Notes on vocalizations of Brazilian amphibians IV: advertisement calls of 20 Atlantic Forest frog species

Bioacoustics is a powerful tool used for anuran species diagnoses, given that advertisement calls are signals related to specific recognition and mate attraction. Thus, call descriptions can support species taxonomy. In spite of that, call descriptions are lacking for many species, delaying advances in biodiversity research. Here, we describe the advertisement calls of 20 anuran species from the Brazilian Atlantic Forest. We accessed 50 digital recordings deposited in the Fonoteca Neotropical Jacques Vielliard. Acoustic analyses were carried out in the software Raven pro 1.5. We provide a general comparison of call structure among species inside taxonomic groups and genera. The vocalizations described here belong to poorly known species, which are representatives of six families: Brachycephalidae, Bufonidae, Ceratophryidae, Cycloramphidae, Hylidae, and Phyllomedusidae. Despite this, still there are 163 species of anurans from Atlantic Forest with calls not formally described. Our work represents an important step in providing data for a taxonomic perspective and improving the knowledge of the Atlantic Forest anuran diversity.


INTRODUCTION
Global biodiversity undergo a substantial crisis caused by human activities leading to a current rate of species extinctions tens to hundreds of times higher than the average across the past 10 million years (Ceballos et al., 2015;Tollefson, 2019). The research effort of

MATERIALS AND METHODS
We reviewed the literature, checking for species from the Atlantic Forest with calls already described, to produce a list of species with calls not formally described. After that, we obtained 50 audio files with advertisement calls of 20 species present in such list, by contacting researchers and surveying the files from the Fonoteca Neotropical Jacques Vielliard (FNJV, Unicamp, Campinas). These files came from different recordists, dates, localities, and equipments, which we specify in the Table 1. We identified species using specimens deposited in biological collections (Célio F.B. Haddad collection-CFBH, Museu de Zoologia João Moojen-MZUFV, Museu de História Natural Capão da Imbuia-MHNCI, Smithsonian Institution Washington-DC-US-Animalia, Museu de Zoologia da Universidade Estadual de Campinas "Adão José Cardoso"-ZUEC-AMP, Coleção de Anfíbios do Instituto Nacional da Mata Atlântica-MBML-Anfibios, Coleção de Anfíbios do Centro de Coleções Taxonômica da UFMG-UFMG-AMP, Coleção de Anfíbios do Museu de Ciências e Tecnologia da PUCRS-MCP), morphological evidence, and geographical distribution (see Table 1).
Before the acoustic analysis, we standardized all sound files to a pattern sample rate of 44.1 kHz and 16 bits of resolution saving the files in Audacity 2.1.1. For each species, we specified a band pass filter to decrease general background noise (Table 1). After the filtering process, calls were individually normalized (peak −0.8 dB) using Audacity 2.  360,363,375,445,296,332,1757,1043,1767,376,1044,359,301,288,377,442,354,1765,337,1572,14427,17874) Least concern

No
Phrynomedusa appendiculata avoiding biases related to the differences in intensity. We carried out acoustic analyses using the software Raven Pro 1.5 (Bioacoustics Research Program, 2011). For call selection, we used the waveform window. For spectral measurements, we adjusted a fast fourier transformation of 1,024 points, with a window of 50% overlap, temporal hope size of 256 samples, and grid spacing of 93.8 Hz. We used the note-centered approach (defining uninterrupted units of sound as notes and their entirety as a call) and the concepts of pulses, notes, and calls as defined by Köhler et al. (2017). Based in the terminology of Gerhardt & Huber (2002) and Wells (2007), we measured the following acoustic properties: (1) number of notes, (2) call/note duration, (3) number of pulses, (4) call rate, (5) harmonic structure, (6) rise time to the maximum amplitude, (7) range frequency, (8) minimum frequency, (9) maximum frequency, (10) fundamental frequency, and (11) dominant frequency. For such measurements we used the following functions: Bandwidth 90% (Hz) (for the range frequency, a band of frequency that includes 90% of the energy of the sound), Frequency 5% (Hz) (for the minimum frequency, ignoring 5% below the total energy in the selected call), Frequency 95% (Hz) (for the maximum frequency, ignoring 5% above the total energy in the selected call), Peak Frequency (Hz) (for dominant frequency), Delta Time (s) (for call/note duration), and Max Amplitude (U) (for finding the time to the maximum amplitude visualizing the limits in the waveform), available in the choose measurements menu in Raven (see also Köhler et al., 2017). We made descriptive statistics (mean, standard deviation, and range) based in individual measurements (by call and/or notes). When we had more than one male for each species we present mean and standard deviation based in mean values by males.

