Frogs originated >200 mya, but the fossil record for Mesozoic anurans is relatively depauperate1. Most extant families of anurans likely originated by the end of the Paleogene with major geographically circumscribed clades originating even earlier in the Cretaceous2. The living diversity of anurans (>6,900 species)3 encompasses a range of morphotypes associated with different microhabitat specializations and various reproductive and locomotor modes4,5,6,7. Present-day tropical forests harbor an extraordinary diversity of frogs in both the number of species and axes of phenotypic variation such as reproductive modes8,9,10. However, there is little direct evidence from the fossil record of this association largely due to the limited paleoecological context for most Mesozoic anurans.

Many species in the diverse clade comprising ‘modern frogs’—the Neobatrachia—are found in tropical forests. However, among other extant frogs not included within the Neobatrachia, only the Pipidae, Megophryidae, and the genus Barbourula (Bombinatoridae) are associated with tropical forests. Time-calibrated molecular phylogenies indicate that both Barbourula and megophryid frogs, and thus likely their association with tropical forests, originated in the mid-Paleogene2,11. Based on the ecology and distribution of extant pipid frogs, this clade of largely aquatic species might have been associated with tropical freshwater habitats in southern Gondwana during the Mesozoic, though the paleoecological context for many relevant fossils remains limited. Recent phylogenetic analysis suggests that all extant families and subfamilies containing arboreal species originated after the Cretaceous2, suggesting that new ecological opportunities shaped anuran diversification as forests rebounded after the massive vegetation loss at the K–Pg extinction event12.

The amber deposits of northern Myanmar provide a unique record of a forest ecosystem during the Upper Albian (~99 mya13,14). Both plants, including mosses and bamboo-like monocots15,16, and invertebrates, including pisaurid spiders, onycophorans, dyspnoid harvestman, and coccoid scale insects13,17,18, preserved in Burmese amber provide evidence that this was a humid, warm, tropical forest ecosystem that contained at least some freshwater habitats. The presence of ammonites and marine ostracods suggest that some of the amber-bearing forests occurred near the shore of a marine environment19. Vertebrates have recently been reported from these deposits, including well-preserved three-dimensional anatomy of skeletons and feathers20,21,22. These amber deposits represent an excellent opportunity for discovering three-dimensionally preserved small vertebrates with a rich associated paleoecological context.

We report the first specimens of frogs preserved in amber from northern Myanmar. These are the oldest records of frogs preserved in amber, with the only two previous reports from Cenozoic amber deposits of the Dominican Republic23,24. These Burmese fossils provide the earliest direct evidence of anurans in a wet tropical forest ecosystem.


Systematic Paleontology

Order Anura Fischer von Waldheim, 1813.

? Alytoidea Fitzinger, 1843.

Family undetermined.

Genus Electrorana gen. nov. LSID, C4047BE5-2894-4B3D-8833-6DE99935D0B6.

Type species Electrorana limoae sp. nov.; by present designation LSID,


Electrorana is feminine and derives from the Latin electrum (amber) and rana (frog). The specific epithet, limoae, is a matronym in the genitive singular for Mrs. Mo Li, who purchased and provided these specimens for study.


Four fossil frog specimens (DIP-V-16119, DIP-V-16121, DIP-V-16127, and DIP-L-0826) are catalogued in the Dexu Institute of Palaeontology (DIP), Chaozhou, China (Figs 1, 2 and 3). A 3D-printed replica of DIP-L-0826 is deposited at the Florida Museum of Natural History (FLMNH VP-312847; Supplemental Fig. 1). The holotype (DIP-L-0826) is a partial skeleton embedded within a block of amber (Figs 1 and 2) that also contains an unidentified beetle (Coleoptera; D. Grimaldi, pers. comm.). Three other specimens preserve either an incomplete anuran forelimb (DIP-V-16121, 16127) or the general body shape of a frog that x-ray computed tomography reveals contains no skeletal material within (DIP-V-16119).

