A New, Morphologically Cryptic Species of Fanged Frog, Genus Limnonectes (Amphibia: Anura: Dicroglossidae), from Mindoro Island, Central Philippines

We describe a new species of fanged frog (genus Limnonectes) from Mindoro and Semirara Islands, of the Mindoro Pleistocene Aggregate Island Complex, of the central Philippines. Although morphologically indistinguishable from its closest relative, Limnonectes acanthi, of the Palawan faunal region, the two species can be readily diagnosed on the basis of spectral (dominant frequency) and temporal (pulse number and structure) properties of their advertisement calls, and their allopatric insular geographic ranges on permanently separate geological platforms which have not been connected by dry land in the recent geological past—all of which we interpret as congruent and independent lines of evidence supporting our recognition of two independently evolving evolutionary lineages (species). Ribosomal RNA mitochondrial gene sequences were used to provide genetic identification of specimens and estimate phylogenetic relationships; genetic divergences between Palawan and Mindoro faunal regions exceed those estimated among other, uncontroversial, phenotypically distinct Philippine species with equivalent levels of allopatry and biogeographic isolation. The recognition of the new species further emphasizes the degree to which even well-studied Philippine landmasses still harbor unrecognized biodiversity, and suggests that other widespread Philippine fanged frogs should be scrutinized for non-traditional diagnostic character differences (mate-recognition signal divergence, ecological differences, larval characteristics, life-history trait variation), especially when their geographic ranges span the archipelago's permanent, deep-water trenches, which define its well-characterized Pleistocene Aggregate Island Complexes.

Over the last two decades, numerous proposed, but heretofore unconfirmed, ''candidate'' species have been identified (Evans et al., 2003;Matsui et al., 2010;McLeod et al., 2011;Setiadi et al., 2011). Additionally, many conspicuously widely-distributed species complexes have been hypothesized to harbor morphologically cryptic evolutionary lineages, presumably composed of multiple independently evolving but undescribed species, characterized by little to no phenotypic variability (McLeod, 2008;Suwannapoom et al., 2016). Discussion of these unconfirmed candidate species have been commonplace in regional summaries of species richness and underestimated biodiversity (Inger, 1999;Zhang et al., 2005;Stuart et al., 2006;Fei et al., 2009;Matsui et al., 2010;Brown and Stuart, 2012), and their potential impact on conservation planning (the expectation of heightened urgency; Brown et al., 2008Ron and Brown, 2008) has been assumed as an almost inevitable outcome of advancing technology, pluralistic approaches to species delimitation (e.g., integrative taxonomy: Padial et al., 2010), and future sampling Amphib-iaWeb, 2020).
The few available taxonomic studies of Philippine Limnonectes have all suggested explicitly, or at least implied, that additional species diversity may await discovery by biologists (Inger, 1954;Brown and Alcala, 1970), especially in widespread species that have been noted for anomalous defiance of PAIC-level distributional expectations (Brown and Alcala, 1977;Evans et al., 2003;Siler et al., 2009). However, taxonomists have been reluctant to identify and name such taxa on the basis of mitochondrial sequence divergences alone (Evans et al., 2003;Brown and Stuart, 2012). As a result, revisionary studies have understandably been held in abeyance, pending the accumulation of samples, genetic data, life history and ecological information, and/or other sources of information which may provide insight into species-level reproductive isolation or evolutionary lineage status, independent of crude, arbitrary, and genetic distance cutoffs derived from single-locus studies (Evans et al., 2003;; see Sanguila et al., 2011, for discussion).
Eleven species of Philippine Limnonectes are currently recognized ; two of these defy the biogeographical predictions of the archipelago's prevailing PAIC-based biogeographic paradigm (Heaney, 1985;review: Brown and Diesmos, 2009;: Limnonectes leytensis, as presently documented (Taylor, 1920(Taylor, , 1923Inger, 1954;Brown and Alcala, 1970), occupies landmasses of the Mindanao, West Visayan PAICs, as well as the Sulu Archipelago and the Romblon Island Group. Another, the subject of this paper, is Limnonectes acanthi, a species originally described from the Palawan PAIC and faunal region (type locality: Busuanga Taylor, 1923;Inger, 1954) and which was later recorded from Mindoro Island (and, eventually, Semirara Island), to the east of the Palawan island and associated Palawan PAIC landmasses Alcala, 1955, 1970;Alcala, 1986).
That interpretation of a closely related Mindoro population, purportedly morphologically indistinguishable from the Palawan nominal population (true L. acanthi), presumes the establishment of a population on Mindoro following over-seas dispersal (Evans et al., 2003) from the northern Palawan PAIC landmasses (Busuanga, Coron, or Culion Islands across Huxley's modification of Wallace's Line; Huxley, 1868;Inger, 1954;Brown and Guttman, 2002). This hypothesis has remained unchallenged and is supported by strong nodal support for a sister relationship between Palawan and Mindoro populations, despite substantial mtDNA sequence divergence between these isolated, allopatric sister lineages (Evans et al., 2003;Setiadi et al., 2011;Figs. 1, 2).
In this study, we reconsider the L. acanthi problem, relying on a consideration of the primary mate-recognition signal of frogs and toads (Wells, 1977), namely male advertisement calls. Together with biogeographic information, and which we interpret as evidence of independent evolutionary lineage status, we conclude that although Palawan and Mindoro PAIC landmasses' populations are morphologically indistinguishable, they each represent distinct, phenotypically ''cryptic'' species. We arrive at this conclusion via our failure to reject hypotheses of population differences, using robust statistical multivariate analyses of continuous mensural variation, and the complete absence of traditionally diagnostic, discrete, character-based differences of external morphology and/or color pattern.
Here, we describe the fanged frog of Mindoro and Semirara Islands and highlight an empirically confirmed case in which species recognition signals appear to have substantially diversified, unaccompanied by phenotypic divergence. Such clear, extreme cases of divergence along one axis of diversification, even in the absence of evolutionary change in another, may exemplify test-case examples of nonadaptive, geographic evolutionary radiation (Evans et al., 2003;Blackburn et al., 2010;Setiadi et al., 2011), especially if axes of differentiation in archipelago settings  differ from those that vary in adjacent mainland source populations (Setiadi et al., 2011;Chan and Brown, 2017;Oliver et al., 2018;Davis et al., 2020).

MATERIALS AND METHODS
Species concept.-In this study, we apply the general lineage species concept as our framework for species delimitation. Under the general lineage model (Wiley, 1978;Frost and Hillis, 1990;de Queiroz, 1998de Queiroz, , 2005, we would consider the Mindoro and Palawan PAIC metapopulations distinct species if they represented allopatric, separately evolving lineages, characterized by cohesive ancestor-descendent series of populations, distinctive and diagnosable via some recognizable trait-based discrete character difference, and with arguably unique evolutionary histories and predictable evolutionary fates (Brown et al., 2000). We test this hypothesis with an integrative taxonomic approach (Padial et al., 2010), assessing genetic, phenotypic, biogeographic, and behavioral (mate-recognition signals) lines of evidence.
