A new nurse frog (Anura: Allobates) from Brazilian Amazonia with a remarkably fast multi-noted advertisement call

Nurse frogs (Aromobatidae: Allobates) are probably the most extensively studied genus by taxonomists in Brazilian Amazonia. The southwestern portion of Amazonia is the most species-rich: as many as seven species may occur in sympatry at a single locality. In this study, we describe a new species of nurse frog from this region. The description integrates data from larval and adult morphology, advertisement calls and DNA sequences. Allobates velocicantus sp. nov. is distinguished from other Allobates mainly by the absence of hourglass-shaped dark marks on the dorsum and dark transverse bars on the thigh; a throat that is white centrally and yellow marginally; basal webbing on toes II and III; finger I longer than finger II; and an advertisement call composed of 66–138 pulsed notes with a note duration of 5–13 ms, inter-note intervals of 10–18 ms and a dominant frequency of 5,512–6,158 Hz. Tadpoles of the new species have 3–4 short, rounded papillae on the anterior labium, 16–23 papillae on the posterior labium, and a labial keratodont row formula 2(2)/3(1). This is the fifth species of Allobates described from the state of Acre, southwestern Brazilian Amazonia.


INTRODUCTION
Nurse frogs of the genus Allobates Zimmermann & Zimmermann, 1988 inhabit the leaf-litter of Neotropical forests and are small and diurnal. They are distributed from northern Central America to the Atlantic Forest of Brazil (Grant et al., 2006(Grant et al., , 2017Frost, 2020). Currently, the genus comprises 55 nominal species (Frost, 2020), 22 of which are distributed in Brazilian Amazonia. Due to the general lack of substantial morphological differences among species, the integration of multiple independent lines of evidence (e.g., larval and adult morphology, vocalization, breeding behavior, and molecular data) have proven indispensable for the reliable discovery, diagnosis and formal description of

MATERIALS AND METHODS
Sampling. Adult specimens were collected 13-14 January 2019 along state highway AC405 (7 38′54″ S, 72 48′59″ W, 211 m a.s.l.), municipality of Mâncio Lima, state of Acre, Brazil. Also collected from this locality were a group of tadpoles obtained from the back of INPAH 41347 (field number APL 21403), an adult male, and a clutch of fertilized eggs obtained soon after oviposition. In addition, an adult male was collected 15 February 2019 along the road (7 34′02″ S, 72 39′32″ W, 181 m a.s.l) connecting the municipality of Cruzeiro do Sul (state of Acre) to Guajará (state of Amazonas). Adults were euthanized with a 2% benzocaine topical solution, fixed in 10% neutral-buffered formalin and preserved in 70% ethanol. Tissue samples were obtained before immersion in formalin. Tadpoles and eggs were reared in the laboratory until they reached Gosner stages 27-37 for morphological description, when they were euthanized with a 10% aqueous benzocaine solution and preserved in 5% neutral-buffered formalin. Adults  The advertisement calls of eight males were recorded. Seven males were recorded in the state of Acre: six along highway AC405 (7 38′54″ S, 72 48′59″ W, 211 m a.s.l.) and one along the road connecting the municipality of Cruzeiro do Sul-Guajará (7 34′02″ S, 72 39′32″ W, 181 m a.s.l). The eighth male was recorded near the municipality of Guajará (7 27′10″ S, 72 35′14″ W, 225 m a.s.l.), Amazonas, Brazil. Recordings were made between 06:30 h and 11:00 h and between 15:00 h and 17:30 h. Air temperature at the time of recording, measured with a digital thermohygrometer positioned 1 m above the ground, ranged from 24 to 27.4 C. Calls were recorded with a directional microphone Shotgun CSR Yoga HT-81 accoupled to a Zoom H4n digital recorder. The microphone was positioned approximately 1 m in front of the focal active male. Recordings were stored with 16 bits resolution, 44.1 kHz and WAV format. Call recordings were deposited in the Fonoteca Neotropical Jacques Vielliard , UNICAMP, Campinas, Brazil.
