W, 145 m). Male QCAZ 53699 was recorded at night on 22 June 2012 (23.9°C / relative humidity ~86.0%). Male CORBIDI 12513 was recorded at night of 18 October 2012 (21.3°C / relative humidity ~95.8%). Recordings were saved in PCM format, with a sample rate of 48000 Hz and 24-bits. Call variables were measured with RAVEN PRO 1.4 (Charif et al. 2010), under a Hanning function, 2048 DFT sample size and a grid spacing of 23.4 Hz. After a preliminary screening of the sound spectrogram, we clipped values below 90.0 db and applied a filter band-pass between 350 and 2420 Hz, to reduce background noise and facilitate the measurement of the acoustic parameters. Fourteen acoustic parameters (modified from Cocroft and Ryan 1995) were measured to describe the structure of each call: (1) Call duration = time from beginning to end of one call, measured from waveform analyzer screen; (2) Call rate = number of calls per minute; (3) Call interval = time from the end of the call to the beginning of the next call; (4) Call rise time = time from beginning of call to point of maximum amplitude; (5) Notes per call = number of notes per call; (6) Note rate = (number of notes-1)/time from beginning of first note to beginning of last note; (7) Note length = average time from beginning to end of a note; (8) Note rise time = time from beginning of note to point of maximum amplitude; (9) Note shape = note rise time/note length; (10) Frequency band = difference between the upper and lower frequencies measured along the entire call; (11) Fundamental frequency = frequency with the highest energy on the 1st harmonic in the call; (12) Dominant frequency = frequency with the highest energy along the entire call. In order to analyze the modulation of dominant frequency, we divided the call in three sections corresponding to the beginning, the middle and the end of the call; (13) Fundamental frequency ratio = number of notes with the dominant frequency in the first harmonic/total number of notes; and (14) Frequency modulation = (frequency with the most energy at end of call – frequency with the most energy at beginning of call)/call duration; values equal or near 0 represent calls lacking frequency modulation. Recordings are deposited in the Sound Archive of Museo de Zoología of Pontificia Universidad Católica del Ecuador (available at AmphibiaWebEcuador website, http://zoologia.puce.edu.ec/vertebrados/anfibios/).

The Maximum Likelihood and the Bayesian analyses yielded similar topologies with differences pertaining only to weakly supported nodes. The phylogeny shows strong support for the inclusion of E. tuberculosa within the tribe Lophiohylini, genus Tepuihyla Ayarzagüena et al. 1993 “1992” (Fig. 1B). Ecnomiohyla tuberculosa is sister to all species of Tepuihyla, except T. warreni from Ayanganna, Guyana. Tepuihyla warreni is polyphyletic because a sample from the Maringma Tepui, Guyana, clusters with T. aecii, T. edelcae, T. obscura, and T. rodriguezi instead of the other samples of T. warreni (Fig. 1A). The distance of the type locality of T. warreni from Ayanganna is approximately 90 km and from Maringma is 15 km. These distances suggest that the Ayanganna population represents an undescribed species.

Figure 1. 

Maximum likelihood phylogram showing the position of Tepuihyla tuberculosa comb. n. and Tepuihyla shushupe sp. n. within Hylidae. Phylogram derived from analysis of 3792 bp of mitochondrial (gene fragments 12S, 16S, ND1, CO1, tRNA ^Leu, tRNA ^Ile, tRNA ^Gln) and nuclear DNA (POM-C). Voucher no. (or, if unavailable, GenBank accession no.) is shown for each sample. Clade posterior probabilities (pp × 100) resulting from Bayesian Markov chain Monte Carlo searches appear above branches in blue. Non-parametric bootstrap (npb) support values, from 200 pseudoreplicates, are shown below. Asterisks represent values of 100. Outgroups are not shown. Abbreviations are: BR = Brazil, EC = Ecuador, FG = French Guiana, GU = Guyana, PE = Peru, VE = Venezuela

There is strong support for the tribe Lophiohylini. However, the relationships between basal clades within Lophiohylini are weakly supported. Within Trachycephalus, the recently described T. cunauaru is sister to T. hadroceps. Trachycephalus typhonius (previously referred as “T. venulosus”) is widely paraphyletic. The new sequences of T. typhonius from the Ecuadorian Chocó are sister to a clade composed of T. cunauaru, T. hadroceps, T. resinifictrix, T. typhonius from Guyana, French Guiana, Venezuela, and the upper Amazon basin in Ecuador and Peru. Populations of T. typhonius from Amazonian Peru and Ecuador are sister to a clade composed of T. resinifictrix and one sample of T. typhonius from Venezuela (Fig. 1A).

