JOURNAL OF MORPHOLOGY 272:149–168 (2011) The Evolution of Jumping in Frogs: Morphological Evidence for the Basal Anuran Locomotor Condition and the Radiation of Locomotor Systems in Crown Group Anurans Stephen M. Reilly* and Michael E. Jorgensen Department of Biological Sciences and Center for Ecology and Evolutionary Studies, Ohio University, Athens, Ohio 45701 ABSTRACT Our understanding of the evolution of frog locomotion follows from the work of Emerson in which anurans are proposed to possess one of three different iliosacral configurations: 1) a lateral-bending system found in walking and hopping frogs; 2) a fore-aft sliding mechanism found in several locomotor modes; and 3) a sagittalhinge-type pelvis posited to be related to long-distance jumping performance. The most basal living (Ascaphus) and fossil (Prosalirus) frogs are described as sagittalhinge pelvic types, and it has been proposed that long-distance jumping with a sagittal-hinge pelvis arose early in frog evolution. We revisited osteological traits of the pelvic region to conduct a phylogenetic analysis of the relationships between pelvic systems and locomotor modes in frogs. Using two of Emerson’s diagnostic traits from the sacrum and ilium and two new traits from the urostyle, we resampled the taxa originally studied by Emerson and key paleotaxa and conducted an analysis of ancestralcharacter state evolution in relation to locomotor mode. We present a new pattern for the evolution of pelvic systems and locomotor modes in frogs. Character analysis shows that the lateral-bender, walker/hopper condition is both basal and generally conserved across the Anura. Long-distance jumping frogs do not appear until well within the Neobatrachia. The sagittal-hinge morphology is correlated with long-distance jumping in terrestrial frogs; however, it evolved convergently multiple times in crown group anurans with the same four pelvic traits described herein. Arboreal jumping has appeared in multiple crown lineages as well, but with divergent patterns of evolution involving each of the three pelvic types. The fore-aft slider morph appears independently in three different locomotor modes and, thus, is a more complex system than previously thought. Finally, it appears that the advent of a bicondylar sacro-urostylic articulation was originally related to providing axial rigidity to lateralbending behaviors rather than sagittal bending. J. Morphol. 272:149–168, 2011. Ó 2010 Wiley-Liss, Inc. about it from the jumping frogs of Calaveras County (Twain, 1867) to the etymology of the oldest known anuran named Prosalirus bitis, meaning ‘‘to leap forward, high over it’’ (Shubin and Jenkins, 1995). The naming of this early frog appears to reflect two kinds of ‘‘leap’’ in the transition to frogs. First, it reflects the apparent design leap from the generalized tetrapod body plan to the set of traits that distinguish the frog morphotype: fewer presacral vertebrae, longer hindlimbs, and long ilia between which lies a shortened, fused tail called the urostyle. The early Triassic amphibian Triadobatrachus has the antecedents of the anuran body plan (moderately longer ilia, fewer trunk, and caudal vertebrae) but was built for lateral undulatory walking (Rage and Roček, 1986, 1989; Sanchiz, 1986). The earliest known frog, Prosalirus and its well-known sister taxon Notobatrachus, made the transition to longer ilia and caudal vertebrae fused into the urostyle (Báez and Nicoli, 2004). Both are morphologically intermediate between Triadobatrachus and modern anurans (Jenkins and Shubin, 1998). The second ‘‘leap’’ reflects the behavioral leap to specialization for bilateral limb extension in terrestrial locomotion that appears to have arisen with the new body plan. The current hypothesis for the evolution of locomotion in frogs follows from the work of Emerson (1979, 1982) in which