RESULTS
A total of 13 Atlantic Forest frog families have species with unknown vocalizations (Fig. 1). Among these, we identified 163 species lacking vocalizations descriptions (nearly 26% of all 624 anuran species described up to April 2019 (L.F. Toledo, 2019, unpublished data); Table S1). Below we describe calls from 20 Atlantic Forest species of six different families.
Melanophryniscus alipioi Langone, Segalla, Bornschein, and de Sá, 2008 This species has a vocalization composed of two segments consisting of series of notes with harmonics (825 ± 165, n = 18, males = 4). The first segment, which we determined as an introductory segment, has a note series composed by longer notes with larger silence intervals than the main segment (Fig. 5). Including both segments, the call duration is 17.7 ± 3.9 s (ranging from 6.7 to 25.8 s, n = 18, males = 4). Each isolated note from the introductory section has mean duration of 0.046 ± 0.012 s (ranging from 0.025 to 0.060 s, n = 18, males = 4) and silence interval between notes of 0.172 ± 0.036 s (ranging from 0.124 to 0.234, n = 18, males = 4). Notes in the main segment have a mean duration of 0.007 ± 0.001 s (ranging from 0.006 to 0.010 s, n = 18, males = 4) and silence interval between notes of 0.009 ± 0.001 s (ranging from 0.006 to 0.013 s, n = 18, males = 4). Notes in the introductory segment show a mean range frequency of 352 ± 127 Hz (n = 18, males = 4), with minimum frequency averaging 2,411 ± 200 Hz (ranging from 2,196 to 2,799 Hz, n = 18, males = 4), maximum frequency of 2,763 ± 260 Hz (ranging from 2,433 to 3,058 Hz, n = 18, males = 4), and dominant frequency of 2,664 ± 248 Hz (ranging from 2,240 to 2,929 Hz, n = 18, males = 4). The rise time to the maximum amplitude in the first segment is 0.0185 ± 0.00811 s (ranging from 0.005 to 0.036 s, n = 18, males = 4). Notes in the main segment have a mean range frequency of 561 ± 262 Hz (n = 18, males = 4), with minimum frequency averaging 2,490 ± 257 Hz (ranging from 2,182 to 2,842 Hz, n = 18, males = 4), maximum frequency of 3,051 ± 321 Hz (ranging from 2,598 to 3,531 Hz, n = 18, males = 4), and dominant frequency of 2,759 ± 229 Hz (ranging from 2,440 to 2,972 Hz, n = 18, males = 4). The rise time to the maximum amplitude in the second segment is 0.0026 ± 0.0007 s (ranging from 0.001 to 0.004 s, n = 18, males = 4). Both types of notes have harmonic structure with a fundamental frequency of the same value of dominant frequency.
Melanophryniscus moreirae (Miranda- Ribeiro, 1920) We recognized two different sections of vocalizations of M. moreirae: a non-harmonic note series (non-pulsed notes repeated in regular interval), which is more commonly emitted and another part composed by an isolated and non-pulsed harmonic note (Fig. 6). It is possible that the note series is the advertisement call, while isolated harmonic notes are aggressive calls. The note series is a repetition of 70 ± 47 notes (ranging from six to 300 notes, n = 14, males = 2), which have 1.83 ± 1.26 s (ranging from 0.21 to 6.08, n = 14, males = 2) of duration. Each note of the note series call has 0.014 ± 0.004 s (ranging from 0.010 to 0.022 s, n = 10, males = 2) of duration, while the call composed by an isolated note averages 0.086 ± 0.032 s (ranging from 0.038 to 0.140 s, n = 16, males = 2) of duration. The more intense note in the note series call vary among the 3rd and 57th unit, with rise time to the maximum amplitude achieved in 0.62 ± 0.70 s (ranging from 0.07 to 3.07 s, n = 14, males = 2). Range frequency of the note series call is 252 ± 9 Hz with minimum frequency averaging 1,733 ± 15 Hz (ranging from 1,637 to 1,809 Hz, n = 14, males = 2), maximum frequency of 1,985 ± 6 Hz (ranging from 1,938 to 2,153 Hz, n = 14, males = 2), and dominant frequency of 1,856 ± 67 Hz (ranging from 1,766 to 1,938 Hz, n = 14, males = 2). The aggressive call (isolated notes) show range frequency of 169 ± 4 Hz, with minimum frequency averaging 1,717 ± 70 Hz (ranging from 1,594 to 1,809 Hz, n = 16, males = 2), maximum frequency of 1,886 ± 74 Hz (ranging from 1,680 to 2,196 Hz, n = 16, males = 2), and dominant frequency (=fundamental frequency) of 1,809 ± 91 Hz (ranging from 1,637 to 1,895 Hz, n = 16, males = 2). The second harmonic is about 3.5 kHz in the aggressive call. The rise time to the maximum amplitude of this aggressive note is 0.025 ± 0.010 s (ranging from 0.012 to 0.049 s, n = 16, males = 2).