As in previous phylogenies (e.g., Pyron 2014) we found a clade composed by Dryaderces + Tepuihyla + Osteocephalus. Two samples of Dryaderces do not cluster together. This is likely a result of the use of short and non-overlapping sequences between both samples. Therefore, our results do not challenge the monophyly of Dryaderces.

The first Principal Component, a proxy for size covariation, explains 97% of the variance and has very similar proportions for all variables (Table 2). The second Principal Component is mostly made up of Eye Diameter, but the variance seems to be higher within rather than among groups (Fig. 2). The third principal component is mainly composed of Tympanum Diameter.

Figure 2. 

Principal Components 2 and 3 from analysis of eight morphological variables. See Table 2 for character loadings on each component.

Morphometric analyses do not show differences in shape among species (Fig. 2). The only morphometric difference pertains to size (Principal Component 1; not shown) as E. tuberculosa is much larger than the other species. The Discriminant Function Analysis graph (not shown) is similar to the PCA in not showing interspecific differences. Group assignment proportions are all very low (between 0.02–0.31), indicating very little power of morphometric discrimination among groups, even when attempting to maximize differences with species/lineage priors.

The four samples of E. tuberculosa form three well-supported clades (pp/npb = 100). The sample from Ere River, Peru, is sister to the Yasuní + Juyuintza in Ecuador. The Juyuintza samples are sister to each other. All genetic distances reported below are uncorrected distances for the gene 12S. Genetic distances between Ere River, Peru, and the Ecuadorian populations (Juyuintza and Yasuní) range from 5.3 to 5.9%; distances between Yasuní and Juyuintza are both 1.2%. Samples from Juyuintza are identical to each other. The genetic distances between Ere River and the other populations are above distances between several closely related species in Tepuihyla and Osteocephalus (e.g., O. buckleyi-O. cabrerai 2.8%, O. oophagus-O. taurinus 1.3–1.8%, O. cannatellai-O. cabrerai 2.6%, O. cannatellai-O. vilmae 2.4%, T. rodriguezi-T. exophthalma 4.5–4.6%, T. rodriguezi-T. edelcae 1.3–1.5%).

Morphologically, the individual from Ere River differs from all other populations (in parenthesis) in having a cream iris with red periphery (entirely cream to reddish cream; Fig. 3G vs. 3H) and dorsum with small tubercles intermixed with abundant large tubercles (small tubercles intermixed with few large tubercles; Fig. 4A vs. 4C and 5). We found differences in advertisement calls between Río Ere and Juyuintza. Calls consist of a cackle of short notes repeated at a fast rate (Table 3; Fig. 6). Note amplitude and note rate increases markedly along the first half of the call. The call from Juyuintza differs from the call from Ere River in temporal and spectral variables. The former has shorter duration, fewer notes, and lower note rate (Ere River 3.37–3.38, 3.37 ± 0.01; Juyuintza 2.63–2.86, 2.74 ± 0.16; Table 3). Harmonic structure is markedly different between calls. The call from Ere River has two harmonics with similar energy content while the call from Juyuintza has one well-defined harmonic with high energy and a second harmonic with low energy, barely evident in the power spectra (Fig. 6).

Figure 3. 

External morphology of Tepuihyla shushupe sp. n. and Tepuihyla tuberculosa comb. n. A–B Tepuihyla shushupe sp. n. CORBIDI 12513 (holotype), adult male, SVL = 85.3 mm, Ere river, Peru C–D Tepuihyla tuberculosa comb. n. QCAZ 55423, adult male, SVL = 83.1 mm, Parque Nacional Yasuní, Tambococha, Ecuador; E–F Tepuihyla tuberculosa comb. n., QCAZ 55413 juvenile, SVL = 58.1 mm, Parque Nacional Yasuní, Tambococha, Ecuador G–H iris coloration of Tepuihyla tuberculosa comb. n. (QCAZ 55423) and Tepuihyla shushupe sp. n., (CORBIDI 12513). Photographs: A–B, H by P. Venegas, C–G by D. Quirola.

Figure 4. 

Holotypes of Tepuihyla shushupe sp. n. (A–BCORBIDI 12513) and Tepuihyla tuberculosa comb. n. (C–EBMNH 1947.2.13.34). Photos of Tepuihyla tuberculosa comb. n. by M. R. Bustamante.

