Bioacoustics Reveal Hidden Diversity in Frogs: Two New Species of the Genus Limnonectes from Myanmar (Amphibia, Anura, Dicroglossidae)

Striking geographic variation in male advertisement calls was observed in frogs formerly referred to as Limnonectes doriae and L. limborgi, respectively. Subsequent analyses of mtDNA and external morphological data brought supporting evidence for the recognition of these populations as distinct species. We describe two new frog species of the genus Limnonectes (i.e., L. bagoensis sp. nov. and L. bagoyoma sp. nov.) from Myanmar. Limnonectes bagoensis sp. nov. is closely related to L. doriae whereas L. bagoyoma sp. nov. is closely related to L. limborgi. Results of this integrative study provide evidence for the presence of additional undescribed species in these species complexes but due to the lack of bioacoustical data, we consider these additional diverging populations as candidate species that need further study to resolve their respective taxonomic status. Both new species are distributed in Lower Myanmar. Limnonectes doriae is restricted to southern Myanmar along the Malayan Peninsula whereas L. limborgi is known to occur in eastern Myanmar and northwestern Thailand. The remaining populations formerly referred to as either L. doriae or L. limborgi are considered representatives of various candidate species that await further study. We further provide a de novo draft genome of the respective holotypes of L. bagoensis sp. nov. and L. bagoyoma sp. nov. based on short-read sequencing technology to 25-fold coverage.


Introduction
The anuran family Dicroglossidae has its center of distribution and greatest species diversity in South, Southeast, and East Asia as well as on the islands of the Sunda Shelf and the Philippines but is also known from northwestern and sub-Saharan Africa across the southern Arabian Peninsula to Pakistan [1]. The great number of new species described every year (e.g., [2][3][4]) shows that this group of frogs is understudied. Obviously, much more field and lab work is needed to evaluate the actual diversity and better understand MgCl 2 , 4 µL 2.5 mM dNTPs (Invitrogen), 0.5 µL (containing 2.5 units) Taq Polymerase (PeqLab), and 1 µL of the primer (16S: forward: L2510, 5 -CGCCTGTTTATCAAAAACAT-3 ; reverse: H3056, 5 -CCGGTCTGAACTCAGATCACGT-3 ; from Eurofins MWG Operon). DNA extraction, PCR, and sequencing were performed at SMF for samples from Myanmar and at CUMZ for samples from Thailand. We generated 43 new sequences for this study (see Appendix B). Additionally, we downloaded relevant 16S sequences from GenBank (Appendix B). Our dataset contains topotypic specimens of Limnonectes limborgi and neartopotypic specimens of L. doriae.
We aligned the sequences with MUSCLE [30] using the default settings in Geneious 6.1.2. [31]. For software applications, sequence data formatting was converted using the online server Alter [32]. The best substitution model for each gene was identified using PartitionFinder2 [33], with linked branch lengths via PhyML 3.0 [34]. Model selection used the corrected (for finite sample size) Akaike Information Criterion (AICc) [35]. Fejervarya limnocharis (GenBank number AF206466) was used as an outgroup [14].
Bayesian Inference analyses (BI) used MrBayes 3.2 [36] with five runs with eight chains. The first 25% of trees were discarded as burn-in. MCMC runs used an initial set of 1,000,000 generations with sampling every 500 generations and adding 500,000 generations until chains reached convergence. Convergence was considered achieved when the standard deviation of split frequencies was 0.015 or less. Additionally, convergence was diagnosed by PSRF (Potential Scale Reduction Factor), which should approach 1.0 as runs converge [37]. We used the IQTree webserver [38] to run a Maximum Likelihood (ML) analysis using 10,000 ultrafast Bootstrap approximation (UFBoot) replicates with 10,000 maximum iterations and minimum correlation coefficient of 0.99 [39], plus 10,000 replicates of Shimodaira-Hasegawa approximate likelihood ratio (SH-aLRT), which proved to be accurate with a high statistical power [34]. We used FigTree 1.3.1 (http: //tree.bio.ed.ac.uk/software/figtree/, accessed on 2 April 2021) for viewing the trees. We estimated evolutionary genetic divergence, computing uncorrected pairwise distances with MEGA 7.0.26 [40] to assess the degree of intra-and interspecific differences, using a Bootstrap estimation method of 10,000 replications. We built a species tree using BEAST 2.4.7 [41,42] with 1,000,000 generations for the MCMC model. The resulting tree was visualized using DensiTree 2.2.6 [43].