Melanophryniscus vilavelhensis Steinbach-Padilha, 2008
This species has the vocalization composed of two segments of non-harmonic note series (546 notes, n = 1). Notes are not pulsed. As in M. alipioi, the first segment (determined as introductory segment) has a note series composed by longer notes with larger silence intervals than the main segment (Fig. 7). Each isolated note from the introductory section has a mean duration of 0.07 ± 0.008 s (ranging from 0.06 to 0.08 s, n = 8) and silence interval between notes of 0.24 ± 0.01 s (ranging from 0.23 to 0.26, n = 7). The rise time to the maximum amplitude in this section is 6.4 s. Notes in the main segment have a mean duration of 0.013 ± 0.001 s (ranging from 0.02 to 0.03 s, n = 8) and silence interval between notes of 0.03 ± 0.002 s (ranging from 0.02 to 0.03 s, n = 7). Notes in the introductory segment show a range frequency of 86.1 Hz, with minimum frequency averaging 3,182 ± 36 Hz (ranging from 3,144 to 3,230 Hz, n = 8), maximum frequency of 3,268 ± 36 Hz (ranging from 3,230 to 3,316 Hz, n = 8), and dominant frequency of 3,230 ± 40 Hz (ranging from 3,187 to 3,273 Hz, n = 8). The first segment notes in the main segment have a range frequency of 188 Hz, with minimum frequency averaging 3,128 ± 22 Hz (ranging from 3,101 to 3,144 Hz, n = 8), maximum frequency of 3,316 ± 0 Hz (n = 8), and dominant frequency of 3,230 ± 0 Hz (n = 8). The rise time to the maximum amplitude in the main section is 10.6 s. Notes do not have harmonic structure and the rise time to the maximum amplitude considering segments combined is 17.5 s.

CERATOPHRYIDAE
Ceratophrys aurita (Raddi, 1823) The vocalization of Ceratophrys aurita is a single note with a mean of 149 ± 19 fused pulses (ranging from 80 to 172, n = 25). Each call has an average of 0.87 ± 0.09 s (ranging from 0.54 to 0.98 s, n = 25), repeated in an interval of 1.23 ± 1.11 s (ranging from 0.11 to 4.95 s, n = 25), and call rate of 27 calls/min. Notes have an unstable spectral modulation, with upward and downward pattern along the call (Fig. 8). Despite some weak sidebands being visible using lower contrasts in spectrogram window, the range frequency is not so large, occupying 479 ± 129 Hz, with minimum frequency averaging 1,044 ± 107 Hz (ranging from 603 to 1,163 Hz, n = 25), maximum frequency of 1,523 ± 46 Hz (ranging from 1,421 to 1,550 Hz, n = 25), and peak dominant frequency of 1,261 ± 69 Hz (ranging from 1,163 to 1,421 Hz, n = 25). The rise time to the maximum amplitude is 0.39 ± 0.18 s (ranging from 0.06 to 0.70 s, n = 25).