Figure 5. 

Variation in dorsal coloration of preserved specimens of adult Tepuihyla tuberculosa comb. n. Left to right, upper row: QCAZ 52855 (male), QCAZ 53542 (male); lower row: QCAZ 53147 (male), QCAZ 55423 (male). See Suppl. material 2 for locality data. Specimens are shown at the same scale.

Figure 6. 

Oscillogram, sound spectrogram (first column), and power spectrum (second column) of advertisement calls of Tepuihyla shushupe sp. n. (A) and Tepuihyla tuberculosa comb. n. (B). Note differences in call length and number of notes (first column). Both species have the dominant frequency (arrows) in the first harmonic but in Tepuihyla shushupe sp. n. the second harmonic has similar energy content (asterisk), while in Tepuihyla tuberculosa comb. n. it has much less.

The combined evidence from genetics, morphology, and bioacoustics indicates the existence of two species within “E. tuberculosa”. Specimens from Yasuní and Juyuintza have similar coloration and dorsal tuberculation to the holotype of E. tuberculosa (BMNH 1947.2.13.34), therefore we assign them to E. tuberculosasensu stricto. The specimen from Rio Ere belongs to an undescribed species that we describe in the following sections. Our species description is based on a single specimen and therefore we cannot characterize the variation in the new species. However, we examined 11 specimens of E. tuberculosa, which allow us to characterize E. tuberculosa morphological variation and show that the single specimen of the new species falls outside the range of variation of E. tuberculosa. Hence, we demonstrate that the new species is, in fact, distinct from E. tuberculosa. Our morphological evidence is corroborated by genetic and bioacoustic characters.

Recent phylogenies of Hylidae (Wiens et al. 2010) and all amphibians (Pyron 2014) show, with strong support, that Ecnomiohyla is part of Hylini, a tribe composed by the genera Acris, Anotheca, Bromeliohyla, Charadrahyla, Diaglena, Duellmanohyla, Ecnomiohyla, Exerodonta, Hyla, Isthmohyla, Megastomatohyla, Plectrohyla, Pseudacris, Ptychohyla, Smilisca, Tlalocohyla, Triprion. Wiens et al. (2010) and Pyron (2014) phylogenies only included 3 out of 14 species of Ecnomiohyla: E. milaria, E. minera, and E. miotympanum. Batista et al. (2014) analyzed one fragment of the mitochondrial gene 16S from seven species of Ecnomiohyla. Their outgroup was limited to two genera within Hylini and was not suitable to assess the phylogenetic position of Ecnomiohyla within Hylidae. Nevertheless, they provided the most complete phylogeny for Ecnomiohyla to date.

In our phylogeny, E. tuberculosa is part of the tribe Lophiohylini. Because all other species of Ecnomiohyla are part of the tribe Hylini (Pyron 2014; Wiens et al. 2010; Suppl. material 2), the phylogenetic position of Ecnomiohyla tuberculosa renders the genus Ecnomiohyla polyphyletic and the genus Tepuihyla paraphyletic (Fig. 1). We solve both problems by assigning Hyla tuberculosa Boulenger 1882 to the genus Tepuihyla. This change results in Tepuihyla tuberculosa (Boulenger 1882) comb. n.

Below we present the species account for T. tuberculosa and we describe the new species from Ere River, Peru. Under this new taxonomy, the genus Tepuihyla contains ten species distributed in the Pantepui region of Venezuela, Guyana and in the Amazon region of Brazil, Colombia, Ecuador, and Peru (AmphibiaWeb 2016; Salerno et al. 2015; Kok et al. 2015).

Tepuihyla tuberculosa (Boulenger, 1882)

Hyla tuberculosa Boulenger 1882. Holotype BMNH 1947.2.13.34

Ecnomiohyla tuberculosa : Faivovich et al. 2005

Holotype

Sub adult female with SVL 67.6 mm (Fig. 4C–E). The description of the holotype provided by Boulenger (1882) is adequate.