Whole-genome sequencing and assembly: Genomic DNA was extracted from muscle tissue of the respective holotypes (SMF 104090 and SMF 106034) of the two new species of Limnonectes described herein according to standard phenol/chloroform procedures [44]. The resulting DNA was then resuspended in TE buffer (10 mM Tris-Cl, 0.1 mM EDTA) and stored at −20 • C. Quality checks for high molecular weight DNA were performed by agarose gel electrophoresis [44]. The DNA samples were shipped on dry ice to Novogene (UK) for short-read Illumina genome sequencing. One genomic library for each of the holotypes (insert size: 350 bp) was prepared and 150 bp paired-end reads were sequenced on an Illumina NovaSeq 6000 platform (San Diego, CA, USA). A long-read genomic sequencing was performed for the holotype SMF 106034 using PacBio Sequel sequencing platform.
A k-mer profile analysis was performed from the raw reads with the help of Jellyfish 2.3.0 software [45] and GenomeScope webserver [46]. The quality checks of the raw reads were performed by trimming the low-quality regions and adapter sequences and filtered for possible contamination using Trimmomatic 0.39 [47] and Kraken 2.0.9 [48], respectively. Nuclear genome assembly of the holotype SMF 106034 was conducted in the SPAdes assembler [49]. Scaffolds smaller than 500 in length were discarded. To assemble the nuclear genome of the holotype SMF 104090, the short reads were mapped over the genome of SMF 106034 with the help of Bowtie2 software [50]. The resulting SAM file was then converted to a sorted bam file by using Samtools [51]. The consensus sequence was then generated with the help of Samtools and BCFtools [52] software. Scaffolds smaller than 500 in length were discarded. The mitochondrial genome of SMF 104090 was assembled from the genomic data separately using NOVOplasty 4.2 [53]. For the holotype, SMF 106034, mitochondrial genome of the holotype SMF 104090 was taken as reference and Pacbio reads which had homology to the reference were fetched using NCBI Blast 2.11 [54]. These reads were then used in Geneious Prime software 2020.2.3 (www.geneious.com) to generate the consensus sequence. The consensus sequence was then polished by Illumina short reads using Pilon v1.23 software [55]. Annotation of the mitochondrial genome was performed by MITOS2 webserver [56] and visualization pictures were generated using Geneious Prime software 2020.2.3.

Results and Conclusions
The alignment of the 16S sequences used in this study had a length of 583 bp. General Time Reversible using a discrete Gamma distribution with five rate categories and by assuming that a certain fraction of sites are evolutionarily invariable (GTR + I + G) was determined to be the best fitting model of sequence evolution. The trees generated by BI, ML, and BEAST analyses agree well at most well-supported nodes, but show differences in branch arrangement at weakly supported nodes (Figures 1 and 2).  Limnonectes doriae species complex. Among the specimens formerly referred to as Limnonectes doriae, three major clades are recognizable ( Figure 1). Clade 1 contains specimens from lower Myanmar in Bago and Yangon Regions ( Figure 3). Clade 2 includes samples from southern Myanmar including the Malayan Peninsula and adjacent regions in extreme western Thailand; only in the ML analysis, these specimens formed a weakly supported clade, whereas, in the Bayes analysis, these are recovered as a polytomy. Finally, Clade 3 contains specimens from western Myanmar in Rakhine state. The genetic distances among these clades vary from 1.9% to 2.4% (Table 1). We consider these three clades as candidate species, that should be assessed by including additional data types. Clade 3 includes specimens from near the type locality of L. doriae and is therefore considered to represent the "true" L. doriae.  Limnonectes doriae species complex. Among the specimens formerly referred to as Limnonectes doriae, three major clades are recognizable ( Figure 1). Clade 1 contains specimens from lower Myanmar in Bago and Yangon Regions ( Figure 3). Clade 2 includes samples from southern Myanmar including the Malayan Peninsula and adjacent regions in extreme western Thailand; only in the ML analysis, these specimens formed a weakly supported clade, whereas, in the Bayes analysis, these are recovered as a polytomy. Finally, Clade 3 contains specimens from western Myanmar in Rakhine state. The genetic distances among these clades vary from 1.9% to 2.4% (Table 1). We consider these three clades as candidate species, that should be assessed by including additional data types. Clade 3 includes specimens from near the type locality of L. doriae and is therefore considered to represent the "true" L. doriae.     