Figure 1
figure 1

Photograph images of four fossil frog specimens referred to Electrorana, including the holotype (A; DIP-L-0826) and three additional specimens (B: DIP-V-16119; C: DIP-V-16127; D: DIP-V-16121). Specimens in (B) and (D) are presented with two views of the amber specimen and the oval in (D) indicates the presence of the anuran specimen. Scale bars equal 5 mm.

Figure 2
figure 2

Holotype (a) of Electrorana limoae (DIP-L-0826) as visualized via microCT scanning. (b) Unidentified coleopteran embedded in amber with holotype. Skull in dorsal (c), left lateral (d), and ventral (e) views. Ventral views of pectoral girdle (f), left hand (g), vertebral column (h), and left foot (i). Abbreviations of anatomical terms are as follows: clav – clavicle; cor – coracoid; dent – dentary; exoc – exoccipital; fpar – frontoparietal; max – maxilla; phyd – parahyoid; pmax – premaxilla; postmed proc – posteromedial process of hyoid; pro – prootic; prsph – parasphenoid; pter – pterygoid; qj – quadratojugal; raduln – radioulna; scap – scapula; sept – septomaxilla; sq – squamosal; supscap – suprascapula; unc – uncinate process. The number of each digit is indicated with roman numerals. Scale bar for (ce), 2 mm; scale bar for (b,fi) 1 mm.

Figure 3
figure 3

Three fossil frog specimens referred to Electrorana visualized via microCT scanning. (A) DIP-V-16119, (B) DIP-V-16127, (C) DIP-V-16121. Scale bars equal 2 mm.

Type locality and horizon

These amber-preserved specimens were acquired in the area of Angbamo in Kachin Province of northern Myanmar in August 2015. Burmese amber derives from late Albian–Cenomanian deposits (approx. 105–95 mya25). Dating of zircons from the volcaniclastic matrix in these deposits provides an age of 98.8 ± 0.6 million years14.


Electrorana differs from extant and extinct anurans by the combination of a premaxilla with a prominent and bifurcated alary process, a V-shaped parahyoid bone, and free ribs (preserved on Presacral vertebra II), as well as the absence of a palatine.


The holotype is a partial skeleton of a small frog (~22 mm snout–vent length) preserving a skull, a nearly complete left forelimb, pectoral girdle, partial right forelimb, a partial preaxial vertebral column, and a partial left hindlimb. The neurocranium has been flattened such that the frontoparietals and sphenethmoid rest alongside the parasphenoid. The sacrum, urostyle, and pelvis are not preserved, nor are the dorsal components of the vertebrae (Supplemental Materials). Ossified sternal elements and posteromedial processes of the hyoid might have been displaced and not preserved.

The skull is approximately as wide as long (~9 mm in length and width) and relatively flat in lateral view. In dorsal view, the jaw joint is approximately even with the occiput. The dermal elements of the skull are not sculptured. The triangular nasals are widely spaced, form the anterior border of the orbit, have a weakly concave anterolateral margin, and taper posteriorly; the nasal does not bear a rostral process or articulate with the maxilla. The frontoparietals are paired and unfused, bordering a large fontanelle. The frontoparietal does not have a supraorbital flange or posterolateral process. The squamosal is present but difficult to discern due to pyritization of overlying soft tissues. The premaxilla is dentigerous (~10–12 teeth) and bears a stout, forked alary process extending dorsally. The palatine process of the premaxilla is weakly developed. A small triradiate septomaxilla is posterolateral to the alary process of the premaxilla. The stout maxilla articulates with the premaxilla, but is relatively shorter, and bears a deep facial process that tapers posteriorly along the orbit. The maxilla bears teeth along three-fourths of its length (~25 teeth). Posteriorly, the maxilla has an overlapping articulation with a slender quadratojugal. The vomer bears pre- and post-choanal processes with a dentigerous process (5–6 teeth) medial to the prechoanal process. There is no palatine. The sphenethmoid is present as a thin posteriorly directed C-shaped mineralization extending across the midline at the anterior margin of the parasphenoid; an ossified nasal septum is not present. The pterygoid is thin and tri-radiate, abutting the maxilla, lateral margin the parasphenoid ala, and the pyritized mass containing the squamosal and quadrate. The parasphenoid is triradiate, tapering anteriorly and nearly reaching the vomers, with alae extending across the ventral surface of the prootics and exoccipitals, which are not co-ossified. The occipital condyles are widely space.