Cycloramphus granulosus Lutz, 1929
This species has a vocalization composed of a single pulsed and harmonic note. The note duration is 1.16 ± 0.25 s (ranging from 0.97 to 1.44 s, n = 3), and the number of pulses varied among 25 and 33 pulses. The harmonic structure appears more clearly related to the more intense pulses (Fig. 9). This call has a large range frequency of 1,321 ± 66 Hz, which is particularly affected by the energy distribution between the first (fundamental) and second harmonics. Minimum frequency averaged 804 ± 25 Hz (ranging from 775 to 818 Hz, n = 3), maximum frequency was 2,125 ± 66 Hz (ranging from 2,067 to 2,196 Hz, n = 3), and Cycloramphus izecksohni Heyer, 1983 The recording of Cycloramphus izecksohni was obtained based in a male recorded inside a plastic bag. The vocalization is a single note with one to three pulses. Sometimes two notes are quickly repeated (interval of 0.1 s). The call duration is 0.28 ± 0.05 s (ranging from 0.17 to 0.36 s, n = 21). The call rate was 20 calls/min, repeated with an interval of 2.25 ± 1.01 s (ranging from 0.09 to 4.80 s, n = 21). A harmonic structure is present with a large spectral interval between the first and the second harmonics (Fig. 10). The dominant frequency is the first harmonic (= fundamental frequency), which has a peak of 1,140 ± 265 Hz (ranging from 861 to 1,594 Hz, n = 21). The range frequency may be excessively high considering the harmonic distribution, which includes 6,936 ± 2,573 Hz. Minimum frequency averages 681 ± 216 Hz (ranging from 43 to 818 Hz, n = 21) and maximum frequency is 7,617 ± 2,649 Hz (ranging from 2,153 to 10,422 Hz, n = 21). The second harmonic is up to eight kHz. The rise time to the maximum amplitude is 0.09 ± 0.05 s (ranging from 0.03 to 0.21 s, n = 21).
Boana leptolineata (Braun & Braun, 1977) We identified two different notes in the vocalization of Boana leptolineata. Note "A" is a train of fused pulses (varying from two to 14), sounding as a trill, while the note "B" is a sequence of pulses (varying from four to 50) with discrete silence interval, sounding as many click-like units (Fig. 13). Both notes have harmonic structure. These notes are frequently combined in a sequence, composing complex calls. It is possible that these notes have different social functions. Note "A" duration is 0.06 ± 0.02 s (ranging from 0.02 to 0.16 s, n = 122, males = 4), with rise time to the maximum amplitude of 0.016 ± 0.005 s (ranging from 0 to 0.10 s, n = 122, males = 4  The vocalization of Bokermannohyla gouveai is a single harmonic note composed by five to 15 pulses (Fig. 14). Call duration is 0.42 ± 0.12 s (ranging from 0.27 to 0.80 s, n = 31), with rise time to the maximum amplitude of 0.18 ± 0.07 s (ranging from 0 to 0.34 s, n = 31). The recorded male emitted a sequence of calls at a rate of 24 calls/min. The range frequency is 860 ± 119 Hz, with minimum frequency averaging 596 ± 140 Hz (ranging from 431 to 1,077 Hz, n = 31), maximum frequency of 1,456 ± 75 Hz (ranging from 1,335 to 1,594 Hz, n = 31), and dominant frequency (= fundamental frequency) of 1,127 ± 201 Hz (ranging from 560 to 1,378 Hz, n = 31). The second harmonic is about 2,800 Hz.
Ololygon flavoguttata (Lutz & Lutz, 1939) We found two different vocalizations: (1) a sequence of harmonic notes (click-like) with a discrete interval and (2) a non-harmonic note composed by fused pulses (Fig. 15). We suggest that the first vocalization is the advertisement call, while the note composed by fused pulses is aggressive. These vocalizations may be emitted isolated or in combination (mixed calls). When combined, the advertisement call always anticipates the aggressive call. In these occasions, the interval between call types is 0.77 ± 0.34 s (ranging from 0.48 to 1.67 s, n = 29). The recorded male has a call rate of nine calls/min. The advertisement call has a variation of one to 25 pulses and duration of 1.70 ± 1.07 s (ranging from 0.03 to 3.66 s, n = 19), with rise time to the maximum amplitude of 1.42 ± 0.99 s (ranging from 0.01 to 3.13 s, n = 19). The range frequency in advertisement calls is 1,763 ± 719 Hz, with minimum frequency averaging 2,149 ± 222 Hz (ranging from 1,766 to 2,541 Hz, n = 19), maximum frequency of 3,912 ± 523 Hz (ranging from 3,273 to 5,039 Hz, n = 19), and peak dominant frequency (= fundamental frequency) of 2,643 ± 146 Hz (ranging from 2,283 to 2,972 Hz, n = 19). The second harmonic is up to four kHz. The aggressive call has a variation of one to seven pulses and duration of 0.19 ± 0.07 s (ranging from 0.08 to 0.41 s, n = 22), with rise time to the maximum amplitude of 0.14 ± 0.08 s (ranging from 0.05 to 0.38 s, n = 22). The range frequency of aggressive calls is 1,159 ± 250 Hz, with minimum frequency averaging 2,300 ± 144 Hz (ranging from 2,110 to 2,584 Hz, n = 22), maximum frequency of 3,459 ± 128 Hz (ranging from 3,230 to 3,704 Hz, n = 22), and peak dominant frequency of 2,786 ± 192 Hz (ranging from 2,498 to 3,230 Hz, n = 22).