Bioacoustical data are available for our Clades 1 and 2 (see also Table 2). The male advertisement calls of these two clades are strikingly different ( Figure 4; Table 2): The call of SMF 104084 from Clade 2 from the region of the species' type locality ( Figure 4A) consists of a single series of 15-26 notes, each note varying from 51 to 103 ms (mean 65 ms) in duration with 11.6-12.1, mean 11.8, notes per second, their amplitude rising during the first 3-8 notes, then remaining about the same level. There is no offset single note anterior to the series of notes. The call has a duration of 1.25-2.24 s and a dominant frequency of 689-990 Hz, a mean of 947 Hz. The analyzed call of a Clade 2 male (ZMKU 1552) from a locality in Mae Hong Son Province, Thailand, about 320 airline km NNE of the SMF 104084 locality (see Figure 3), is similar in structure (i.e., a single series of notes without an off-set single note anterior to this series), dominant frequency (840-1055 Hz, mean 902 Hz), note duration (60-72 ms, mean 68 ms), and number of notes per second (11.7). It differs in the length of the call (6.75-7.71 s) and thus in the number of notes per call (79-90, mean 83.7). Whether this difference in call duration is of individual or geographical nature can only be clarified once the calls of additional individuals and localities can be included in the analyses. The call of Clade 1 males ( Figure 4C,D) consists of two portions that are separated by a break of 573-1043 ms (mean 792 ms). The first portion of the call is a single unpulsed sound emission (note duration 225-291 ms) and with a slightly weaker amplitude than the second portion which consists of 4-9 notes. Depending on the number of notes, the second portion of the call has a duration of 1202 (4 notes) to 1954 ms (9 notes), mostly around 1600 ms (7 notes); each note varying from 155 to 239 ms in duration with 4. from the different localities agree very well in all examined characters (see Table 2). The mean genetic distance (16S) between Clade 1 and 2 is 1.9% (Table 1).   In external morphology, specimens of these three clades are very similar but show differences in some morphometric values ( Figure 5) as well as in the amount of toe webbing and in some details of coloration (see Table 3 and respective diagnosis sections below). Additionally, there is sexual dimorphism evident in several morphometric characters such as SVL, IOD, TYD, and HL. In particular, individuals of Clades 1 and 2 differ in the pattern on the posterior surface of the thigh (distinct and contrasting in Clade 1 versus weak and indistinct in Clade 2), development of a dermal fringe along the outer edge of Toe 5 (absent or weak in Clade 1 versus distinct in Clade 2), and the gular coloration in adult males (pale in Clade 1 versus dark gray in Clade 2). However, since our sample sizes are small for characters of external morphology, these putative differences between the clades need to be evaluated with more specimens in order to test their actual diagnostic value.   Thus, based on the results of the integrative analyses of mtDNA data, morphology, and bioacoustics, we recognize Clades 1 and 2 as species-level units. Clade 3 remains a candidate species awaiting further study, especially based on nuclear DNA data and bioacoustics. Based on the respective type locality, our Clade 2 is assigned to Limnonectes doriae whereas our Clade 1 represents an undescribed species for which no name is available. Thus, we describe the Limnonectes related to L. doriae from lower Myanmar (Bago Region and Yangon Region, our Clade 1) as a new species below.
Limnonectes limborgi species complex. Among the specimens formerly referred to as Limnonectes limborgi, we found nine major clades that we recognize as candidate species (Figures 1 and 2). The genetic distances among these clades vary from 1.8% to 4.9%. (Table 4). Of these clades, five (i.e., Clades 1,2,4,8,9) are represented only by sequences (mostly from GenBank) and we are unable to study these in any detail due to the lack of morphological or bioacoustical data. Therefore, these clades were not included in the integrative taxonomic analysis. Table 4. Mean genetic distances (in %) of Clades 1-12 (see Figure 1) of frogs referred to as Limnonectes limborgi as well as L. hascheanus based on the 16S dataset. Cl = Clade; hasch. = hascheanus. See text for details.

