A new species of Dendropsophus (Anura, Hylidae) from southwestern Amazonia with a green bilobate vocal sac

Abstract Recent studies have shown that species diversity of the South American frog genus Dendropsophus is significantly underestimated, especially in Amazonia. Herein, through integrative taxonomy a new species of Dendropsophus from the east bank of the upper Madeira River, Brazil is described. Based on molecular phylogenetic and morphological analyses, the new species is referred to the D. microcephalus species group, where it is differentiated from its congeners mainly by having a green bilobate vocal sac and an advertisement call comprising 1–4 monophasic notes emitted with a dominant frequency of 8,979–9,606 Hz. Based on intensive sampling conducted in the study area over the last ten years, the new species is restricted to the east bank of the upper Madeira River, although its geographic range is expected to include Bolivian forests close to the type locality.


Introduction
The genus Dendropsophus Fitzinger, 1843 is a taxonomically difficult group of small and, for the most part, morphologically similar species. The group exhibits high species diversity -108 species are currently recognized, of which 66 occur in Brazilian Amazonia (Segalla et al. 2016;Frost 2019) -as well as cryptic diversity due to a high degree of both phenotypic similarity among species and intraspecific polymorphism (Gehara et al. 2014;Caminer et al. 2017). Whereas traditional morphological methods have often failed to reveal cryptic species and accurately delimit species boundaries, nonmorphological methods (e.g., molecular phylogenetics and bioacoustics) have proven to be very useful for reliably documenting the full extent of species diversity in the genus (e.g., Fouquet et al. 2015;Rivanadeira et al. 2018).
The advertisement call is the most common mate-recognition signal among anurans; it has a direct impact on sexual selection and speciation (e.g., Sullivan et al. 1995;Boul et al. 2006). Consequently, advertisement call characteristics are widely used to identify anuran species both in field-based faunal inventories and in taxonomic studies (Schneider and Sinsch 2007;Köhler et al. 2017). During herpetological surveys of the amphibian and reptile fauna in the vicinity of the upper Madeira River (southwestern Amazonia, Rondônia, Brazil) in 2011-2013, we recorded several anuran advertisement calls that were markedly different from calls of all described species of Dendropsophus known from Brazilian Amazonia. The frogs emitting these calls morphologically resemble members of the D. microcephalus species group (sensu Faivovich et al. 2005), and preliminary bioacoustic analyses revealed that their calls are monophasic, i.e., they consist of only one call type (e.g., Orrico et al. 2014), and have a remarkably high dominant frequency (above 8 kHz).
In the Dendropsophus microcephalus species group, a similarly high dominant frequency has been reported only for two "monophasic" species: D. meridianus (Lutz, 1954) and D. ozzyi Orrico, Peloso, Sturaro, Silva, Neckel-Oliveira, Gordo, Faivovich &Haddad, 2014 (Pombal andBastos 1998;Orrico et al. 2014). A high dominant frequency (~ 9 kHz) was also reported for D. minusculus (Rivero, 1971) from Belem, Brazil, by Duellman and Pyles (1983), but a low dominant frequency (~ 3 kHz) was recorded by Tarano (2011) from a population of the same species closer to the type locality in Venezuela. We suspect that the report by Duellman and Pyles (1983) may represent a species misidentification, and that the population referred to D. minusculus instead likely corresponds to D. ozzyi.
We believe that the unknown Dendropsophus with high dominant frequency calls represent at least two new species, which differ markedly in body shape, coloration and molecular characters. Herein, we provide formal description of the most strikingly distinct of these species, which to date is known only from the east bank of the upper Madeira River. In addition to its distinctive advertisement call, the species is characterized by a green bilobate vocal sac. Our description combines morphological, bioacoustic and molecular data.