Given (1) the persistent sea-barrier between the Palawan and Mindoro PAICs (Inger, 1954;Voris, 2000;Yumul et al., 2009b) and (2) the relatively low overwater dispersal abilities presumed for frogs (Stebbins and Cohen, 1995;Brown and Guttman, 2002;Wells, 2007;Brown, 2016), we would consider substantive genetic divergence, accompanied by phenotypic differentiation (either fixed, traditional, diagnostic trait differences or continuous mensural and/or shape differences), acoustic mate recognition communication system attributes (male advertisement calls), or both, to constitute strong contrary evidence sufficient for rejecting the hypothesis that L. acanthi represents a single widespread species, with a distribution spanning throughout two faunal regions, and the numerous islands constituting the Palawan and Mindoro PAIC landmasses.
Molecular data and phylogenetic analyses.-We accessed preserved tissue samples from field-collected Philippine specimens of Limnonectes in the University of Kansas Biodiversity Institute cryogenic facility (fresh frozen in LN 2 or preserved in 95% ethanol and then frozen, 1 mo following collection); specimens were collected between 1991 and 2014. We extracted DNA from tissue samples using a Promega Maxwellt RSC extraction robot with Maxwellt RSC Tissue DNA kits. We amplified a~900 base-pair fragment of the 16S rRNA mitochondrial gene using Polymerase Chain Reaction (PCR) under a protocol adapted from Quah et al. (2017). We summarize the PCR conditions here: 1.0 ll DNA extract, 1.0 ll (10 lM concentration) forward primer 16Sc 5 0 -GACCTGTGATMTGAAAAACCAYC-3 0 , 1.0 ll (10 lM concentration) reverse primer 16Sd 5 0 -GACCTGTGATMTGAAAAACCAYC-3 0 (Evans et al., 2003), 2.0 ll 5x buffer (1.5 lM), 1.0 ll MgCl 2 10x buffer (1.5 lM), 1.0 ll deoxynucleotide triphosphates (1.5 lM), 0.1 ll Promega Taq polymerase (5u/ll), and 7.4 ll water. We implemented the following thermal PCR protocol in a Bio-Rad gradient thermocycler: initial denaturation at 958C for 2 min, second denaturation at 958C for 35 s, annealing at 568C for 35 s, followed by an extension cycle at 728C for 95 s þ 4 s per cycle for 34 cycles. We visualized PCR amplicons via electrophoresis in a 1.0% agarose gel in order to confirm the quality and molecular weight of PCR products. We sent our amplified DNA fragments to GENEWIZt for contract sequencing (purification, cycle sequencing, clean up, and nucleotide sequence determination). GENEWIZt produces independent sequences from the 3 0 and 5 0 ends of the DNA fragment. We edited and assembled the resulting forward and the reverse sequences with the software Geneioust version 11.1.5 (Kearse et al., 2012).
In addition to our newly generated sequences, we downloaded sequences from GenBank of Philippine Limnonectes (original data from Evans et al. [2003]). GenBank sequences consisted of an approximately 2,400 base-pair fragment of the mitochondrial gene regions tRNA phe , 12S rRNA, tRNA val , and 16S rRNA, which overlaps the region we sequenced here. We present a list of all sequences included in this study along with their GenBank numbers and associated information in Supplemental Table 1 (see Data Accessibility; GenBank accession numbers for new sequences from this study: MW020111-MW020160).
We employed MAFFT v7.0 (Katoh et al., 2019) to generate a 2,437 base-pair alignment utilizing the q-ins-I strategy because it accounts for the secondary structure of RNA. The 12S and 16S rRNA gene regions of the mitochondrion contain hypervariable regions that are difficult to align with confidence. Therefore, we used the program GBLOCKS v0.91b (Castresana, 2000) to excise any ambiguously aligned regions, resulting in a final alignment of 2,399 base pairs. We analyzed this alignment as a single locus in two Maximum likelihood (ML) and one Bayesian Inference (BI) phylogenetic analyses. We performed the first ML analysis in IQ-TREE on the web server (Trifinopoulos et al., 2016;Minh et al., 2020). The Bayesian Information Criterion employed in model selection within IQ-TREE selected TIM2þFþIþC4 as the model of evolution. We summarized nodal support in IQ-TREE with 1,000 bootstrap pseudoreplicates via the integral ultrafast bootstrap (UFBS) approximation algorithm (Hoang et al., 2018) and the Shimodaira-Hasegawa-like approximate likelihood ratio test (SH-LR). We considered nodes with UFBS values of 95 and above and SH-LR support values of 0.8 and above to be well supported (Hoang et al., 2018). We performed the second ML analysis using RAxML-HPC Black-Box 8.2.1 (Stamatakis, 2014) on the CIPRES Science Gateway (Miller et al., 2010) and employed the default GTRþCat model with RAxML rapid bootstrapping set to halt automatically. We considered nodes with rapid bootstrap support values of 70 and above to be well supported (Hillis and Bull, 1993;Wilcox et al., 2002).
We employed BI phylogenetic analysis in BEAST 2.5 (Bouckaert et al., 2019) to estimate a time-calibrated ultrametric phylogeny, and we designated the guardian frogs (L. finchi, L. palavanensis, and L. parvus) as outgroups based on prior studies of the evolutionary relationships of Limnonectes (Evans et al., 2003;Setiadi et al., 2011). We employed a Yule tree prior and a relaxed log normal clock with rate of mitochondrial evolution set to 0.0065 to estimate a timecalibrated phylogeny. This rate corresponds to a rate of 1.3% sequence divergence per million years and has been used to estimate divergence times in Limnonectes (Reilly et al., 2019).
We employed the reversible jump (RJ) model to select the most appropriate substitution model while simultaneously estimating model parameters and estimating tree topology; the General Time Reversible (GTR) was selected and implemented in subsequent analysis. We implemented two independent runs, each for 100 million generations and sampled every 10,000 generations. We evaluated convergence by visualizing the log file in Tracer v1.5 (Rambaut et al., 2014) to ensure that all parameter effective samples sizes reached stationarity (accepted ESS values 200). We combined the two runs after discarding the first 10% of each run as burn-in and used TreeAnnotator v1.8.0 to generate a maximum clade credibility tree using median heights. We considered nodes with Bayesian posterior probability (PP) support values of 0.95 and above to be well supported (Wilcox et al., 2002;Bouckaert et al., 2019). Finally, we calculated uncorrected pairwise sequence divergences among clades using MEGA v7.0 (Kumar et al., 2016).