anurans are proposed to possess three different iliosacral configurations that are correlated with locomotor mode (burrowing, walking, hopping, and jumping) KEY WORDS: frog evolution; jumping; sacrum; urostyle; bicondylar *Correspondence to: Stephen M. Reilly, Department of Biological Sciences, Ohio University, Athens, OH 45701. E-mail: [email protected] INTRODUCTION Received 10 May 2010; Revised 10 June 2010; Accepted 19 June 2010 The advent of frog jumping is one of the most enigmatic and paradoxical evolutionary transitions in vertebrate evolution. Much has been written Published online 8 November 2010 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/jmor.10902 Ó 2010 WILEY-LISS, INC. Additional Supporting Information may be found in the online version of this article. 150 S.M. REILLY AND M.E. JORGENSEN Fig. 1. Three basic pelvic designs in frogs (Emerson 1979, 1982) based on the shape of the sacral diapophyses, the presence/absence of iliac ridges (CT images), and the nature of the dorsal iliosacral ligaments (bottom, modified from Emerson, 1982). A: Emerson’s fore-aft slider morph with expanded flat-sided sacral diapophyses (flat sides either bony or cartilaginous) and a large ilio-ilial ligament. B: The lateral-bender morph with expanded but rounded diapophyses (with rounded cartilaginous caps, not shown) and long dorsal iliosacral ligaments. C: The sagittal-hinge morph with rod-like elevated (white arrows) diapophyses, iliac ridges, and short dorsal iliosacral ligaments. and ecological (terrestrial and arboreal) groups. One configuration (with greatly expanded flatsided sacral diapophyses) is proposed to be a predominantly fore-aft-sliding system in derived aquatic forms but is also scored for some climbing and burrowing forms. Another configuration (with bowtie-like sacra and narrow to moderately expanded sacral diapophyses) is mostly a lateralbending system in walker/hoppers and some burrowers. The third (with rod-like, nonexpanded diapophyses) is hypothesized to function as a sagittalhinge system, limiting the iliosacral movement to vertical rotation in long-distance jumping forms. Journal of Morphology These pelvic functions are based on limited functional studies of a few model species (movement or cineradiography of anesthetized specimens: Whiting, 1961; Emerson, 1979; one EMG study: Emerson and De Jongh, 1980; and cineradiographic frames of three sagittal-hinge jumps: Jenkins and Shubin, 1998) and are illustrated in videos in the Supporting Information S1–S5. The ‘‘high over it’’ connotation of P. bitis’ etymology reflects the current hypothesis that long-distance jumping with the sagittal-hinge (as observed in ranid frogs) is the ancestral state for Anura (Fig. 1C; Shubin and Jenkins, 1995; Jenkins and THE EVOLUTION OF LOCOMOTION IN FROGS Shubin, 1998). This hypothesis has not been proposed from any formal phylogenetic analysis but is based entirely on interpretations of iliosacral traits in the most basal fossil (Prosalirus) and living (Ascaphus) frog taxa as evidence that the sagittalhinge-type pelvis appeared in the first frogs. The extinct Prosalirus was interpreted as having the sagittal-hinge morph and ranid-like jumping ability based on two partial ilia and one partial sacrum among the remains of about four individuals (Shubin and Jenkins, 1995; Jenkins and Shubin, 1998). Although Emerson did not comment on the basal anuran sacral condition, she did score the basal living frog family (Leiopelmatidae) as possessing her sagittal-hinge iliosacral traits based on dissection of one specimen of Ascaphus truei (Emerson, 1979). With the ‘‘sagittal-hinge jumping is primitive’’ hypothesis in mind, we were perplexed by the question of how ranid-like jumping evolved so rapidly in the first frogs. A gradual transition seems more plausible, one changing from a lateral-bending mode (posited