First record of Vitreorana ritae (Anura, Centrolenidae) for southern Amazonia inferred from molecular, reproductive and acoustic evidence

Frogs of the family Centrolenidae are adapted to live at high altitudes, and so it is not surprising that their greatest diversity is in the Andes mountain range. Here we extend the known geographic distribution of Vitreorana ritae by more than 550 km towards southern Amazonia. The species was identified by morphological, acoustic and molecular characteristics (mitochondrial marker 16S rRNA). The records reported here are the first for the genus Vitreorana in the state of Mato Grosso, Brazil, as well as for the transition zone between the Amazonia and Cerrado biomes. These records are consistent with the hypothesis that the low number of species of centrolenids reported in the lowlands of the Amazon region may be the result of limited sampling. Even though V. ritae is distributed throughout the Amazon, most of its diagnostic morphological characteristics are conserved. Thus, the great rivers of the Amazon Basin do not seem to act as geographical barriers for this species; however, due to the limited sample size, further acoustic and molecular studies are needed to confirm this hypothesis.


INTRODUCTION
The family Centrolenidae comprises 157 species of eleven genera of what is commonly referred to as glass frogs (Guayasamin et al. 2009). They are distributed from the tropics of Central America, throughout the tropical Andes, the Sierra Nevada in Colombia, the coast of Venezuela to Bolivia, Trinidad and Tobago, the Guiana Shield, and the Atlantic Forest of southeastern Brazil and northeastern Argentina (Frost c2019). Fourteen centrolenid species are currently known to occur in Brazil: Hyalinobatrachium cappellei Van Lidth de Jeude, 1904;H. carlesvilai Castroviejo-Fisher, Padial, Chaparro, Aguayo-Vedia, and De la Riva, 2009;H. iaspidiense (Ayarzagüena, 1992); H. mondolfii Señaris and Ayarzagüena, 2001; H. muiraquitan Oliveira and Hernández-Ruz, 2017;H. munozorum (Lynch and Duellman, 1973); Teratohyla adenocheira (Harvey and Noonan, 2005); T. midas (Lynch and Duellman, 1973); Vitreorana baliomma Pontes, Caramaschi, and Pombal, 2014;V. eurygnatha (Lutz, 1925); V. franciscana Santana, Barros, Pontes, and Feio, 2015;V. parvula (Boulenger, 1895); V. ritae (Lutz, 1952); and V. uranoscopa Although the number of Brazilian centrolenid species has increased in recent years with new records for previously isolated sites, mainly in the Amazon (ICMBio c2009, Toledo et al. 2009, Venâncio et al. 2014, there still remain large knowledge gaps regarding the distribution of several species, most of which are known from isolated localities within a large geographic range (Frost c2019). Moreover, only 79 % of Brazilian centrolenids have had their advertisement call described (Guerra et al. 2018). Specific information about the distribution and vocalization of anuran species is of great importance for taxonomic and conservation purposes (Castroviejo-Fisher et al. 2011, Guerra et al. 2018. Bioacoustic information has been particularly relevant in studies of anuran behavior (Gerhardt and Huber 2002), community ecology (Oseen and Wassersug 2002) and evolution (Cocroft andRyan 1995, Goicoechea et al. 2010), and has contributed to the discovery of many morphologically-conservative cryptic species (Padial et al. 2008, Glaw et al. 2010. Vitreorana ritae (Lutz, 1952) occurs in eastern Ecuador, Colombia, Peru, southern Guyana, eastern Suriname, French Guiana and from the state of Roraima south to central Amazonia and west to extreme western Amazonas State in Brazil (Frost c2019). The species has been recorded at different localities in the state of Pará, such as the Caxiuanã and Tapajós National Forest (Gonçalves 2013), the Xingu River (Vaz-Silva et al. 2015), and the Altamira National Forest (ICMBio c2009). Vitreorana ritae is the senior synonym of Cochranella oyampiensis (Lescure, 1975) and Cochranella amertasia (Flores, 1987). It is morphologically similar to V. helenae, which is considered its sister species (Kok and Castroviejo-Fisher 2008, Castroviejo-Fisher et al. 2014, Moraes et al. 2017) and occurs in the Tepuis of the Guiana Shield, although V. ritae is distinguished by its gray iris and green dorsum with black spots (yellow and uniformly green dorsum, respectively, for V. helenae; Guayasamin 2008).
The present study extends the distribution of Vitreorana ritae with sites in the municipalities of Cotriguaçu and Sinop in the state of Mato Grosso, Brazil. The advertisement call and tadpole morphology and molecular data from a male collected in the transition zone between the Cerrado and the Amazon are compared with that of previously known populations.