Phyllomedusa iheringii Boulenger, 1885
The vocalization of Phyllomedusa iheringii is a single note composed of seven to 36 non-fused pulses (Fig. 20). Call duration is 0.80 ± 0.59 s (ranging from 0.27 to 1.87 s, n = 18), with rise time to the maximum amplitude of 0.27 ± 0.26 s (ranging from 0.08 to 1.06 s, n = 18). The recorded male has a call rate of six calls/min (n = 1). The range frequency is 1,074 ± 219 Hz, with minimum frequency averaging 975 ± 41 Hz (ranging from 818 to 991 Hz, n = 18), maximum frequency of 2,031 ± 196 Hz (ranging from 1,766 The vocalization of Pithecopus rusticus has two acoustic units: (1) a short pulsed note with duration of 0.045 ± 0.002 s and (2) a long pulsed note with duration of 1.196 ± 0.308 s (Fig. 21). These notes may be combined or emitted in an isolated way. When combined the interval between notes is 4 ± 0.87 s (n = 17, males = 3). The mean call rate was 14 ± 2 calls/min (males = 3). The short note has two to three pulses with rise time to the maximum amplitude of 0.011 ± 0.008 s (ranging from 0.003 to 0.036 s, n = 17, males = 3) and range frequency of 900 Hz, with minimum frequency of 1,112 ± 81 Hz (ranging from 302 to 1,249 Hz, n = 17, males = 3), mean maximum frequency of 2,012 ± 8 Hz (ranging from 1,981 to 2,110 Hz, n = 17, males = 3), and a peak dominant frequency of 1,623 ± 64 Hz (ranging from 1,249 to 1,852 Hz, n = 17, males = 3). The long note has 50-80 pulses with rise time to the maximum amplitude of 1.126 ± 0.300 s (ranging from 0.784 to 1.345 s, n = 3) and range frequency of 631 Hz, with minimum frequency of 1,363 ± 108 Hz (ranging from 1,249 to 1,464 Hz, n = 3), mean maximum frequency of 1,995 ± 138 Hz (ranging from 1,895 to 2,153 Hz, n = 3), and peak of dominant frequency of 1,751 ± 49 Hz (ranging from 1,723 to 1,809 Hz, n = 3).