Analyses of morphological data.-We performed all analyses of external phenotypic data in R v3.4.1 (R Core Team, 2019). To test the null hypothesis of a single species (L. acanthi) on both the Mindoro PAIC and Palawan PAIC landmasses, we recorded morphometric data from formalin-fixed, alcoholpreserved specimens of L. acanthi from the islands of Palawan, Busuanga, Mindoro, and Semirara (Material Examined). Morphometric data were collected by a single observer (RO, working with guidance from RMB, to ensure compatibility with past studies) to avoid inter-observer bias and/or extraneous error quantification of body dimensions (Lee, 1982;Hayek et al., 2001). We determined sex by inspecting gonads and/or secondary sexual characters and measured specimens with digital calipers to the nearest 0.1 mm, examined type series material (holotype, paratypes) of all Philippine species, and compared name-bearing type specimens to each species' published descriptions (Stejneger, 1910;Taylor, 1920Taylor, , 1923Inger, 1954;Brown and Alcala, 1977;Siler et al., 2009). We incorporated Köhler's (2012) color standardization scheme into our descriptions and interpretation of older (pre-2006) RMB and CDS color field notes, using digitized color images of specimens in life (deposited in KU Digital Archive; see KU Specify database: https://collections.biodiversity.ku.edu/KUHerps/).
We measured the following 15 continuous morphometric characters: snout-vent length (SVL), head length (HL), snout length (SL), tympanum diameter (TYM), head width (HW), forearm length (FLL), thigh length (THL), tibia length (TL), tarsus length (TAR), foot length (FL), hand length (HAL), eyenostril distance (EN), internarial distance (IND), fang length (FANGL), and fang height (FANGH). All character definitions and acronyms follow Watters et al. (2016) except for fang length and height. Definitions of odontoid process/fang dimensions (fang ''length'' and ''height'') follow McLeod (2008) and Emerson (2001) and are defined here for clarity. We measured the length of recurved, posterodorsally projecting fangs as the distance from the dorsal surface of the mandible, immediately anterior to the base of the odontoid, diagonally, to the tip of the posteriorly projecting dorsal tip of the fang. Alternately, we measured the height of the tip of the fang, as the perpendicular distance from the tip of the fang to the ventral surface of the mandible just below the base of the odontoid. Before proceeding with statistical procedures, we transformed data to account for differences in body size by performing separate linear regressions between SVL and each of the remaining 14 variables. We then substituted residuals of these regressions for the raw data for those 14 characters in all further univariate and multivariate analyses. We did not transform the SVL data themselves but did include this measure of body size in subsequent univariate analyses. We then ensured that data conformed to assumptions of normality by performing separate Shapiro-Wilk tests on each variable (results not shown; P values 0.05). We tested whether Palawan and Mindoro PAIC populations display mean differences in single morphometric characters with two-way analyses of variance (ANOVAs), using populations from separate PAICs (Palawan vs. Mindoro) and sex (males, females) as factors, and an alpha level of 0.05/15 ¼ 0.0033 to adjust for multiple comparisons (Rice, 1989).
We followed these univariate comparisons with an exploration of continuous phenotypic data, surveying morphological data for recognizable group structure in multivariate space. Prior to multivariate analyses, we excluded SVL and scaled all characters (residuals) to their standard deviation to avoid overleveraging. We then performed principal component analysis (PCA) and discriminant analysis of principal components (DAPC) using the ADEGENET package in R (Jombart et al., 2010).
We employed PCA, which produces the optimal lowdimensional representation of morphological variation in the sample, to visualize variation initially, and assess whether geography (Palawan vs. Mindoro) or sex (adult male vs. female specimens) forms the basis of discernable group structure in continuous morphological data. In accordance with Kaiser's Criterion (Kaiser, 1960), we retained components with eigenvalues above 1.0 for subsequent DAPC analysis, which utilizes PCA-transformed data to identify the linear combinations of variables with the smallest withingroup and largest between-group variance. Unlike PCA, group membership is designated, a priori, in DAPC; our prior, PCA-transformation ensured that variables were uncorrelated and reduced in numbers relative to the total number of observations (specimens). We used this procedure to determine whether proposed groups (Palawan vs. Mindoro populations, and specimens of both sexes) could be distinguished in morphospace, as visualized via bivariate ordination of PC axes.
Bioacoustic analysis of male vocalizations.-Calls were archived at Macaulay Library of Animal Sounds, at the Cornell University Laboratory of Ornithology. Calls were recorded holding the microphone at a distance ranging from 1.0 to 3.5 m, and ambient temperatures were recorded immediately after recording; cloacal temperatures and snout-vent lengths were measured immediately after recording when capture and collection of specimens was possible. Advertisement calls of adult males from Palawan (n ¼ 4) were recorded in the field using a Sonye WM DC6 Professional Walkman with a Sennheiser ME80 condenser microphone equipped with K3U power module ( We did not correct for differences in ambient temperature due to the narrow range of temperature variation (less than 3.08C). Calls were analyzed using Raven Pro v.1.5 for Mac OS X (Bioacoustics Research Program, 2017). We analyzed oscillograms (waveforms) and spectrograms for the following temporal and spectral variables: (1) dominant frequency (Hz) measured along the entire call, (2) call duration (s), (3) number of notes (k) in the call, and (4) note repetition rate ([number of notes -1]/call duration). Spectrograms were generated using a 1,024-point fast Fourier transformation (FFT) with a frequency grid resolution of 43.1 Hz. We analyzed several calls per individual (Mindoro: ML 224427 , calculated average values per individual for each variable measured, and used these values to calculate average values for Mindoro vs. Palawan populations. We report the mean 6 one standard deviation for each acoustic variable measured. In addition, we tested for a relationship between body size (SVL) and dominant frequency, and also SVL vs. note repetition rate, using Pearson's correlation coefficients. All statistical acoustic analyses were conducted using R v3.4.1 (R Core Team, 2019).

RESULTS
Phylogenetic analyses.-Both of our ML analyses, IQ-TREE and RAxML, generated phylogenetic trees with identical, well-resolved, and strongly supported topologies ( Fig. 1; RAxML results not shown); therefore, hereafter we interpret results of our IQ-TREE analysis only. The outgroup topology of the ML tree is well supported and consistent with previous phylogenetic studies of Philippine species of Limnonectes ( Fig.  1; Evans et al., 2003;Setiadi et al., 2011). Limnonectes acanthi is strongly supported as monophyletic and sister to L. cf. acanthi from Mindoro and Semirara Islands; together these lineages are most closely related to a lineage containing L. diuatus, L. visayanus, L. macrocephalus, and L. magnus (lineages C/E sensu Evans et al. [2003]; Setiadi et al. [2011]). Additionally, we estimated mean uncorrected pairwise sequence divergence between the Mindoro and Palawan clades of 5.0%, 3.4%, and 5.1% between the comparison of L. diuatus-L. magnus and L. visayanus-L. macrocephalus species pairs, respectively (Table 1).
