Introduction

Panama’s herpetofauna is known to be the most diverse in consideration of its size in Central America, with only Mexico being more diverse in absolute species count (Myers and Duellman 1982; Jaramillo et al. 2010). Although herpetological research has been conducted in Panama for more than a hundred years (Ibáñez et al. 2001), the knowledge of amphibian species diversity is still far from being completed. This is demonstrated impressively by the multitude of amphibian species described from this country within the last years (e.g. Wake et al. 2005, 2007; Köhler et al. 2007; Mendelson et al. 2008; Bolaños and Wake 2009; Crawford et al. 2010b; Mendelson and Mulcahy 2010; Ryan et al. 2010a, 2010b).

The genus Diasporus (Hedges et al. 2008) is the closest relative of the predominantly Caribbean genus Eleutherodactylus. The species of Diasporus are distributed from eastern Honduras to western Ecuador (Frost 2011; Köhler 2011). The genus contains nine described species, five of which (Diasporus diastema Cope, Diasporus hylaeformis Cope, Diasporus tigrillo Savage, Diasporus ventrimaculatus Chaves, García-Rodríguez, Mora and Leal, and Diasporus vocator Taylor) are currently known to occur in western Panama and/or southern Costa Rica. The remaining four species (Diasporus anthrax Lynch, Diasporus gularis Boulenger, Diasporus quidditus Lynch, and Diasporus tinker Lynch) are distributed in Panama east of the Canal, and further along the Pacific coast of northern South America southwards to northwestern Ecuador (Frost 2011; IUCN 2011). However, differences in body size, male advertisement call, and coloration in the genus Diasporus suggest that there are several undescribed species (Lynch and Duellman 1997; Ibáñez et al. 1999; Savage 2002). Recent fieldwork in the Serranía de Tabasará of western Panama resulted in the discovery of a remarkable new species of Diasporus. The purpose of this paper is to describe this new species.

Materials and methods

Field work was carried out between May and August 2010 at several sites along both slopes of the Serranía de Tabasará between the Fortuna depression and Santa Fé, Veraguas, Panama. All specimens were encountered during opportunistic searches at night. Preparation and preservation of voucher specimens follows Köhler (2001). Tissue samples, usually the left forearm, were stored in 98% undenatured ethanol and deposited in the tissue collection of the Senckenberg Forschungsinstitut und Naturmuseum Frankfurt, Germany (SMF). Geographic coordinates and altitude above sea level were recorded with a handheld Garmin GPS MAP 60 CSx GPS receiver. All georeferences are in the geographical coordinate system and WGS 1984 datum, given in decimal degrees rounded to the fourth decimal place. Elevations are rounded to the next tenth. The map was created using ArcGIS 10 (ESRI).

We took additional morphological data from all Central American species currently assigned to the genus Diasporus in the SMF collection. We list all specimens examined for comparison in the Appendix I. Abbreviations for museum collections follow those of Sabaj Pérez (2010) except MHCH (Museo Herpetológico de Chiriquí, the herpetological collection of the Universidad Autónoma de Chiriquí, David, Panama). Specimens in the Appendix labeled with AH field numbers will be deposited at MHCH.

The sex of the male holotype and the paratypes was determined by the presence of vocal slits and vocal sac. Measurements were made with a dial caliper with the aid of a dissecting microscope and rounded to the nearest 0.1 mm. Measurements of the holotype (LACM 146212) and paratype (LACM 146241) of Diasporus tigrillo were taken from Savage (1997), those of Diasporus ventrimaculatus from Chaves et al. (2009). Additionally, we examined photographs of the type specimens of Diasporus tigrillo. If possible, missing measurements were calculated on the basis of data presented in the respective descriptions. In the case of Diasporus ventrimaculatus, no individual measurements are available, for which reason calculations were made using the average values given for the paratypes by Chaves et al. (2009). We follow Savage (1997, 2002) in the terminology of disk cover and disk pad shape, dorsal outline of head, and snout profile shape. Abbreviations used for measurements are: snout-vent length (direct line distance from the tip of the snout to the posterior margin of the vent): SVL; length of Finger III (from distal end of the Finger III including disk to the base of the second subarticular tubercle): LF III; length of Toe IV (from distal end of the toe IV including disk to the base of third subarticular tubercle): LT IV; disk width at Finger III (at greatest width): DWF III; disk width at Toe IV (at greatest width): DWT IV; head length (from quadratojugal region to tip of the snout): HL; head width (between angles of jaw): HW; tibia length (straight length of the tibia): TL; horizontal eyelid diameter (greatest length of the upper eyelid): EL; interorbital distance (the width of frontoparietal bone between the orbits): IOD; horizontal tympanum diameter (at greatest length): TD; and horizontal eye diameter (at greatest length): ED. The capitalized colors and color codes (the latter in parentheses) in color descriptions in life are those of Smithe (1975–1981). Drawings of head, hands, and feet were made using a camera lucida attachment for a Leica MZ 12 dissecting microscope. Values provided for morphometric and acoustic para- meters are minimum, maximum, and mean value ± standard deviation.