Specimens examined
We examined adult specimens of three forms of Dendropsophus collected in nine longterm ecological research (hereafter RAPELD) sampling sites (Magnusson et al. 2013) on the east and west banks of the upper Madeira River (Fig. 1, Table 1). Collected individuals were killed by topical application of a 2% benzocaine solution. Tissue samples were then taken from all specimens and stored in 100% ethanol. Finally, all specimens were fixed in 10% neutral-buffered formalin and stored in 70% ethanol. Voucher specimens are deposited in the herpetological collection of the Instituto Nacional de Pesquisas da Amazônia, Manaus, Brazil (INPA-H). Dendropsophus species used for comparisons are listed in Appendix 1.

Morphological characters
The format for the description follows Moravec et al. (2008). Specimens were sexed based on the presence or absence of secondary sexual characters (e.g., vocal sac and vocal slits) in males. Morphometric measurements were taken to the nearest 0.1 mm using a dissecting microscope and digital calipers. Thirteen morphometric measurements follow Duellman (1970) and Heyer et al. (1990): SVL, snout-vent length; HL, head length; HW, head width; EN, eye-nostril distance; ED, horizontal eye diameter; TD, horizontal tympanum diameter; HAL, hand length; 3FD, third finger disk diameter; 4TD, fourth toe disk diameter; TL, tibia length; THL, thigh length; FL, foot length; TAL, tarsus length. Webbing formulae of toes follow Savage and Heyer (1967) as adapted by Myers and Duellman (1982). Field notes and photographs taken by A. P. Lima were used to describe coloration in life.

Molecular analysis
We included samples of three forms of small-sized Dendropsophus collected in the area of the upper Madeira River during surveys in 2011-2013. The aim of those surveys was to detect phylogenetic diversity of Dendropsophus species distributed in this region. For the final dataset, we retrieved additional sequences of Dendropsophus from GenBank to locate phylogenetic positions of our new material in relation to DNA sequences published earlier, most importantly by Faivovich et al. (2005) and Jansen et al. (2011Jansen et al. ( , 2019. We included species representing all Dendropsophus species groups (sensu Faivovich et al. 2005). Primary attention was paid to Amazonian species of the D. microcephalus species group. In concordance with earlier published phylogenies, we used Xenohyla truncata (Izecksohn, 1959) as an outgroup. The final dataset comprised 63 samples representing 34 nominal taxa, three new forms and the outgroup. All sequences acquired from GenBank are identified by GenBank accession numbers (Appendix 2).  Genomic DNA was extracted from muscle tissue of 16 specimens of the three new forms. DNA extractions were obtained using the Wizard Genomic DNA Purification Kit (Promega Corporation, USA) following the manufacturer's protocols. We used the 16sbr (GCCGTCTGAACTCAGATCGCAT) and 16sar (CGCCTGTTTAT-CAAAAACAT) primers (Palumbi et al. 1991) to amplify a fragment of the 16S rRNA containing 495 base pairs (bp). The reaction conditions had a pre-heating step at 73 °C for 60 s, 35 cycles of denaturation at 92 °C for 10 s, primer annealing at 50 °C for 35 s, and primer extension at 72 °C for 90 s, followed by a final extension step of five minutes at 72 °C. PCR products were purified through Exonuclease I and Thermosensitive Alkaline Phosphatase (Thermo Fisher Scientific, USA) and followed ABI BigDye Terminator Cycle Sequencing Kit protocols (Life Technologies, USA) as recommended by the manufacturer. Amplicons were sequenced using the forward primer in Macrogen (Macrogen Inc., Seoul, Korea).
Sequences were visually checked and edited with GENEIOUS 7.1.7 (GeneMatters Corp, Minneapolis, MN, USA). The final 16S rRNA matrix was composed of 63 terminals and 495 bp. BIOEDIT (Hall 1999) was used to align the final matrix through the ClustalW algorithm (Thompson et al. 1994). The most probable evolutionary model explaining sequence divergence was estimated using the Akaike Information Criterion (AIC) in JMODELTEST 2.1.7 (Darriba et al. 2012), which recovered the GTR+G+I as the most probable evolutionary model.