Our time-calibrated Bayesian analysis recovered a tree with identical interspecific topology as that recovered in our ML analyses (Fig. 2), with two reciprocally monophyletic sister lineages (a Palawan clade and a Mindoro clade). The BI chronogram estimated that the Palawan and Mindoro clades diverged from one another approximately 4.4 million years ago (Ma; 95% HPD 2.5-6.6 Ma). This wide time interval suggests caution in use of temporal estimates-but as a broad comparison, we note that mean estimated divergence of the Palawan and Mindoro clades of L. acanthi is comparable or older than that of two other, readily diagnosable, and uncontroversial distinct sister-species pairs (L. diuatus-L. magnus and L. visayanus-L. macrocephalus; Fig. 2).
Analysis of morphological data.-We obtained measurements for a total of 100 specimens of L. acanthi from Palawan, and L. cf. acanthi from Mindoro PAIC landmasses (Table 2). Following transformation (except SVL which was not transformed), our Shapiro-Wilk tests confirmed that all variables conformed to assumptions of normality (individual P-values not shown).
Two-way ANOVAs found that the Palawan and Mindoro clades differed from one another in mean values for SL, TYM, HW, TAR, HAL, END, and IND (P values 0.043). However, these comparisons were non-significant when adjusting our alpha level for multiple comparisons (P values 0.0033) and box plots revealed substantial overlap among ranges for these seven characters ( Fig. S1; see Data Accessibility). Males and females did differ significantly in mean values for HL and HW, FANGL and FANGH (P values 0.00071). Additional visualization of all characters with bean plots (Fig. S2; see Data Accessibility) reveals the full overlapping distribution of values for each character by population (both sexes).
The first four principal components (PCs) had eigenvalues . 1.0 and were retained; together these components accounted for 70.3% of the total variation (Table 3). The factor loadings for the first PC were all positive, with the heaviest loadings on THL and TL. The loadings of the second PC were weighted heavily positive for TAR and FL, and negatively for HL and HW. Our PCA showed no group-based structure in the data, and our DAPC with a priori group designations was unable to maximize group dispersion or  (Fig. 3).
Analysis of acoustic data.-The advertisement call of Limnonectes cf. acanthi from Mindoro Island ( Fig. 4A-C) is a loud, low-frequency trill (call with multiple sequential notes), which differs from that of L. acanthi   5C) and lower dominant frequencies (Fig. 5D). We did not find significant relationships between SVL and dominant frequency or SVL and calling rate (all P values 0.05).  Conclusion and justification for the recognition of a new cryptic species.-We find that the totality of evidence requires us to reject the hypothesis of conspecificity for L. acanthi from Palawan and Mindoro PAICs. Thus, given that the Palawan PAIC population corresponds to the nominal species, (i.e., true L. acanthi [Taylor, 1923]), then the population from the Mindoro PAIC represents an undescribed, morphologically cryptic species. Analyses of DNA sequences do not contradict this view. Although we do not use genetic distances to diagnose the new species, all three phylogenetic analyses identified two reciprocally monophyletic lineages corresponding to Mindoro and Semirara versus Palawan populations. Additionally, the divergence between these two clades is comparable to or larger than observed differences between  other closely related, phenotypically distinctive, uncontroversial, formally recognized species of Philippine Limnonectes (e.g., L. macrocephalus-L. visayanus, L. magnus-L. diuatus).
Given that the Mindoro versus Palawan PAIC populations did not differ significantly in multiple body-size related dimensions after adjusting for multiple comparisons, their lack of traditional fixed character differences (non-overlapping character state ranges, anatomical character states, color pattern difference, etc.; Inger, 1954), as well as their high degree of overlap in multivariate morphospace, renders unequivocal diagnosis impossible on the basis of external morphology alone (Figs. 3, 4). However, the Mindoro and Semirara population can be readily diagnosed from its sister species, L. acanthi, of the Palawan PAIC, on the basis of acoustic data. Because male advertisement call is the primary mate recognition signal in anurans (Wells, 1977(Wells, , 2007, and the two lineages differ unambiguously in call spectral structure and temporal characteristics (note structure, note repetition rate, and dominant frequency), we do not hesitate to use this variation, plus biogeographical information (below), to form the basis of our diagnosis of the new species.
Our results are also entirely consistent with predictions based on geology and biogeography (see also Brown and Guttman [2002] and  for discussion of other species pairs from Palawan and Mindoro). The Mindoro and Palawan PAICs are separated by a deep-water channel and have not been connected by land (Yumul et al., 2003(Yumul et al., , 2009a(Yumul et al., , 2009b). Because amphibians generally are poor long-distance overwater dispersers (Stebbins and Cohen, 1995;Wells, 2007), we interpret the permanent deep-water marine barrier represented by the Mindoro Straits to have most likely severely limited gene flow between fanged frog populations on the two PAICs. This is consistent with the reciprocal monophyly and deep divergences between clades in our phylogeny (Evans et al., 2003;Setiadi et al., 2011). In contrast, we would expect much higher rates of historical gene flow among islands within each PAIC because such islands were repeatedly connected to one another by land during Pleistocene glaciations. Consistent with that expectation, we found no evidence of genetic structure between the islands of Mindoro and Semirara, both of which are part of the Mindoro PAIC.
