Species delimitation of crab-eating frogs (Fejervarya cancrivora complex) clarifies taxonomy and geographic distributions in mainland Southeast Asia

Abstract The taxonomy and geographic distributions of species of crab-eating frogs (Fejervarya cancrivora complex) in mainland Southeast Asia have been highly uncertain. Three taxonomic names are used in recent literature (F. cancrivora, F. raja, and F. moodiei) but the applications of these names to localities has been inconsistent, especially owing to the lack of available molecular data for F. raja. Morphometric and mitochondrial DNA variation was examined in these frogs, including name-bearing types and topotypes of all three species. Findings corroborate evidence for the existence of two species in coastal mainland Southeast Asia, with F. moodiei having a wide geographic distribution and F. cancrivora sensu stricto occurring only in extreme southern Thailand and peninsular Malaysia. Fejervarya raja is shown to be only a large-bodied population of F. cancrivora sensu stricto and is synonymized with that species. Revised descriptions of F. moodiei and F. cancrivora sensu stricto are provided.


Introduction
Southeast Asia harbors high levels of amphibian species diversity and endemism (Brown and Stuart 2012), and new species continue to be discovered and described (e.g., Geissler et al. 2014;Phimmachak et al. 2015;Sheridan and Stuart 2018). Moreover, recent evaluations of morphological and molecular diversity of Southeast Asian amphibians have routinely shown that long-recognized geographically widespread single species actually represent complexes of cryptic species (Stuart et al. 2006b;Aowphol et al. 2013;Phimmachak et al. 2015;Sheridan and Stuart 2018). The presence of cryptic species in Southeast Asian amphibians has hindered accurately assessing species boundaries and, ultimately, efforts to conserve them (Bickford et al. 2006;Sheridan and Stuart 2018). Even geographically widespread, human commensalist species may contain unrecognized diversity that alters their priority for conservation (Wogan et al. 2016).
Species of frogs in the genus Fejervarya Bolkay, 1915 have been subject to numerous investigations into cryptic diversity in efforts to resolve species boundaries and uncertain taxonomy in South, Southeast and East Asia (e.g., Vieth et al. 2001;Matsui et al. 2007;Islam et al. 2008;Kotaki et al. 2010;Sanchez et al. 2018). A notable challenge remains with the crab-eating frog, F. cancrivora (Gravenhorst, 1829), a species that is remarkable in its ability to thrive in brackish or salt water (e.g., Gordon et al. 1961;Balinsky et al. 1972;Wright et al. 2004;Hopkins and Brodie 2015). Fejervarya cancrivora occurs in coastal areas throughout much of Southeast Asia, and as expected owing to its large geographic range, recent molecular investigations have hypothesized the existence of cryptic species and discordance between taxonomy and species diversity within the taxon (Kurniawan et al. 2010(Kurniawan et al. , 2011. Historically, the name F. cancrivora had been erroneously applied to larger members of the F. limnocharis complex, but application of the name was stabilized following designation of a neotype specimen from Cianjur, West Java, Indonesia, by Dubois and Ohler (2000). Taylor (1920) described the Philippine populations of F. cancrivora as a distinct species, F. moodiei (originally Rana moodiei Taylor, 1920) based on an adult female collected at Manila, Luzon, Philippines. Smith (1930) described a population of F. cancrivora specimens having large body sizes from Pattani, Thailand, as F. raja (originally R. cancrivora raja Smith, 1930).
Two of these species, F. cancrivora and F. raja, have been reported from Thailand, where they occur in the vicinity of sea shores or river mouths (Smith 1930;Taylor 1962;Nutphud 2001;Chan-ard 2003;Chuaynkern and Chuaynkern 2012). However, these designations have been uncertain. Iskandar (1998) suggested that F. raja from Thailand might just be unusually large individuals of F. cancrivora. Other authors have questioned the distinctiveness of the Philippine F. moodiei from F. cancrivora, and have synonymized them (Smith 1927;Inger 1954) or considered F. moodiei to be invalid ). Analyses of morphological and molecular variation, as well as laboratory crossing experiments, revealed three distinct "types" (= forms) of F. cancrivora across its large geographic range: a large type considered to be true F. cancrivora, a mangrove type considered to be F. moodiei, and a Sulawesi type that might belong to an undescribed species (Kurniawan et al. 2010(Kurniawan et al. , 2011. Their results also inferred that F. raja might be conspecific with F. cancrivora. A lack of molecular data from true F. raja and examination of type specimens in the F. cancrivora complex (Islam et al. 2008;Kurniawan et al. 2010Kurniawan et al. , 2011 have hindered resolving species boundaries and taxonomy within the crab-eating frogs.