Cl_4
Cl_8 Of the remaining four clades (i.e., 3, 5-7), we also examined voucher specimens for characters of external morphology. Clade 7 includes topotypic specimens from the neotype locality of L. limborgi and are therefore considered to represent the "true" L. limborgi. Bioacoustical data are available for our Clades 6 and 7 (see also Table 5). The male advertisement calls of these two clades differ clearly (Figure 7): the call of males of "true" L. limborgi (Clade 7) consists of a series of three to five notes whereas males from Clade 6 emit a single "quaaak". The two calls differ also in the length of the notes (0.218-0.252 s, usually <0.250 s, in males of Clade 6 versus 0.241-0.419 s, usually >0.260 s, in males of Clade 7). The genetic distance (16S) between these two clades is 2.8% (Table 5). Clade 7 includes topotypic specimens from the neotype locality of L. limborgi and are therefore considered to represent the "true" L. limborgi. Bioacoustical data are available for our Clades 6 and 7 (see also Table 5). The male advertisement calls of these two clades differ clearly (Figure 7): the call of males of "true" L. limborgi (Clade 7) consists of a series of three to five notes whereas males from Clade 6 emit a single "quaaak". The two calls differ also in the length of the notes (0.218-0.252 s, usually <0.250 s, in males of Clade 6 versus 0.241-0.419 s, usually >0.260 s, in males of Clade 7). The genetic distance (16S) between these two clades is 2.8% (Table 5).    In external morphology, specimens of these five clades differ in details such as in the shape and orientation of the supratympanic fold, the development of the tarsal fold as well as in the amount of toe webbing, and in some details of coloration (see Table 3 and respective diagnosis sections below). Additionally, they show differences in some morphometric values (Figure 8), Finally, there is sexual dimorphism evident in several morphometric characters, most evident in relative head width. In external morphology, specimens of these five clades differ in details such as in the shape and orientation of the supratympanic fold, the development of the tarsal fold as well as in the amount of toe webbing, and in some details of coloration (see Table 3 and respective diagnosis sections below). Additionally, they show differences in some morphometric values (Figure 8), Finally, there is sexual dimorphism evident in several morphometric characters, most evident in relative head width.
Thus, based on the results of the integrative analyses of mtDNA data, morphology, and bioacoustics, we recognize Clades 6 and 7 as species-level units whereas the other clades remain candidate species awaiting further study, especially based on nuclear DNA data and bioacoustics. Based on the respective type locality, our Clade 7 is assigned to Limnonectes limborgi whereas our Clade 6 represents an undescribed species for which no name is available. Thus, we describe the Limnonectes related to L. limborgi from lower Myanmar (Yangon and Bago Regions) as a new species below.
This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the International Commission on Zoological Nomenclature (ICZN). The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information can be viewed through any standard web browser by appending the LSID to the prefix http://zoobank.org. The LSID for this publication is as follows: urn: lsid:zoobank.org:pub:4C463126-CD59-4935-96E3-11AF4131144C. The LSID registration and any associated information can be viewed in a web browser by adding the LSID to the prefix "http://zoobank.org/".   Diagnosis. A species of the genus Limnonectes to which it is assigned because of its inferred phylogenetic position, males with hypertrophied heads, and the presence of odontoids on the lower jaw in adult males [19,57]. Limnonectes bagoensis sp. nov. is assigned to the subgenus Elachyglossa [17] because of its close phylogenetic position to the type species of this subgenus (i.e., L. gyldenstolpei). Limnonectes bagoensis sp. nov. differs from all congeners by having (1) a medium body size (males 32-49 mm; females 30-47 mm); (2) slightly enlarged toe disks; (3) adult males with small odontoids in the lower jaw; (4) adult males without a knob-like or flap-like structure (caruncle) in the interorbital and parietal region; also lacking swelling in the occipital region; (5) in adult males head not conspicuously enlarged; (6) webbing formula I 1-2 II 1-2.4 III 1.2-3 IV 3.3-2 V to I 1.8-2.2 II 1.5-2.8 III 2.2-3.2 IV 3.5-2 V; (7) a feeble dermal fringe along the outer edge of Toe 5; (8) male advertisement call consists of two portions that are separated by a break of slightly less than 1 s. The first portion of the call is a single note with a duration of 225-291 ms whereas the second portion consists of 4-9 notes with a duration of 1200 (4 notes) to about 2000 ms (9 notes) with about 4.5 notes per second; the dominant frequency is at 861-990 Hz (mean 945 Hz). Limnonectes bagoensis differs from all other species currently assigned to the subgenus Elachyglossa except L. doriae by lacking a knob-like or flap-like structure (caruncle) in the interorbital and parietal region in adult males (versus such structure present in adult males), and also lacking swelling in the occipital region (versus such swelling present in adult males). Limnonectes bagoensis differs from L. doriae in having a male advertisement call that consists of two portions, separated by a break of slightly less than 1s, the second portion with a series of 4-9 notes at about 4.5 notes per second (versus a single series of notes in L. doriae with about 12 notes per second); gular coloration in adult males not dark (versus gular region dark gray in adult males of L. doriae).