Molecular analysis
Total genomic DNA was extracted from the liver of the male collected in the municipality of Sinop using the 2 % CTAB protocol (Doyle and Doyle 1990). A 16S mtD-NA fragment was amplified through PCR by applying 16Saf and 16Sbr primers (Palumbi et al. 1991). The sequencing reaction was performed according to the manufacturer's recommendations for the sequencing mix ABI BigDye Terminator using the primer 16Saf. The sequences (ca. 600 Pb) were aligned using the ClustalW algorithm (Thompson et al. 1994) with BioEdit 7.2 software (Hall 1999). The uncorrected pairwise distance (p-distance) between the Vitreorana ritae populations of this study and those of GenBank from French Guiana and V. helenae from French Guiana and Venezuela were calculated using MEGA 6.0 (Tamura et al. 2013).

RESULTS
The specimens collected were identified as Vitreorana ritae based on the vomerine teeth, a white hepatic peritoneum partially covering the liver, a white gastrointestinal peritoneum, green bones in life, dorsal green coloration with dark green spots in life and lavender with dark spots in preservative, and a rounded snout in dorsal view and slightly inclined in lateral view (Fig. 1). The new records of V. ritae in the municipalities of Cotriguaçu and Sinop, Mato Grosso ( Fig. 2), extends the distribution of the species by 530 and 560 km, to the southeast and south, respectively, from the nearest known distribution of the species (Altamira National Forest, Pará, Brazil; ICMBio c2009), and 1500 and 1700 km, respectively, to the southeast of the type locality (Colombia) (Flores 1987, Cisneros-Heredia 2013).
The egg clutch was found on the adaxial face of a leaf hanging at a height of 1.5 m over a small permanent stream (1.2 m width). The clutch contained 21 embryos wrapped in an off-white, translucent gelatinous capsule (Fig. 3a).
The embryos were at stage 21 (Gosner 1960) and exhibited a greenish vitellum and head and a greenish brown tail. After five days in an aquarium, the gelatinous capsule became whitish and opaque (Fig. 3b) and the tadpoles hatched at approximately stage 25. At this stage, the Mendes de Conservação da Biodiversidade (SISBIO n º 30034-1).
An egg clutch in the final stages of embryonic development was also found on the same shrub where the male was found in the municipality of Sinop. The clutch was photographed, collected and taken to the UFMT laboratory, Campus Sinop, where it was kept in an aquarium (40 x 60 x 50 cm). After hatching, two tadpoles were euthanized using a 2 % lidocaine solution, fixed in 5 % formaldehyde and deposited in ABAM (ABAM-H 50, ABAM-H 52). The external morphology of the tadpoles was compared with descriptions of other Vitreorana tadpoles (e.g. Menin et al. 2009). The developmental stage of the tadpoles was determined according Gosner (1960).
Two approaches were used to identify the adult frogs. The first combined morphological data (presence of vomerine teeth; presence and pigmentation of hepatic and intestinal peritoneum; bone and dorsum coloration; and face shape) and bioacoustic data described by several authors (Lutz and Kloss 1952, Lescure 1975, Lescure and Marty 2000, Guayasamin et al. 2006, Cisneros-Heredia and Mcdiarmid 2007, Guayasamin et al. 2009, Menin et al. 2009, Cisneros-Heredia 2013. The second employed BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi), to compare DNA sequence data of one of the collected specimens with V. ritae sequences found in GenBank.