In this study, we examined morphology and mitochondrial DNA variation in historical and newly-collected museum specimens of the F. cancrivora complex from Thailand and adjacent Asian countries to evaluate and clarify the taxonomic status of F. cancrivora, F. moodiei and F. raja. Importantly, our analyses included molecular and morphological data of topotypes of F. raja, and morphological data from the namebearing type specimens of F. cancrivora and F. moodiei.

Sampling
During 2015-2017, specimens of F. cancrivora were collected at 12 localities and F. raja at two localities in Thailand (Fig. 1). Specimens were humanely euthanized using tricainemethanesulfonate (MS-222) solution. Liver or muscle tissue was removed from each individual, preserved in 95% ethyl alcohol, and stored at -20 °C for molecular analysis. Voucher specimens were initially fixed in 10% buffered formalin and later transferred to 70% ethyl alcohol for long-term preservation. Tissue samples and voucher specimens were deposited in the herpetological collection of the Zoological Museum, Kasetsart University, Bangkok, Thailand (ZMKU). Comparative material was also studied in the holdings of ZMKU, Carnegie Museum of Natural History (CM), Field Museum of Natural History [FMNH; formerly Chicago Natural History Museum (CNHM)], and Thailand Natural History Museum (THNHM; Table 1; Appendix 1).

Figure 1.
Map of sampling localities of the Fejervarya cancrivora complex, including F. cancrivora neotype (yellow pentagon), F. cancrivora sensu stricto (yellow circles), F. moodiei holotype (blue diamond), F. moodiei (blue triangles), and F. cancrivora samples that were referred to F. raja (red circles) prior to this study. Open symbols indicate molecular data only, shaded symbols indicate morphological data only, and shaded symbols with center dots indicate both molecular and morphological data were studied.

Morphological study
Morphological analyses were performed on 108 sexually mature individuals (61 males, 47 females) of F. cancrivora, F. moodiei, and F. raja (Table 1; Appendix 2, 3). Importantly, these included the neotype (FMNH 256688) and topotypes of F. cancrivora from Java, Indonesia; the holotype of F. moodiei (CM 3724) from Luzon, Philippines; and topotypes of F. raja from Pattani, Thailand (Table 1). Sexual maturity was determined by presence of secondary characteristics, including nuptial pads or vocal sac folds in males, and convoluted oviducts or mature ova in females. Webbing formulae follow Savage and Heyer (1967).
Measurements were taken with digital Vernier calipers to the nearest 0.1 mm. Twenty-three morphological characters were measured following Djong et al. (2007) and Islam et al. (2008) Qualitative characters were taken on the presence and condition of the vomerine ridge, skin on dorsum, coloration and pattern on dorsum, vocal sac pigmentation, fejervaryan lines (conspicuous ventrolateral lines on the ventral side of the body), tubercles on forelimbs and hindlimbs, dermal fringe on fingers II and III, inner tarsal ridge, dermal flap on outer side of Toe V, and foot webbing.
To correct for body size, each mensural character was divided by SVL to a ratio (r) and then converted to a percentage. Specimens were assigned to group (= species) based on their mtDNA assignment (below). Principal component analysis (PCA) was performed separately by sex using FactoMineR and factoextra R package (Lê et al. 2008;Husson et al. 2017) in the R programs v.3.4.3 (R Core Team 2017) to assess morphometric differences between groups. All variables were tested for normality using Shapiro-Wilk's test. Statistical differences between species were tested by t-test for parametric data and Mann-Whitney U test for non-parametric data at a significance level of 95%.

