The first species of Aplastodiscus endemic to the Brazilian Cerrado (Anura, Hylidae)

Abstract The genus Aplastodiscus includes 14 nominal species in four monophyletic groups with occurrence in the Atlantic Forest and Brazilian Cerrado (Brazilian Savanna) of South America. A recent study reviewed the taxonomy and phylogenetic relationships of the genus and suggested a third species for the Aplastodiscus perviridis Group. Herein, on the basis of morphology and advertisement call, we describe this species and test its monophyly. The new species is the only Aplastodiscus with endemic occurrence in the Cerrado Biome. In addition, its geographical distribution and conservation status are discussed.


Introduction
The genus Aplastodiscus includes 14 nominal species in four monophyletic groups (Berneck et al. 2016) with occurrence mainly in Atlantic Forest in Brazil and Argentina (Frost, 2016).The species are stream-breeding treefrogs, usually of green color and medium size.The A. perviridis species Group includes A. perviridis Lutz 1950 andA. cochranae (Mertens 1952), which share, among other characters, bicolored irises, no webbing between toes I and II, and reduced webbing among their remaining toes (Garcia et al. 2001).Berneck et al. (2016) recently reviewed the taxonomy and phylogenetic relationships of all Aplastodiscus species and suggested a third species for the A. perviridis Group, the "Aplastodiscus sp.3".This species was previously called A. perviridis by the previous authors (Garcia et al. 2001;Bastos et al. 2003;Morais et al. 2012;Valdujo et al. 2012).Herein, on the basis of morphology and advertisement call, this species is described as new, and its monophyly tested.In addition, its geographical distribution and conservation status are discussed.

Descriptions of adults and their calls
The following measurements follow Duellman (1970): SVL snout-vent length, HL head length, HW head width, ED eye diameter, TD tympanum diameter, END eye-nostril distance, IOD interocular distance, THL thigh length, FL foot length; including tarsus.
However, the tibia length (TBL) follows Heyer et al. (1990).Measurements are in millimeters and were taken with a digital caliper and, except for SVL, HL, HD, THL, and TBL, under a stereomicroscope.The webbing formula follows Savage and Heyer (1967) and Myers and Duellman (1982).Adult males were collected while calling and/ or recognized by the presence of vocal slits.
Males of the new species were recorded in the Municipality of Alto Paraíso de Goiás, Goiás State (N = 6) and Brasília, Distrito Federal.For comparative purposes, males of A. perviridis were recorded at the type-locality (N = 5), in Serra da Bocaina, São José do Barreiro, São Paulo State.Calls were recorded with a Marantz PMD 671, a Boss BR-864 (both with a Sennheiser ME67/K6 microphone) or a MicroTrack (ME66/ K6 microphone), all set at 44.1 kHz and 16-bit resolution.Calls were recorded from 50 to 150 cm from calling males, and 10 to 15 calls were analyzed for each male.Acoustic variables were analyzed with RAVEN PRO 1.5, 64-bit version (Bioacoustics Research Program 2014); terminology used for call features were according to Raven's manual (Charif et al. 2010).A 500 Hz high pass filter was applied prior to call analyses and figuring to reduce wind noise interference.Sound figures were obtained with the SEEWAVE 1.6.4(Sueur et al. 2008) R package (R Development Core Team, 2012, v. 2.15.1), settings used were the Hanning window, 85% overlap, and 256 points resolution.Measured call parameters were: 1) call duration (CD), 2) peak of dominant frequency (PDF), 3) lower dominant frequency (LDF), 4) higher dominant frequency (HDF), 5) time to frequency peak (TFP) (expressed as % of call duration).All calls used in descriptions are housed at the AAG-UFU collection (Suppl.material 1, Table 1).