Advertisement calls were recorded using a Sennheiser ME 66 shotgun microphone capsule with a Sennheiser K6 powering module in combination with the Marantz PMD 620 solid-state recorder. A minimum distance from microphone to frog of one meter was kept while recording to prevent disturbance. As needed, the microphone was attached to branches with the aid of a Joby Gorillapod in order to minimize disturbance of the calling frog. Calls were recorded in PCM format at a sampling rate of 48 kHz with 24 bit resolution and stored as wav files on a SD Card. Call editing and analysis were performed using SOUND RULER 0.9.6.0 (Gridi-Papp 2003–2007) for frequency analysis and to generate figures of oscillograms and audiospectrograms. We measured temporal parameters by hand using ADOBE AUDITION 3.0, because SOUND RULER has difficulties in accurately and precisely measuring temporal parameters (Bee 2004). Frequency information was obtained through Fast Fourier Transformation (FFT length 512 points, overlap between FFTs 0.8) at Hanning window function. Air temperature and humidity were measured immediately after each sound recording using the digital device Voltcraft HT-200 to the nearest 1°C and 3.5% relative humidity (RH). An alcohol extraction of skin secretions of the new species has been examined for alkaloids using Liquid Chromatography-Time-Of-Flight-Mass Spectrometry (LC-TOF-MS) at the Center of Forensic Medicine of the Goethe-University Frankfurt am Main.

For the complementary molecular analysis, we extracted DNA following the protocol of Ivanova et al. (2006). To eliminate potential PCR-inhibiting contaminants, the tissue samples were incubated for one hour in TE-buffer (10 parts Tris-HCl (pH = 8.0) and one part EDTA) before digestion for at least 10 hours in 50 µl Vertebrate Lysis Buffer and 5.2µl Proteinase K at 56 °C. After extraction, DNA was eluted in 50 µL TE buffer. A fragment of the mitochondrial 16S rRNA gene was amplified in an MJ Research Dyad Deciple™ Peltier Thermal Cycler using the following program: initial denaturation for 180 s at 94 °C; followed by 39 cycles with denaturation for 15 s at 94 °C, hybridization for 60 s at 51 °C, and elongation for 60 s at 72 °C; final elongation for 120 s at 72 °C. Reaction mix for each sample contained 1 µL DNA template, 2.5 µL Reaction Buffer (PeqGold), 2.5 µL 2.5 mM dNTPs, 0.5 µL Taq Polymerase (PeqLab), 16.5 µL H₂O, and 1 µL of each primer (forward: 16SA-L, 5’-CGCCTGTTTATCAAAAACAT-3’; reverse:16SB-H, 5’-CCGGTCTGAACTCAGATCACGT-3’). The achieved 16S sequences were deposited in GenBank. We compared 21 sequences of the genus Diasporus in our analysis, three of the type series of the new species and one referred specimen, four specimens referred to as Diasporus aff. hylaeformis from Cerro Pando, as well as all16S sequences of the genus Diasporus available on GenBank. We used an additional 16S sequence of Pristimantis ridens as an outgroup (see Appendix II for examined specimens and GenBank accession numbers). Sequences were aligned with ClustalW (Larkin et al. 2007) using the default settings in Geneious (Drummond et al. 2010). The manually refined final alignment contained 535 positions. Using MEGA5 (Tamura et al. 2011), we computed uncorrected pairwise genetic distances, determined the Tamura 3-parameter model (Tamura 1992) as the best-fitting substitution model, and conducted Maximum Likelihood as well as Maximum Parsimony analyses (each with 10000 bootstrap replicates). Using TCSv1.21 (Clement et al. 2000), we conducted a statistical parsimony network analysis, with gaps considered as a fifth character state and a connection limit of 95%.