Phylogenetic trees were inferred through Maximum Likelihood (ML). The ML phylogenetic tree was calculated under the GTRGAMMA model with IQTREE webserver (Trifinopoulos et al. 2016). Clade support was estimated through 5,000 ultrafast bootstrap approximation replicates. MEGA 6.06 (Tamura et al. 2013) was used in order to estimate the uncorrected-pairwise genetic distance (p-distance) and Kimura-2-Parameters genetic distance (K2P; Kimura 1980) between the new Dendropsophus forms and other members of the D. microcephalus species group included in the phylogenetic analyses.

Bioacoustics
Advertisement calls of three males of the new Dendropsophus species (INPA-H 41302, 41303, 41304) were recorded in the sampling site Jaci-Novo during the rainy season on 15 February 2013. Calls were recorded with a Marantz PMD660 digital professional recorder (Marantz, Japan) and a Sennheiser K6/ME66 directional microphone (Sennheiser, Germany). The microphone was positioned approximately 1 m from each male. Recordings were made in wave format at a sampling rate of 44.1 kHz with 16bits resolution. Air temperature taken with a digital thermometer during the recording was 25-26 °C (N = 3). Recordings are housed in the bioacoustic repository of the Amazonian Biodiversity Studies Centre at INPA (CENBAM 706,707,708).
Seven advertisement calls were analyzed for each recorded male. Advertisement call parameters were measured in RAVEN 1.5 (Bioacoustics Research Program 2015). Raven parameters were set as follows: window type = Blackman window, 3 dB filter bandwidth = 82 Hz, FFT window size = 2048 samples; FFT overlap = 80%, hop size = 4 ms. The following temporal and spectral parameters were inferred: call duration, inter-call interval, call period, number of notes, note duration, number of pulses per note, pulse duration, inter-pulse interval, dominant frequency (measured trough the function Peak Frequency), and bandwidth. The bandwidth was measured 20 dB below the peak frequency to avoid the overlap with background noise. Terminology of call measurements follows Köhler et al. (2017) while terminology of call structure follows Littlejohn and Harrison (1985). Graphic representation of the advertisement calls was produced in the R environment (R Core Team 2016) through the package seewave v.2.1 (Sueur et al. 2008). Seewave was set as follows: window = Hanning, FFT size = 150 samples, FFT overlap = 85%.

Molecular analysis
The Maximum Likelihood (ML) analysis based on 16S rRNA recovers several wellsupported clades within Dendropsophus (Fig. 2). Samples collected in the area of the upper Madeira River form three monophyletic lineages nested within a major clade (ML support = 96), which includes species of the D. microcephalus species group sensu Faivovich et al. (2005).
Dendropsophus specimens from the west bank of the upper Madeira River cluster in sister position to D. reichlei Moravec, Aparicio, Guerrero-Reinhard, Calderon & Köhler, 2008 from Bolivia (ML support = 98). These frogs sort into two well-supported sister lineages (ML support = 98; Fig. 2). The first lineage (hereafter referred to as Dendropsophus sp. A) comprises specimens collected in the RAPELD Teotonio and M11 sampling sites (distance apart ~ 250 km). The second lineage (Dendropsophus sp. B) comprises specimens from the RAPELD Pedras and Bufalos sampling sites (distance apart ~ 20 km). Genetic distances between D. sp. A and D. sp. B range from 1.6% (p-distance) to 1.7% (K2P). Genetic distances between D. reichlei and D. sp. A (K2P and p-distance = 3.1%) are slightly higher than those between D. reichlei and D. sp. B (K2P and pdistance = 2.8%). The average intraspecific genetic distance is higher in Dendropsophus sp. A (K2P and p-distance = 0.4%) than in D. sp. B (K2P and p-distance = 0.1%).
Because Dendropsophus bilobatus sp. nov. also differs from other congeneric species by its remarkably distinct morphology, we here describe it as a new species. Resolution of the taxonomic status of D. sp. A and D. sp. B is pending the results of additional species delimitation tests, which will be treated in a future study. Generic placement. We assign this species to Dendropsophus based on our molecular phylogenetic analysis (Fig. 2) and on its general morphological similarity to other members of the genus.