DISCUSSION
About half of the Brazilian amphibians occur in the Atlantic Forest (Toledo & Batista, 2012;Haddad et al., 2013;Segalla et al., 2014). Even though the rate of call descriptions has increased exponentially in the last years, a recent review (Guerra et al., 2018) showed that most Brazilian species with undescribed advertisement calls are concentrated in the Amazon Basin and mainly in the Atlantic Forest. Many species have restricted distribution or are rare (Toledo et al., 2014) and these factors together potentially affect the high number of species that remain with calls to be described. Below we present a general view about the acoustics knowledge inside each genus or species group with call described in this paper:

Ischnocnema lactea species group
Ischnocnema concolor and I. melanopygia are members of the I. lactea species series (Padial, Grant & Frost, 2014;Taucce et al., 2018), and both species have calls with harmonic notes, differing strictly in note duration and spectral band. While some species as I. concolor and I. vizitoi have calls composed only by one non-pulsed harmonic note, I. melanopygia may have calls formed by a sequence of three to five of these notes (Martins & Haddad, 2010). Calls of I. lactea are composed by one multipulsed note (Silva-Soares et al., 2018) and other species of the species series, as I. nigriventris and I. randorum have calls composed by more than one note with no apparent harmonics (Berneck, Targino & Garcia, 2013;Heyer et al., 1990). Rocha et al. (2017) and Silva-Soares et al. (2018) provided a detailed comparison of acoustic properties among species.

Genus Dendrophryniscus
The genus Dendrophryniscus is composed of 16 species, occurring in the Atlantic Forest (Frost, 2019). All of them with undescribed calls. Like other congeneric species, D. berthalutzae is a bromeliad phytotelmata specialist (Malagoli et al., 2017). Further effort should be employed to obtain recordings of different species trying to improve the taxonomic resolution for this genus. As it is difficult to record this species, maybe the use of autonomous recorders can contribute for obtaining such data. The sequence of pulsed notes that compose the vocalization in D. berthalutzae can be a conserved feature of the Bufonidae family (Alonso & Rodríguez, 2003;Martin, 1972).

Genus Melanophryniscus
The vocalizations of M. alipioi, M. moreirae, and M. vilavelhensis are equally complex, with two different types of notes. As observed for M. moreirae, many species of the genus combine isolate notes with a sequence of notes (Duré, Schaefer & Kehr, 2015). The social function of these call sections is unknown, but it is possible that they are advertisement and aggressive signals. Playback experiments should be performed to elucidate such functions.
However, M. alipioi and M. vilavelhensis have two different note sequences (segments) mainly determined by variations in silence interval, being that the introductory segment has notes with longer duration and a larger interval among them than the notes in the main segment. This configuration is similar to M. atroluteus, M. pachyrhynus, M. krauczuki, M. montevidensis, and M. dorsalis (Caldart, dos Santos & Maneyro, 2013).

Genus Ceratophrys
Among the Neotropical horned frogs, Ceratophrys is the most diverse genus of the family Ceratophryidae (Frost, 2019); however, they are the less known regarding acoustic descriptions (Lescano, 2011;Zaidan & Leite, 2012). Vocalizations composed by a single note with multiple pulses seem to be a common feature for the Ceratophryidae family (Lescano, 2011;Zaidan & Leite, 2012). The vocalization of Ceratophrys aurita has the lowest frequency among the species of the genus with calls already described (see a complete comparison of acoustic traits in Zaidan & Leite (2012)).