Diagnosis

In this section coloration pertains to preserved specimens unless otherwise noted. A large-sized Tepuihyla differing from other known species in the genus by the following combination of characters: (1) SVL in males 79.3–86.2 mm (n = 5), SVL in females 67.6–85.7 mm (n = 2); (2) skin on dorsum coarsely tuberculate, covered by small tubercles with scattered large tubercles; tubercles without keratinized tips; (3) skin on flanks covered by large tubercles; (4) webbing between fingers extensive but without reaching the proximal border of the disks, hand webbing formula I2 —2 II1 —2 III1½—1IV to I2—2+II1—2III2 —1IV (Fig. 7A); webbing between toes extensive and reaching the proximal border of the disks on at least three toes, foot webbing formula I0+—1+II0+—1 III0+—1+IV1+— 0+V to I1+—1½II1—1+III1+—1½IV1½— 0+V (Fig. 7B); (5) dorsal coloration in life greenish cream or brownish cream with scattered dark brown reticulations, anterior and posterior surfaces of thighs and hidden surfaces of shanks yellowish orange; (6) ventral coloration whitish cream and webbing between fingers and toes pale orange; (7) suborbital mark absent, clear labial stripe present, faint and with some brown spots; (8) coloration on flanks similar to dorsal coloration except for greenish yellow axilla and groins and whitish gray ventrolateral area; (9) skin on upper surface of head not co-ossified with underlying cranial elements, cranial crest slightly exostosed; (10) in life, bones green; (11) small triangular dermal flaps form serrate fringe along the ventrolateral margin of the forearm and along the outer edge of Finger IV; heels bearing two or three enlarged fleshy conical tubercles surrounded by few smaller round tubercles; small triangular dermal flaps form serrate fringe along ventrolateral margin of tarsus and outer margin of Toe V; (12) adults, in life, have cream iris with a coppery hue and scattered thin black reticulations; (13) vocal sac is white, single and subgular, (14) a juvenile is similar to adults in coloration but with dorsal background coloration brownish cream and a black medial stripe in the iris; (15) larvae unknown.

Figure 7. 

Ventral views of the left hand and foot of Tepuihyla. A–B Tepuihyla tuberculosa comb. n. (Parque Nacional Yasuní, Ecuador, SVL = 86.2 mm, QCAZ 52855, adult male) C–D Tepuihyla shushupe sp. n. (Ere River, Peru, SVL = 85.4 mm, CORBIDI 12513, adult male).

Tepuihyla tuberculosa differs from all congeneric species (in parenthesis), except T. shushupe sp. n., in having a larger size: SVL in males 79.3–86.2 mm (n = 5), SVL in females 67.6–85.7 mm (n = 2) (maximum SVL 59.2 mm in other Tepuihyla), having extensive webbing in the hands (basal webbing) and a serrate fringe along the ventrolateral margin of the forearm (absent). Other differences are listed on Table 4.

Table 4.

Download as

CSV

XLSX

Qualitative morphological characters of frogs of the genus Tepuihyla. Data was obtained from Ayarzagüena et al. 1992; Duellman and Hoogmoed 1992; Gorzula and Señaris 1998; Kok and Kalamandeen 2008; Kok et al. 2015; Mijares-Urrutia et al. 1999).

	Serrate fringes on limbs	Maximum size females (mm)	Dorsum	Tubercles on jaw	Webbing between fingers	Vocal sac
T. aecii	absent	36.8	smooth (females) to spiculate (males)	absent	absent	subgular
T. edelcae	absent	45.7	smooth (females) to spiculate (males)	absent	absent	subgular
T. exophthalma	absent	42.5	smooth with few tubercles	absent	absent	subgular
T. luteolabris	absent	59.2	granular	absent	absent	subgular
T. rodriguezi	absent	50.3	smooth (females) to spiculate (males)	absent	absent	subgular
T. shushupe sp. n	forelimbs and hindlimbs	--	tuberculate	present	extensive	subgular
T. tuberculosa comb. n.	forelimbs and hindlimbs	85.7	tuberculate	present	extensive	subgular
T. obscura	absent	38.4	smooth (females) to spiculate (males)	absent	absent	subgular
T. warreni	absent	36.2	smooth	absent	basal	subgular