Diagnosis. A species of the Dendropsophus microcephalus species group, distinguished from other species of Dendropsophus by the following combination of characters: (1) small size, SVL 18.8-20.8 mm (N = 8) in males (females unknown), head slightly wider than body; (2) snout short, truncate in dorsal and lateral views; (3) tympanum evident, round, about one third of eye diameter, tympanic annulus distinct anteriorly, ventrally and partly posteriorly; supratympanic fold barely evident; (4) dentigerous processes of vomers small, barely prominent, and separated medially between posterior halves of choanae; (5) skin on dorsal surfaces smooth; (6) tarsal fold and tubercles on outer edge of tarsus absent; ulnar folds and tubercles absent; (7) axillary membrane extensively developed; (8) fingers about half webbed; toes about three-fourths webbed; (9) bifid distal subarticular tubercle under fourth finger; (10) pectoral glands absent; (11) generally darker coloration of the loreal-tympanic region contrasts sharply with the lighter dorsal head coloration, one or two white spots below the eye; (12) in life, ground coloration of dorsum light brown; head greenish brown laterally; flanks ventrally and posteriorly a translucent pinkish white without chromatophores; hidden surfaces of thighs yellow without melanophores; (13) in life, throat green in males; belly yellowishwhite in pectoral and central parts, translucent pinkish-white in posterior and lateral parts; ventral surfaces of thighs translucent pinkish white; (14) in life, iris pale to dark brown with barely visible tiny brown veins, iris periphery dark brown to black; bones white; (15) advertisement call consisting of 1-4 notes (usually 1-2 notes), emitted regularly in series of 7-35 calls; high-pitched, monophasic, pulsed notes (3-8 pulses) with a duration of 12-24 ms and a dominant frequency of 8,979-9,606 Hz.
Comparisons. Dendropsophus bilobatus sp. nov. is readily distinguished from all congeners by having a green bilobate subgular vocal sac (some members of the D. marmoratus species group have a bilobate vocal sac, but not green) and a monophasic advertisement call with a remarkably high dominant frequency (8,979-9,606 Hz). Below we describe additional important differences between the new species and other members of the D. microcephalus species group (sensu Faivovich et al. 2005) that occur in Brazilian Amazonia and surrounding areas of Bolivia, Colombia, Peru and Ecuador. Characters of D. bilobatus are set in parentheses if not otherwise stated.
The dark-greenish-brown coloration of the loreal-tympanic region of Dendropsophus bilobatus , which sharply contrasts with the light brown dorsal head coloration, resembles the head color pattern of D. coffea (Köhler, Jungfer & Reichle, 2005), D. cruzi (Pombal & Bastos, 1998) (Melin, 1941), and D. reichlei, but the new species is easily distinguished from each named species as follows (species already distinguished above are not listed here): D. coffea lacks white subocular spots (present) and has dark brown dorsal stripes (absent; Köhler et al. 2005); in D. cruzi, the thigh is longer than the tibia (tibia longer than thigh; Pombal and Bastos 1998); D. studerae has tuberculate dorsal skin (smooth; Carvalho-e-Silva et al. 2003); D. juliani has an acutely rounded snout in dorsal view (truncate), absence of white subocular spots (present), and greenish yellow plantar surfaces (orange; Moravec et al. 2006); D. microcephalus has maximum male SVL 24.5 mm (20.8 mm), an acutely rounded snout in dorsal view (truncate), an ovoid tongue (cordiform), and a weak tarsal fold (absent; Duellman 1970); D. shiwiarum has the discs of finger III and toe IV with pointed tips (pointed tips absent), a prominent conical tubercle on the dorsal surface of fingers III and IV (tubercle absent), both palmar and plantar surfaces unpigmented (palmar surface greenish yellow, plantar surface orange), and a lower dominant frequency of the advertisement call (3,254 Hz in D. shiwiarum vs. 8,606 Hz in D. bilobatus ; Ortega-Andrade and Ron 2013); D. tintinnabulum has a triangular-to-rounded snout in dorsal view (truncate) and orange palmar surfaces (greenish; Teixeira and Giaretta 2017), and lacks white subocular spots (present); and D. reichlei has a rounded snout in dorsal view (truncate) and a distinct canthus rostralis (absent), and lacks a glandular nuptial pad (present; Moravec et al. 2008).