The results of our phylogenetic and acoustic analyses, concordant with historical biogeography, provide strong evidence that L. cf. acanthi, of Mindoro and Semirara Islands, represents a cohesive, separately evolving lineage, recognizable under any modern lineage-based species concept (review: de Queiroz, 1998de Queiroz, , 2005, and we describe this lineage as a new species. Limnonectes beloncioi, new species urn:lsid:zoobank.org:act:4CFCEF76-99D6-40DC-9B53-58DECB1027CF Mindoro Fanged Frog Figures 1, 6, 7 Rana macrodon blythi Boulenger, 1920 (partim). Rana acanthi Taylor, 1923;Taylor and Elbel, 1958. Rana macrodon acanthi (Inger, 1954). Rana macrodon macrocephala (Inger, 1954)  Diagnosis and comparisons.-Limnonectes beloncioi is a medium-sized fanged frog, assigned to the genus Limnonectes (family Dicroglossidae), on the basis of its prominent, sexually dimorphic odontoid processes characteristic of the genus among other osteological synapomorphies (Inger, 1954(Inger, , 1966Emerson and Berrigan, 1993). The new species can be distinguished from all other known congeners based on a combination of its single-pulse/note advertisement call (vs. dual-pulses/note in L. acanthi from Palawan Island faunal region and PAIC landmasses), its phylogenetic position (sister to L. acanthi from Palawan PAIC; Evans et al., 2003;Setiadi et al., 2011), and its geographic distribution on Mindoro and Semirara Islands (vs. Palawan PAIC); it is the only species of Limnonectes known to occur on Mindoro and Semirara Islands and associated small satellite islands and, therefore, has no sympatric congeners. The new species is morphologically similar to its closest relative, L. acanthi; however, it may be distinguished from this allopatric congener by its male advertisement call. The note pulse substructure of L. beloncioi is singular (1 pulse per note vs. 2 pulses per note in L. acanthi), and the new species has a slower note repetition rate (11.5-12.8 notes per second vs. 14.2-15.4 notes per second), and has a higher dominant frequency (2,067.2-2,799.3 Hz vs. 1,335.1-1,679.6 Hz in L. acanthi).
With the exception of the morphologically indistinguishable L. acanthi, the new species can be distinguished from all other Philippine species of Limnonectes (L. diuatus, L. ferneri, L. leytensis, L. macrocephalus, L. magnus, L. micrixalus, L. palavanensis, L. parvus, L. visayanus, and L. woodworthi) by a combination of body size, fang (odontoid) length, snout shape, relative lengths of the first and second finger, dorsal skin rugosity, restriction of white-tipped dermal asperities to the sacral region (not aggregated in radial clusters), the presence of irregular, elongate, discontinuous dorsolateral ridges (absence of a continuous dorsolateral fold), complete interdigital webbing of the foot, and the absence of a dark inverted ''V''-shaped mark on the dorsum. We provide morphological comparisons below, based on our data, in conjunction with or with consideration of the descriptions of Stejneger (1910), Taylor (1923), Inger (1954), Brown and Alcala (1977), and Siler et al. (2009).
Limnonectes beloncioi differs from L. diuatus and L. ferneri by its rounded snout in lateral aspect (vs. posteroventrally sloping ), its Finger I . Finger II relative finger lengths (vs. approximately equivalent length), and by restriction of white-tipped dorsal asperities to dorsal sacral region, and not distributed in radial clusters (vs. asperities not posteriorly restricted in L. diuatus; and densely distributed across entire of dorsum, and concentrated in radial sacral clusters in L. ferneri); and the presence of irregular dorsolateral ridges (absent in L. diuatus and L. ferneri); from L. leytensis, the new species can be distinguished by its larger adult body size (SVL range 54.2-83.1 vs. 25.8.2-34.0 mm in L. leytensis), rounded snout (vs. snout moderately pointed in lateral aspect in L. leytensis), Finger I . Finger II (equivalent length in L. leytensis), complete webbing (webbing incomplete/reduced in L. leytensis), and the absence of an inverted ''V''-shaped mark on the anterior dorsum (vs. present in L. leytensis); from a large male specimen of L. macrocephalus (the species endemic to the Luzon PAIC landmasses of Luzon, Polillo, Catanduanes, and Marinduque), the new species can be distinguished by the observation that it attains a considerably smaller maximal adult male body size (SVL range 54.2-83.1 vs. 78.9-144.6 mm in L. macrocephalus); the new species also possesses a fully exposed tympanum (vs. dorsal and/or posterior edge of tympanum hidden beneath overlapping supratympanic dermal ridge skin in L. macrocephalus) and lacks sexual size dimorphism (Table 2), whereas L. macrocephalus exhibits reverse sexual size dimorphism (males larger); from a large adult male specimen of L. magnus, the new species can similarly be distinguished by its smaller adult body size (SVL range 54.2-83.1 vs. 66.3-164.4 mm in L. magnus), rugose middorsal skin texture (vs. smooth to shagreened in L. magnus), fully exposed tympanum (vs. dorsal and/or posterior edge of tympanum hidden beneath overlapping supratympanic dermal ridge skin in L. magnus), rounded snout (vs. pointed in L. magnus), and by the absence of reverse sexual size dimorphism (present in L. magnus); from L. parvus, L. micrixalus, and L. palavanensis, the new species can be distinguished by its larger adult body size  in L. palavanensis), rugose (vs. smooth) dorsal skin, Finger I . Finger II (equivalent length), the presence (vs. absence) of white-tipped sacral asperities and the presence of irregular, discontinuous dorsolateral ridges (vs. asperities absent, dorsolateral folds continuous), complete webbing (vs. reduced), and by the absence (vs. presence) of an inverted ''V''-shaped middorsal marking (vs. present); the new species is additionally distinguished from these species by the absence of sexual size dimorphism (vs. females larger in L. parvus, L. micrixalus, and L. palavanensis); from L. visayanus, L. beloncioi is readily diagnosed by its rounded snout (vs. pointed in L. visayanus), a tendency towards longer adult male fangs (2.6-5.6 vs. 1.8-3.0 mm in L. visayanus), by the presence (vs. absence) of white-tipped sacral asperities; finally, the new species can be distinguished from L. woodworthi by its longer male fangs (2.6-5.6 vs. 1.3-1.6 mm in L. woodworthi), moderately rugose middorsal skin (vs. smooth in L. woodworthi), rounded snout (vs. snout moderately pointed in L. woodworthi), the presence (vs. absence) of white-tipped sacral asperities, and by the absence (vs. presence) of continuous dorsolateral folds.