Sequencing and phylogenetic analysis. Total genomic DNA was extracted from tissue samples of five specimens of the new species using the commercial kit Wizard (Promega Corp., Madison, WI, USA) following the manufacturer's instructions. Fragments of the 16S rRNA mitochondrial gene were amplified by polymerase chain reaction (PCR) using universal primers 16sar (5′-CGCCTGTTTATCAAAAACAT-3′) and 16sbr (5′-CCGGTCTGAACTCAGATCACGT-3′) (Palumbi, 1996). Extractions, PCR reactions and sequencing were conducted with both forward and reverse primers (Maia, Lima & Kaefer, 2017). Sequences were visually checked and manually edited in Geneious 5.3.4 (Kearse et al., 2012). Sequences were deposited at GenBank and can be accessed under numbers MT446458-62.
We used BLAST to search for sequences already deposited in GenBank that might correspond to the new species. Two 16S rRNA sequences, KY886579 and MF624181, matched our sequences with high identity. Additionally, we retrieved 68 additional 16S sequences that represent the 37 species of Allobates currently available in GenBank (Table S1). Sequences of four species of other genera in the family Aromobatidae were used to root the tree: Anomaloglossus stepheni (Martins, 1989), Aromobates nocturnus Myers, Paolillo & Daly, 1991, Mannophryne collaris (Boulenger, 1912 and Rheobates palmatus (Werner, 1899). For phylogenetic reconstruction, we downloaded additional sequences of three other mitochondrial genes (12S ribosomal RNA; cytochrome oxidase I, COI; and cytochrome b, cytb) and six nuclear genes (28S ribosomal RNA, 28S; histone H3, HH3; recombination activating gene 1, RAG1; rhodopsin, RHO; seventh in absentia, SIA; and tyrosinase, TYR) from the same specimens (when available) for which 16S sequences were initially downloaded. Sequences were aligned in Bioedit 7.2.5 (Hall, 1999) using the CLUSTAL W algorithm (Thompson, Higgins & Gibson, 1994). Alignments were concatenated in Mesquite 3.04 (Maddison & Maddison, 2015), which yielded a final alignment consisting of 5,915 base pairs (bp) and 79 terminals. Vouchers and GenBank accession numbers are listed in Table S1.
The best-fit partition scheme and most probable nucleotide evolution model, considering codon partitioning for protein-coding genes, were inferred with PartitionFinder 2.1.1 (Lanfear et al., 2016) via the CIPRES webserver (phylo.org) using Bayesian Inference Criterion and the PhyML algorithm (Guindon et al., 2010). The best-fit partition schemes and evolution models for each partition are shown in Table 1. Phylogenetic relationships were reconstructed using Maximum Likelihood inference in IQ-TREE (Trifinopoulos et al., 2016). Clade support was calculated through 10,000 ultrafast bootstrap approximation replicates with 10,000 maximum iterations, a 0.99 minimum correlation coefficient and 10,000 replicates of the Shimodaira-Hasegawa approximate likelihood ratio. Kimura 2-parameter distance (Kimura, 1980) and uncorrected genetic distance were calculated using the 16S mitochondrial gene through MEGA 6.0 (Tamura et al., 2013).
