IntraspecIfIc varIatIon In acoustIc traIts and body sIze , and new dIstrIbutIonal records for PseudoPaludicola giarettai carvalho , 2012 ( anura , leptodactylIdae , leIuperInae ) : ImplIcatIons for Its congenerIc dIagnosIs

In this paper, we provide an updated diagnosis for Pseudopaludicola giarettai based on the morphometric and acoustic variation observed with the assessment of new populations, plus an expansion of its distribution range. Our results support that all acoustic variation observed might be attributed to intraspecific variation. The variation in body size and dorsal stripe patterns observed for Pseudopaludicola giarettai reinforces that the distinctive whistling advertisement call pattern is the most reliable evidence line to diagnose it from its congeners, whereas morphological (robust body, glandular dorsum) and morphometric (body size) features vary considerably within and among populations so that they should no longer be employed as diagnostic features of Pseudopaludicola giarettai. Key-Words: Bioacoustics; Minas Gerais; Taxonomy. 1971, 1989). The monophyly of the genus was later assessed through phylogenetic analyses based on external morphology and osteology (Lobo, 1995), and, more recently, molecular data (Veiga-Menoncello et  al., 2014). In both Lobo’s (1995) and Veiga-Menoncello et al.’s (2014) hypotheses, a few species were INTRODUCTION The genus Pseudopaludicola currently comprises 18 species distributed throughout South America (Frost, 2014). This genus was traditionally defined by external morphology and osteological features (Lynch http://dx.doi.org/10.1590/0031-1049.2015.55.## Volume 55(##):A‐J, 2015 1. Universidade Federal de Uberlândia – UFU, Laboratório de Taxonomia, Sistemática e Ecologia Comportamental de Anuros Neotropicais, Faculdade de Ciências Integradas do Pontal. Rua Vinte, 1.600, Bairro Tupã, CEP 38304-402, Ituiutaba, MG, Brasil. 2. Universidade de São Paulo – USP, Faculdade de Filosofia Ciências e Letras de Ribeirão Preto, Departamento de Biologia, Programa de Pós-Graduação em Biologia Comparada. Avenida dos Bandeirantes, 3.900, Vila Monte Alegre, CEP 14040-901, Ribeirão Preto, SP, Brasil. 3. Universidade Federal de Viçosa – UFV, Departamento de Biologia Animal, Museu de Zoologia João Moojen. Avenida Peter Henry Rolfs, s/no, Campus Universitário, CEP 36570-000, Viçosa, MG, Brasil. 4. E-mail to corresponding author: thiago_decarvalho@yahoo.com.br not tested for their phylogenetic position, including P. giarettai Carvalho, 2012, which possesses a unique, whistled advertisement call pattern in the genus, associated with a large size, robust body, warty dorsum, yellow vocal sac (Carvalho, 2012), a suite of features that is apparently more related to the “2n = 20” clade from Veiga-Menoncello et  al. (2014), formed by P. ameghini Pansonato, Strüssman, Mudrek & Martins, 2013 and P. ternetzi Miranda-Ribeiro, 1937. During recent field surveys (2013/2014) in the Cerrado of southeastern Brazil, we recorded and collected specimens of Pseudopaludicola assigned in the field to P.  giarettai through its distinctive advertisement call pattern. We herein provide an updated diagnosis for P. giarettai based on the morphometric and acoustic variation observed with the assessment of new populations, plus an expansion of its distribution range. MATERIAL AND METHODS Field work was conducted in six municipalities of Minas Gerais State, southeastern Brazil: Buritis (15°21’29”S, 46°29’45”W; approx. 905 m a.s.l.); Buritizeiro (16°52’32”S, 45°05’02”W; approx. 700 m a.s.l.); Chapada Gaúcha, at the Parque Nacional Grande Sertão Veredas (15°10’51”S, 45°46’10”W; approx. 710  m  a.s.l.); Coromandel (18°07’35”S, 47°11’60”W; approx. 730  m  a.s.l.); Curvelo (18°46’09”S, 44°26’54”W; approx. 620 m a.s.l.); Unaí (16°43’33”S, 46°29’30”W; approx. 525 m a.s.l.). Examined specimens are listed in Appendix 1. Calls were recorded using digital equipment (M-Audio Microtrack II, and Marantz PMD 671) set at a sampling rate of 48.0 kHz and a resolution of 16 bits (mono WAVE file format), coupled to directional microphones (Sennheiser K6/ME66 and K6/ME67, respectively). At two study sites (Buritizeiro and Unaí), calls were recorded using an Olympus DM-420 Digital Voice Recorder coupled to an Audio-Technica Pro stereo condenser microphone set at a sampling rate of 32 kHz and a resolution of 24 bits (stereo WMA file format). Sound files were converted to mono WAVE file format prior to perform any sound analysis. Calls from two municipalities were provided by Diego J. Santana: i) Muriaé (Minas Gerais State; see Santana et al., 2009), and ii) Flores de Goiás (Goiás State); recordings perfectly fitted P. giarettai distinctive calling pattern, so both populations were assigned by us to P. giarettai, but we excluded these data from acoustic comparisons since just a few calls of one male from each site were recorded. Calls were analyzed with the software Raven Pro version 1.5, 32-bit version (Bioacoustics Research Program, 2012). Temporal and spectral traits were measured from spectrograms; temporal traits were manually measured, and spectral traits (dominant and other frequencies) were obtained with the Peak Frequency measurement. Raven Pro settings: window type = Hann; window size = 1024 samples; 3 dB filter bandwidth = 67.4 Hz; brightness = 50%; contrast = 50%; overlap (locked) = 85%; hop size (temporal resolution = 3.21 ms); DFT size (locked) = 1024 samples (spectral resolution  = 46.9  Hz). Calls of P.  giarettai from its original description (Carvalho, 2012) were reanalyzed under the analytical standards and software settings presented earlier. Acoustic terminology is defined in Table  1. Mean and standard deviation (SD) given along the text and tables were calculated from individual mean values. The overall thermal dependency of temporal variables of the advertisement call of P.  giarettai was performed in two steps using Vegan package (Oksanen et  al., 2013) version 2.0-10, R  platform (R Core Team, 2014): a Principal Component Analysis (PCA) was performed on a correlation matrix from untransformed individual mean values. Then, a correlation analysis between recording air temperature and the first component scores (PC1) was performed to identify potential temperature-dependent variables. This statistical approach was not likewise applied to calling male size (SVL) dependency of spectral traits of its call due to the lack of either a minimum sample size or a non-correspondence of voucher males and recordings. See Appendices 2-3 for the first three principal component scores, eigenvalues, eigenvectors, proportional explained variation, and mean individual values. Further, the potential association between linear geographic distance and variation in acoustic traits was assessed through a Mantel test applied on Euclidean distance matrices with 10,000 Monte Carlo permutations, using ade4 package (Dray & Dufour, 2007) version 1.6-2, R platform. Specimens are housed in the Collection of Anurans of the Museu de Biodiversidade do Cerrado, Universidade Federal de Uberlândia (AAG-UFU), Municipality of Uberlândia, and at the Museu de Zoologia João Moojen, Universidade Federal de Viçosa (MZUFV), Municipality of Viçosa, both in Minas Gerais, Brazil. Call voucher males were registered under the following accession numbers: AAG-UFU 1772 (Buritis), AAG-UFU 1920-1924 (Chapada Gaúcha), AAG-UFU 3565 (Coromandel), AAG-UFU 0309-0313 (Curvelo), MZUFV 14207, 14210 (Buritizeiro), MZUFV 14878, 14882 (Unaí). Carvalho, T.R. et al.: Intraspecific variation and distribution of PseudoPaludicola giarettai B