Description of the holotype (Figure 9). Adult male, as indicated by the presence of odontoids, broad head, and folds along mandibular arch; SVL 40.8 mm; habitus robust; head broad, about as wide as long, ratio HL/HW 1.07; snout nearly rounded in dorsal view, projecting beyond lower jaw, rounded in profile; nostril dorsolateral, closer to tip of snout than eye; canthus rounded; ratio EYD/SVL 0.10; IOD (4.8 mm) greater than width of upper eyelid (3.1 mm); tympanum distinct, slightly recessed relative to skin of temporal region, tympanic rim distinctly elevated relative to tympanum; ratio TYD/EYD 0.85; four to five vomerine teeth on slightly raised oblique ridges between choanae, separated from choanae by about 1/4 length of one group, gap between groups about 1/2 length of one group; tips of all four fingers rounded, not expanded into discs; relative finger lengths III > I > IV > II; no webbing; distinct subarticular tubercles; supernumerary tubercles ab-sent; palmar tubercle distinct, flat, bifid; thenar tubercle large, elongate, raised, about twice the size of palmar tubercle; tips of toes rounded, slightly expanded into discs; relative toe lengths IV > III > V > II > I; feet almost fully webbed, webbing formula I 1-1.5 II 0.5-2 III 1-2.5 IV 2.5-1 V; a well-developed dermal ridge along lateral side of Toe V from level of outer metatarsal tubercle to distal subarticular tubercle; large, flap-like inner metatarsal tubercle, >0.5 length of first toe; distinct fold on distal one-half of tarsus beginning at inner metatarsal tubercle; outer metatarsal tubercle tiny; skin on dorsal and lateral surfaces of head, body, and limbs smooth except for a few scattered tubercles in sacral region and on dorsal surface of shank; skin on throat and venter smooth; distinct supratympanic ridge from    Etymology. The species name "bagoensis" refers to the city of Bago where the holotype of this species was collected, and ensis denoting place.
Natural history notes. At the type locality at night time, adult males were heard calling on the surface hidden under flat rocks. Figure 11 shows the holotype (SMF 104090) that continued calling even after the covering rocks had been removed. Additional individuals were encountered sitting on the forest floor at night without cover. Etymology. The species name "bagoensis" refers to the city of Bago where the holotype of this species was collected, and ensis denoting place.
Natural history notes. At the type locality at night time, adult males were heard calling on the surface hidden under flat rocks. Figure 11 shows the holotype (SMF 104090) that continued calling even after the covering rocks had been removed. Additional individuals were encountered sitting on the forest floor at night without cover. Figure 11. Holotype of Limnonectes bagoensis n. sp. (SMF 104090) in life, at its calling site after the covering rocks had been removed. Figure 11. Holotype of Limnonectes bagoensis n. sp. (SMF 104090) in life, at its calling site after the covering rocks had been removed. Geographic Distribution and Conservation. As currently known, Limnonectes bagoensis sp. nov. is known from several localities in Bago Region, Myanmar. Given how little we know about this species, we classify Limnonectes bagoensis sp. nov. as Data Deficient based on the IUCN Red List Categories and Criteria [58].