Call description. Parameters were measured using Raven 1.5 (Bioacoustics Research Program, 2015). The following parameters were measured: call duration (CD), inter-call interval (ICI), notes per calls (NNC), note duration (ND), inter-note interval (INI), note repetition rate (NRR), call dominant frequency (DF), and first-note dominant frequency (FNDF). We analyzed 1-5 calls for each recorded male and 10 notes and their respective inter-note intervals in each call. Measured notes were equally distributed along each call. Bioacoustic terminology follows Köhler et al. (2017). Spectral parameters were measured by spectrograms set as Window = Blackman; Discrete Fourier Transform = 2,048 samples; 3 dB filter bandwidth = 82 Hz. Graphic representation of the advertisement calls was generated in the R environment (R Core Team, 2016) using the Seewave package 2.0.5 (Sueur, Aubin & Simonis, 2008

RESULTS
Our phylogeny is mostly consistent with previously published phylogenetic hypotheses (Grant et al., 2006;Grant et al., 2017;Melo-Sampaio, Oliveira & Prates, 2018) (Fig. 2). The monophyly of Allobates is highly supported (ML = 94%), with A. olfersioides emerging as the sister group to all other congeners (ML = 100%). It is followed by A. undulatus (ML = 100%) and a clade formed by A. niputidea and A. talamancae (ML = 100%). Despite the differences between the phylogenetic relationships of some clades recovered in this study and those of previously published phylogenies, most known clades are recovered in our analyses, such as the one composed of A. nunciatus, A. nidicola and A. masniger, and the one that groups A. subfolionidificans, A. conspicuous, A. insperatus and A. juami (Fig. 2). Samples of the new species nest with two other GenBank samples from Cruzeiro do Sul (Acre, Brazil), MF624181 and KY886579, which together form a well-supported clade (ML = 100%). While Allobates carajas is retrieved as sister to the new species, support for this relationship is very weak (ML = 24%). Although they share a similar call structure, the new species is not closely related to A. crombiei, A. amissibilis or A. juami. Allobates insperatus and A. juami are recovered as sister species with strong support (ML = 100%), whereas A. crombiei is sister to A. grillisimilis with weak support (ML = 76%).
Allobates trilineatus clade A (6.6%) and the largest is to A. olfersioides (13.5%). Distances between the new species and A. insperatus (9.7%), A. juami (9.5%) and A. crombiei (9.9%) are similar to or larger than the average distance between the new species and the entire dataset. See Table 2  Phylogenetic placement. The new species is assigned to the genus Allobates based on its phylogenetic position presented in the present study and in Grant et al. (2017).
Diagnosis. Allobates velocicantus sp. nov. is characterized by (1) small size, SVL 14.9-16.2 mm in males and 16.0-17.4 mm in females; (2) dorsal color pattern predominantly light brown, with no dark patches and marks; (3) dorsum granular; (4) dark brown lateral stripe from the tip of the snout to the groin; (5) light dorsolateral stripe absent or inconspicuous in living specimens but present in preserved specimens; (6) light but incomplete ventrolateral stripe; (7) oblique lateral line diffuse; (8) snout slightly rounded in dorsal view; (9) tympanum inconspicuous; (10) Table 3. Head 23% wider than long. Head width and length equal 35% and 30% of SVL, respectively. Eye diameter larger than distance from eye to nostril (EN/EL = 0.59). Eye diameter 50% of head length. Interorbital distance 81% of head width. Tympanum inconspicuous to the naked eye. Snout rounded in dorsal and lateral views. Nares located laterally at the tip of the snout. Canthus rostralis barely defined, slightly straight in dorsal view and rounded in lateral view; loreal region slightly concave. Maxillary teeth absent. Median lingual process absent. Tongue three times longer than wide (3.6 mm vs. 1.2 mm). Vocal sac single and subgular. Lateral vocal slits present at the height of the jaw angle. Palmar tubercle rounded, conspicuous, 0.6 mm wide. Thenar tubercle elliptical, conspicuous, 0.25 mm wide. Thenar tubercle width 42% of palmar tubercle diameter. Distal subarticular tubercle on finger III laterally positioned, tiny and barely noticeable. Distal subarticular tubercle on finger IV absent. Additional subarticular tubercles on fingers III and IV small and rounded, not exceeding phalangeal width. Subarticular tubercles on fingers I and II oval, protuberant and as wide as phalanges, corresponding to 2 and 1.4 times the width of the thenar tubercle, respectively. Supernumerary tubercles absent. Finger fringes and hand webbing absent. Tip of finger IV does not reach the distal subarticular tubercle of finger III when juxtaposed. Preaxial expansion of finger III, from the base to the tip of the finger. Finger I slightly longer than finger II. Relative length of fingers: IV < II < I < III. Discs of fingers II, III and IV slightly wider than the third phalange; disc of finger I of similar width to the distal phalange. Paired digital scutes present.