Results Discussion

Diasporus citrinobapheus is easily distinguishable from all other known frogs of the genus in Lower Central America by its bright yellow to orange coloration. The only described species of the genus that somewhat resemble the new species in coloration are Diasporus gularis from Colombia and Ecuador and Diasporus tigrillo from the Caribbean slopes of the Costa Rican part of the Serranía de Talamanca. The latter species is known only from two specimens, both collected in 1964 at a single locality and there are no photographs of the species in life, tissue samples, or call recordings available to clarify the systematic relationships of this species. The different ground coloration in preservative between Diasporus tigrillo and Diasporus citrinobapheus is certainly due to different preservation techniques, because the fixation process in 10% formalin darkens the complete specimen. However, this does not influence the general color pattern, so we treat the dark brown spots on the dorsum of Diasporus tigrillo as a diagnostic feature to differentiate between Diasporus tigrillo and Diasporus citrinobapheus. Additional material is required, preferably from near the type locality of Diasporus tigrillo to conduct further studies. In contrast, Diasporus gularis is known from a number of specimens from Colombia and Ecuador. However, the presence and development of papillae at the apex of the pad on the underside of the disk cover, one of the main characters that has been used to distinguish this species from its congeners, is a controversial issue. Lynch (1976, page 12, Fig. 3 B) provided a drawing of the left hand of a specimen (LACM 73239) from Chocó, Colombia, that shows long papillae. In a later work, the same author presented drawings of finger disk pads of two Diasporus gularis (ICN 45168, 45171) from Valle del Cauca, Colombia, that show knobbed disk covers (Lynch 2001, Figs 2 A, B), but he also noted that there are specimens lacking this character. He further considered the presence or absence of this knob at the underside of the disk cover might be due to preparation technique. Lynch and Duellman (1997) noted that the holotype of Diasporus gularis from Ecuador does not have papillae at the tip of any digit, while they stated that specimens from southern Colombia and Ecuador have a rounded knob at the apex of the pad on the underside of the disk cover of toe II–IV. In Chocoan Colombia, specimens referred to as Diasporus gularis have larger papillae. Depending on their diagnosis, Lynch and Duellman (1997) argued that several species might currently be referred to Diasporus gularis, three in western Colombia and only one in Ecuador, where the type locality is. Based on our genetic analysis, Diasporus citrinobapheus is closely related to, and may even be conspecific with, the candidate species Diasporus aff. diastema from El Copé (Crawford et al. 2010a). Albeit the comparably small p-distance, the haplotype network analysis yields a separate network for each of the two clades, supporting the assumption of two distinct species. However, genetic evidence revealed only from mitochondrial markers alone is not strong enough to support either one or the other hypothesis (Vences et al. 2005). Further integrative taxonomic studies, including morphology, bioacoustics, and nuclear genes are needed to clarify this matter.

Besides various records of other amphibians and reptiles, we found no additional species of the genus Diasporus at the type locality. At Willie Mazú, a locality approximately 120 km NW of the type locality of Diasporus citrinobapheus, we collected a single specimen of Diasporus that we refer to Diasporus vocator based on size, coloration, disk shape, and male advertisement call. At Cerro Negro, Diasporus citrinobapheus occurs sympatrically with Diasporus diastema. Based on our current concept of its distribution, the possibility remains that also Diasporus vocator occurs at this locality, although the species has not been recorded from this site.

The eponymous, readily soluble yellow coloration of Diasporus citrinobapheus lead us to the assumption that this might serve a defensive function against predators. On this account, an alcohol extraction was analyzed for alkaloids, but no active substances were found. Probably, the yellow pigment is just highly soluble and therefore easily washed out. Nevertheless, one could speculate that it has a bitter or otherwise unpalatable taste that might deter certain predators.