Tepuihyla tuberculosa is most similar to T. shushupe sp. n. It differs from T. shushupe sp. n. (character states in parenthesis) in having cream iris in life (iris cream with reddish periphery), dorsum covered with small tubercles intermixed with few large tubercles (dorsum covered with small tubercles intermixed with abundant large tubercles; Figs 3–5); in preservative, its dorsum is light cream or cream with a coppery hue and brownish cream to creamy coppery large tubercles (dorsum light brown with large brownish cream or creamy coppery tubercles with a dark brown posterior border). Tepuihyla tuberculosa also resembles the Amazonian Trachycephalus cunauaru and T. resinifictrix but differs from both species in having serrate fringes on the limbs and tubercles on the lower jaw (both absent in Trachycephalus; Fig. 8). Tepuihyla tuberculosa is easily distinguished from all other large Amazonian treefrogs in having fully webbed hands and feet, coarsely tubercular skin on dorsum, and serrate dermal fringes on the outer margin of the forearm and foot. Cruziohyla craspedopus and Dendropsophus marmoratus also have dermal fringes on the outer margin of the foot, but the former is distinguished by its deep green dorsum, vertically elliptical pupils in a bicolored iris (pale silver with yellow borders), smooth skin on dorsum and elongate calcar on the heel. The smaller Dendropsophus marmoratus exhibits a bright yellow belly with black spots or mottling (uniform yellowish-tan in T. tuberculosa).

Figure 8. 

Dorsolateral views of Amazonian Trachycephalus. A Trachycephalus helioi, Juruti, Pará, Brazil B Trachycephalus resinifictrix, Juruti, Pará, Brazil C Trachycephalus cunauaru, Santa Izabel do Rio Negro, Amazonas, Brazil D–E Trachycephalus typhonius, Village de Ouanary, French Guiana F Trachycephalus typhonius, Matiti, French Guiana. Photographs: A–C by M. Gordo, D–E by Daniel Baudain, F by Vincent Premel.

Variation

In this section, coloration refers to preserved specimens unless otherwise noted. Morphometric data for adult specimens are summarized in Table 5, whereas variation in dorsal coloration of preserved specimens is shown in Figure 5. Dorsal coloration varies between light cream (e.g., QCAZ 55423) and cream suffused with a coppery hue (e.g., QCAZ 52855) with brownish cream or creamy coppery enlarged tubercles, bearing scattered dark brown reticulations or marks. In all specimens the belly and ventral areas of thighs are lighter than those of the holotype probably because they were collected more recently. Head is rounded in dorsal view, wider than long; the snout is truncate in dorsal view and truncate (e.g. QCAZ 32716) in profile. The largest male has 86.2 mm SVL (average = 82.9 ± 2.7 mm, n = 5) and the largest female 85.7 mm (Table 5).

Table 5.

Download as

CSV

XLSX

Descriptive statistics for measurements of adult Amazonian Tepuihyla. Mean ± SD is given with range in parenthesis. Abbreviations are: SVL = snout-vent length; FOOT = foot length; HL = head length; HW = head width; ED = eye diameter; TD = tympanum diameter; TL = tibia length; FL = femur length. All measurements are in mm.

	Tepuihyla shushupe sp. n. Male CORBIDI 12513	Tepuihyla tuberculosa comb. n. Female QCAZ 32716	Tepuihyla tuberculosa comb. n. males (n = 5)
SVL	85.4	85.7	82.9 ± 2.9 (79.3–86.2)
FOOT	35.5	35.8	35.53 ± 2.4 (33.5–39.5)
HL	26.6	25.7	26.0 ± 1.0 (25.1–27.5)
HW	28.9	29.2	28.9 ± 1.6 (27.8–31.7)
ED	7.7	7.5	7.5 ± 0.9 (6.3–8.9)
TD	5.5	4.8	5.3 ± 0.5 (4.9–5.8)
TL	46.9	45.6	46.3 ± 2.2 (44.1–49.9)
FL	35.9	36.6	36.8 ± 1.3 (35.0–38.3)

Coloration in life

Dorsal coloration is greenish cream with scattered brown marks or reticulations, top and sides of head are more greenish in some specimens (e.g. QCAZ 52855); lips greenish with scattered small dark brown marks; supratympanic fold delineated by a thin brown edge, tympanic membrane lighter than the background; flanks with coloration similar to dorsum; axilla and groin pale orange, ventrolateral region whitish cream, limbs with faint brown transversal bands; dorsal surfaces of hands and feet, including webbing, greenish cream with scattered dark brown marks; discs pale green. Throat, chest and belly are white but some specimens have an orange hue on the belly (e.g. QCAZ 55423); serrate fringes along forearms and tarsus white; webbing, thighs and concealed surfaces of shanks orange. Iris is cream with a light coppery hue bearing some scattered black venations. Juvenile specimen QCAZ 55413 is similar to adult specimens but has a brownish cream dorsum with scattered faint brown blotches (Fig. 3E–F).