Phylogenetic analyses
The aligned dataset contained 61 individuals and 981 characters. The standard deviation of split frequencies was 0.003331 among the two Bayesian runs, and the Estimated Sample Sizes (ESS) of parameters were ≥ 200. The Bayesian analysis recovered the F. cancrivora complex as monophyletic with strong support, and to contain two major clades referred to as Clades A and B (Fig. 2). Clade A contained subclade A1 consisting of F. cf. cancrivora from Indonesia (Pelabuhan Ratu and Sulawesi) and subclade A2 consisting of F. cancrivora from Indonesia (Sumatra, Java, Bali) and Malaysia (Selangor), as well as F. raja from Thailand (Phatthalung, Nakhon Si Thammarat). Clade B contained subclade B1 consisting of F. cancrivora from Thailand (Trat, Nakhon Si Thammarat, Surat Thani, Prachuap Khiri Khan, Rayong), Philippines and China, and subclade B2 consisting of F. cancrivora from Thailand (Phuket, Phang-nga, Ranong, Satun, Krabi) and F. moodiei from Bangladesh (Cox's Barza, Khulna).

Morphological analyses
PCA analysis of males revealed morphometric differences between F. cancrivora Group A and F. cancrivora Group B, with no overlap on a plot of the first two axes (Fig. 3A). The first three principal components (PC) of males with Eigenvalues > 1.0 accounted for a cumulative 61.2% of the total variance (29.6% by PC1, 19.2% by PC2 and 12.4% by PC3; Table 3). PC1 was heavily and positively loaded on rTL, rHW, rFOL, rTHIGHL, rTFOL, and rSL. PC2 was heavily and positively loaded on rEL, rTD, rNTL, and negatively on SVL, suggesting a strong negative correlation between these characters. PC3 was heavily and positively loaded on r1FL. These results indicated that PC1 and PC2 were strongly influenced by body size. Males of F. cancrivora Group A had larger SVL, rTL, rHW, rFOL, rTHIGHL, rTFOL, and rSL, but smaller rEL, rTD, and rNTL than males of Group B based on scores of the first two axes (Fig. 3A).  (10.9-12.5) (12.8-13.7) (12.3-13.5) (9.5-10.2) (9.8-11.0) (8.9-11.0) (4.5-7.9) (0.6-6.0) 11.7 13.4 13.2 9.8 10.5 9.3 6.5 3.6 PCA analysis of females revealed morphometric differences between F. cancrivora Group A and F. cancrivora Group B, with only slight overlap on a plot of the first two axes (Fig. 3B). The first three PCs of females with Eigenvalues > 1.0 accounted for a cumulative 35.3% of the total variance (35.3% by PC1, 14.5% by PC2 and 9.3% by PC3; Table 3). PC1 was heavily and positively loaded on rSL, rFOL, rTFOL, rSTL, rTL, rITL, rEN, and rHAL, indicating that it was strongly influenced by body size. PC2 was heavily and positively loaded on rEL and negatively on SVL, implying a strong negative correlation between these characters. PC3 was moderately and positively loaded on rUEW and negatively on r1FL. Females of F. cancrivora Group A had larger SVL rSL, rFOL, rTFOL, rSTL, rTL, rITL, rEN, and rHAL, but smaller rEL than females of Group B based on scores of the first two axes (Fig. 3B).

Species accounts
The genetic and morphometric data provide congruent, independent lines of evidence to support the hypothesis that F. cancrivora Groups A and B represent two separate species. Specifically, Group A consists of a composite of F. cancrivora from Indonesia and Malay-sia, and F. raja from Thailand (Smith 1930;Taylor 1962;Chan-ard 2003;Chuaynkern and Chuaynkern 2012), while Group B consists of a composite of F. cancrivora and F. moodiei from Thailand, Philippines, China, and Bangladesh (Smith 1930;Taylor 1962;Chan-ard 2003;Kurniawan et al. 2010Kurniawan et al. , 2011Chuaynkern and Chuaynkern 2012; Table 1). We propose that Group A be referred to as F. cancrivora sensu stricto, with F. raja treated as a junior synonym of F. cancrivora. We propose that Group B be referred to as F. moodiei, with specimens of F. "cancrivora" in this clade reallocated to that species. The two species, F. cancrivora (Group A) and F. moodiei (Group B), can be recognized as follows.  (Table 4; Appendix 2, 3); (2) head length slightly greater than head width; (3) skin on dorsum and flank with spinules and glandular warts, with irregular skin folds not arranged in series; (4) relative finger lengths II < IV < I < III; (5) dermal fringe on Finger II and III; (6) prepollax indistinct; (7) palmar tubercles indistinct; (8) foot moderately webbed with webbing formula I1-11/2II1-2III1-2IV2-1V; (9) dermal flap on postaxial side of Toe V; (10) Fejervaryan lines absent; (11) inner metatarsal tubercles prominent; (12) inner tarsal ridge prominent on distal half to two-thirds of tarsus, and (13) vocal sacs in adult males with wrinkled skin covered by triangular, very dark brown blotches on each side of throat.