in Triadobatrachus) to a walker/ hopper mode of other basal living frogs (such as Rhinophrynus and Bombina) to dorsoventral bending during locomotion posited for phylogenetically derived jumping frogs. We also questioned the parsimony of resulting inferences from the ‘‘sagittalhinge jumping is primitive’’ hypothesis that many basal and crown taxa subsequently and repeatedly ‘‘abandoned’’ the hinge joint to evolve the other horizontal-moving iliosacral configurations for swimming, walking, hopping, and burrowing (Shubin and Jenkins, 1995; Přikryl et al., 2009). The purpose of this study was to reexamine the anatomy of the pelvic region in frogs to better understand historical patterns and correspondence of iliosacral traits in relation to Emerson’s pelvic functional complexes and basic behavioral locomotor modes. We studied two bony characters Emerson used that are well defined and easily scorable. Emerson (1979, 1982) defined the sagittal-hinge type as having two bony features: 1) narrow (termed cylindrical, nonexpanded) sacral diapophyses extending posterolaterally and 2) large raised dorsal ridges on the ilia (Fig. 1). In contrast, Emerson’s lateral-bender and fore-aft slider morphs have slightly to moderately expanded or greatly expanded flat-sided sacral diapophyses, respectively, and lack iliac ridges (Fig. 1). We examined the traits that defined Emerson’s (1982) three iliosacral functional units in three samples. First, we studied larger sample sizes of the same (or closely related) species Emerson used. Second, we included a larger taxonomic sample of the most basal living frogs (Leiopelmatidae), which are considered to be very similar to the last common ancestor of living frogs (Roelants and Bossuyt, 2005). Finally, we added extinct taxa (including Prosalirus) for which pelvic traits could be scored 151 and their phylogenetic placement is known. We scored taxa using literature descriptions, new dissections, cleared and double-stained preparations, and micro-CT and radiographic imagery. In addition, we scored two new traits related to pelvic function: the form of the sacro-urostylic articulation and nature of the dorsal ridge on the urostyle. Basic locomotor states were taken from the locomotor modes scored for the study taxa by Emerson (1979, 1982), and data on jumping performance in relation to body length were used to provide a quantitative distinction for hopping versus jumping following Zug (1978) and Emerson (1979, 1982). MATERIALS AND METHODS We sampled pelvic characters in 75 species representing five paleotaxa and a resampling of 54 taxa used by Emerson (1979, 1982) in her description of the iliosacral functional complexes of frogs (Table 1). Thirty-six were the same species, 13 were different species of the same genus, two were different genera in the same family, and 18 were new species we added. Three taxa were omitted (where Emerson sampled two closely related species within families). We also sampled the two species of Ascaphus currently recognized, one species of Leiopelma (n 5 3) and five extinct basal taxa (Table 1) for which phylogenetic relationships have been proposed (Gao and Wang, 2001). The phylogeny of the extant taxa primarily follows Frost et al. (2006) with topology of the microhylid subfamilies following Van der Meijden et al. (2007), and we recognized the family status of the Eleutherodactylidae following Hedges et al. (2008). Emerson’s original sampling now covers a broader array of higher anuran taxa because of advances in our understanding of anuran phylogeny (Frost et al., 2006). Taxonomic changes in generic names are indicated with the new usage in parentheses in the middle column of Table 1. Museum specimen numbers are listed in Appendix. Characters were studied and scored from dorsal X-ray images (n 5 except where noted in Table 1) from each of the extant taxa and