The lower jaw does not bear teeth. Separate elements of the lower jaw, including the mentomeckelians, are difficult to discern but it appears to be comprised of a dentary and angulosplenial. Medial and anterior to the clavicles, there is a thin V-shaped parahyoid bone. Posteromedial processes of the hyoid are present posterolateral to the parahyoid and dorsal to the coracoids.

The pectoral girdle was likely arciferal, as suggested by the anteriorly curved clavicles. The clavicle tapers medially and is approximately twice the length of the stout scapula. The coracoid is approximately the same length as the clavicle, and is weakly expanded medially. Both the coracoid and clavicle articulate with the scapula. The clavicles are widely spaced, and there are no ossified sternal elements. There may be a cleithrum present but it is not clearly demarcated from the suprascapula. The humerus and fused radiulna are approximately equal in length. The metacarpals are relatively straight and lack processes along their lateral margins; metacarpal III is the longest. The carpals are not fully ossified and thus difficult to discern. The phalangeal formula of the manus is 2-2-3-3, and a prepollex is absent. The terminal phalanges of the manus are subtriangular and weakly expanded at their distal tip.

The vertebral column is incomplete, but portions of presacral vertebrae I–VII are preserved. Neural arches are not preserved and a transverse process is only preserved on Presacral II. The atlas does not bear transverse processes, and the atlantal cotyles are widely spaced and not contiguous. A free rib is preserved in articulation with the right transverse process of Presacral II, possibly with a posteriorly directed uncinate process. The centra of the presacral vertebrae are difficult to discern.

The hindlimb is incompletely preserved, but longer than the forelimb. Neither the femur nor the fused tibiofibula is preserved in its entirety. The tarsals are not fully ossified. An enlarged prehallux is absent. The metatarsals are all similar in length. The phalangeal formula of the pes is 2-2-3-4-3.


The holotype of Electrorana is likely not an adult. This is based on the absence of the columella and incomplete ossification of the carpals, prootic-exocciptal, and sphenethmoid. These often form or fully ossify after metamorphosis26. For example, in Bombina ossification of the columella may not happen until late in ontogeny (>2 years of age)27.

Phylogenetic analyses

Because previous phylogenetic analyses of Mesozoic anurans have produced conflicting results for the relationships of some taxa, we included Electrorana in three different recent matrices that differ in both their characters and taxa. Analysis of the 72 characters and 27 taxa (including Electrorana) based on the matrix from Báez28 resulted in nine equally parsimonious trees (score = 260); the matrix contains 20% missing data, including 41 of 72 characters for Electrorana. Analysis of the 97 characters and 52 taxa (including Electrorana) based on the matrix from Gao & Chen29 resulted in 598,547 equally parsimonious trees (score = 381); the matrix contains 28% missing data, including 47 of 97 characters for Electrorana. Analysis of the 66 characters and 27 taxa (including Electrorana) based on the matrix from Henrici et al.30 resulted in four equally parsimonious trees (score = 243); the matrix contains 20% missing data, including 36 of 66 characters for Electrorana.

The topology based on the matrix from Gao & Chen29 clearly indicates that Electrorana is a crown-group anuran and supports an affinity with the extinct taxon Aerugoamnis from the Early Eocene Green River Formation of Wyoming (Fig. 4). However, analyses based on the matrices of Henrici et al.30 and Báez28 suggest that Electrorana may be an earlier diverging lineage of crown-group anurans. Notably, analysis of the matrix of Henrici et al.30 which was used in the recognition and description of Aerugoamnis, does not suggest a close relationship between Electrorana and Aerugoamnis. Because of both the large amount of missing data in these matrices, as well as the relatively small number of characters (ranging from 66 to 97 characters), the phylogenetic affinities of Electrorana remain uncertain. However, based on the phylogenetic analysis, we are confident that Electrorana is not within crown-group Acosmanura, which includes Anomocoela and Neobatrachia. Further, among extant taxa, Electrorana has strong anatomical similarities to extant alytoids (formerly referred to as Discoglossoidea31).


