Southeast Asia is recognized as a major global biodiversity hotspot (Myers et al. 2000; Mittermeier 2004; Corlett 2014), and the amphibian order Anura is well represented in this region. However, it seems that the actual diversity of anurans in this part of the world is still grossly underestimated due to poor sampling of areas with difficult or dangerous access and because most of the supposedly wide-spread species actually constitute species complexes of two or more distinct species (e.g., Funk et al. 2011; Hasan et al. 2012; Köhler et al. 2019). During field work in Myanmar and Thailand we came across variation in the advertisement call of the common mud or puddle frogs, in this region all currently referred to as Occidozyga martensii Peters, 1867 (Niyomwan et al. 2019; Frost 2020). These observations prompted us to collect this group of frogs to obtain a reasonable geographic coverage with the goal to evaluate whether these really are a single species occurring in a large geographic area or if these actually represent a complex of several – morphologically similar (cryptic) – species with smaller geographic ranges.

The generic placement of martensii in either Phrynoglossus Peters, 1867 or Occidozyga Kuhl and van Hasselt, 1822 has caused some controversy among recent authors. It seems that the majority of authors have treated Phrynoglossus as a synonym of Occidozyga (e.g., Duellman and Trueb 1994; Nguyen et al. 2009; Frost 2020) whereas others have recognized them as two valid genera (e.g., Manthey and Grossmann 1997; Ohler and Dubois 1999; Ziegler, 2002; Köhler et al. 2018). Here we provide a short summary of the taxonomic history of Phrynoglossus and Occidozyga:

The genus name Occidozyga was introduced by Kuhl and van Hasselt 1822. Its type species (by subsequent designation of Stejneger 1925) was described as Rana limaGravenhorst 1829 a few years later. In 1859, Günther described Oxyglossus laevis based on material from the Philippines. OxyglossusTschudi 1838 is an objective synonym of Occidozyga Kuhl and van Hasselt 1822 according to Ohler and Dubois (1999) whereas Ooeidozyga, Oxydozyga, Occidogyna are incorrect subsequent spellings of Occidozyga (Dubois 1981, Frost 2020). The genus name OreobatrachusBoulenger 1896 is a synonym of PhrynoglossusPeters 1867 according to Smith (1931).

In 1896, Boulenger introduced his new species Oreobatrachus baluensis from „Mount Kina Balu, North Borneo [= Sabah]“, Malaysia (Borneo). Smith (1931) placed OreobatrachusBoulenger 1896 in the synonymy of PhrynoglossusPeters 1867. With the introduction of the latter genus, Peters (1867) described the new species P. martensii, the holoype of which he had received from Bangkok, Thailand. In 1877, Peters described Microdiscopus sumatranus from Sumatra, Indonesia. Smith (1916) discussed the taxa laevis, lima and martensii and considered martensii to be a subspecies of laevis. Also, he described and illustrated the tadpoles of lima and martensii and provided natural history notes for the two taxa. He stated (Smith 1916: 175) that “O. lima is strictly aquatic in its habits” whereas martensii, “although never far from water, is seldom to be found in it.” Taylor (1922) introduced the new species Micrixalus diminutiva based on specimens from “near Pasananka, Zamboanga, Mindanao”, Philippines. Inger (1954) transferred the latter taxon to the genus Ooeidozyga. In 1927, Smith described two new species from Sulawesi, Indonesia: Ooeidozyga semipalmata from „Lowah, near Mt. Bonthain“, and Ooeidozyga celebensis from “Djikoro, Mt. Bonthain”. Based on specimens from “Rana Mese, 1200 m. H., West-Flores”, Indonesia, Mertens (1927) described Oxydozyga floresiana. In 1958, Taylor and Elbel introduced their new species Occidozyga magnapustulosa with type locality in “Ban Na Phua (subvillage), Kan Luang (village), Na Kae (district), Nakhon Phanom (province), Thailand, elevation approx. 200 m”. More recently, Iskandar et al. (2011) analyzed the morphological variation in populations of frogs related to Occidozyga semipalmata and recognized the populations from „Bantayan, Mount Tompotika, Balantak Mountains …, Desa (=Village) Bualemo, … Sulawesi Province, Indonesia“ as a distinct species that they named Occidozyga tompotika.