Genus Cycloramphus
Vocalizations in the genus Cycloramphus may be emitted in three configurations regarding notes and pulses organization: (1) one non-pulsed note, (2) one pulsed note, and (3) a sequence of unpulsed or pulsed notes (Lingnau et al., 2008;Lima et al., 2010). Cycloramphus granulosus and Cycloramphus izecksohni emits a type (2) call, however, the second species can combine two pulsed notes as a call unit. Many species of this genus reproduce in small waterfalls in the Atlantic Forest (Heyer, 1983) and have to deal with an intense low frequency background noise. It is possible that spectral call traits have been modulated by such environmental condition and this should be a subject of interest in future research involving this taxonomic group. Lima et al. (2010) presented a detailed comparison of calls traits among different species.

Genus Bokermannohyla
The genus Bokermannohyla has 32 species, 19 of them belonging to the Bokermannohyla circumdata species group (Frost, 2019). Despite recent effort describing several new species in the last 10 years, this group needs more attention to improve its taxonomic resolution by an integrative view (Faivovich et al., 2005). Among the Atlantic Forest species of the Bokermannohyla circumdata species group, only Bokermannohyla caramaschii remains with undescribed calls, since Bokermannohyla izecksohni is considered voiceless (Toledo et al., 2014). Vocalizations in this group of species may vary between isolated pulsed notes to complex calls with different notes (Gaiga et al., 2013). Similar to the vocalization of Bokermannohyla gouveai are the calls of Bokermannohyla circumdata, which emit a single and harmonic note (De Carvalho, Giaretta & Magrini, 2012). Many other species of the group, as Bokermannohyla astartea, Bokermannohyla luctuosa, and Bokermannohyla nanuzae have two types of notes in their calls (Heyer et al., 1990;Napoli & Caramaschi, 2004;De Carvalho, Giaretta & Magrini, 2012), possibly with different social functions. Gaiga et al. (2013) provide a thorough comparison of acoustic traits among species.

Ololygon catharinae species group
Many species of the O. catharinae group may show mixed calls (possible advertisement + aggressive calls) (Hepp, Lourenço & Pombal, 2017), as we describe for O. flavoguttata and O. tripui. Species may have calls composed by click-like, long and short squawk-like notes as classified by Hepp, Lourenço & Pombal (2017). Squawk-like notes, supposedly phylogenetically conserved among Ololygon species (Bang & Giaretta, 2017), were not identified in the vocalization of O. flavoguttata and O. tripui. Considering the O. catharinae species group restricted to the Atlantic Forest, now the calls of 21 species are formally described (Hepp, Lourenço & Pombal, 2017;present Napoli & Fonseca, 2008), and Phasmahyla lisbella (Pereira et al., 2018). Calls of these three species are very similar to those of Phasmahyla cochranae and Phasmahyla jandaia. All five species have calls composed by a single note with non-fused pulses (Cruz, Napoli & Fonseca, 2008;Dias et al., 2011;Pereira et al., 2018). Phasmahyla jandaia has the longest notes (achieving 0.09 s), while Phasmahyla spectabilis and Phasmahyla timbo have notes around 0.03 s (Cruz, Napoli & Fonseca, 2008;Dias et al., 2011). Phasmahyla lisbella shows high variation in note duration, from 0.007 to 0.087 s (Pereira et al., 2018). Probably this difference is due to the larger number of pulses in Phasmahyla jandaia than in other species. Harmonic structure is only visible in Phasmahyla cochranae. Dominant frequency in all species ranges between 1,700 and 2,200 Hz. Phasmahyla timbo has a call with lower frequencies compared to the other species. New recording efforts should be made to obtain recordings of the other species without calls described, as Phasmahyla exilis and Phasmahyla guttata.

Genus Phrynomedusa
The current knowledge of acoustics in this genus, before our work, was limited to two species (out of six): Phrynomedusa marginata (Weygoldt, 1991) and Phrynomedusa dryade (Baêta et al., 2016). The last species was recently described by Baêta et al. (2016), but Weygoldt (1991) presented only superficial acoustic data for a male Phrynomedusa marginata recorded from a terrarium. Summed with Phrynomedusa appendiculata both species have calls composed by a single note with short duration (Weygoldt, 1991), while Phrynomedusa dryade has longer calls composed by a series of pulsed notes (Baêta et al., 2016). Call descriptions still remain to be known for Phrynomedusa bokermanni, Phrynomedusa fimbriata, and Phrynomedusa vanzolini, although, Phrynomedusa fimbriata is considered as an extinct taxon (IUCN, 2017).