INTRODUCTION
The genus Pseudopaludicola currently comprises 18 species distributed throughout South America (Frost, 2014).This genus was traditionally defined by external morphology and osteological features (Lynch not tested for their phylogenetic position, including P. giarettai Carvalho, 2012, which possesses a unique, whistled advertisement call pattern in the genus, associated with a large size, robust body, warty dorsum, yellow vocal sac (Carvalho, 2012), a suite of features that is apparently more related to the "2n = 20" clade from Veiga-Menoncello et al. (2014), formed by P. ameghini Pansonato, Strüssman, Mudrek &Martins, 2013 andP. ternetzi Miranda-Ribeiro, 1937.During recent field surveys (2013/2014) in the Cerrado of southeastern Brazil, we recorded and collected specimens of Pseudopaludicola assigned in the field to P. giarettai through its distinctive advertisement call pattern.We herein provide an updated diagnosis for P. giarettai based on the morphometric and acoustic variation observed with the assessment of new populations, plus an expansion of its distribution range.
Calls were recorded using digital equipment (M-Audio Microtrack II, and Marantz PMD 671) set at a sampling rate of 48.0 kHz and a resolution of 16 bits (mono WAVE file format), coupled to directional microphones (Sennheiser K6/ME66 and K6/ME67, respectively).At two study sites (Buritizeiro and Unaí), calls were recorded using an Olympus DM-420 Digital Voice Recorder coupled to an Audio-Technica Pro stereo condenser microphone set at a sampling rate of 32 kHz and a resolution of 24 bits (stereo WMA file format).Sound files were converted to mono WAVE file format prior to perform any sound analysis.Calls from two municipalities were provided by Diego J. Santana: i) Muriaé (Minas Gerais State; see Santana et al., 2009), and ii) Flores de Goiás (Goiás State); recordings perfectly fitted P. giarettai distinctive calling pattern, so both populations were assigned by us to P. giarettai, but we excluded these data from acoustic comparisons since just a few calls of one male from each site were recorded.
Calls were analyzed with the software Raven Pro version 1.5, 32-bit version (Bioacoustics Research Program, 2012).Temporal and spectral traits were measured from spectrograms; temporal traits were manually measured, and spectral traits (dominant and other frequencies) were obtained with the Peak Frequency measurement.Raven Pro settings: window type = Hann; window size = 1024 samples; 3 dB filter bandwidth = 67.4Hz; brightness = 50%; contrast = 50%; overlap (locked) = 85%; hop size (temporal resolution = 3.21 ms); DFT size (locked) = 1024 samples (spectral resolution = 46.9Hz).Calls of P. giarettai from its original description (Carvalho, 2012) were reanalyzed under the analytical standards and software settings presented earlier.Acoustic terminology is defined in Table 1.Mean and standard deviation (SD) given along the text and tables were calculated from individual mean values.
The overall thermal dependency of temporal variables of the advertisement call of P. giarettai was performed in two steps using Vegan package (Oksanen et al., 2013) version 2.0-10, R platform (R Core Team, 2014): a Principal Component Analysis (PCA) was performed on a correlation matrix from untransformed individual mean values.Then, a correlation analysis between recording air temperature and the first component scores (PC1) was performed to identify potential temperature-dependent variables.This statistical approach was not likewise applied to calling male size (SVL) dependency of spectral traits of its call due to the lack of either a minimum sample size or a non-correspondence of voucher males and recordings.See Appendices 2-3 for the first three principal component scores, eigenvalues, eigenvectors, proportional explained variation, and mean individual values.Further, the potential association between linear geographic distance and variation in acoustic traits was assessed through a Mantel test applied on Euclidean distance matrices with 10,000 Monte Carlo permutations, using ade4 package (Dray & Dufour, 2007)