Laboratory protocols and genetic distance calculation
The extraction of DNA was carried out using ethanol-preserved tissues and the DNeasy isolation kit (Qiagen, Valencia, CA, USA).We carried out DNA amplification in a 25 µL volume reaction using master mix Fermentas Taq Polymerase and reagents.The Polymerase chain reactions (PCR) included an initial denaturing step of 30s at 94 °C, followed by 35 cycles of amplification with a final extension step at 72 °C for 6 min.The products of PCR were sent for sequencing to Macrogen, South Korea.We sequenced DNA fragments in both directions to minimize potential errors.The chromatograms were read and edited using SEQUENCHER 3.0 (Gene Codes, Ann Arbor, MI, USA) and complete sequences were edited with MEGA 6.06 (Tamura et al. 2013).The distance estimations of genetic p-distances were calculated in MEGA 6.06 for the regions delimited by the primers 16sAR (Palumbi et al. 1991), Wilk2 (Wilkinson et al. 1996), and COI (Jungfer et al. 2013), considering d:transitions + transversions, uniform rates among sites, and gaps/missing data as complete deletion.A list of vouchers, GenBank accession numbers, and locality data is available in Suppl.material 2. Berneck et al. (2016) studied Aplastodiscus in a wider context and consequently included only one specimen of the species described here.Therefore, we carried out a reduced phylogenetic analysis that included all terminals of the A. perviridis Group analyzed by Berneck et al. (2016) and four specimens of the species described here.As outgroups, we included two terminals of the A. albosignatus Group and two of the A. albofrenatus Group, rooting the tree in the A. sibilatus Group (see Berneck et al. 2016).The dataset used for the analysis were the fragments delimited by the primers 16sAR, Wilk2, and COI forward and reverse.

Phylogenetic analysis and taxon sampling
Sequence alignments were performed in Clustawl (Thompson et al. 1994) under MEGA 6.06.For the phylogenetic analysis T.N.T Willi Hennig Society Edition was employed (Goloboff et al. 2008) with searches by "new technology", search level 50, sectorial searches included, tree drift, and tree fusing (Goloboff 1999), hitting the best length 100 times.Parsimony Jackknife absolute frequencies (Farris et al. 1996) were also estimated using "new technology" and requesting 10 hits, driven searches, and 1000 replicates.Edition of trees were made with FIGTREE (Rambaut 2014).Diagnosis.Aplastodiscus lutzorum sp.n. belongs to the A. perviridis Group and thus bears bicolored irises, lacks webbing between toes I and II, has reduced webbing among the remaining toes, and lacks peri-cloacal ornamentation.The new species is diagnosed by its small SVL for the A. perviridis Group (30-36 mm) and by the advertisement call 2.5 times longer.