Fejervarya cancrivora (Gravenhorst, 1829)
Description of neotype. Dubios and Ohler (2000) designated and described the neotype adult male, FMNH 256688, from Java, Indonesia (Fig. 4A-B; Table 1). We supplement their description of the neotype, as follows: rather large size, body rather slender; head narrow, slightly longer than wide; snout oval in dorsal view, round in lateral view, projecting beyond lower jaw; nostril dorsolateral, pointed oval, with small lateral flap, closer to tip of snout than eye; canthus indistinct, rounded; loreal region concave and obtuse; eye diameter about 60% snout length; interorbital space flat, less than width of upper eyelid and internarial distance; pineal body visible; tympanum distinct, rounded [oval according to Dubois and Ohler (2000)], about 90% of eye diameter, not depressed relative to skin of temporal region, tympanic rim weakly elevated relative to tympanum, dorsoposterior margin obscured by supratympanic fold; two vomerine ridges bearing a few small teeth between choanae, obliquely oriented at an angle of 45° to body axis, closer to choanae than to each other; tongue large, cordate, emarginate [based on Ohler and Dubois (2000), not examined by us]; distinct supratympanic fold extending from eye to axilla, not obscuring dorsoposterior margin of tympanum.
Forelimbs short, rather stout [rather thin according to Dubois and Ohler (2000)], slightly longer than hand; fingers rather long, thin; tip of fingers slightly rounded and swollen [pointed according to Dubois and Ohler (2000)], but not expanded into discs; relative length of fingers II < IV < I < III; fingers II and III with dermal fringe; webbing on fingers absent; subarticular tubercles prominent, rounded; supernumerary tubercles absent; prepollex indistinct, oval; palmar tubercles indistinct.
Hindlimbs moderately short, robust; tibia longer than thigh, but shorter than distance from base of inner metatarsal tubercle to tip of Toe IV; toes long, thin; tips of toes rounded [pointed according to Dubois and Ohler (2000)], not expanded into discs; relative length of toes I < II < V < III < IV; webbing moderate, deeply excised between toes, formula I1-11/2 II1-1III1-2IV2-1V, Toe I webbed to base of distal phalanx; preaxial side of Toe II webbed to point between distal subarticular tubercle and distal phalanx, continuing as narrow fringe to base of distal phalanx; postaxial side of Toe II webbed to base of distal phalanx; preaxial side of Toe III webbed to distal subarticular tubercle, continuing as narrow fringe to base of distal phalanx, postaxial side of Toe III webbed to base of distal phalanx; preaxial side of Toe IV wedded to distal subarticular tubercle, continuing as narrow fringe to base of distal phalanx, postaxial side of Toe IV webbed to distal subarticular tubercle, continuing as narrow fringe to base of distal phalanx, Toe V webbed to base of distal phalanx; dermal flap well developed, extending along postaxial side of Toe V from level of inner metatarsal tubercles to distal phalanx; subarticular tubercles prominent; inner metatarsal tubercle prominent, oval, less than length of Toe I; distinct dermal ridge extending along inner metatarsal tubercle to distal phalanx of Toe I; distinct inner tarsal ridge on distal two-third of tarsus (Fig. 5A); outer metatarsal tubercles absent; supernumerary tubercles absent; tarsal tubercle absent.
Skin on snout and interorbital region shagreen; skin on eyelid with glandular warts and spinules; skin on dorsum with irregular skin folds, with intervening glandular warts and spinules; dorsolateral fold extending posteriorly to two-thirds length of dorsum; skin on side of head with small spinules; skin on flank with glandular warts; skin on cloacal region with dense glandular warts; skin on forelimbs, thigh, tibia and tarsus with glandular warts and spinules; skin on ventral surfaces smooth, except dense, fine spinules on chin. Nuptial pad with small translucent spinules on dorsal and medial surface of Finger I from base of distal phalanx to slightly over the base of prepollax; vocal sac present on both sides of throat, with wrinkled skin covered by triangular dark brown blotches. Fejervaryan lines absent.