Advertisement call redescription (type locality: Curvelo, Minas gerais)
Eight males were recorded (N = 400 calls).Advertisement call (Table 4; Figs.3-4) consists of a whistle, i.e. non-pulsed signal, with a remarkable ascendant frequency modulation throughout its duration.Calls may have a slight upward amplitude modulation in their half or last third duration (see oscillogram in Fig. 4) or upward and downward amplitude modula-tions in their first third or half duration (see oscillogram in Fig. 3).Fundamental frequency (1 st harmonic) always corresponds to the carrier frequency, and up to four higher harmonically related frequencies can be observed.

Acoustic comparisons
Despite the advertisement call pattern shared among all populations, i.e. non-pulsed signal with notable ascendant frequency modulation throughout its duration, dominant frequency coincident with fundamental frequency, and higher harmonically related frequencies typically observed, quantitative (temporal and spectral) traits of P. giarettai call varied within and among populations (Table 4; Fig. 5).The first three PC axes explained approximately 87% of the overall variation.Type specimens and from Buritizeiro were partially separated from the other populations along PC1 (ca.55% explained variation) mainly by temporal traits of call (higher call rate, and shorter call duration and interval), as well as faster call rise time, whereas type specimens and from Chapada Gaúcha were completely separated along PC2 (ca.21% explained variation) from specimens from Buritizeiro and Unaí, mainly by a wider frequency modulation.
Considering acoustic variation represented in Fig. 5, it is possible to assume that the between-population variation was not great enough to discriminate any as a potential independent group with respect to acoustic information, thus we attributed it to intraspecific variation only.Populations from Coromandel, Buritis, and Unaí showed within-population acoustic variation that should possibly be explained by an insufficient sample size of recorded males (N = 2 males).The Mantel test showed no significant association between geographic distance and acoustic variation (Mantel r = -0.04;p = 0.77).

Effect of temperature on temporal traits of call
Air temperature could not explain the overall variation observed in all three temporal traits analyzed of P. giarettai call (F 1,24 = 3.61; df = 24; p = 0.07).