Evolutionary relationships

Electrorana bears strong similarities to extant taxa that form the clade Alytoidea, including Bombinatoridae and Alytidae. These are the only extant taxa exhibiting the unique combination of a V-shaped parahyoid bone, free ribs, and lacking a palatine. Among extinct Mesozoic anurans, Electrorana is similar to anurans from the older Jehol Biota. These are recognized as either the single genus Liaobatrachus32 or a collection of crown-group anurans with affinities to extant Alytoidea33. Other taxa with similarities to Electrorana include Eodiscoglossus from the Jurassic and Lower Cretaceous of Europe1,34,35 and two taxa from the Lower Cretaceous of Japan36. However, phylogenetic analyses conducted by different authors result in conflicting patterns of relationships for these Mesozoic taxa, including whether these are all within crown-group Anura28,32,33,34,35,36. The incompleteness of the holotype of Electrorana, especially the lack of a sacrum, urostyle, and pelvis, complicates determining its affinities among these Mesozoic anurans. It is diagnosable as a member of crown-group Anura by the presence of a V-shaped parahyoid which is not known in stem anurans.


Among extant anurans, Electrorana is most similar to taxa in the families Alytidae and Bombinatoridae, which together comprise the Alytoidea. While today representing only five genera, many extinct taxa extending into the late Jurassic have been referred to this clade34,37. In comparison to Ascaphus (Ascaphidae) and Leiopelma (Leiopelmatidae), which comprise the earliest diverging lineage of extant anurans, Electrorana has more robust premaxillae that are approximately as tall as wide in anterior view. In both Ascaphus and Leiopelma, the premaxillae are wide with a narrow alary process. The alary process of the premaxilla in Electrorana is robust, comprising approximately half the length of the bone, which is most similar to the condition of Alytes (Alytidae) and Bombina (Bombinatoridae). The alary process is also forked dorsally, a condition similar to that recently reported for Genibatrachus by Gao & Chen29. The nasals of Electrorana are widely spaced and slender, similar to Ascaphus and Bombina (Bombinatoridae) and believed to be the ancestral condition for anurans38. The frontoparietals of Electrorana are widely spaced as in Ascaphus, Leiopelma, and Alytes, though unlike Bombina and Barbourula (Bombinatoridae) in which the fontanelle is partially or fully covered. The dentigerous vomer of Electrorana is more complicated in shape than the simple vomer found in Ascaphus (which does not have a postchoanal ramus of the vomer38), but similar in structure to the vomer of other anurans including Leiopelma, Alytes, Bombina, or Pelodytes (Pelodytidae). Unlike these other taxa, the vomer of Electrorana is relatively small and displaced anteriorly, which would suggest a relatively anterior location for the choana; however, this may be a result of either the holotype not being an adult specimen or being displaced postmortem. While thin and small, the sphenethmoid of Electrorana does appear to be contiguous across the midline, unlike the condition found in Ascaphus and Leiopelma in which it is generally poorly ossified38. The sphenethmoid is clearly not as robust as seen in Alytes, Discoglossus, Bombina, and many other anurans, but this is likely due to the holotype of Electorana not being an adult specimen. The pterygoids lack the enlarged ventrolaterally projecting flange found in Barbourula.

There is a thin V-shaped parahyoid bone, as found in Alytes, Discoglossus, and Pelodytes38, but there are no plate-like parahyoid bones as found in Bombina, Barbourula, or Rhinophrynus (Rhinophrynidae). A similar V-shaped parahyoid is found in the various anuran taxa referred to as Liaobatrachus32,34, including Callobatrachus, Mesophryne, Dalianbatrachus, and Yizhoubatrachus33. Unlike the condition in Pelodytes, the parahyoid bone does not appear ‘x-shaped’ with posterolateral processes that approach the bony posteromedial processes of the hyoid.