Coloration of neotype in preservative. Dorsum and side of head medium brown with indistinct dark brown markings; dark brown band between outer margins of upper eyelids; tympanum brown with inferior half more translucent, lighter in coloration than head; flank creamy white with dark brown marbling; three wide dark brown vertical spots on upper lips; wide light brown mid-dorsal stripe continuous from tip of snout to vent; dorsal surfaces of forelimbs, thigh, tibia, and foot brown with dark brown transverse spots; posterior surface of thighs with irregular pattern of dark brown marbling on white background; chin mottled dark brown, throat with triangular dark brown blotches on each side; chest, belly and ventral surfaces of hindlimbs creamy white with indistinct dark brown mottling; ventral surfaces of forelimbs creamy white; ventral surfaces of hand and foot brown; lower lip creamy white with dark brown spots.
Coloration of referred Thai specimen in life. Adult male ZMKU AM 01426 ( Fig. 6A-E) from Khuan Khanun District, Phatthalung Province, Thailand. SVL 60.3 mm. Dorsum dark brown with indistinct darker markings, side of head lighter brown; dark brown band between outer margins of upper eyelids; lower half of tympanum with brown blotches; dark brown streak on canthus rostralis from tip of snout to eye; dark brown streak from eye along supratympanic fold to posterior rim of tympanum; flank creamy white with dark brown marbling; three wide dark brown spots on upper lips; a wide beige mid-dorsal stripe continuous from tip of snout to vent; dorsal surfaces of forelimb, thigh, tibia, and foot dark brown with darker transverse spots; posterior part of thigh with irregular pattern of dark brown marbling on light brown background; chin and chest creamy white with dark brown mottling; throat with triangular dark brown blotches on each side; ventral surfaces of forelimbs and belly creamy white; ventral surfaces of hindlimbs creamy white with dark brown mottling; ventral surfaces of hand and foot brown; lower lip creamy white with dark brown spots.
Variations. Females are distinctly larger in size (Table 4; Appendix 3), lack nuptial pads and vocal sacs, and have fewer spinules and glandular warts on dorsum and flanks than males. Two male specimens (ZMKU AM 01511 from Nakhon Si Thammarat Province, Thailand and CNHM 131100 from Java, Indonesia) have nuptial pads extending to the base of prepollax. Most male specimens have dense fine spinules over the entire surface of the chest, belly, and ventrolateral surface.
The examined male and female specimens closely resemble the neotype in morphology, with most observed variation pertaining to coloration. Dorsal coloration in preservative varied from medium to very dark brown with darker markings. Markings or spots on dorsum, and transverse spots on dorsal surface of forelimbs and hindlimbs fainter than neotype in some individuals. Flank pale brown with dark brown marbling in some individuals. Ventral coloration pale brown in some individuals, with dark mottling on chin and chest. Ventral surface of hand pale brown or creamy white in some individuals. Dorsal vertebral stripe present (n = 18, 41%) or absent (n = 26, Figure 6. Adult male Fejervarya cancrivora (ZMKU AM 01426) from Khuan Khanun District, Phatthalung Province, Thailand (SVL = 66. 9 mm) immediately prior to preservation in A right lateral B dorsal C ventral D right palmar, and E right plantar views. Photographs by Attapol Rujirawan. 59%). Two specimens from Nakhon Si Thammarat Province, Thailand (ZMKU AM 01509 and ZMKU AM 01513), have a narrow light brown stripe on tibia. Pineal body not visible in one male specimen from Pattani Province, Thailand (THNHM 21248).
Distribution. Based on a combination of the morphological and genetic studies of F. cancrivora large type (Kurniawan et al. 2010;, the reported distribution of F. raja (Chan-ard 2013;Chuaynkern, and Chuaynkern 2012), and localities of specimens examined in this study, F. cancrivora is distributed from south of the Isthmus of Kra in Thailand, West Malaysia, Kalimantan (Borneo), Sumatra, West and Central Java, and Bali in Indonesia, with introduced populations in Papua New Guinea and Guam (Christy et al. 2007;Frost, 2019). In Thailand, F. cancrivora was confirmed to occur at Phatthalung, Nakhon Si Thammarat, Pattani, Songkhla, and Narathiwat Province ( Fig. 1; Table 1).