Description of holotype.-A mature male, specimen in excellent condition; small portion of liver preserved separately for genetic material; habitus robust; head broader than body, its length 98.3% of its width, 39.0% of SVL; snout tip rounded in dorsal and lateral aspect (Fig. 6); supralabial region markedly swollen, increasingly protuberant towards angle of jaw; interorbital region and dorsal rostrum nearly flat; eye diameter 62.0% snout length, 97.5% eye-nares distance, 1.43 eye-tympanum distance; pupil horizontally sub-elliptical with discontinuous posterior margin; canthus rostralis distinct, slightly medially bowed in dorsal aspect; loreal region concave; nostrils oriented dorsolaterally, narial openings visible in dorsal view; internarial region slightly convex; tympanum exposed, annulus slightly distinct, diameter 55.7% of eye diameter; supratympanic fold thick, strongly protuberant, moderately rugose, extending from posterior corner of eye, extending horizontally over (concealing) dorsoposterior corner and posterior margin of tympanic annulus, turning ventrally at nearly right angle, to end in supra-axillary region, where it is discontinuous with postrictal tubercular swelling at angle of jaw. Tongue elongate, tapered anteriorly, with narrow anterior attachment and laterally expanded, free, bilobed posterior margin at rest (anterior edge when tongue projected); choanae situated at anterolateral edge of palate, subcircular, their anterolateral edge partially concealed by palatal shelf of maxilla in ventral view; choanae widely separated by distance five or six times greater than diameter of single choana, each located just anterolaterally to (in contact with) lateral tip of dentigerous process of vomer; dentigerous process of vomer distinct, with four or five conical teeth on each side; dentigerous process angled anterolaterally (rostrally), approximately at 458 inclination, with closest (posterior) points separated by distance approximately equal to one choana, their most distant (anterior) ends separated by distance equal to three choanae; enlarged odontoid ''fangs'' large, recurved, unsheathed by oral mucosa for . distal half their length, situated on either side of mandibular symphysis/medial bulge, their tips sharply pointed, total length 4.6 mm (perpendicular distance from ventral edge of mandible), inclined dorsoposteriorly, tips 2.9 mm perpendicular from dorsal mandible surface; maxillary fang ''sockets'' anteromedial to choanae, large, round, similar in size to one choana; vocal apertures large, elongate, surrounded by extensive mucosal invaginations, situated at posteroventral margin of buccal floor, just medial to angle of jaw.
Hand length 46.8% foot length; foot 93.4% tibia length; tibia length 58.3% SVL; fingers laterally, irregularly ovoid in cross section, due to presence of slight lateral dermal flange, extending from base of each digit, on either side, to proximal margins of terminal finger discs; terminal discs not expanded beyond widths of penultimate phalanges (Fig. 6), their relative descending lengths: III . I . II ¼ IV; subarticular tubercles prominent, their ventral surfaces convex and velvety in texture; one subarticular tubercle below Fingers I and II, two tubercles under Fingers III and IV; terminal discs and subarticular tubercles with gray, velvety, thickened surfaces; distal margins of tubercle more distinct than their proximal margins, and supernumerary tubercles absent, but articular surfaces of fingers between subarticular tubercles of digits, and at base of all digits covered medially with thickened tubercular surface; palmar surface with large, elongate, thenar tubercle (ventromedial surface of Finger I), enlarged, flattened, squarish medial ''inner'' (base of Finger III) metacarpal tubercle, and small, ovoid, convex outer (base of Finger IV) metacarpal tubercle; surface of these palmar structures, intervening, and surrounding surfaces all covered with similar, thickened, velvety (matte) tubercular dermis layer; nuptial excrescences or pads, asperities, and webbing absent; forearm musculature not hypertrophied.
Tarsus folds and flaps absent; terminal discs of toes slightly expanded, with distinct circummarginal grooves; plantar surfaces of foot with well-developed, prominently protruding (ventrally), rounded subarticular tubercles (Fig. 6); plantar surfaces of foot smooth, with velvety-textured subarticular tubercles; relative lengths of toes: I , II , V , III , IV; toes fully webbed (interdigital webbing extending to proximal edges of terminal discs of all toes); postaxial flap of skin running along entire outer edge of Toe V; inner metatarsal tubercle prominent, elongate, with raised ventral edge; outer metatarsal tubercle absent.
Skin of dorsal surfaces of trunk and head smooth to slight shagreened texture, bearing low but clearly evident fleshy dermal tubercle clusters in supra-and post-tympanic regions, and tuberculate dorsolateral ridges, immediately following supratympanic region, and continuing posteriorly to the scapular region approximately to the points of forearm insertion (Fig. 6); similar dorsolateral tubercular ridges extend from this point, along more lateral (flanks) and slightly medial (dorsal) surfaces, and extend back to the sacral region; on posterior half of trunk and sacral region, dermal tubercles present mid-dorsally, consisting of single, raised tubercles or short, raised tubercular ridges; in sacral region, some tubercles capped with round, weakly keratinized dermal asperities; not arranged in clusters, or rows, but lightly dispersed in sacral region and upper one-third of dorsal surface of thigh and supra-cloacal region; ventral surfaces of head smooth; lateral and ventral surfaces of limbs smooth; remaining dorsal surfaces of limbs smooth to lightly shagreened, with occasional low tubercles; tarsus smooth on dorsolateral surface; cloacal region rugose (wrinkled), with smooth laterally and ventrally surrounding skin. Measurements of holotype (mm). HL 29.2;SL 13.0;TYM 4.4;HW 29.7;FLL 16.1;THL 40.2;TL 43.7;TAR 18.9;FL 40.8;HAL 19.1;EN 7.2;IND 5.7;FANGL 4.6; FANGH 4.1.
Variation.-Summaries of univariate morphological variation in the type series are presented in Table 2. We observed no apparent sexually dimorphic color variation in our large series of paratypes; similarly, although comparisons among older specimens from particular localities (CAS, FMNH, and USNM paratypes, collected over the past century) and more recent material from some of the same, but also novel localities on Mindoro (KU, and more recent USNM specimens) revealed some variation suggestive of geographically variable color variation, these initial observations could conceivably reflect circumstances of preservation and color shifts with time. As a result, we emphasize color variation across all specimens available to us (from throughout Mindoro Island): three general ground-pattern dorsal coloration types are immediately apparent, including (1) specimens with very dark brown to black dorsal surfaces, and transverse hindlimb bars, forearm blotches, and interorbital bar indistinct or barely evident (Sepia 286, Jet Black 300;e.g., KU 302095, 303432, 308303, 335866), (2) specimens with medium brown to gray dorsal surfaces (Glaucous 291 to Brownish Olive 292 or Dark Neutral Gray 299) and darker transverse hindlimb bars, forearm blotches, interorbital bar, and labial bars all evident (the holotype [PNM 9870;Fig. 6] and paratypes [KU 302084-88, 302100, 302105-07, 303451, 303353], plus nearly all older CAS, FMNH, and USNM paratypes and referred specimens, collected 40-100 y ago), and (3) specimens with light gray (Pale Neutral Gray 296 to Pratt's Gray 293) dorsal body surfaces, with a dense network or reticulum of dark gray (Plumbeous 295 to Dark Neutral Gray 299; e.g., KU paratypes 302084, 302085) throughout the body, and densely congregated into hindlimb bars, forearm blotches, a darkly pigmented interorbital bar, and labial bars. Ventral body surfaces are generally white (Buff White l to Cream White 52) and transitioning posteriorly to more yellowish white (Pale Horn 11 to Light Yellow Ocher 13) in inguinal region and posterior surfaces of the thighs. Darker pattern elements situated around posterolateral surfaces of head (chin and throat lateral edges), and lateral margins of ventral limb surfaces (lateral edges of ventral forelimbs and hindlimbs) correspond to the same general dorsal coloration patterns (above), which wrap laterally onto ventral surface and sharply fade to white. Thus, the specimens with dark dorsal coloration are also the specimens with the darkest ventral pattern elements (infralabial blotch-es, lateral forearm ventral coloration, ventral shank color, and anteroventral thigh pattern), and the specimens with the medium brown to gray dorsal surfaces (including most paratypes and the holotype) have corresponding medium pigmentation in these same characters contributing to the ventral pattern (Fig. 6). Ventral surfaces of hands and feet are surprisingly invariant and do not differ from those described in the holotype (Fig. 6), and no ontogenetic color variation was evident in our series. The one ventral color pattern character that seemed to depart from the above was the concentration of pigment contributing to throat and pectoral region. Throats ranged from homogeneously medium gray (offset and darker from ventral body surface; males KU 302104, 303432, females KU 302105, 303451) to darkly spotted brown (males KU 302095, 302106, 308313, PNM 9870 [holotype], females KU 302100, 302118, 335866), to pale cream (not differing from remainder of ventral body coloration; males KU 302085, 302086, 302089, 303353, 304132, females KU 302084, 302387, 302088). In life, some specimens have more yellowish hues in lateral head surfaces, the flanks, inguinal region, ventral hindlimb and cloacal color (Fig. 7); these are lost and fade to shades of cream to light gray in preservative; iris gold above and silver below pupil.