Thigh and tibia lengths similar, each 48% of body length. Foot length 88% of tibia length. Tarsal keel conspicuous, curved, straightening until almost forming a line towards, but not reaching, the inner metatarsal tubercle. Inner metatarsal tubercle protuberant and elliptical. Outer metatarsal tubercle small and rounded, protuberant, four times the diameter of the inner metatarsal tubercle. Metatarsal fold absent. Fringes on toes absent. Basal webbing present only between toes III and IV. Subarticular tubercle on toe I oval and protuberant. Subarticular tubercles on toes II-V rounded and barely evident; one subarticular tubercle on toe II, two on toes III and V, and three on toe IV. Disc of toe I same width as phalanx, tip rounded. Discs of toes II-V expanded laterally, tips rounded. Paired digital scutes present. Skin on dorsum granular. Skin on throat, chest, belly and ventral surfaces of limbs smooth.
In preservative, dorsum of the body cream, with elevated concentration of dark granules and melanophores extending from the interorbital region to the urostyle (Fig. 3A). Dorsolateral stripe present. Dorsal region of arms, forearms, thighs and paracloacal marks cream (Fig. 3A). Tibia with dark brown patches of varied sizes. A dark brown lateral stripe surrounds the whole body, but it is narrower in the loreal region and of constant width in the lateral part of the body. Anterior and posterior regions of the thigh dark brown. Ventrolateral stripe present but barely noticeable and composed of irregular small whitish patches (Fig. 3B). Belly, chest, forearms and throat cream; melanophores grouped in the anterior portion of the throat. Ventral surface of arms, thighs and tibia cream, with melanophores present in the distal region of the thighs and peripheral regions of the tibia (Fig. 3C). Ventral surfaces of the hands and feet dark brown (Figs. 4A and 5B).
Variation. On average, females are slightly larger (SVL 16.7 mm ± 0.7, 16.0-17.4; n = 3) than males (SVL 15.4 mm ± 0.3, 14.9-16.2; n = 10) (Table 3). However, the small number of females impedes statistical tests to verify sexual dimorphism. Females differ from males by having thinner phalanges on finger III. In both males and females, the central region of the dorsum varies in the concentration of dark granules and melanophores: five specimens show a dense concentration of dark granules (thus resembling the holotype), whereas in six other specimens such a concentration is less evident. The dorsolateral stripe varies in width and visibility; in both males and females, it is absent or barely noticeable in life but more evident in preservative. The ventrolateral stripe is conspicuous in living specimens but almost invisible to the naked eye in preserved specimens, being formed by whitish patches with no melanophores. The anterior region of the throat has a variable amount of melanophores in both males and females. The tibia shows dark patches, which form a transverse line visible in five specimens but not in six others. The tarsal keel varies in size, shape and length; it reaches the external edge of the inner metatarsal tubercle in only one specimen, INPAH 41341. In life, dorsum brown with dark brown granules concentrated medially. Dorsal surface of arms and forearms orange brown. Thighs light brown dorsally but dark brown at anterior and posterior ends. Supraorbital region light brown. Dark brown lateral stripe of constant width, narrower in the loreal region and wider in the flanks; small whitish dots inside the brown lateral line at the inguinal region. Dorsolateral stripe, when present, light brown. An irregular ventrolateral stripe is iridescent white. It is formed by dots in some specimens; in others, it is well defined and complete or incomplete (Figs. 6B, 6E, 6H and 6K). Ventrolateral surfaces of thighs and arms, as well as the flanks, are translucent beige. Belly white in males and females; chest white in females and translucent in males.
Throat of females translucent with the edges of the jaw yellowish. In males, the vocal sac, when deflated, is whitish centrally and yellow peripherally. When inflated, it is mostly white with translucent yellow edges. Iris metallic bronze; pupils black, horizontal semielliptical.