Habitat, ecology and natural history. Specimens were collected in Thailand (Khuan Khanun District, Phatthalung Province and Pak Panang District, Nakhon Si Thammarat Province) at night (1900-2200 h) following light rain during May and October 2016. At Khuan Khanun, frogs were sampled in grasslands, rice paddy fields near standing or slow flowing ditches, and ponds at 1-24 m elevation (Fig. 7A). These were found sitting on the ground near water bodies, or hiding within grass or in mud cracks in the ground, and jumped to water bodies when disturbed. Other anuran species found in syntopy at this locality included Duttaphrynus melanostictus (Schneider, 1799), F. limnocharis, Hoplobatrachus rugulosus (Wiegmann, 1834), Hylarana erythraea (Schlegel, 1837), Polypedates leucomystax (Gravenhorst, 1829) and Microhyla butleri Boulenger, 1900. At Pak Phanang District, frogs were collected at night (1900-2100 h) after heavy rain in November 2017. These were found on the bank or in the water of brackish shrimp ponds near the Pak Phanang River at 0 m asl (Fig. 7B). No other anuran species were found in syntopy at this locality, although F. moodiei was sampled at a site approximately 4.5 air-km, or 5.2 km following the river course, upriver (below).
Description of holotype. Taylor (1920) described the species based on an adult female, CM 3724, from Manila, Luzon, Philippines (Fig. 8A, B; Appendix 3). We supplement his description of the holotype, as follows: rather large body size; head narrow, slightly longer than wide; snout tip oval in dorsal view, round in lateral view, projecting beyond lower jaw; nostril dorsolateral, oval, with small lateral flap, closer to tip of snout than eye; canthus indistinct, rounded; loreal region slightly concave and oblique [loreal region broadly sloping, not concave according to Taylor (1920)]; eye diameter about 60% snout length [eye diameter equal to snout length according to Taylor (1920)]; interorbital region flat, about half width of upper eyelid and slightly  less than internarial distance; pineal body present; tympanum distinct, rounded, about 90% of eye diameter, not depressed relative to skin of temporal region, tympanic rim weakly elevated relative to tympanum, dorsoposterior margin obscured by supratympanic fold; vomerine ridge present in two strongly oblique series, very slightly closer to each other than to choanae [based on Taylor (1920), not examined by us].
Forelimbs short, rather robust; fingers rather long, slightly swollen; tips of fingers slightly rounded, terminus slightly swollen but not expanded into discs; relative finger lengths II < IV < I < III [first finger longer than second and fourth according to Taylor (1920)]; dermal fringe on fingers absent; webbing on fingers absent; subarticular tubercles distinct; supernumerary tubercles absent; prepollex indistinct, oval; palmar tubercles indistinct.
Hindlimbs moderately short, robust; tibia slightly longer than thigh, but shorter than distance from base of inner metatarsal tubercle to tip of Toe IV; toe long, stout; tips of toes rounded, not expanded into discs; relatively toe lengths I < II < III < IV, webbing moderate, deeply excised between toes, formula I1-11/2II1-1III1-2IV2-1V, Toe I webbed to base of distal phalanx; preaxial side of Toe II webbed to point between distal subarticular tubercle and distal phalanx, continuing as narrow fringe to base of distal phalanx; postaxial side of Toe II webbed to base of distal phalanx; preaxial side of Toe III webbed to distal subarticular tubercle, continuing as narrow fringe to base of distal phalanx, postaxial side of Toe III webbed to base of distal phalanx; preaxial side of Toe IV wedded to webbed to proximal distal subarticular tubercle, continuing as narrow fringe to base of distal phalanx, postaxial side of Toe IV wedded to webbed to proximal distal subarticular tubercle, continuing as narrow fringe to base of distal phalanx, Toe V webbed to base of distal phalanx; dermal flap well developed, extending along postaxial side of Toe V from level of inner metatarsal tubercles to distal phalanx; subarticular tubercles prominent, inner metatarsal tubercle prominent, oval, length about 30% that of Toe I; distinct dermal ridge extending along inner metatarsal tubercle to distal phalanx of Toe I; indistinct inner tarsal ridge on distal two-third of tarsus (Fig. 7C); outer metatarsal tubercles absent; supernumerary tubercles absent; tarsal tubercle absent.