Distribution.-Limnonectes beloncioi is known only from Mindoro and Semirara Islands, central Philippines (Fig. 1). Minor land-bridge and deep-water islands to the east and southwest of Mindoro have not yet been surveyed for herpetofauna; as such, the identity of any populations of Limnonectes that might eventually be discovered on Ilin, Sibay, Ambulong, Caluya, Maestre de Campo, and/or the Cuyo Islands group cannot yet be confirmed. The genetic identification of populations of Limnonectes on Busuanga, Coron, and Culion (currently considered L. acanthi [Taylor, 1923[Taylor, , 1928Inger, 1954;Brown and Alcala, 1970;; thus, we anticipate these populations will be more closely related to Palawan Island L. acanthi than to L. beloncioi) would be of particular interest to herpetologists, biogeographers, and speciation geneticists .
Etymology.-We derive the specific epithet, a patronym, to honor the Philippine army scout and freedom fighter Esteban Beloncio, who contributed substantially to the armed resistance against the Japanese WWII occupation of Mindoro Island, 1942-1945. Suggested common name ¼ Mindoro Fanged Frog.
Ecology and natural history.-The new species is commonly encountered in riparian habitats, and in the vicinity of moving water in galley forests; it has been collected from rocks, sandy shingles, and muddy river banks of streams and rivers of Mindoro Island and its land-bridge associated island, Semirara; the species also is anticipated to be present on Ilin and Ambulong Islands, which are separated from Mindoro by shallow channels. On Mindoro, the new species has been collected from just above sea level in coastal lowland habitats associated with river mouths (the type locality: Barangay Carmundo, Sitio Paypay-Ama, at the mouth of the Paypay-Ama River; the Municipality of Victoria, Barangay Loyal, Sitio Panguisan, Panguisan River), and in coastal habitats in the northern Mindoro Municipalities of Puerto Galera, Naujan, Calapan, and San Teodoro, southern Municipalities of San Jose, Bulalacao, and Roxas, and western coast Municipalities of Bongabong, Gloria, and Pinamalayan; Mindoro's interior regions, in which the new species has been recorded, collected, or observed include the relatively undisturbed low to mid-elevation forests of the Municipalities of Sablayan (Fig. 8), San Teodoro, Calapan, and Baco (Brown and Guttman, 2002;. Other native species of amphibians encountered on Mindoro Island  include Platymantis corrugatus (Ceratobatrachidae), Fejervarya moodiei, F. vittigera, Occidozyga laevis (Dicroglossidae), Kaloula conjuncta, K. picta (Microhylidae), Leptobrachium mangyanorum (Megophryidae), Pulchrana mangyanum (Ranidae), Polypedates leucomystax, Philautus schmackeri, and Rhacophorus pardalis (Rhacophoridae). Invasive species of frogs introduced to Mindoro include Rhinella marina (Bufonidae), Kaloula pulchra (Microhylidae), Hylarana erythraea (Ranidae), and Hoplobatrachus rugulosus (Dicroglossidae). The last of these is a voracious predator, which we have observed on numerous occasions, feeding on juveniles and metamorphs of Limnonectes.
With very little original vegetation present at lower elevations on Mindoro Island, and with the rising human population, the loss of original forest, development, and agriculture throughout most the island's lower elevations, the loss of forest cover must be viewed as a primary threat to all native forest-associated amphibians, including L. beloncioi. Estimates of forest loss or severe alteration between 2003 and 2010, for example, range from 8-9% annually, of 53-61% cumulatively for over this 7-yr period (ELBR, pers. comm. with Center for Conservation Innovations [unpubl. data, 2020]). Despite the resilience of the new species in heavily disturbed and highly degraded riparian habitats (Brown and Guttman, 2002;ELBR and RMB, pers. obs.), its ubiquitous persistence in Mindoro's chronically degraded lowland watersheds (Lizuka et al., 2009) suggests that poor water quality may be less of a threat to larval development in L. beloncioi than might otherwise be expected of amphibians with aquatic larval development (Wells, 2007). An important source of protein in the diets of many indigenous communities (including the eight ethnolinguistic Mangyan tribal groups of Mindoro; Lopez, 1976;Kikuchi, 1984), species of fanged frogs are widely hunted and consumed in alarming numbers throughout their ranges in the Philippines (RMB, CDS, and ELBR, pers. obs.). Although the new species is one of the most common species of frogs on Mindoro Island (as evinced by numbers of specimens available for this study in collections), it is heavily hunted and widely consumed by humans on Mindoro (Schult, 1991), including the migrant Tagalog (Luzon-derived) majority in the lowlands (Scheffers et al., 2012) and interior highlands indigenous peoples (Mangyans) who target river frogs (''Palakang Ilog'') during the rainy season (ELBR, pers. obs.). We have evaluated this species against the IUCN conservation status classification rubric (IUCN, 2020), and find that it does not qualify for Critically Endangered (CR), Endangered (EN), Vulnerable (VU), or Near Threatened (NT) status. Limnonectes beloncioi has a geographic distribution that spans multiple islands, including the large landmass of Mindoro, and is quite abundant at all sampled localities. We therefore classify this species as Least Concern (IUCN, 2020), but acknowledge that some other, currently undefined, threatened status may be conceivable, following consideration of unrelated, or secondary sources of information which do not bear directly on the biology of the new species (Gonzalez et al., 2018).