Advertisement call

Quantitative characteristics of the advertisement call of T. tuberculosa (QCAZ 53699) are detailed in Table 3. The call consists of a cackle of short notes repeated at a fast rate with amplitude modulation (Fig. 6). Note amplitude and note rate increase markedly along the first half of the call, decreasing at the end. The call has two harmonics but most of the energy located on the first. Fundamental frequency of the notes ranges from 515.6 to 796.9 Hz (mean = 619.1, SD = 77.2). Two notes located at ~50% of the call duration have their greatest energy in the second harmonic, at 1078.1 and 1195.3 Hz. All other notes have their greatest energy in the first harmonic. The dominant frequency of the entire call ranges from 562.5 to 632.8 Hz. Interestingly, the call of this species is feared by natives in the lower Pastaza basin (Shiwiar, Sapara, Shuar, Achuar people), because it is commonly confused with the “calling” of the bushmaster Lachesis muta (Squamata: Viperidae). This belief almost certainly incorrect as L. muta cannot vocalize. See also the Etymology section in E. shushupe sp. n. description.

Distribution and natural history

Localities documented for this species are shown in Figure 9. Duellman (1974) reported four specimens: two from Pastaza province in Ecuador, one from the mouth of Rio Santiago in northern Peru, and one juvenile from Rio Uaupes at junction of Rio Querari, Amazonas, Brazil. Given its geographic location, the latter probably corresponds to T. shushupe sp. n., but its identity requires confirmation. We recorded five additional localities in the Amazon lowlands of Ecuador and two in Peru (Fig. 9). Elevation range is 132 to 1076 m above sea level. The southernmost and highest locality is Cordillera Escalera, San Martin department, in northeastern Peru (CRBIIAP 1252).

Figure 9. 

Records of Tepuihyla shushupe sp. n. and Tepuihyla tuberculosa comb. n. Tepuihyla shushupe sp. n., red star; Tepuihyla tuberculosa comb. n., blue triangles and blue star for type locality. Locality data from specimens deposited at Natural History Museum (BMNH), London, Centro de Ornitología y Biodiversidad (CORBIDI), and Museo de Zoología, Pontificia Universidad Católica del Ecuador (QCAZ).

All specimens with ecological data were collected at night perching on vegetation 1.5 to 3.0 m above the ground. One specimen from Yasuní (QCAZ 52855) and one from Juyuintza (QCAZ 53542; Fig. 10B) were collected on primary forest (M. Read and H. M. Ortega-Andrade field notes). The specimen recorded at Juyuintza was calling from a tree hole 1.5 m above the ground in Terra Firme forest. The hole had a depth of ~15 cm and a diameter of 10 cm, and was ¾ flooded with water. Two individuals were calling nearby, at distances of approximately 100 to 200 m. No amplectant pairs, clutches, or tadpoles have been observed.

Figure 10. 

Life individuals and habitat of Tepuihyla shushupe sp. n. and Tepuihyla tuberculosa comb. n. A habitat of Tepuihyla tuberculosa comb. n. at Juyuintza, Provincia Pastaza, Ecuador B Adult male of Tepuihyla tuberculosa comb. n. (QCAZ 53542) sitting at the entrance of the tree hole where he was calling C Adult male of Tepuihyla shushupe sp. n. (CORBIDI 12513, holotype) partly submerged in the water accumulated on the tree hole where he was calling D Tepuihyla tuberculosa comb. n. (CRBIIAP 1839) from the vicinity of Puerto Almendras, Loreto, Peru. Photographs: A, B by M. Ortega, C by A. Del Campo, D by G. Gagliardi-Urrutia.

Conservation status

The scarcity of records of E. tuberculosa, even at thoroughly sampled localities, like Yasuni National Park, could be a consequence of small populations sizes or an artifact of low capture probabilities. Our finding of a male calling from a tree hole containing water suggests that E. tuberculosa breeds on tree holes and is unlikely to be found on ground-level surveys. Therefore, lack of records probably results from low capture probabilities consequence of microhabitat preferences. Because population status is unknown, we suggest that the Red List category of T. tuberculosa is Data Deficient according to IUCN (2001) guidelines. Searches of the species will benefit from audio sampling to detect the advertisement call described here.