Skin on snout and between the eyes shagreened; skin on eyelid shagreened with glandular warts; skin on dorsum shagreened with glandular warts and irregular skin folds; dorsolateral fold extending posteriorly to two-thirds length of dorsum; skin on side of head smooth; skin on flank with glandular warts; skin on cloacal region with glandular warts; forelimbs shagreened; thigh with indistinct glandular warts; tibia, tarsus, throat, chest and belly smooth.
Coloration of holotype in preservative. Coloration mostly lost in preservative. Dorsum and side of head medium brown with a few dark brown markings; tympanum translucent brown with pale brown spot in center; flank pale brown with faint brown marbling; three wide brown vertical spots on upper lips; dorsal surfaces of forelimbs, thigh, tibia, and foot medium brown with a few dark brown spots, posterior surface of thigh with irregular pattern of indistinct dark brown marbling on light background; chin, chest, belly, and ventral surfaces of forelimb and hindlimb pale brown; ventral surfaces of hand and foot pale brown; lower lip pale brown with a few dark brown spots; vertebral and tibial stripes absent; Fejervaryan lines absent.
Coloration of referred Thai specimen in life. Adult male ZMKU AM 01390 ( Fig.  9A-E) from Mueang Phang-nga District, Phang-nga Province, Thailand. SVL 44.7 mm. Dorsum and side of head light brown with indistinct olive brown marking; olive-brown band between outer margin of upper eyelids; tympanum with orange-brown blotches in center; olive-brown streak on canthus rostralis from tip of snout to eye; dark brown streak from eye along supratympanic fold to posterior rim of tympanum; flank creamy white with dark brown marbling; three wide dark brown spots on upper lips; dorsal part of limbs: forelimbs, thigh, tibia, and foot light brown with olive-brown transverse spots, posterior part of thigh with irregular pattern of dark brown marbling on creamy yellow background; ventral part of body: chin creamy white with indistinct mottled dark brown, triangular dark brown blotches and mottling on each side of throat; forelimbs, chest, belly creamy white and hindlimbs with indistinct dark brown mottling, hand brown and foot dark brown; lower lip creamy white with dark brown spots; Fejervaryan lines absent.
Variations. Vomerine ridges slightly closer to choanae than to each other in some individuals. Most adult males have nuptial pads with small translucent spinules on dorsal and medial surface of Finger I from base of distal phalanx to base of prepollax, but some individuals have the nuptial pad extending to slightly over the base of prepollex. Most adult males have dense, fine spinules covering only the chin, but some individuals have dense, fine spinules on the chin and chest. Adult males have vocal sac present on each side of throat with wrinkled skin covered by triangular, very dark brown blotches. Adult males with larger spinules and glandular warts on dorsum, dorsal surfaces of forelimbs, flank, hindlimbs and vent region. Females are distinctly larger in size (Table 4, Appendix 3), lack nuptial pads and vocal sacs, having fewer spinules and glandular warts on dorsal surface of body and flank than males.
Dorsal coloration in preservative varies in males and females from brown to dark brown with darker markings. Two female specimens from Trat Province, Thailand (ZMKU AM 01444 and 01451) have dark orange markings on anterior part of dorsum. Markings or transverse spots on dorsum and dorsal surfaces of forelimbs and hindlimbs usually distinct, but faint in a few individuals. Coloration on flank usually creamy white, but pale brown, with dark brown marbling, in some individuals. Ventral coloration usually creamy white, but pale brown with indistinct dark mottling on chin and chest in some individuals. Hand usually creamy white, but light brown in some individuals. Most specimens have dermal fringe on fingers II and III (males N = 21, 70%; females N = 23, 69.7%), but some individuals lack this fringe (males N = 9, 30%; females N = 10, 30.3%). One specimen from Narathiwat Province, Thailand (THNHM 19720) has a vertebral stripe.
Distribution. Based on a combination of morphological and genetic studies of F. cancrivora mangrove type (Kurniawan et al. 2010(Kurniawan et al. , 2011 and Fejervarya Bangladesh mangrove type (Islam et al. 2008), the reported distribution of F. cancrivora (Chan-ard 2003;Chuaynkern and Chuaynkern 2012), F. moodiei (Brown et al. 2013), and F. cf. cancrivora (Harikrishnan and Vasudevan 2018), and specimens studied here, F. moodiei occurs in coastal areas from eastern India, the Andaman and Nicobar Islands, and southern China, southward through Vietnam, Thailand, Myanmar, Malaysia and Luzon Island in the Philippines. In Thailand, F. moodiei was documented in all coastal regions except the extreme southeastern Gulf of Thailand coast, where it is replaced by F. cancrivora (Fig. 1).