DISCUSSION
Although we did not diagnose the new species on Mindoro from its closest relative (L. acanthi from Palawan) via genetic distance thresholds, we note that the mean pairwise mitochondrial sequence divergence from its sister species is 5.0%, which is commensurate with, or exceeds, divergences between uncontroversial and phenotypically distinct sisterspecies pairs (Table 1), such as L. macrocephalus (Luzon PAIC) versus L. visayanus (West Visayan PAIC), and L. magnus versus L. diuatus (both of the Mindanao PAIC). Limnonectes beloncioi (Mindoro PAIC) and L. acanthi (Palawan PAIC) occupy allopatric insular distributions on separate Philippine geological platforms (PAICs; ) as do most other species-pairs comparisons used for reference here, and it does not surprise us that genetic divergence between the new species and its sister taxon is equivalent to or exceeds divergences observed among other Philippine sisterspecies pairs.
With the description of Limnonectes beloncioi, and the recognition of a distinct evolutionary lineage of fanged frog from the Mindoro PAIC, the number of species of Philippine Limnonectes climbs to 12. We consider the formal description of a new, endemic Mindoro PAIC fanged frog to represent a long-overdue step, which ameliorates a persistent taxonomic shortfall allowed to lie unresolved for more than half a century (Inger, 1954;Brown and Alcala, 1970;Evans et al., 2003). The numbers of native amphibian species, previously thought of as distributed on both Palawan and Mindoro PAICs, has steadily declined through time; this trend has been a consequence of successive taxonomic revisions involving one or both populations (Taylor, 1920(Taylor, , 1923Inger, 1954;Brown and Alcala, 1970;Brown and Stuart, 2012;Diesmos et al., 2014Diesmos et al., , 2015. Moreover, as the geographic ranges of Philippine amphibians have become properly characterized (Brown and Alcala, 1970;Diesmos et al., 2015), cases of true, widely distributed, single species exhibiting geographic ranges that span PAIC boundaries has become increasingly uncommon Diesmos et al., 2015). With one exception, all other Philippine Limnonectes are restricted to the landmasses encompassed by single Pleistocene Aggregate Islands (as defined by ; the single exception is the remarkably widespread Limnonectes leytensis, which occupies four PAICs (the Sulu, Mindanao, West Visayan, and Romblon faunal regions). The hypothesis of conspecificity of all these populations of L. leytensis, however, remains untested (Evans et al., 2003;Siler et al., 2009;. It is interesting that, although Taylor (1923) and Inger (1954) both recognized the Palawan faunal region lineage (L. acanthi) as a clearly distinct species, the close affinities of the Mindoro population escaped their attention. However, as pointed out by Alcala (1955, 1970), large sample sizes of herpetological specimens from Mindoro did not come available until much later; thus, it is not surprising that Inger (1954) tentatively identified three Mindoro specimens available at the time of his review as L. macrocephalus. As explained by Brown and Guttman (2002), the prevailing view of Mindoro's land vertebrate biogeography for much of the late 20 th century was that its fauna was assembled primarily by colonization of lineages from Luzon (Taylor, 1928;Inger, 1954;Leviton, 1963;Brown and Alcala, 1970;Heaney, 1985), which has proven to be an oversimplification (Inger, 1999;Brown and Guttman, 2002;Blackburn et al., 2013).
Although traditional, character-based, taxonomic approaches were unable to convincingly identify fixed diagnostic differences between Palawan and Mindoro populations, even under a polytypic species concept (Inger, 1954;Brown and Alcala, 1970;Brown et al., 2000), the identification of other amphibian lineages that have been capable of dispersing across Huxley's modification of Wallace's Line have been associated with the eventual recognition of species of amphibians now considered endemic to Mindoro Island (Brown and Guttman, 2002;. However, like Limnonectes beloncioi, these Mindoro endemics were only recognized after a considerable delay (80-100 yrs after their populations were first surveyed by herpetologists; Taylor, 1920Taylor, , 1923Inger, 1954), due almost certainly to their phenotypic similarity with other species from neighboring landmasses adjacent to Mindoro (e.g., Pulchrana mangyanum [Brown and Guttman, 2002]; Leptobrachium mangyanorum ). It is unclear why some Mindoro amphibian populations historically were recognized early (Inger, 1954;Brown, 2007) and their taxonomic status never questioned (e.g., Philautus schmackeri [Boettger, 1882]; Alcala, 1970, 1994), whereas others have gone historically confused and never had their status clarified-despite suggestions of apparent phenotypic and/or genetic distinctiveness on this island (e.g., Mindoro Island Platymantis corrugatus and Kaloula conjuncta; Inger, 1954;Blackburn et al., 2013;Brown et al., 2015). Suffice it to say, Mindoro's amphibian populations have been historically understudied, most likely due to a combination of logistical challenges and untested, indiscriminately applied taxonomic expectations ascribed to its land vertebrates by the last two generations of biogeographers who assumed Mindoro's fauna to be derived from, and closely related to, that of Luzon (Inger, 1954;Leviton, 1963;Brown and Alcala, 1970;Heaney, 1985). Evans et al. (2003) first identified a deep,~5.0% divergence in mitochondrial gene sequences (12S-16S) and hypothesized that the Mindoro population might constitute a distinct evolutionary lineage, or unrecognized species. Although this level of genetic divergence certainly is commensurate with those empirically estimated between other, uncontroversial, recognized Philippine fanged frogs separated on isolated PAICs (e.g., Luzon PAIC L. macrocephalus, West Visayan PAIC L. visayanus, and Mindanao PAIC L. magnus; Table 2), the use of unreliable single-locus genetic distances, combined with arbitrarily assigned cut-offs, or divergence thresholds, has been criticized on multiple grounds, which need not be repeated here. As such, to reconsider Evans et al.'s (2003) finding, we required data from other sources (character data, mensural data, advertisement calls, biogeographic information) before taxonomic recognition of the Mindoro population might be justified by our rejection of the single-species hypothesis. Unfortunately, legal restrictions preventing faunal surveys and voucher specimen collection and other logistical obstacles to fieldwork on Mindoro have historically hampered the amassing of the necessary specimen-associated data. We are encouraged that the factors preventing the realization of this long-anticipated eventuality (the recognition of a species of fanged frog, endemic to Mindoro; Evans et al., 2003;Siler et al., 2009;Diesmos et al., 2015) have been alleviated by the dedicated efforts of field biologists with a shared commitment to international collaboration and conservation of Mindoro's unique and imperiled amphibian biodiversity (Alcala and Brown, 1998;Stuart et al., 2008;Diesmos and Brown, 2011;Alcala et al., 2012).