Acoustic analysis
Two advertisement calls of a male were recorded using a Sennheiser ME 66 directional microphone coupled to a professional Marantz PMD 660 digital recorder set at a sampling rate of 44.1 kHz and 16-bit resolution and positioned 1.0 m from the specimen. The calls were analyzed using Raven Pro 1.5 (Bioacoustics Research Program c2014). Temporal properties were obtained from oscillograms (temporal resolution of 500 ms), while spectral information was obtained from spectrograms using Fast Fourier Transformation (FFT) with a 256-point window and a frequency resolution of 90 Hz. Seven parameters of the advertisement call were Measured: three temporal -(1) call duration (s), (2) interval between calls (s) and (3) call rate (call / min.); and four spectral -(1) number of harmonics, (2) minimum frequency (5 %) (Hz), (3) dominant frequency (highest peak of energy) including all harmonics (Hz), and (4) maximum frequency (95 %) (Hz). Terminology for the acoustic parameters followed Köhler et al. (2017). Sonograms and spectrograms were produced tadpoles remained with green vitellum, but with a slightly translucent greenish brown stem and head ( Fig. 3c-d).
The tadpoles did not develop any further after falling into the aquarium water, at which point they had an elongated and flat body with a subterminal oral disc, a labial tooth row formula (LTRF) of 0/1-2, large marginal papillae and unserrated or arched jaw sheaths.
The calling male was located on the top of a leaf (25 x 15 cm) at a height of approximately 3.5 m over a small stream. The advertisement call consists of a single pulsed note with two visible frequencies (Fig. 4). The duration of the two recorded calls were 0.29 and 0.30 s (N = 2), with a call interval of 1.43 min. The average minimum frequency was 4392 Hz (4306 and 4478 Hz, N = 2), the maximum frequency was 4995 Hz (for both calls), and the average dominant frequency was 4737 Hz (4651 and 4823 Hz, N = 2). The frequency of the second harmonic averaged 9990 Hz (9646 and 10335 Hz, N = 2) (Fig. 4).

Molecular analysis of 16S mtDNA (ABAM-H 3085)
from Mato Grosso revealed a 1 % difference with V. ritae collected in French Guiana and 4 -5 % difference with vouchers of its sister species V. helenae collected in French Guiana and Venezuela (Table 1).