Several authors have discussed the considerable morphological variation in T. typhonius (Duellman 1971; McDiarmid 1968; Savage 2002) and have suggested that it may represent a species complex (Cochran and Goin 1970; Duellman 2005). The paraphyly of Trachycephalus typhonius (Fig. 1A) also suggests the existence of undescribed species. The resolution of the taxonomy of T. typhonius is challenging given the large number of available synonyms (Duellman 1971; Lavilla et al. 2010). However, the identification of the type locality of T. typhonius as Paramaribo, Suriname, by Lavilla et al. (2010) allows inferring the phylogenetic position of T. typhoniussensu stricto. The type locality is geographically close to samples from Guyana (AMNH-A 141142) and French Guiana (163MC), Clade C in Figure 1A. Therefore, we consider Clade C to represent T. typhoniussensu stricto. Two linages from Ecuador and Peru are distinct from Clade C and each has an available name. Below we discuss their status.

Populations from western Ecuador are sister to a clade composed of T. cunauaru, T. hadroceps, T. resinifictrix and T. typhonius. Their phylogenetic position demonstrates that they are not conspecific with T. typhoniussensu stricto or other species of Trachycephalus. One binomial available for Trachycephalus from western Ecuador is Hyla quadrangulum Boulenger 1882 (type locality “western Ecuador”). Duellman (1971) synonymized Hyla quadrangulum under T. coriaceus stating that the holotype is “indistinguishable from P. coriacea”. He also questioned its type locality because “several species contained in the Fraser collections supposedly from western Ecuador have been found subsequently only in the Amazon Basin”. Our examination of a large series of specimens from western Ecuador and T. coriaceus from the Amazon basin show unequivocally that the holotype of H. quadrangulum (Fig. 11G–H) is not conspecific with T. coriaceus. Although the holotype has a quadrangular dorsal blotch, also present in T. coriaceus, we found the same pattern in specimens from western Ecuador (e.g., QCAZ 23427, 31302, 39361, 51746). Moreover, the holotype lacks the dark blue blotch above the arm characteristic of T. coriaceus. We conclude that the type locality is correct and Hyla quadrangulum is a valid species. The resurrection of Hyla quadrangulum Boulenger 1882 results in Trachycephalus quadrangulum (Boulenger 1882) comb. n.

Figure 11. 

External morphology of Trachycephalus quadrangulum comb. n. from the Chocó region, western Ecuador. AQCAZ 39360, adult male, SVL = 74.1 mm, Caimito, Provincia de Esmeraldas, Ecuador BQCAZ 39361, adult male, SVL = 73.2 mm, same locality as A C–DQCAZ 51746, adult female, SVL = 80.3, Bosque Pacocha, Provincia Manabí, Ecuador EQCAZ 35406, adult male (SVL = 69.5), Bosque Protector Puyango, Provincia El Oro, Ecuador FQCAZ 28530, adult male (SVL = 64.5), Hacienda Santa Teresita, Provincia del Guayas, Ecuador; (G–H) Holotype of Hyla quadrangulum, adult female; scale, in cm, shown for reference. Photographs: A–F by S. R. Ron, G–H by Jeff Streicher.

Names under T. typhonius that could be senior synonyms to T. quadrangulum include Hyla spilomma Cope 1877 (type locality Veracruz, México), Hyla paenulata Brocchi 1879 (“versant occidental du Guatemala”), and Hyla nigropunctata Boulenger 1882 (several localities in Mexico). We exclude from this list synonyms with type localities east of the Andes based on our phylogeny, which shows that T. quadrangulum is restricted to the Chocó region. The phylogeny is consistent with a general biogeographic pattern showing that humid lowlands east and west from the Andes do not share amphibian species as result of the vicariant effect of the Andes (dos Santos et al. 2015). The phylogenetic and biogeographic evidence indicates that it is highly unlikely that available binomials with type localities in the Amazon basin or Guianan region are conspecific with T. quadrangulum. The resolution of the taxonomic identity of populations from Central America and the Chocó region is beyond the scope of this article. Nevertheless, the resurrection of T. quadrangulum provides a name to an evolutionary lineage separate from T. typhonius. Even if provisional, this new arrangement better reflects the evolutionary uniqueness of Chocoan populations and contributes to a more accurate characterization of the species diversity of the Chocoan amphibiofauna. It also improves our understanding of the conservation status of the T. typhonius species complex.