In our phylogenetic analysis, the clade that contains Dendropsophus bilobatus is closely related to D. bipunctatus (Spix, 1824), D. meridianus and D. berthalutzae (Bokermann, 1962) from the southern and southeastern Brazilian coast (Fig. 2). In addition to differences in shape and color of the vocal sac and in advertisement call, these three species can be distinguished from D. bilobatus as follows: D. bipunctatus has a granulate dorsum (smooth), maximum SVL in males 25 mm (maximum male SVL 20.8 mm) and several small spots surrounded by a dark network that are distributed across the subocular area and lateral snout (spots only on subocular area and are not surrounded by dark network; Lutz 1973); D. berthalutzae has a snout that is slightly mucronate in dorsal view (truncate) and longer than eye diameter (snout shorter than eye diameter), and a minute outer metatarsal tubercle (absent; Lutz 1973).
The two unnamed forms of Dendropsophus in the D. microcephalus species group from the west bank of the upper Madeira River (D. sp. A and D. sp. B) differ from D. bilobatus in having a single yellow subgular vocal sac (bilobate, green) and pointed discs on toes and fingers (rounded).
Holotype description. INPA-H 41300. Adult male (Figs 3, 4A, B), SVL 18.8 mm; body moderately robust; head slightly wider than long (HW/HL = 1.08); snout truncate in dorsal and lateral views; snout short, eye-nostril distance shorter than eye diameter (END/ED = 0.68); canthus rostralis rounded in dorsal and lateral views; loreal region slightly concave; internarial area slightly depressed; nostrils barely protuberant, directed dorsolaterally; interorbital area flat, slightly depressed in the central portion; interorbital distance equal 34% of head width; eyes large, strongly protuberant, ED/ TD = 3.30, ED/HL = 0.42; tympanic membrane small, round, clearly distinct, its di- ameter 30% of eye diameter and 13% of head length; tympanic annulus distinct ventrally and anteriorly; supratympanic fold barely evident, slightly obscuring the upper edge of the tympanum. Arms slender and not hypertrophied; ulnar tubercles and fold absent; axillary membrane reaches the second third of the upper arm; hand relatively long, about 30% of SVL, approximately the same size as the forearm; fingers long, slender, bearing small discs; finger III twice as wide medialy than anteriorly; relative length of fingers I<II<IV<III; discs rounded on fingers; diameter of disc on finger III about the size of the tympanum; subarticular tubercles of fingers I and IV medium to largesized, round, prominent, bifid in finger IV; subarticular tubercles of fingers II-III small, round, prominent; supernumerary tubercles barely evident; palmar tubercle small, flat, oval, barely evident proximally; prepollical tubercle large, flat, ovoid; nuptial pad white, glandular, covering the dorsolateral portion of the thumb but not reaching the ventral surface; webbing formula of fingers I 2 + -2 II 1 1/2 -2 2/3 III 2 --2 IV. Legs moderately long, slender (THL/SVL = 0.55; TL/SVL = 0.56); tibia slightly longer than thigh (TL/THL = 1.02); tarsal fold and tarsal tubercles absent; calcar tubercles absent; toes moderately long, bearing discs slightly smaller than those on fingers; toe IV length equals 60% of foot length; relative length of toes I<II<III<V<IV; toes I, II and V slender; toes III and IV widened by elongated flat glandular structures on both sides, glandular structures forming a continuous elongated glandular patch along toe IV, small glandular aggregations present also on fingers II and V; discs rounded on toes; diameter of the disc on toe IV equals diameter of the disc on finger III; subarticular tubercles round, prominent, penultimate tubercle on toe V bifid; supernumerary tubercles on toes III-IV small, round, barely evident; inner metatarsal tubercle elliptical, flat; outer metatarsal tubercle barely distinct; webbing formula of toes I 1 + -2 -II 1 + -1 1/2 III 1 1/2 -2 -IV 2 --1 + V.