In his treatment of the herpetofauna of Mount Kinabalu, Borneo, Smith (1931) pointed out the unique tongue morphology of Ooeidozyga lima, strikingly different from the other species of that genus, and proposed to retain only O. lima in the genus and to transfer the remaining species to the genus Phrynoglossus. However, this proposal was not widely accepted by subsequent authors (e.g., Inger 1954, 1996) who preferred to maintain all species in a single genus (i.e., Occidozyga respectively its synonym Ooeidozyga). Inger (1996) argued that of the five supposedly diagnostic characters listed for Phrynoglossus, only one, the tongue morphology, would distinguish the species of the latter genus from Occidozyga.

Further evidence for considering Phrynoglossus as a synonym of Occidozyga was provided in the large-scale molecular phylogeny of amphibians of Pyron and Wiens (2011) where the two species of Phrynoglossus were nested within the species of Occidozyga, thereby rendering the latter genus paraphyletic if Phrynoglossus was recognized as a valid genus. Basal to the clade containing the species of Phrynoglossus and Occidozyga is Ingerana in the tree of these authors. However, aside from tongue morphology, there is additional evidence in favor of recognizing these two groups as distinct genera (see our Results section), and we therefore recognize Phrynoglossus as a valid genus distinct from Occidozyga.

Specimens examined for this study were personally collected by GK, NLT, and PT (see Appendix 1 for specimens examined). Specimens labeled with GK field numbers were deposited in the collections of Senckenberg Forschungsmuseum Frankfurt (SMF) or at East Yangon University (EYU), Thanlyin, Myanmar, and those with PT field numbers were deposited in the collection of the Chulalongkorn University, Museum of Natural History (CUMZ), Bangkok.

Prior to preservation of collected specimens in the field, we took color photographs of each individual in life. We euthanized the frogs with a pericardial injection of T61. We cut tissue samples from one forelimb or from the tongue and preserved these in 98% non-denatured ethanol for DNA extraction. The tissue samples were deposited in the collection of SMF and CUMZ. Specimens were then preserved by injecting a solution of 5 mL absolute (i.e., 36%) formalin in 1 L of 96% ethanol into the body cavity, and stored in 70% ethanol. Coordinates and elevation were recorded using Garmin GPS receivers with built-in altimeters. All coordinates were recorded in decimal degrees, WGS 1984 datum. Capitalized colors and color codes (in parentheses) followed Köhler (2012).

In evaluating species’ boundaries within and among populations, we followed the evolutionary species concept (Simpson 1951; Wiley 1978). As lines of evidence for species delimitation, we applied a phenotypic criterion (external morphology), the genetic distinctness of a mitochondrial genetic marker as well as a criterion for reproductive isolation (bioacoustic data).

Abbreviations used are EYD (eye diameter); FL (foot length); HL (head length); HW (head width); IND (internasal diameter); IOD (interorbital diameter); NED (nostril–eye distance); HNL (hand length); SHL (shank length); SL (snout length); SVL (snout–vent length); TED (tympanum–eye distance); THL (thigh length); TYD (longitudinal tympanum diameter). Webbing formulae follow (Savage and Heyer 1997). Terminology of snout shape follows Heyer et al. (1990).

We recorded anuran vocalizations using a digital audio recorder (Olympus LS-12) with a Sennheiser ME 66 shotgun microphone capsule and a Sennheiser K6 powering module. The microphone was positioned between 0.5 and 1.5 m from the calling frog. Aside from the GPS coordinates and elevation above sea level of the locality we also noted ambient air temperature and determined SVL of the recorded individual. Files were recorded as uncompressed 24-bit WAV files at a sampling frequency of 96 kHz. Audio files were deposited in the Fonoteca Zoológica, Museo Nacional de Ciencias Naturales, Madrid, Spain.