The one preserved rib of Electorana is relatively longer than seen in Alytes and Bombina. In Bombina, the ribs of Presacral II and III can bear a similar posteriorly directed uncinate process to what appears to be present in Electrorana. Unlike taxa in the Megophryidae, Pelobatidae, Pelodytidae, and many neobatrachian frogs, Electrorana does not exhibit ossified sternal elements along the midline.


The paleoecological context for most Mesozoic anurans is either unknown or not described as being a tropical forest. The paleoecology of the Jehol Biota in China (~130–122 mya39) is probably the best documented for a Cretaceous faunal assemblage that contains anurans. It is characterized as having seasonal and mesic environments40 and possibly open forested habitats41. The Lower Cretaceous Crato Formation of Brazil is another example with anuran fossils coming from lacustrine deposits that were also likely formed in a semiarid environment42. Other examples of paleoenvironments include Notobatrachus from the Jurassic La Matilde Formation of Argentina43 and the late Cretaceous Paralatonia from the Haţeg Basin of Romania44, both of which were associated with low-energy freshwater environments including floodplains and swamps. The earlier Triassic stem-anuran Triadobatrachus likely lived in semi-arid, near-shore riparian forests near floodplains45,46. Transitions to more arid paleoenvironments have even been cited as a possible driver of morphological evolution in the earliest anurans47. Electrorana from the mid-late Cretaceous amber deposits in Myanmar provides the oldest definitive association of anurans with tropical forests and suggests that lineages today associated with temperate regions may have occupied a larger diversity of habitats in the past.


We performed high-resolution x-ray computed tomography (CT-scanning) at the University of Florida’s Nanoscale Research Facility. We used a Phoenix v|tome|x M (GE Measurement & Control, Boston, USA) scanner with a 180 kv x-ray tube with a diamond-tungsten target and with the following settings: 75 kV, 150 mA, a 0.5 second detector time, averaging of three images per rotation and a voxel resolution of 8.6–10.7 µm. Raw 2D x-ray data were processed using the datos|x software v. 2.3 with post-processing, analyses (including segmentation), and visualization conducted using VG StudioMax v. 3.1 (Volume Graphics, Heidelberg, Germany).

A replica of DIP-L-0826 (enlarged by 300%) was created by 3D-printing.stl files that were extracted from the CT volumetric data. The replica was printed using rigid polypropylene materials using an Objet260 Connex2 3D printer (Stratasys, Eden Prairie, USA) based at the University of Florida’s Nanoscale Research Facility.

Morphological comparisons to other anurans were based on CT-scans of extant species also scanned at the University of Florida and fossil specimens housed at the Institute for Vertebrate Paleontology and Paleoanthropology in Beijing, China (see Supplementary Table 1), as well as published descriptions of extinct anurans.

To evaluate potential evolutionary relationships of Electrorana, we coded this new taxon into three different recent matrices used in phylogenetic analyses of extinct and extant anurans28,29,30. We conducted phylogenetic analyses using parsimony as an optimality criterion in PAUP* v4.0a (build 158)48. For the analysis of each matrix, we conducted 10 replicate heuristic searches with starting trees obtained through random stepwise addition. Missing (?) or inapplicable (−) character states were both treated as missing data in the analyses; taxa with multistate characters were treated as uncertainties (rather than polymorphisms). We summarized the results using majority rule consensus and these are presented in Fig. 4.

Figure 4
figure 4

Majority-rule consensus trees representing phylogenetic analyses using parsimony of matrices from (a) Henrici et al.30, (b) Báez28, and (c) Gao & Chen29 including Electrorana from the Cretaceous of Myanmar. Numbers adjacent to nodes represent the percentage of equally parsimonious trees containing that node. The phylogeny based on Henrici et al.30 is rooted using the extant alytoid clade, that based on Báez28 is rooted with Ascaphus, and that based on Gao & Chen29 is rooted with Triadobatrachus. Extinct taxa are indicated with a dagger, and Electrorana is indicated in boldface.

Data accessibility

Tomograms (TIFF) and shape files (STL) are available from MorphoSource ( See Supplemental Materials for further details.