According to the phylogeny (Fig. 1A), populations from Amazonian Ecuador and Peru (Clade B) are sister to T. resinifictrix and this clade is sister to T. typhoniussensu stricto. Clade B is morphologically distinct from T. resinifictrix and comparisons of live individuals of clade B with photographs of the holotype of T. typhonius published by Lavilla et al. (2010) reveal that they belong to separate species as well. The holotype of T. typhonius (UUZM 134) shows the clear iris with dark reticulations, also characteristic of populations from French Guyana (Fig. 8D–F). Clade B, in contrast, is characterized by having a dark brown iris without reticulations (Fig. 12).

Figure 12. 

External morphology of Trachycephalus macrotis comb. n. A–BQCAZ 39565, adult female, SVL = 101.1 mm, Reserva Zanjarajuno, Ecuador C calling male, Reserva Zanjarajuno, Ecuador D amplectant pair, QCAZ 39565, adult female (SVL = 101.12) and QCAZ 39566, adult male (SVL = 81.38), Reserva Zanjarajuno, Ecuador ECORBIDI 9544, adult male, (SVL = 82.9), Peru FQCAZ 43017, adult male, SVL = 91.5 mm, Parque Nacional Yasuní, km 22 Pompeya-Iro road, Ecuador. Photographs by S. R. Ron except for (E) by A. Catenazzi.

Given its phylogenetic position and distinct external morphology, we conclude that Clade B is not conspecific with T. typhonius. We propose a solution for the status of those populations by resurrecting the binomial Hyla macrotis Andersson 1945 from its synonymy under T. typhonius proposed by Duellman (1971). The resurrection is based in the similarity between the holotype of H. macrotis (Fig. 13) and specimens from Clade B (Fig. 12). Within the region where the holotype was collected, “Río Pastaza watershed, Ecuador”, only clade B is morphologically similar. This allows us to confidently conclude that the holotype of H. macrotis and clade B are conspecific. The resurrection of Hyla macrotis results in Trachycephalus macrotis (Andersson 1945) comb. n.

Figure 13. 

Holotype of Hyla macrotis Anderson 1945, NHRM 1958. A lateral view of the head B ventral view C dorsal view. The holotype is an adult female, SVL = 120 mm (Anderson 1945). Photographs by Bo Delling.

It seems unlikely that other synonyms under T. typhonius are senior synonyms to T. macrotis because most of them are from the Guianan region and Central America and are geographically distant from the range of T. macrotis. The resurrection of T. macrotis provides a name to an evolutionary lineage separate from T. typhonius. As in T. quadrangulum, the recognition of T. macrotis better reflects the biology of its populations.

We could not examine the voucher specimens for the GenBank sequences of Trachycephalus used in our phylogeny. Some of those identifications could be incorrect and should be revaluated. For example, specimen KU 217753 is identified as T. venulosus (= T. typhonius) from the Chocó region of Ecuador. However, in the phylogeny it is nested within T. cunauaru from Amazonian Ecuador. Both the identification and locality are incongruent with its phylogenetic position suggesting that the tissue actually belongs to T. cunauaru from the Amazon region. The same error applies to GenBank sequence DQ347027, a sample that was obtained from the pet market. GenBank sequence JQ868532 (field number PS013) is identified as T. venulosus (= T. typhonius). The specimen is likely lost and its correct identification appears to be T. resinifictrix. Despite these potential identification errors, our taxonomic conclusions are only dependent on the correct identifications of the samples of T. typhonius from Guyana and French Guyana, which are reliable. We obtained photographs of populations from French Guyana (Fig. 8) from the same collector who provided the tissues for individual 163MC. Therefore, the identification of the tissue should be correct. Sequences for the specimen from Guyana were published by Faivovich et al. (2005). They have a voucher specimen (AMNHA 121142), which should be correctly identified.

The external morphology of T. quadrangulum is distinct from that of T. macrotis. Populations from western Ecuador (Fig. 1; Fig. 11) are smaller (adult males average SVL = 69.0 mm, range 53.4–76.9, n = 9; adult females SVL = 75.5 mm, 60.5–80.8 mm; n = 10; Ron and Read 2011), have brown to cream disks on the fingers, and a bronze iris with irregular black reticulations (Fig. 11); T. macrotis (clade B, Fig. 1) are larger (adult males average SVL = 84.58 mm, range 69.8–91.5, n = 12; adult females SVL = 103.2 mm, 93.9–118.7, n = 4), have green discs on the fingers, and a dark brown iris (Fig. 12). The morphologic and genetic differences between both clades corroborate that they represent separate species.