Aplastodiscus
Description of holotype.Adult male: head 20% wider than long; snout rounded in profile, nearly rounded in dorsal view; canthus rostralis curved; loreal region concave; nostrils ovoid; internarial region grooved.Supratympanic fold distinct, from posterior corner of orbit to insertion of forearm; tympanum distinct, almost circular; tympanum diameter 48.5% of eye diameter.Upper eyelid smooth as the dorsum.Thoracic fold just discernible.External vocal sac single, subgular, and expanded.Fingers long, slender, no lateral fringe, bearing discret round terminal discs; relative lengths of fingers I, II, IV, III; similar sized discs on fingers II, III and IV, larger than disc of Finger I. Diameter of disc of Finger III equals to diameter of Toe IV, about 42% eye diameter.Subarticular tubercles well defined, rounded; supernumerary tubercles poorly defined on palm, rounded; inner metacarpal tubercle large, rounded, about four times the size of subarticular tubercles; other metacarpal tubercle barely defined; spine of prepollex absent; no nuptial pads; ulnar crest barely defined.Hand webbing formulae: I-II 2 --3 -III 2 + -2 IV.Tarsal texture smooth; tarsal fold discrete, extending to the entire length of tarsus; heel smooth; inner metatarsal tubercles large, elongate, three times the size of foot subarticular tubercles; outer metatarsal tubercle absent; subarticular tubercles distinct, large, and rounded, diameter about 3/4 width of terminal disc on the same toe; supernumerary tubercles absent; toes long, slender, without lateral fringe; toes bearing rounded discs, smaller in diameter to those of fingers II-IV.Foot webbing formula: I 2 + -3 -II 2 + -3 1/2 III 2 + -4 -IV 3 + -2V.Supra cloacal fold absent.Skin on dorsum smooth; skin on throat, belly, ventral surface of  2).
Color in life of the type-series: Dorsal head surface dark green, almost olive.Dorsal body surface and flanks yellowish green with small and scattered melanophores.The superior third of eye is golden, whereas the inferior 2/3 is red copper.Eye surrounded by a black ring.Vocal sac bluish green.In preservative, colors fade to pale beige and the dorsum shows several dark brown dots, making it darker than other parts of the body.The belly is uniformly pale yellow.
Variation in the type series.The main variation in type series is the body size (Table 2).Small brown chromatophores are along the dorsal skin; but the amount of these chromatophores is variable, ranging from sparse to dense.
Calls.Advertisement calls of A. lutzorum sp.n. (Figure 3 and 4, Table 1) are long regularly-spaced single notes released at a mean rate of 39 calls/minute (SD = 8; n = 12 males).Calls resemble a whistle lasting around 0.26-0.40s.Most of the energy is between 1,494-1,732 Hz and 2,334-2,647 Hz, peaking between 1,884-2,156 Hz.Call exhibits an ascending frequency modulation, reaching its maximum between 49-70% of the call duration.Aplastodiscus perviridis' advertisement call (Figure 3 and 4, Table 1) is released in groups of 1-11 (mode = 5-7); within groups, call released at a rate of 140/min.Calls resemble a short whistle lasting around 0.12-0.15s.Most of the energy is between 1,587-1,806 Hz and 2,419-2,750 Hz, peaking between 1,981-2,153 Hz.Call with an ascending frequency modulation, reaching its maximum between 23-38% of the call duration.The advertisement call of A. cochranae is described by Garcia et al. (2001) as being barely indistinguishable from the call of A. perviridis.
Natural history and geographic distribution.All specimens of Aplastodiscus lutzorum sp.n. were collected along gallery forests with scattered buriti (Mauritia flexuosa) palm trees within the Cerrado Biome (see also Brandão and Araujo, 2002;Morais et al. 2012;Santoro and Brandão, 2014) (Figure 5).A female bearing large oocytes seen by the transparency of the body walls was found in mid-December and calling males were found from December to March.Males call during the night in proximity of riverine forests (< 2m), perched on leaves or branches from the water level to 5 m high (Figure 6).Aplastodiscus perviridis males call during the night along swamps in open areas, perched on grass leaves or bushes bordering streamlets, from 0.5 m to 3 m high.Tadpoles are unknown.Sometimes, the species also uses places with wet and soft mud soil, covered by a layer of dense bush, in places where the forest was removed, but is under secondary growth.Sympatric frog species were Hypsiboas ericae (Caramaschi and Cruz, 2000) and H. albopunctatus (Spix, 1824).All localities where A. lutzorum sp.n. was found are 1000 m above sea level or more.
Etymology.The new species is named after the Brazilian scientists Adolfo and Bertha Lutz, who were pioneers in discovering and studying Aplastodiscus and some species of Hyla now included in the genus Aplastodiscus.Comparison with congeneric species.Aplastodiscus lutzorum sp.n. can be distinguished from the species of other groups of Aplastodiscus (A. albosignatus, A. albofrenatus, and A. sibilatus groups) by the presence of bicolored irises, the lack of the webbing between toes I and II, the webbing among the remaining toes reduced, and the absence of peri-cloacal ornamentation.The new species is diagnosed from A. perviridis and A. cochranae by having smaller SVL (30-36 mm SVL in the new species, 38-44.7 mm in A. perviridis, and 41-46.4mm in A. cochranae) and longer advertisement calls (0.38-0.42 in new species, 0.14-0.17 in A. perviridis and 0.10-0.18 in A. cochranae).From A. cochranae it can be also distinguished by the green dorsal color in life (A.cochranae is brown) and by the absence of a brown line on the loreal region and a white line in the cloacal region (both present in A. cochranae).(Figures 1-4; Tables 1 and 2).
Phylogenetic relationships and genetic p-distances.The two DNA fragments aligned and concatenated resulted in a matrix of 1,227pb.Our analysis recovered four most parsimonious trees with 808 steps each (Figure 7).The differences in topology among these trees are the position of A. lutzorum sp.n. specimens from different locali-  ties.Aplastodiscus lutzorum sp.n. were recovered as a monophyletic group with 100% Parsimony Jackknife Support (hereafter PJS) being the sister species of A. perviridis plus A. cochranae.The A. perviridis plus A. cochranae clade is low supported (54% PJS) and both species are supported by 99% of PJS each.The p-distances calculated for 16s of species in the A. perviridis Group range from zero to 5.9% (for all distances see Table 3).The internal distances among specimens of the new species range from zero to 0.93%.The p-distance in 16s between the new species and A. perviridis is 4.4-5.8%and between the new species and A. cochranae is 4.0-4.5%.our analysis.Berneck et al. (2016) recovered Aplastodiscus lutzorum as a sister species of A. cochranae, a topology not corroborated by the present work, where the new species is a sister species of A. cochranae plus A. perviridis.In the present work, the node of A. perviridis plus A. cochranae is supported by less than 70% of PJS while in Berneck et al. (2016) the node of A. lutzorum plus A. cochrane was poorly supported (also less than 70%).Those are possibly the reason of the instability in the internal relationships of members of the A. perviridis Group.However, our dataset and taxon sampling is very reduced in relation to that of Berneck et al. (2016) and so the analysis of these authors is preferable for relationships of Aplastodiscus species.
The scope of this paper was not to test biogeographic hypotheses.However, Aplastodiscus lutzorum is the only species of Aplastodiscus that occurs deep in the Cerrado Biome (see Silva et al. 2006) with a disjunctive distribution from its sister species of the Atlantic Forest (Valdujo et al. 2012).Therefore, it seems interesting to point out some remarks on its geographic distribution pattern (Figure 5).The new species has been reported in several localities in the Brazilian Central Plateau and our topology suggests an origin of the A. perviridis Group in the Brazilian Central Plateau (Figure 7).However, the low PJF support of the clade A. perviridis + A. cochranae and the incongruence between our topology and that of Berneck et al. (2016) make any further inference premature.
A population from the dam of Queimado in the municipality of Unaí, in the state of Minas Gerais, Brazil (an area flooded by the construction of a hydroelectric station) was included as "referred specimens" for A. lutzorum The conservation status of this population is unknown.We consider the new species to be listed as a "Least Concern", due to the fact that most of its area of occurrence is in protected places, such as the Parque Nacional da Chapada dos Veadeiros, Área de Relevante Interesse Ecológico (ARIE) do Capetinga/Taquara (Fazenda Água Limpa), Estação Ecológica de Águas Emendadas, and Floresta Nacional de Silvânia.Goin (1961) suggested that Chorophilus cuzcanus Cope, 1878 should be an Aplastodiscus, but had not stated that it was Aplastodiscus perviridis (as pointed out by Frost, 2016).Lutz (1968) suggested that C. cuzcanus was possibly a second species of Aplastodiscus at that time.According to Frost (2016), the combination Chorophilus cuzcanus is a junior synonym to both A. perviridis and Gastrotheca marsupiata (Duméril and Bibron 1841).We recognize only the synonym of Duellman and Fritts (1972) for G. marsupiata as valid; therefore A. perviridis has no junior synonyms.comments on a draft.We thank to John Karpinsk for English review.The editor A. Herrel and reviewers Helio da Silva and Ivan Nunes improved this manuscript.BVMB would like to thank: J. Faivovich for suggestions and support, D. Baêta for discussions on taxonomy, the CFBH staff from 2008-16, and the São Paulo Research Foundation (FAPESP) for grants #2008/55235-4, #2013/18807-8, and #2013/50741-7.Financial support was provided to AAG by CNPq and FAPEMIG; a grant by CNPq.AAG collection permits: IBAMA 29954-3 and IBAMA 02015.008064/02-51.Financial support to CAGC and CFBH by CNPq.