The spectral and temporal parameters were analyzed and the power spectra were calculated in RAVEN PRO 1.4. (Bioacoustics Research Program 2011). Spectrograms were obtained using the Blackman window function at 3db Filter Bandwidth of 141 Hz; grid spacing of 21.5 Hz; overlap 90%. Frequency information was generated through Fast Fourier Transformation (FFT, width 2,048 samples). Temporal measurements of calls such as repetition rates, duration of notes, and number of pulses, were measured on the waveforms. Terminology in call descriptions follows Köhler et al. (2017). The map was created using ArcMap 10.4. Additionally to the specimens examined by us, we also plotted specimens from the Field Museum (FMNH), Chicago, not examined by us.

The final 16S alignment for 69 samples (genera Ingerana, Occidozyga, and Phrynoglossus) was 599 nucleotide positions long. GTR+G was determined to be the best fitting model of sequence evolution. The trees obtained by BI, ML, *BEAST, and ABGD analyses show a high degree of congruence at well-supported nodes, with some differences in branch arrangement at poorly supported nodes (Figs 1, 2 and 3). Our final phylogenetic analyses recover the deep divergences between the genera Ingerana, Occidozyga, and Phrynoglossus (Figs 1, 2 and 3). These three genera form monophyletic groups in all analyses.

Ingerana is the sister clade to a clade containing Occidozyga plus Phrynoglossus. There is much genetic structure in the Occidozyga clade indicating the possible presence of cryptic species among the populations currently referred to as O. lima (Gravenhorst, 1829). Among the specimens here referred to the genus Phrynoglossus, four major clades are recognizable and also supported by the ABGD analysis (Fig. 3). Clade 1 (= Group 7 in our ABGD analysis) contains specimens from southern Thailand including Bangkok (= type locality of P. martensii) and are therefore considered to represent the “true” P. martensii. Clade 2 (= Group 8 in our ABGD analysis) contains specimens from northern Thailand, southern China, Vietnam, Laos, and Cambodia. Clade 3 (= Group 11 in our ABGD analysis) is geographically restricted to the Khorat Plateau, Thailand, and includes specimens from Ban Kan Luang (= type locality of P. magnapustulosus Taylor and Elbel, 1958) and are therefore considered to represent the latter species. Finally, Clade 4 (= Group 10 in our ABGD analysis) contains specimens from central and lower Myanmar. We consider these four clades as candidate species. The genetic distances among these clades vary from 3.5% to 5.9% (Table 1).

In bioacoustics, the advertisement calls of males from Clades 1, 3, and 4 can be readily distinguished whereas the calls of males from Clades 1 and 2 are exceedingly similar (Table 2; Fig. 4). Males of P. magnapustulosus (our Clade 3) emit an advertisement call that sounded “mourning” to our ears (“maaaaaaa”). It has a duration of 243–437 ms (mostly in the 300 ms range) with its dominant frequency mostly in the range of 3500–3700 Hz (3187–3790 Hz) and a gap duration between calls of 2.3–5.3 s, mostly 3–4 s. On the contrary, the call of P. martensii (Clade 1) is very short (sounding like “bic”) with a duration of 40–48 ms and a dominant frequency of 3467–3596 Hz. The gap duration between calls varies from 2.8–3.9 s. The calls of males from Myanmar (Clade 4) are somewhat similar to the calls of P. martensii but are of longer duration (82–114 ms) and lower pitched (dominant frequency 2454–2885 Hz). The gap duration between calls varies from 1.8–4.8 s. The advertisement call of our “cf. martensii” (Clade 2) is very similar to the call of P. martensii (Clade 1) and differs only in having mostly a shorter gap duration between calls (0.5–3.0 s, mostly < 2.5 s in “cf. martensii” versus 2.8–3.9 in P. martensii).

In external morphology, specimens of these four clades are somewhat conservative in external morphology but show differences in some morphometric values (Fig. 5; Table 3) as well as in the amount of toe webbing and in some details of coloration (see respective diagnosis sections below). Also, there is sexual dimorphism evident in several morphometric characters such as SVL, SHL, FL, and HL. Phrynoglossus magnapustulosus have the smallest SVL (males < 20 mm versus > 20 mm in the other clades) and also have a much larger ratio TYD/SVL compared to those from the other clades.