Habitat, ecology, and natural history. In Thailand, specimens were collected at night (1900-2200 h) in a variety of coastal habitats at elevations ranging from 0-16 m asl. Most specimens were observed in marshes near slow flowing ditches, ponds, or canals in mangrove forest (Fig. 7C). The species was also found in man-made environments such as agricultural fields adjacent to mangroves. In Kleang District, Rayong Province, most specimens were collected in and around brackish fish ponds and ditches in mangrove areas near the mouth of the Prasae River (Fig. 7D). Specimens from Pak Phanang District, Nakhon Si Thammarat Province were found around brackish shrimp ponds and ditches near the mouth of the Pak Phanang River. Frogs were observed sitting on the ground, under tree roots, or in or on the bank of water bodies. When disturbed, they usually escaped into holes in the ground or jumped into brackish water. No other anuran species were found in syntopy at this locality, although F. cancrivora was sampled at a site approximately 4.5 air-km, or 5.2 km following the river course, downriver (above).

Discussion
Our study clarifies that two species of crab-eating frogs (Fejervarya cancrivora complex) occur in mainland Southeast Asia: F. moodiei in coastal regions throughout mainland Southeast Asia, with replacement by F. cancrivora sensu stricto in extreme southern Thailand (on the Gulf of Thailand coast) and peninsular Malaysia. These findings corroborate those of Kurniawan et al. (2010; that the name F. moodiei is the correct name to apply to populations of the F. cancrivora complex throughout most of coastal mainland Southeast Asia. Our study provides the first molecular evidence that F. raja from southern Thailand represents only a large-bodied population of F. cancrivora sensu stricto, as suspected but untested by Iskandar (1998) and Kurniawan et al. (2010Kurniawan et al. ( , 2011. Both F. cancrivora and F. moodiei have wide geographic distributions that span coastlines of both mainland and insular Southeast Asia (Fig. 1), a likely testament to their remarkable tolerance of salt and brackish water (e.g., Gordon et al. 1961;Balinsky et al. 1972;Wright et al. 2004;Hopkins and Brodie 2015). Although our findings of two Southeast Asian frog species having wide geographic distributions is inconsistent with many recent analyses of other taxa (e.g., Stuart et al. 2006;Aowphol et al. 2013;Geissler et al. 2014;Phimmachak et al. 2015;Wogan et al. 2016;Sheridan and Stuart 2018), the conserved morphology of the F. cancrivora complex has long hindered accurately understanding species diversity and distributions of these frogs, as evidenced by the conflicting interpretations of experienced systematic herpetologists (e.g., Smith 1930;Inger 1954;Taylor 1962). Hence, the integrative taxonomic approach used here that incorporated both molecular and morphological data, including from topotypes and name-bearing type specimens, respectively, proved to be imperative for resolving these uncertainties.
This study provides a basis for revising the identifications of historical and contemporary records (both museum vouchers and literature descriptions) of crab-eating frogs to improve the finer-scale details of the geographic ranges, as well as the natural histories, of F. cancrivora and F. moodiei in mainland Southeast Asia. Our sampling did not reveal F. cancrivora and F. moodiei to occur in sympatry, but did find the two species to occur in shrimp ponds that were separated by only approximately 4.5 air-km (or 5.2 km following the river course) along the Pak Phanang River in Pak Phanang District, Nakhon Si Thammarat Province, Thailand ( Fig. 1; Appendix 1). The Pak Phanang locality of F. moodiei (8°19.850'N, 100°11.870'E) lies closer to the river mouth and has higher saltwater intrusion than does the Pak Phanang locality of F. cancrivora (8°17.454'N, 100°11.229'E) that lies further upstream of a complex system of water gates and irrigation canals that were constructed in the 1960s to reduce saltwater intrusion and facilitate rice production (Boromthanarat et al. 1991). It is not known if the two species were separated at these shrimp ponds because the two localities are coincident with the boundaries of their geographic ranges, or if the two species differ in saltwater tolerance and other aspects of their ecology. Future sampling to clarify the fine-scale partitioning of the two species where their ranges come into contact is warranted.