Figure 1 .
Figure 1.Holotype of Aplastodiscus lutzorum sp.n. (AAG-UFU 864).A Lateral view of head B dorsal view of head C plantar view D palmar view.Scale bar 12 mm.

Figure 3 .
Figure 3. Above, audiospectrogram and oscillogram of three advertisement calls of the holotype of Aplastodiscus lutzorum sp.n. (Chapada dos Veadeiros, 12 December 2011, air temperature 20 o C); the background calls are from another male calling in antiphony.Bellow, audiospectrogram and oscillogram of three advertisement call of A. perviridis (Serra da Bocaina, 10, January 2012, air temperature 16 o C).

Figure 4 .
Figure 4.A comparison between duration and frequency peak time of Aplastodiscus lutzorum (N = 12 males) and A. perviridis (N = 5 males).In both samples, calls are of topotypes.

Figure 6 .A
Figure 6.A The habitat of A. lutzorum sp.n.: flooded gallery forests B A male in calling activity at Fazenda Água Limpa, Brasília, Distrito Federal, Brazil.

Figure 7 .
Figure 7.One of the four most parsimonious trees recovered.Asterisks indicate 100% Parsimony Jackknife absolute frequency; only values above 75% are shown.See Suppl.material 2 for details and complete locality names.

Table 1 .
Acoustic variables of the advertisement call of topotypes Aplastodiscus perviridis and A. lutzorum sp.n. n = number of recorded males.