Comparisons with other species. Allobates velocicantus sp. nov. easily differs from A. femoralis and A. hodli by the absence of bright yellow, orange or red marks on thighs (present). Cryptically colored species of Allobates show either of two striking dorsal color patterns, one with and the other without large dark patches or marks, a characteristic that helps to quickly differentiate species. Allobates velocicantus sp. nov. lacks a dark hourglass-shaped mark on the dorsum and is thus easily distinguished from A. brunneus, A. carajas, A. crombiei, A. flaviventris, A. gasconi, A. magnussoni, A. ornatus, A. pacaas, A. tapajos, and A. trilineatus (all of which have dark patches or marks on the dorsum with rhomboid-, diamond-, hourglass-or triangle-shaped patterns). Among Allobates species that lack dark patches or marks on the dorsum, throat color is a reliable characteristic for differentiating species. Males of Allobates velocicantus sp. nov. has a white throat with yellow edges, which distinguishes it from males of A. bacurau (light to dark gray), A. fuscellus (dark to solid dark), A. juami (pinkish to translucent), A. masniger (dark gray), A. melanolaemus (black), A. nidicola (dark gray), A. nunciatus (violaceous), A. paleovarzensis (grayish-violet) and A. vanzolinius (light to dark gray).  Allobates velocicantus sp. nov. could be mistaken for A. caeruleodactylus, A. conspicuus, A. grillisimilis, A. subfolionidificans or A. tinae, which have similar ventral color patterns in preservative and adult males of similar size (SVL). However, Allobates velocicantus sp. nov. differs from A. caeruleodactylus by lacking basal webbing between fingers II-III and by scutes of fingers white (basal webbing between fingers II-III present, scutes of fingers blue); from A. conspicuus by having finger I slightly longer than finger II and by lacking both fringes on toes and transverse bars on thighs (finger I considerably longer than finger II, fringes on toes present and transverse bars on thighs present); from A. grillisimilis by the absence of a conspicuous transverse bar on legs and arms, finger III with phalanges of uniform width, edges of the throat yellow with center whitish (varying number of transverse bars on legs and arms present, finger III with phalanges of non-uniform width, and throat translucent white); and from A. subfolionidificans by having a white belly and a whitish throat with yellow edges, a ventrolateral stripe, and by lacking conspicuous transverse bars on the thighs (belly and throat opaque white, ventrolateral stripe absent, and transverse bars on thighs present).
The advertisement call of A. velocicantus sp. nov. has only one type of temporal arrangement of notes (trills), which distinguishes it from the calls of A. brunneus, A. carajas, A. flaviventris and A. tapajos (each of these species has two types of temporal arrangement of notes: notes emitted continuously, or in trills of unpulsed notes). Color of freshly laid eggs and larvae description. Freshly laid eggs of A. velocicantus sp. nov. have dark gray animal pole covering approximately two thirds of the animal hemisphere. The dark gray animal pole shows a well-delimited edge above the whitish equatorial portion. The vegetal pole is completely whitish. Eggs are deposited in a cloudy jelly.
Morphometric measurements of tadpoles are presented in Table 5. See Figs. 8 and 9 for body form and oral disc. Descriptions of quantitative characters are based on tadpoles at Gosner stage 34 (n = 5). Body ovoid, rounded anteriorly and posteriorly in dorsal view, flattened in lateral view (BH/BW = 0.7-0.8). Body length and tail length 30.7-33.0% and 66.9-69.3% of TL, respectively; HWLE 81.5-90.9% of BW; snout rounded in lateral and dorsal views; END 60.0-76.5% of ED; eyes dorsal and directed laterally; IOD 54.8-58.6% of HWLE. Small nares located dorsolaterally and directed anterolaterally, visible in dorsal and lateral views; internarial distance 40.0-45.2% of HWLE. Fleshy ring on inner margin of nostrils rounded, not ornamented. Spiracle single, sinistral, tubular, 0.6-0.9 mm long, attached to body ventrolaterally, slightly below mid-body length. Gut coiled and visible through the skin to the naked eye, with its axis directed to the left side of the body. Vent tube dextral, 1.4-2.2 mm long. Maximum tail height 2.5-2.9 mm. Dorsal fin emerges after 2.0-2.5 mm from the limit between tail and body; dorsal edge shallow and straight anteriorly (along approximately 10% of its length) but deeper posteriorly, reaching maximum height at two thirds of the tail length. Dorsal fin slightly deeper than ventral fin. Ventral fin does not exceed body height. Tail tip slightly acuminate. Caudal musculature in dorsal view attains 47.4-51.5% of body width. In lateral view it reaches 54.8-75.0% body height. Oral disc located anteroventrally, emarginated laterally (Fig. 8); width of oral disc 1.4-1.6 mm, corresponding to approximately 41.6-45.5% of body width at the level of spiracle. Anterior labium with 3-4 short, pyramidal, rounded papillae distributed in a single row on each lateral margin. Posterior labium with a single row of 16-21 marginal papillae of similar size and shape to those on anterior labium. Submarginal papillae absent. Upper jaw sheath arc-shaped, longer than lower jaw sheath, with no medial notch. Lower jaw sheath V-shaped, slightly Table 5 Morphometric measurements (in millimeters) of 14 tadpoles of Allobates velocicantus sp. nov., Gosner stages 27-37, collected in the municipality of Mâncio Lima, state of Acre, Brazil.