Phyllomedusa burmeisteri species group
The Phyllomedusa burmeisteri group is represented by five species (Faivovich et al., 2010) and the vocalizations, now (including Phyllomedusa iheringii), are totally described. Calls composed by a single note with non-fused pulses are common to all species (Haddad, Pombal & Batistic, 1994;Silva-Filho & Juncá, 2006). Phyllomedusa iheringii has longer calls (0.80 s) than other species, which vary among 0.20 and 0.40 s (Haddad, Pombal & Batistic, 1994;Silva-Filho & Juncá, 2006). According to Haddad, Pombal & Batistic (1994), advertisement calls of Phyllomedusa distincta and Phyllomedusa tetraploidea are spectrally indistinguishable, they occupy a range frequency of 700-2,500 Hz, making the bioacoustics an apparently weak feature for species recognition (Köhler et al., 2017). Possibly, this high similarity in sexual signals among different species promote extensive cases of hybridization between these sympatric species (Haddad, Pombal & Batistic, 1994). De Andrade et al. (2018) compared calls of Phyllomedusa burmeisteri and Phyllomedusa bahiana and also defined that calls of these species cannot be distinguished by qualitative or quantitative acoustic properties. However, a focused and standardized study based in a robust data set for comparing the vocalizations among species of the Phyllomedusa burmeisteri group is still necessary for a more reliable understanding.

Genus Pithecopus
The genus Pithecopus comprises 11 species, with only three occurring in the Atlantic forest: Pithecopus nordestinus, Pithecopus rohdei, and Pithecopus rusticus (Frost, 2019). All species belong to the Pithecopus hypochondrialis group (Faivovich et al., 2005). Males of many species in this taxonomic group have calls composed by two different notes as described in Pithecopus rusticus, then an acoustic repertoire with short and long pulsed notes are exhibited by Pithecopus ayeaye, Pithecopus azureus, Pithecopus centralis, Pithecopus hypochondrialis, Pithecopus nordestinus, and Pithecopus rohdei (Guimarães et al., 2001;Wogel, Abrunhosa & Pombal, 2004;Brandão et al., 2009;Vilaça, Silva & Solé, 2011;Nali, Borges & Prado, 2015;Haga et al., 2017b). It is possible that these notes have different social function, with the short note being an advertisement signal and long notes emitted in an aggressive context. However, we suppose that such assumptions still should be tested using playback experiments. Advertisement calls of Pithecopus araguaius have only one acoustic unit as an isolated pulsed note, similar to the longer note by Pithecopus rusticus, but with less pulses (five to eight pulses) (Haga et al., 2017a). In Pithecopus palliatus the advertisement call is one or two notes with indistinct pulses (Köhler & Lötters, 1999). Calls of Pithecopus megacephalus and Pithecopus oreades remain undescribed.

CONCLUSION
Our work extends the acoustic knowledge for anuran species from the Atlantic Forest, describing the vocalization of 20 species. Despite such progress, a further effort increasing the sample of recorded males for species represented by only one male in our analysis should improve the perception of call variation in these species.
Descriptions of hylid calls have been the focus of many recent papers, probably because it is the most diverse family of frogs in the Atlantic Forest. However, the families Phyllomedusidae and Cycloramphidae were well represented in our results despite the fact that these families are not as diverse as Hylidae. Despite there being 163 species of Atlantic anurans with calls not described yet, our work represents an important step in providing data for an integrative taxonomy and the best knowledge of such rich biodiversity. A future geographical analysis linking the distributions of these species should be helpful, which may point us in new directions to reduce this gap. Finally, we argue that sound files should always be deposited in sound archives, in order to promote the rapid access to such biodiversity component, neglected even by nowadays taxonomists (Toledo, Tipp & Márquez, 2015).