Notes on natural history
Males tend to call during daytime and decrease or cease call activity after nightfall.Males call on the margins of artificial ponds or slow-flowing streamlets with clean water and muddy/sandy bottom, either on water surface among grassy vegetation or at their border on the ground, almost always in association with Buriti (Mauritia flexuosa) palm grove marshes (Fig. 6).

DISCUSSION
A better sample of calls from each population should be assessed prior to any assumption on the association of temperature and temporal traits of P. giarettai call, as well as to test the effect of water temperature on them, given that this species may call on water surface in association with vegetation.Although the exploratory test (PCA) recovered acoustic variation both within and among populations of P. giarettai, it was insufficient to raise doubts on the assignment of all populations enclosed in a single taxonomic unit.Thus, our results support that all acoustic variation observed might be attributed to intraspecific variation.
The remarkable morphological and morphometric intraspecific variation observed for P. giarettai reinforces that the distinctive advertisement call pattern is the most straightforward evidence line, perhaps the unique unambiguous character, to diagnose it from its congeners, whereas morphological (robust body, glandular dorsum) and morphometric (body size) features show such variation, especially for non-topotypical populations, so that they should no longer be employed as reliable diagnostic features for P. giarettai.Still, the color feature of yellow vocal sac was observed in all specimens of P. giarettai, which might be helpful to distinguish this species besides its distinctive whistling advertisement call.
Morphological/morphometric features might have little contribution to assess specific identity of Pseudopaludicola, whereas acoustic information has revealed cryptic diversity and has contributed to the reassessment of taxonomic status within this Neotropical frog group.In the last four years, six species were described (in chronological order: P. giarettai, P. hyleaustralis, P. facureae, P. Parnaiba, P. pocoto, and P. atragula ;Carvalho, 2012;Pansonato et al., 2012Pansonato et al., , 2014;;Andrade & Carvalho, 2013;Roberto et al., 2013;Magalhães et al., 2014), one revalidated (P.ameghini; Pansonato et al., 2013), and two invalidated (P.riopiedadensis and P. serrana; Cardozo & Toledo, 2013;Pansonato et al., 2014b) taking into account this line of evidence.Therefore, it is essential to obtain, whenever it is possible, the association of advertisement call pattern to Pseudopaludicola specimens to unambiguous assessment of their species identity.

TABLE 2 :
Presence of vertebral (pin-stripe) and/or dorsolateral stripes in adult specimens of Pseudopaludicola giarettai from six localities of Minas Gerais, southeastern Brazil.N = number of examined specimens; * Type specimens.

TABLE 4 :
Acoustic traits for Pseudopaludicola giarettai from Curvelo (type locality), and five additional municipalities of Minas Gerais, southeastern Brazil.Mean ± SD (minimum-maximum).N = number of recorded males (number of analyzed calls).H = harmonic peak frequency.* Dominant frequency (= Fundamental frequency); ** Obtained from calls of one male.

FIgURE 3 :
FIgURE 3: Advertisement call of P. giarettai from the type locality (Curvelo, Minas Gerais).From top to bottom: oscillogram section (ca.1.6 s) depicting calling pattern, waveform, spectrogram, and power spectrum of the second advertisement from oscillogram section.Sound energy with relative amplitude below 40 dB was clipped to zero dB to remove background noise from power spectrum.

FIgURE 4 :
FIgURE 4: Advertisement call of P. giarettai from the Parque Nacional Grande Sertão Veredas (Chapada Gaúcha, Minas Gerais).From top to bottom: oscillogram section (ca.1.6 s) depicting calling pattern, waveform, spectrogram, and power spectrum of the second advertisement from oscillogram section.Sound energy with relative amplitude below 40 dB was clipped to zero dB to remove background noise from power spectrum.

FIgURE 6 :
FIgURE 6: Typical breeding habitat of P. giarettai: a permanent pond associated with a Buriti palm grove marsh at the Parque Nacional Grande Sertão Veredas (Municipality of Chapada Gaúcha), northwestern Minas Gerais, southeastern Brazil.

TABLE 3 :
Summary of snout-vent length (SVL) ranges (mm) for adult specimens of Pseudopaludicola giarettai from six localities of Minas Gerais, southeastern Brazil.Sample sizes in parentheses; * Type locality.
Carvalho, T.R. et al.: Intraspecific variation and distribution of PseudoPaludicola giarettai D complete in spectrogram screen [maximum frequency measurement allowed by a 48-kHz sampling rate (Nyquist of sampling rate) = 24 kHz].