Comparisons with tadpoles of other species. Tadpoles of Allobates velocicantus sp. nov. have short, rounded and pyramidal papillae on each side of the labium. There are 3 or 4 papillae on the anterior labium and from 16 to 21 on the posterior labium. Hence, this species is distinct from congeners that have more-elongate papillae on the posterior labium-A. caeruleodactylus (6 long papillae), A. grillisimilis (approximately 8 very long papillae), A. subfolionidificans (40 papillae on the posterior labium, which are longer than those on the anterior labium) and A. tapajos (8-10 long papillae)-and from those that have a larger number of short papillae on the anterior labium: A. brunneus (5 papillae), A. carajas (5-8 papillae), A. magnussoni (12-13 papillae) and A. nunciatus (8-9 papillae).
Natural history notes. Allobates velocicantus sp. nov. inhabits the litterfall of primary and secondary lowland ombrophilous open forest (Fig. 10A). Populations of this species were found in terra firme forest either close to or distant from small forest streams. The new species uses leaves of small shrubs in the forest understory as egg deposition sites; two clutches (nine tadpoles and 13 eggs) were found in this situation. On 13-15 February 2019, during the middle of the rainy season, males were found calling between 8:00 and 18:00 h, and always perched between 10 and 30 cm above the ground. A pair of adults was observed courting: the male was leading the female to an egg deposition site, and he emitted a courtship call while both frogs were moving. As the male started emitting another advertisement call, the female stopped following. Arriving at the egg deposition site, the male jumped on to a leaf (adaxial surface), which was located around 20 cm above the ground, and continued to emit advertisement calls interspersed with courtship calls. After a few moments, the female jumped to the leaf with the male and approached him, initially touching snouts before turning her back to him. The male then jumped onto the female's dorsum, while also sliding its hand to her head. The male remained in that position-cephalic amplexus-for 2 min before jumping from the leaf. The female remained in place for a few moments but then began moving, turning clockwise 30 , stopping, and repeating that pattern several times, always turning clockwise. The length of time the female stopped between movements ranged from 2 to 10 min, while she deposited eggs (Fig. 10C). Overall, egg deposition lasted 40 min before the female abandoned the leaf, leaving behind a clutch of 13 eggs. Each egg had a distinctly pigmented animal pole and was encased within cloudy jelly (Fig. 10D). The clutch was collected immediately after the female abandoned it, the embryos maintained alive through hatching, and the resulting tadpoles reared to Gosner stages 27-37 before being sacrificed and preserved.
The second egg clutch had nine tadpoles at aquatic-transport stage. The male (INPAH 41347) jumped onto the egg clutch and made circular movements inside it, at which time the tadpoles began to wriggle towards his dorsum, eventually climbing onto it (Figs. 10E-10G). This event lasted for 5 min.

DISCUSSION
Allobates velocicantus is the twenty-sixth species of Allobates described from Brazil and the twenty-third species from Brazilian Amazonia. The species is distributed in southwest Amazonia, a region of high species richness for several taxonomic groups (Jenkins, Pimm & Joppa, 2013), including anurans (Souza, 2009;Bernarde, Machado & Turci, 2011).
The type locality is located in the Area of Relevant Ecological Interest Japiim Pentecostes, a conservation unit intended for sustainable use. Yet, the conservation of A. velocicantus, and of regional biodiversity in general, is not guaranteed. Deforestation in this part of Amazonia, including in conservation units, has increased with the expansion of illegal logging, cattle ranching, and agriculture (Kröger, 2020;Montibeller et al., 2020). Moreover, deforestation and degradation rates in Amazonia are increasing substantially due to the anti-environmental agenda of the current Brazilian government (Ferrante & Fearnside, 2019;Kröger, 2020).