DISCUSSION
The new records of Vitreorana ritae extend the known distribution of the species 560 km south of the nearest known locality (ICMBio c2009), and thus represents the southernmost known population of this species. These new records, combined with previous records for glass frogs in the Brazilian Amazon (i.e., Noronha et al. 2012, Pontes and Mattidi 2013, Oliveira et al. 2017, Araújo et al. 2018, indicate that there is an insufficient sampling in many regions and that the geographic distribution of most species is not fully known. Furthermore, many of these species lack natural history data (Noronha and Rodrigues 2018). Adult individuals of V. ritae are generally difficult to locate and males call sporadically (Zimmerman and Bogart 1984). Reproductive activity of V. ritae occurs mainly between January and May, when males can be observed calling from vegetation above streams and where females deposit clutches on upper or lower leaves. Eggs hatch after several days and the embryos fall into the water where they complete their development as free-swimming tadpoles (Menin et al. 2009). The streams where V. ritae reproduces support strong water currents during the rainy season, and so the tadpoles possess a fossorial habit. This likely explains why few centrolenid tadpoles have been observed and collected throughout the vast region of the Brazilian Amazon. The morphological description of the tadpoles originating from the clutch found in Mato Grosso agrees with that of Menin et al. (2009).
The advertisement call of V. ritae recorded in southern Amazonia (municipality of Sinop, MT) is a single pulsed harmonic note, which differs from the number of notes of the calls described for the species by Zimmerman and Bogart (1984) for central Amazonia (two notes), Señaris and Ayarzagüena (2005) for Venezuela (two notes), and Lescure and Marty (2000) in Guyana (four notes). Call duration also differed from the calls described from central Amazonia (average 130 ms, Zimmerman and Bogart 1984) and Guyana (average 107 ms, Lescure and Marty 2000).
The dominant frequency reported here for southern Amazonia (average 4737 Hz, 4651 Hz and 4823 Hz, N=2) was similar to that reported for males of central Amazonia (4640 -5160 Hz, Zimmerman and Bogart 1984) and Venezuela (3810 -4920 Hz, Señaris and Ayarzagüena 2005), but differed from that described for males of Guyana (5570 -7470 Hz, Lescure and Marty 2000). The calling male reported here exhibited a long period of silence between calls, which was also reported by Zimmerman and Bogart (1984). Although the acoustic parameters of the advertisement calls differ among regions, all these studies analyzed a very small number of individuals and calls, which limits the ability to understand intra-and inter-individual variation and variation among populations. In addition, environmental (e.g., temperature) and social (e.g., number a b c d Giaretta, andToledo, 2016 (Andrade et al. 2016) and Proceratophrys ararype Mângia, Koroiva, Nunes, Roberto, Ávila, Sant'Anna, Santana, andGarda, 2018 (Mângia et al. 2018). The low genetic distance among populations of V. ritae supports the hypothesis that some anuran species can maintain low genetic distances even though they have wide geographic distributions (Wynn andHeyer 2001, Fouquet et al. 2007).
The present work contributes to the knowledge of V. ritae and demonstrates that even though it is widely distributed throughout the Amazon and Cerrado transition zone, the genetic, morphological and acoustic characteristics of the species are highly conserved, which may indicate the maintenance of gene flow among populations.
The results presented here also demonstrate the capacity of the species has to adapt to different climates and environments by eradicating to lowland forests of the Amazon in drier areas such as the ecotone with the Cerrado. Finally, we emphasize the need for greater sampling in southern Amazonia because the diversity of the region is of calling males) contexts can influence temporal parameters, such as the number of notes, for example (Morais et al. 2012, Köhler et al. 2017. Since anurans can have wide vocal repertoires (Toledo et al. 2015), it is also possible that the calls described for V. ritae represent different types of calls.
Amphibians generally have a low vagility and high philopatry, which are characteristics that promote differentiation and ultimately speciation (Wells 2012. In this sense, although most anuran species would be expected to have small geographic distributions with distantly separated populations having greater genetic differences, there are some cases of widely distributed species with low genetic distances, such as Dendropsophus nanus (Boulenger, 1889) (Fouquet et al. 2007), Allophryne ruthveni Gaige, 1926(Castroviejo-Fisher et al. 2012, and Adelphobates galactonotus (Steindachner, 1864) (Rojas et al. 2020). There are also cases of low genetic distances between distinct species, such as Pseudopaludicola jaredi Andrade, Magalhães, Nunes-de-Almeida, Veiga-Menoncello, Santana, Garda, Loebmann, Recco-Pimentel,  underestimated. Furthermore, the region has been experiencing major environmental impacts, from accelerated expansion of agribusiness and industrial exploitation of natural resources, which may cause the extinction of numerous populations or even endemic species before they are even known scientifically.

LITERATURE CITED
AUTHOR'S CONTRIBUTION MP was responsible for collecting data, identifying the specimens, acoustic analysis and writing the text; SFA identified the specimens and revised the text; EAO was responsible for molecular analysis, wroted the text and created the image; EJHR wroted the text; VGB was responsible for acoustic analysis and writing the text; LRRR was responsible for molecular analysis and reviewing the text; DJR was responsible for collecting data, identified the specimens and reviewing the text.

CONFLICT OF INTEREST
The authors declare that they have no conflict of interest.