Genomic characterization. Whole-genome sequencing and assembly (Table 6): Illumina sequencing produced 302,829,670 short-read pairs with a total sequence amount of 90 Gb. K-mer analysis estimated the genome size of 2.16 Gb with 0.78% heterozygosity. (Figure 14) The mitochondrial genome was assembled into one 17,480 bp long circular contig. A total of 13 protein-coding genes, two rRNAs, 22 tRNAs were annotated in the mitochondrial genome ( Figure 12). A D-loop region could not be annotated in the mitochondrial genome sequence. Geographic Distribution and Conservation. As currently known, Limnonectes bagoensis sp. nov. is known from several localities in Bago Region, Myanmar. Given how little we know about this species, we classify Limnonectes bagoensis sp. nov. as Data Deficient based on the IUCN Red List Categories and Criteria [58].
Genomic characterization. Whole-genome sequencing and assembly (Table 6): Illumina sequencing produced 302,829,670 short-read pairs with a total sequence amount of 90 Gb. K-mer analysis estimated the genome size of 2.16 Gb with 0.78% heterozygosity. (Fig 14) The mitochondrial genome was assembled into one 17,480 bp long circular contig. A total of 13 protein-coding genes, two rRNAs, 22 tRNAs were annotated in the mitochondrial genome (Figure 12). A D-loop region could not be annotated in the mitochondrial genome sequence.   The nuclear genome obtained by reference-based assembly consisted of 1,941,736,683 bases fragmented in 647,264 scaffolds. The longest scaffold was 42,771 bases and the N50 scaffold was 4270 bases whereas GC content was only 34.74%. Except for the high-quality mitochondrial genomes, the simple assembly from Illumina short-reads was predicted to be 26.2% complete as per the BUSCO gene space [59]  Diagnosis. A species of the genus Limnonectes to which it is assigned because of its inferred phylogenetic position (Figure 1), males with hypertrophied heads, and the presence of odontoids on the lower jaw in adult males [19,57]. Limnonectes bagoyoma sp. nov. is assigned to the subgenus Taylorana because of its close phylogenetic position to the type species of this subgenus (i.e., L. limborgi; Figure 1). Limnonectes bagoyoma sp. nov. differs from all congeners by having (1) a small body size (males 29.3-30.7 mm; females 23.1-26.6 mm); (2) males with slightly enlarged odontoids; (3) inner metatarsal tubercle large, raised, usually >0.5 length of the first toe; (4) webbing formula I 1-2 II 1-2.4 III 1.2-3 IV 3.3-2 V to I 1.8-2.2 II 1.5-2.8 III 2.2-3.2 IV 3.5-2 V; (5) females with clutches of enlarged, non-pigmented eggs; (6) male advertisement call consisting of a single note with a duration of 220-250 ms and a dominant frequency mostly in the 1300-1400 Hz range. Limnonectes bagoyoma differs from the remaining species of the subgenus Taylorana (see also [15]) as follows: from L. limborgi by having a male advertisement call consisting of a single note with a duration of 220-250 ms (versus call consisting of a series of 3-5 notes, each note with a duration of usually >260 ms in L. limborgi), and no dark lateral face mask (versus dark face mask present). Limnonectes bagoyoma differs from L. hascheanus by its larger body size, reaching 30 mm SVL in adult males (versus SVL <26 mm in both sexes in L. hascheanus), and by having more toe webbing, usually three phalanges free of webbing on the medial side of Toe 4 (versus more than three phalanges free of webbing on the medial side of Toe 4). Limnonectes bagoyoma differs from L. liui by its smaller body size, SVL not exceeding 31 mm in both sexes (versus SVL 32.0-38.5 mm in males of L. liui, only female in type series measuring 32.7 mm), toe webbing well-developed (versus rudimentary), and inner metatarsal tubercle usually >0.5 length of the first toe (versus "just half" of the first toe in L. liui according to [15]). Limnonectes bagoyoma differs from L. medogensis by having the nostril positioned closer to the snout than to the eye (versus closer to eye in L. medogensis), vomerine teeth present (versus absent), venter patternless (versus marbled markings present), toe webbing well-developed (versus rudimentary), inner metatarsal tubercle distinctly shorter than the first finger (versus about length of Toe 1), and tarsal fold indistinct but present (versus absent). Limnonectes bagoyoma differs from L. xizangensis by having the toe tips enlarged into small disks (versus toe tips pointed), vomerine teeth present (versus absent), toe webbing well-developed (versus absent), and the presence of tiny tubercles on the dorsal surface of the shank (versus smooth), venter patternless (versus black reticulated markings present), and smooth skin on venter (granular skin on venter). Limnonectes bagoyoma differs from L. alpinus by having the toe webbing well-developed (versus toe webbing absent in L. alpinus), vomerine teeth present (versus absent), and tympanum distinct (versus hidden).