The high note-repetition rate call of Allobates velocicantus is a remarkable feature. Four species of Allobates emit calls with similar structure: A. bacurau, A. crombiei, A. juami and A. insperatus. However, the note-repetition rate in A. velocicantus is strikingly higher than the rates of these other species. Furthermore, the advertisement call of A. velocicantus has the highest number of notes among all species of Allobates with documented calls. These characters make the species unique and easily distinguished from congeners. While most species of Allobates share a conserved external morphology, vocalizations are highly variable and species-specific. They represent essential characters for use in the formal diagnosis and description of new taxa, as well as in future species discovery.
Information on egg coloration in Allobates is scarce except for species recently described or redescribed (Lima, Sanchez & Souza, 2007;Lima, Caldwell & Strüssmann, 2009;Lima et al., 2010;Lima, Simões & Kaefer, 2014Moraes, Pavan & Lima, 2019;Simões et al., 2013;Simões, Rojas & Lima, 2019). Allobates velocicantus oviposits on green leaves of small shrubs. Currently, only two other species in Brazilian Amazonia use this breeding strategy exclusively: A. subfolionidificans and A. carajas. Unlike in A. velocicantus, however, the animal pole of eggs in these two species is white (Lima, Sanchez & Souza, 2007;Souza, Kaefer & Lima, 2017;Simões, Rojas & Lima, 2019). Despite both A. velocicantus and A. subfolionidificans inhabiting the open-canopy lowland forest with considerable incidence of sun rays above canopy, eggs of A. subfolionidificans are protected from direct solar radiation by being deposited on the abaxial surfaces of green or dry leafs of small shrubs (Lima, Sanchez & Souza, 2007;Souza, Kaefer & Lima, 2017), while A. velocicantus oviposits on the adaxial surface of green leaves of small shrubs. On the other hand, A. carajas oviposits white eggs on the adaxial surface of green leaves (Simões, Rojas & Lima, 2019), but eggs are protected by a dense-canopy forests with low levels of solar incidence on the lower understory vegetation (P.I. Simões, 2014, personal communication). Intense solar radiation may cause mortality and abnormal embryo development in anurans (Beudt, 1930;Gurdon, 1960;Blaustein et al., 1997), so the amount of melanin on eggs in these species might be an ecophysiological response to the solar radiation to which eggs are exposed. Differences in the amount of melanin on eggs were reported from other two species of the A. tinae species complex inhabiting forests with distinct canopy openness in the Brazilian Amazonia (Lima, Ferrão & Silva, in press). The color of jelly may have a protective function against solar radiation. Jelly of A. velocicantus is cloudy, while the jelly of A. subfolionidificans and A. carajas are translucent. The effect of solar radiation on eggs of Allobates with different concentrations of melanin and deposited within jellies of different colors needs physiological tests.

CONCLUSION
Allobates velocicantus is differentiated from its congeners based on external morphology of adults and tadpoles, advertisement call and molecular analyses. The species represents an excellent model to study the ecological and physiological adaptations to solar radiation on eggs of Allobates. However, the conservation of A. velocicantus is threatened by the expansion of illegal logging, cattle ranching, and agriculture.

APPENDIX I. SPECIMENS EXAMINED Field Study Permissions
The following information was supplied relating to field study approvals (i.e., approving body and any reference numbers): Specimens were collected under collection permit number 1337-1 provided by the Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renováveis (IBAMA).

Data Availability
The following information was supplied regarding data availability: Specimens are housed in the herpetological section of the Zoological Collection of the Instituto Nacional de Pesquisas da Amazônia (INPAH), Manaus, Amazonas, Brazil, under numbers INPAH 41339 to INPAH 41351. Sequences are available at GenBank under accession numbers MT446458 to MT446462. Morphometric measurements are provided in a Supplemental

Supplemental Information
Supplemental information for this article can be found online at http://dx.doi.org/10.7717/ peerj.9979#supplemental-information.