Skin on head, dorsum, dorsal surfaces of limbs and flanks smooth; vocal sac and ventral surfaces of arms smooth; belly smooth laterally, coarsely granular medially; lower surfaces of thighs and surroundings of cloaca slightly granular. Cloacal opening directed posteroventrally at midlevel of thigh, covered dorsally by a wide cloacal sheath. Choanae small, vertically oval; dentigerous processes of vomers small, three vomerine teeth present on the right process, absent on the left process. Tongue cordiform, posterior third not attached to the floor of the mouth. Vocal slits long, extending from the midlateral base of the tongue to the angle of the jaw; anterior part covered by the lateral margin of the tongue. Vocal sac bilobate, subgular (Figs 3A, 4A, C, D).
In life (Fig. 4A, B), the dorsum and dorsal surfaces of the limbs are light brown with an irregular pattern of yellow spots; the head has a large triangular yellow blotch that extends from the tip of the snout to the anterior interorbital region, including the anterior margin of the upper eyelids; the lateral sides of the head are greenish brown with two white horizontally elongate subocular spots on the left side and one elongate and one round white spot on the right side. The iris is pale to dark brown with barely visible tiny brown veins; its outer edge is brown to black. Proximal dorsal surfaces of fingers I-III are greenish white to yellowish white; the proximal dorsal surface of finger IV is brown; distal dorsal surfaces of the fingers are yellowish orange; nuptial pads are white. The upper part of the flanks is a light pinkish brown; the posterior part of the flanks and the groin are pinkish white. Hidden dorsal surfaces of the thighs are yellow. The vocal sac is green when deflated but translucent greenish white when inflated. The chest and belly are yellowish white medially but translucent pinkish white laterally and posteriorly. Ventral surfaces of arms and legs are translucent pinkish white; the anteroventral side of the thigh is yellow, the posteroventral side is pinkish white; palmar surfaces are greenish yellow; plantar surfaces are orange. Bones are white.
In alcohol (Fig. 3), the head and dorsum are cream to brown with numerous tiny black melanophores and irregular white spots and blotches; dorsal surfaces of the limbs are light cream or translucent; ventral surfaces are translucent to cream, the chest and medial area of the belly are white. Bones are white.
Distribution and natural history. Our research team has sampled frogs at more than 150 permanent sampling sites distributed on both banks of the upper Madeira River and along the Purus-Madeira Interfluve. Yet, we have only observed Dendropsophus bilobatus in the lowland ombrophilous open forest on the east bank of the upper Madeira River. This area is close to the border between Brazil and Bolivia, and we expect that the new species also occurs in Bolivian lowland ombrophilous open forest, as do other anuran species that are known exclusively from the east bank of the upper Madeira River (e.g., Hydrolaetare caparu [Jansen, Gonzales-Álvarez & Köhler, 2007] and Hamptophryne alios [Wild, 1995]; Simões et al. 2011, Ferrão et al. 2014. To date, specimens of Dendropsophus bilobatus have been observed only in the rainy season (early November to late March), which coincides with the species' breeding season. Calling males were observed in flooded areas connected to rivers of moderate (Jaci-Parana River) to large size (Madeira River). Males typically call in a large chorus while perched on leaves and tiny trunks that range in height from just a few centimeters above the water surface to ~ 2 m high. Males start calling in the crepuscule (~ 18:00 hs) and call activity has been observed at least to approximately midnight. When call activity ends remains unknown. Amplexus has not been observed. Other sympatric frogs include Rhaebo guttatus (Schneider, 1799), Boana cinerascens (Spix, 1824), B. lanciformis (Cope, 1871), Scinax sp. 6 (sensu Ferrão et al. 2016) and an uncollected Scinax with an advertisement call that resembles that of S. garbei (Miranda-Ribeiro, 1926).