Thus, based on the results of the analyses of mtDNA data, morphology, and bioacoustics, we recognize three of the four clades as defined above as species level units (i.e., our OTUs 1, 3 and 4) whereas we tentatively consider our OTU 2 to be conspecific with our OTU 1 since we are unable to find any diagnostic differences in morphology or bioacoustics that would separate these two OTUs. Based on the respective type localities, our Clade 1 is assigned to Phrynoglossus martensii whereas our Clade 3 is referred to as P. magnapustulosus. Clade 4 represents an undescribed species for which no name is available. Thus, we describe the Phrynoglossus from central and lower Myanmar as a new species below.

Regarding the recognition of Phrynoglossus as a valid genus distinct from Occidozyga, we find support by the results of our genetic and morphologic analyses as well as our field observations on the amplexus mode of these frogs. Phrynoglossus has an inguinal amplexus whereas Occidozyga has an axillary amplexus (Fig. 6). The two genera differ readily in tongue morphology with Occidozyga having a very slender, worm-like tongue whereas the tongue is fleshy and thick in Phrynoglossus (Fig. 7a,b) Also, species of the two groups differ strikingly in skin texture (Fig. 7c,d) and mucosome, at least judged from touching these frogs with bare hands: species of Phrynoglossus have a smooth skin and are extremely slimy and thus difficult to constrain manually. On the other hand, Occidozyga has a rough skin and is not slimy at all. Finally, the two groups differ in ecology: species of Phrynoglossus are terrestrial frogs that are usually found on mud at the shore of small water bodies whereas Occidozyga is a fully aquatic species that always remains in the water body. For the issue of possible paraphyly of this group of frogs see also our results of the analysis of the mtDNA data used in this study.

Thus, we formally resurrect Phrynoglossus from the synonymy of Occidozyga and define the two genera as follows (data also from Ohler and Dubois 1999):

Our study provides support for the general assumption that most geographically wide-spread species contain cryptic diversity with one or more unrecognized species “hiding” behind another taxon name due to its similarity in external morphology (Hasan et al. 2012; Köhler et al. 2019). Typically, these anuran species differ from one another in bioacoustics, i.e., in the male advertisement call. Often these species are poorly differentiated in external morphology which was the reason for lumping them before in the first place. The reason for the poor degree of morphological differentiation is possibly the lack of ecological differentiation of the members of such a cluster of frogs. In the case of the Phrynoglossus studied here, it is most likely that they went through the process of speciation in allopatry as indicated by their present distribution pattern. A single prezygotal isolating mechanism is enough to finalize the speciation process no matter how similar the diverging populations are in other characters such as external morphology or ecological traits, e.g., habitat preference, diet, or activity patterns (Wilson, 2001). In anurans the male advertisement call serves as a very effective isolating mechanism to avoid hybridization among similar species under natural conditions (Köhler et al. 2017). Thus, the discovery that there are three distinct male advertisement calls among the species referred to as P. martensii in mainland Southeast Asia is strong evidence for the presence of three distinct species that are not compatible reproductively. In this sense it could be expected that the variation in mtDNA data would be congruent with the results of the bioacoustical analyses. The mtDNA data found a fourth clade of frogs (our Clade 2 with specimens from northern Thailand, southern China, Vietnam, Laos, and Cambodia) that differed from the other three clades by a mean genetic distance of 16S of 3.5–5.1% and therefore qualifies as candidate species (sensu Fouquet et al. 2007). However, since we did not find additional support for the recognition of this clade as a distinct evolutionary species, we tentatively assigned it to the species it is genetically most closely related too (i.e., P. martensii). Additional research including nuclear genetic data is needed to evaluate these northeastern populations.