Description of the holotype (Figure 13). Adult female, as indicated by the rather slender head, absence of odontoids, and absence of folds along mandibular arch; SVL 26.8 mm; habitus robust; head narrow, slightly longer than wide, ratio HL/HW 1.14; snout subovoid in dorsal view, projecting beyond lower jaw, rounded in profile; nostril dorsolateral, closer to tip of snout than eye; canthus rounded; ratio EYD/SVL 0.13; IOD (2.7) greater than width of upper eyelid (1.8 mm); tympanum distinct, slightly recessed relative to skin of temporal region, tympanic rim distinctly elevated relative to tympanum; ratio TYD/EYD 0.59; four to five vomerine teeth on slightly raised oblique ridges between choanae, separated from choanae by about 1/2 length of one group, gap between groups about 1/2 length of one group; tips of all four fingers rounded, not expanded into discs; relative finger lengths III>I>II>VI; no webbing; distinct subarticular tubercles; supernumerary tubercles absent; palmar tubercle distinct, flat, bifid; thenar tubercle large, elongate, raised, about twice the size of palmar tubercle; tips of toes rounded, slightly expanded into discs; relative toe lengths IV>III>V>II>I; feet almost fully webbed, webbing formula I 1.5-2 II 1.5-2.5 III 2. Coloration after about one and half years preservation in 70% ethanol was recorded as follows: Dorsal and lateral ground color of the head and body Drab (19), but paler, with a suffusion of Smoke Gray (266) on the dorsal head and with indistinct Vandyke Brown (281) stipples on dorsum and as a broken interorbital bar as well as a band below canthus rostralis and along the supratympanic ridge; Glaucous (289) vertical bars on the lower lip; dorsal surfaces of the thigh and shank Drab (19) with indistinct Glaucous (289) transverse bars; ventral surfaces of head, body, and limbs Cream White (52); palmar surfaces Ground Cinnamon (270); plantar surfaces Natal Brown (49).
Variation. The paratypes agree well with the holotype in general appearance; morphometrics and coloration (see Table 3). The two adult paratype males both have broad heads, folds along the mandibular arch, distinct odontoids, and four to five vomerine teeth. The coloration of a topotypic adult male (SMF 106035; Figure 14B Etymology. The species name "bagoyoma" refers to the type locality Bago Yoma, a large but relatively low mountain range that runs in a north-south direction between the Irrawaddy (=Ayeyarwady) and the Sittaung River in Myanmar. It is likely that Limnonectes bagoyoma is geographically restricted to this mountain range. Diversity 2021, 13, x FOR PEER REVIEW 28 of 39  Coloration after about one and half years preservation in 70% ethanol was recorded as follows: Dorsal and lateral ground color of the head and body Drab (19), but paler, with a suffusion of Smoke Gray (266) on the dorsal head and with indistinct Vandyke Brown (281) stipples on dorsum and as a broken interorbital bar as well as a band below canthus rostralis and along the supratympanic ridge; Glaucous (289) vertical bars on the lower lip; Natural history notes. At the type locality, the specimens were collected during the daytime in leaf litter in the rainforest. The two males we collected were heard calling before we caught them. Each male was sitting in a shallow depression, covered by leaf litter. Other frog species collected at the type locality include Duttaphrynus melanostictus, Fejervarya orissaensis, Ingerana tenasserimensis, Limnonectes bagoensis, and Leptobrachium smithi. Geographic Distribution and Conservation. As currently known, Limnonectes bagoyoma sp. nov. is only known from the vicinity of its type locality. Given the little we know about this species, we classify Limnonectes bagoyoma sp. nov. as Data Deficient based on the IUCN Red List Categories and Criteria [58].