Etymology. The specific name bilobatus is derived from the Latin noun bilobate. The name refers to the characteristic bilobate shape of the vocal sac of males of the new species.

Discussion
The upper Madeira River is characterized by high biodiversity; several priority areas for conservation are identified in this region (Capobianco 2001). Yet, recent studies have revealed that knowledge of the species diversity of amphibians inhabiting forests in the upper Madeira River is still incomplete, and description of new species is ongoing (e.g., Simões et al. 2010;Brcko et al. 2013;Ferrão et al. 2016;Ferrão et al. 2018). The new taxon described herein is the first species of Dendropsophus described from the Brazilian portion of the upper Madeira River, and at least one other new Dendropsophus species is awaiting formal description (work in preparation). At the same time, many of these species are highly threatened by increasing levels of deforestation caused by both illegal expansion of pastureland and infrastructure development associated with human settlements (e.g., reconstruction of the BR-319 highway and construction of large hydroelectric powerplants: Fearnside and Graça 2006;Fearnside 2015). Forests adjacent to the type locality of D. bilobatus sp. nov. were illegally deforested in 2016 and the paratype locality is now surrounded by pastures.
Based on general morphological similarity with specimens in the type series, we tentatively refer specimens from Jirau-Direito and Morrinhos, two localities in the east bank of the upper Madeira River, to D. bilobatus. However, these specimens differ from the type material in some otherwise conserved characters (e.g., dorsal skin texture and color of iris), and there are no molecular data from them that might clarify their taxonomic relationship to the sequenced type specimens. Therefore, we have chosen not to include these specimens in the type series of D. bilobatus to avoid confounding specimens of the new species with what might turn out to be a second, but undescribed, cryptic species.
Intrageneric variation in vocal sac shape is not unusual in anurans, e.g., single subgular median in Osteocephalus subtilis and O. oophagus, single subgular expanded laterally in O. vilarsi, and paired lateral in O. taurinus (Jungfer and Schiesari 1995;Jungfer et al. 2013;Ferrão et al. 2019). As a result, vocal sac shape is a widely used character in anuran taxonomy and systematics. Aside from some members of the Dendropsophus marmoratus species group, D. bilobatus is the only species in the genus to possess a bilobate subgular vocal sac. All other congeners differ in the size and extent of a single subgular vocal sac, e.g., medium size in D. bromeliaceus, large size in D. juliani, well developed and extending laterally over the forearms in D. ozzyi (Moravec et al. 2006;Orrico et al. 2014;Ferreira et al. 2015). Given that the D. marmoratus and D. microcephalus species groups are not close relatives (Faivovich et al. 2005;Jansen et al. 2019;present study), the bilobate vocal sac evolved at least twice within Dendropsophus. However, this conclusion is tentative pending a fuller evaluation of the diversity and evolutionary history of vocal sac structures in Dendropsophus in a phylogenetic context.
Intrageneric differences in vocal sac shape have been associated with different breeding strategies in some neotropical anurans. For example, a small or indistinct single subgular vocal sac in phytotelmata-associated Osteocephalus is hypothesized to be a morphological adaptation for calling in small cavities relative to the large paired sacs of pond-breeding congeners (Jungfer and Hödl 2002;Moravec et al. 2009;Jungfer et al. 2013). We have not observed, however, any unusual feature of the breeding behavior or habitat of Dendropsophus bilobatus that might explain its remarkable bilobate vocal sac versus the single sac of most of its congeners.
de Desenvolvimento Científico e Tecnológico -CNPq (PDJ process 154325/2018-0) and by an Edward O. Wilson Biodiversity Postdoctoral Fellowship from the Museum of Comparative Zoology, Harvard University. JM was financially supported by the Ministry of Culture of the Czech Republic (DKRVO 2019-2023/6.VI.b, National Museum Prague, 00023272).