In previous publications, the geographic distribution of Phrynoglossus magnapustulosus was given as “Nakhon Phanon, Ubon, Loei, and Chiang Mai provinces, Thailand“ (e.g., Taylor 1962; Frost 2020) or “northern and northeastern Thailand” (Khonsue and Thirakhupt 2001). According to our molecular genetic data and also based upon our field observations in Chiang Mai, this statement is erroneous and probably based on misleading characters of external morphology that were considered to be diagnostic for this species such as the presence of “pearly tubercles” on the dorsum (Taylor 1962). In our series of topotypic specimens of P. magnapustulosus we found that dorsal skin texture varies considerably. Some specimens virtually have only a few low tubercles on the dorsum whereas other have distinct ones, pearl-tipped or not (see Fig. 12). A similar degree of variation in dorsal skin texture was observed in individuals of P. martensii (Fig. 13). Thus, we dismiss this character as useful for distinguishing among species of Phrynoglossus. As far as known, P. magnapustulosus is restricted to the Khorat Plateau. Other herpetofaunal species endemic to the Khorat Plateau include Enhydris subtaeniata and Malayemys khoratensis (Murphy and Voris 2014; Ihlow et al. 2016).

The species of Phrynoglossus (or Occidozyga) that are supposedly endemic to the islands of the Sunda Archipelago and the Philippines were beyond the scope of this study. Future studies need to address the phylogenetic relationships and verify their generic assignment as proposed here. The integrative taxonomic approach including also bioacoustic and genetic data along with a thorough analysis of the geographic variation of external morphology will be useful in order to clarify the taxonomic status of these populations.

The evidence for recognizing Phrynoglossus and Occidozyga as distinct evolutionary units at genus level is supported by the monophyly of the two taxa based on the analyses of mtDNA in this study. Furthermore, there are several diagnostic morphological characters that define each genus, some of which may turn out to be apomorphic for one or the other genus. The conspicuous skin morphology of Occidozyga as well as its unique tongue shape are such candidates. Regarding the documented mode of amplexus, finding two different modes in these supposedly closely related genera is interesting. The inguinal amplexus mode that we documented in P. magnapustulosus and P. myanhessei is supported as an autapomorphy of the genus Phrynoglossus by the observation of an inguinal amplexus mode also in P. sumatranus by Eto and Matsui (2012). According to a large-scale analysis of the amplexus mode across all groups of anuran amphibians, reveal the inguinal amplexus as the basal state to all Anura (Carvajal-Castro et al. 2020). However, the latter authors gave the axillary amplexus as the only mode found within the family Dicroglossidae obviously not being aware of the inguinal amplexus mode in Phrynoglossus. Given the phylogenetic position of Phrynoglossus in the amphibian tree of life (Pyron and Wiens 2011), we interpret the amplexus mode in this genus as a reversal from the axillary mode, and thus as an autapomorphy of this genus.

With this description of a new species we provide the complete, annotated mitochondrial genome, a draft genome and 25× coverage of short-read genome resources of the holotype. This fundamental genomic characterization of the species based on the name-bearing specimen will be an invaluable genomic resource for future taxonomic and evolutionary studies. The quality of the nuclear genome assembly provided for the holotype of Phrynoglossus myanhessei is sufficient to genetically characterize the new species on the basis of the name-bearing specimen. The genome data describe the entire genetic variation, including heterozygosity of the holotype individual and make it possible to extract any genetic locus for future taxonomic or phylogenetic studies specifically of occidozygine frogs and frogs in general.

Following the first example from Köhler et al. (2021) we reiterate our proposition that wherever feasible, descriptions of new species should be accompanied by genomic data and a draft genome assembly from the respective holotype as an important resource to future biodiversity research. The investment in a draft genome of newly described holotypes ensures that the genetic fingerprint of newly described species is captured and preserved long-term based on the name-bearing specimen. Furthermore, Köhler et al. (2021) suggest to balance between quality and investment regarding draft genome quality. Therefore the quality of the nuclear genome assembly from 25× coverage is not very high by purpose and still provides a basis for further studies. In addition the complete mitochondrial genome is a valuable resource e.g., for further phylogenetic studies. In the near future, adding genomic data to traditional phenotypic studies will likely become a standard, as costs and efforts to produce them drop dramatically. And the time to start is now.