Genomic characterization. Whole-genome sequencing and assembly (Table 6): Pacbio sequencing resulted in 16,749,860 reads of size 90 Gb whereas Illumina sequencing produced 278,081,641 short read pairs with a total sequence amount of 83.4 Gb. K-mer analysis estimated the genome size of 2.4 Gb with 0.78% heterozygosity (Figure 15). A mitochondrial genome of 17,406 bases was assembled using Pacbio reads which matched to the complete annotated mito-genome of the holotype SMF 104090. The annotation resulted in 13 protein-coding genes, three rRNAs, 22 tRNAs and three replication origin sites. A D-loop region was also annotated in the mitochondrial genome sequence. (Figure 16)

Discussion
The present study presents further examples for putative geographically wide-spread species that actually represent species complexes of several taxa that are similar in external morphology but distinct in bioacoustics and genetics ( [5,21,60]). To resolve the diversity in these complexes, additional field research is needed to obtain fresh voucher specimens from across their range for genetic and bioacoustics characterization of the geographically distinct populations. The case of Limnonectes bagoensis reported here, shows that in some cases the genetic differentiation in mtDNA can be rather low (i.e., just 2% in 16S between L. bagoensis and its putatively most closely related sister species L. doriae). However, given their rather drastically different male advertisement calls, there is no doubt that these two forms are distinct species. This hypothesis is reinforced by the observation that the male advertisement call of L. bagoensis does not vary across a large geographic distance (i.e., almost 200 air-line km). In the case of the species pair L. limborgi and L. bagoyoma, the genetic differentiation in mtDNA data (16S) is close to 3%, and the species are distinct in bioacoustics, too. In the course of this study, we identified additional candidate species in either species complex. Due to the lack of additional evidence lines we refrained from naming any of these groups defined only by mtDNA data. For "L. limborgi" from northern Vietnam (which corresponds with our Clade 10) the male advertisement call has been described as "single notes, repeated in very large intervals from 30 s to several minutes", an observation confirmed by [61]. The latter author reports the call to have a dominant frequency of 1000-1100 Hz and a duration of 278-331 ms. Both accounts describe the male sitting in a hole with just the head visible after the covering leaves had been removed. Except for some details, this call description agrees very well with our analyses of male advertisement calls of L. bagoyoma and is unlike the call of the geographically closer L.
limborgi from the neotype locality of the latter species. Both the Vietnamese "L. limborgi" and L. bagoyoma have a male advertisement call consisting of a single note as opposed to a series of 3-5 notes in "true" L. limborgi. The calls of males from Vietnam have a dominant frequency of 1000-1100 Hz and a duration of 278-331 ms versus a dominant frequency mostly in the 1300-1400 Hz range and a duration of 220-250 ms in L. bagoyoma. The genetic distance in 16S between these two clades with similar calls (i.e., our Clades 6 and 10) is 3.2% whereas it is 2.5% between the Vietnamese "L. limborgi" and "true" L. limborgi. In order to determine whether the two similar calls are cases of convergence or have both remained in the ancestral state, it would be essential to know which call type represents the plesiomorphic and which one the apomorphic character state. A more robust genetic phylogeny, including nuclear markers as well as bioacoustical data from additional populations of this species complex, is needed to gain a better understanding of call evolution in this group of frogs. In frogs, bioacoustic data and analyses are as useful as are molecular genetics and often yield initial information on the presence of cryptic taxonomic diversity [60,62,63]. The male advertisement call in frogs works as an effective isolating mechanism to avoid hybridization among similar species under natural conditions [28].
With this description of two new species, we provide the complete, annotated mitochondrial genome as well as a draft genome of short-read genome resources of the holotypes of both species described herein. This genomic characterization of the species based on the name-bearing specimen will be a genomic resource for any forthcoming studies.
Funding: This research was funded partly by Chulalongkorn University: CU_GR_63_66_23_10 and also partly financially supported by the Sci-Super VI fund from Faculty of Science, Chulalongkorn University. The genomic work was funded by the Centre for Translational Biodiversity Genomics (LOEWE-TBG) through the program "LOEWE-Landes-Offensive zur Entwicklung wissenschaftlichokonomischer Exzellenz" of Hesse's Ministry of Higher Education, Research, and the Arts.

Institutional Review Board Statement:
The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Chulalongkorn University Animal Care and Use Committee of Chulalongkorn University, Bangkok (protocol code 2123010; 1 January 2021).

Data Availability Statement:
No new data were created or analyzed in this study. Data sharing is not applicable to this article.