Various studies have shown that the advertisement call represents a premating isolating mechanism in anurans (e.g., Duellman and Trueb 1986), which makes it a valuable tool in taxonomy. Having in mind that there are great morphological overlaps between members of the genus Diasporus, analyses of vocalizations might form the most powerful taxonomic approach to decipher its species diversity. Unfortunately, the calls of most species have never been formally described. Fouquette (1960) was the first to describe the call of Diasporus diastema from the Panama Canal area, about 10 km northwest of Panama City, not far from the type locality (Cope 1876; Dunn 1942). Later, Wilczynski and Brenowitz (1988) presented another call description based on calls recorded in the surroundings of Gamboa in Central Panama, about 24 km NNW of Panama City, and even closer to the type locality. Interestingly, the call descriptions of Fouquette (1960) and Wilczynski and Brenowitz (1988) are incongruent in terms of call duration, frequency range, and dominant frequency, rendering it possible that different species were recorded. Unfortunately, none of these papers cited any voucher specimens, so it is impossible to determine which species they actually recorded. The most recent contribution on vocalizations of Diasporus diastema is that of Ibáñez et al. (1999), also from the environs of the Panama Canal. They provided a rough sonogram, but did not give any numerical values. The dominant frequency in all three papers is roughly described as 3000–4000 Hz, thus considerably higher than in Diasporus citrinobapheus. Furthermore, all three papers (Fouquette 1960; Wilczynski and Brenowitz 1988; Ibáñez et al. 1999) present sonograms, which show an obvious frequency modulation expressed by a rapid rise of frequency over time with approximately 1000 Hz difference between beginning and end of the call. In contrast, the call of Diasporus citrinobapheus is characterized by only a moderate frequency rise over time, on average 350–400 Hz. Confusing are the data of call duration provided by Fouquette (1960) and Wilczynski and Brenowitz (1988), respectively. Fouquette (1960) reports on mean call duration of 0.2 s for Diasporus diastema. Yet, in the accompanying sonogram (Fig. 2 A in Fouquette 1960), the call seems to be only slightly longer than 0.1 s. Wilczynski and Brenowitz (1988) even mentioned a call duration of more than 0.3 s, but in the accompanying oscillogram (Fig. 1 B in Wilczynski and Brenowitz 1988), the call does not exceed 0.1 s on the time axis. Although difficult to assess precisely, the duration of the call pictured in the sonogram provided by Ibáñez et al. (1999) is clearly shorter than 0.2 s. Furthermore, Ibáñez et al. (1999) present a sonogram of Diasporus vocator, recorded also in the Canal Zone. According to their analysis, the call of Diasporus vocator has a frequency range between 6000 and 7000 Hz, is very short, and shows a strong frequency modulation, thus being very different from the calls of Diasporus diastema and Diasporus citrinobapheus. However, the type locality of Diasporus vocator is Agua Buena in the Peninsula de Osa, Costa Rica. Thus, it is advisable to record comparative call material from the type locality for future analyses. The most recent contribution on Diasporus vocalizations was made by Chaves et al. (2009) in the course of the original description of Diasporus ventrimaculatus. This species’ voice differs in all standard parameters from that of Diasporus citrinobapheus, as it has much shorter call durations of about 0.08 s, a low dominant frequency of about 2550 Hz, and a lower frequency range between 2140 and 2995 Hz. Furthermore, the dominant frequency is reached at the very beginning at the call. The same authors presented a preliminary analysis of calls emitted by specimens assigned to Diasporus hylaeformis. According to this analysis, call duration in Diasporus hylaeformis is on average 0.214 s, while it resembles Diasporus ventrimaculatus in spectral parameters. Regarding the vocalizations of Diasporus tigrillo, the least known species in the genus, only a field note citation appearing in the original description describes it as “similar to the dink dink of Eleutherodactylus [Diasporus] diastema” (Savage 1997).

Nevertheless, there is also an intraspecific variation among calls of specimens referred to Diasporus citrinobapheus. The call of the single male recorded at Willie Mazú (SMF 89817) differs from the calls of the members of the type series in temporal parameters, such as shorter call duration and call interval that result in a higher call rate. These differences are minor, but lead us to not include specimens from localities other than the type locality in the type series. However, various studies have shown that call parameter variation is linked to ambient temperature (e.g., Zweifel 1959; Schneider 1977; Gerhardt 1978). According to these studies, call duration and call interval are negatively correlated with temperature, which in turn leads to an increased call rate at higher temperatures. As SMF 89817 was recorded at lower temperatures than for the other three specimens, one would expect the opposite pattern. Nevertheless, these studies used data from many individuals, built scatter diagrams of parameters against temperature and fitted least-squares regression, and there are always outliers that do not follow the general trend. In our case, individual differences may be stronger than temperature-related ones, but this assumption needs further research to be reliably assessed.

Apart from morphology which apparently is not the best tool to identify species of Diasporus, neither DNA nor bioacoustics, both of paramount importance for contemporary anuran taxonomy, have been consistently analyzed among geographically and taxonomically wide-ranging samples. While the Panamanian and Costa Rican 16S barcodes compared in this study reveal the existence of more infrageneric lineages than names are available, the doubtless assignation of a given Diasporus “aff. hylaeformis” or “aff. diastema” is likely to be highly challenging if one is to rely on the existing treatments, which mostly provide only partial or even contradicting information. In conclusion, the complex and cryptic diversity within the genus Diasporus requires a thorough revision of as many “quality vouchers” (collected specimens associated with both well-preserved tissue samples and call recordings) from as many localities throughout the generic range as possible.

Key to the species of Diasporus in Central America