verified where needed by dissection, from cleared and double-stained specimens (bone and cartilage), from whole-body micro-CT images of 36 taxa (asterisks in Table 1), and using illustrations and/or descriptions in the literature for extant (Stephenson, 1952; Ritland, 1955; Estes and Reig, 1973; Duellman and Trueb, 1994; Evans and Borsuk-Bialynicka, 1998; Brown and Crespo, 2000; Pugener, 2002; Přikryl et al., 2009; Pugener and Maglia, 2009) and extinct taxa (Rage and Roček, 1986, 1989; Sanchiz, 1986; Roček and Nessov, 1993; Báez and Basso, 1996; Jenkins and Shubin, 1998; Gao and Wang, 2001; Pugener, 2002; Gao and Chen, 2004; Roček, 2008; Pugener and Maglia, 2009). Micro-CT images were scanned at the Ohio University Micro-CT facility using a GE eXplore Locus small animal scanner at a slice thickness of 45 or 90 lm. Data were output as DICOM images, and 3D reconstructions were produced in the program MeVisLab (version 2.0). Ancestral Character State Reconstruction Ancestral character states of crown group taxa were reconstructed for each of our four pelvic characters (sacral diapophyseal expansion, dorsal iliac ridge, sacro-urostylic articulation, and urostyle dorsal ridge) using maximum likelihood (ML) in Mesquite (version 2.72; Maddison and Maddison, 2009). The Mk1 model of evolution was used in the ML analysis because this treats changes between all character states as equally likely (Schluter et al., 1997). Character probability trees were used to visualize patterns of evolution in individual traits. Journal of Morphology 152 S.M. REILLY AND M.E. JORGENSEN TABLE 1. Species sampled in order of taxa listed on phylogenetic trees Sampled taxa yTriadobatrachus yProsalirus yNotobatrachus yMesophryne Leiopelmatidae Rhinophrynidae yPalaeobatrachus Pipidae Emerson’s (1979, 1982) study species Ascaphus truei Rhinophrynus dorsalis Xenopus laevis Pipa pipa Alytidae Discoglossus pictus Bombinatoridae Bombina orientalis Scaphiopodidae Pelobatidae Megophryidae Scaphiopus couchii Pelobates fuscus Megophrys monticola Myobatrachidae Pseudophryne occidentalis Limnodynastidae Eleutherodactylidae Hylidae: Pelodryadinae Phyllomedusinae Notaden bennetti Eleutherodactylus punctariolis Litoria rubella Agalychnis callidryas Pachymedusa dacnicolor Hylinae Hyla cinerea Hyla gratiosa Smilisca phaeota Phrynohyas (Trachycephalus) venulosa Hylinae Pseudacris triseriata Hyla (Pseudacris) regilla Leptodactylus pentadactylus Leptodactylidae Ceratophryidae Cycloramphidae Leiuperidae Bufonidae Telmatobius marmoratus Rhinoderma darwinii Physalaemus (Engystomops) pustulosus Bufo (Rhinella) marinus Bufo (Anaxyrus) boreas Bufo (Anaxyrus) americanus Bufo (Rhaebo) blombergi Bufo (Epidalea) calamita Melanophryniscus stelzneri Atelopus zeteki Dendrobatidae Dendrobates tinctorius Hemisotidae Arthroleptidae Hemisus marmoratus Leptopelis bocagii Leptopelis aubryi Kassina senegalensis Hyperolius marmoratus Hyperoliidae Microhylidae: Kalophryninae incertae cedis Gastrophryninae Phyrnomerinae Microhylinae Journal of Morphology Kalophrynus pleurostigma Myersiella subnigra Gastrophryne carolinensis Hypopachus variolosus Dasypops shirchi Phrynomerus (Phrynomantis) annectens Microhyla rubra Kaloula pulchra Glyphoglossus molossus Species in this study 1, 2 (see Fig. 6) 3 (see Figs. 2 and 3) 4, 5 (see Figs. 2 and 3) 6 Same species*, 7 Ascaphus montanus* (n 5 3) Leiopelma hochstetteri* (n 5 3), 8 Same species* 6 Same species (x 5 4) Same species Pseudhymenochirus merlini* Same species* Alytes obstetricans* Same species* (n 5 3) Bombina maxima Scaphiopus hammondii* Pelobates cultripes (n 5 4) Megophrys nasuta Ophryophryne microstoma* Scutiger boulengeri* Pseudophryne bibronii Myobatrachus gouldii* Same species Eleutherodactylus coqui Litoria nasuta* Same species Same species* Phyllomedusa sauvagii* (n 5 1) Same species (n 5 4) Hyla versicolor (n 5 3) Smilisca baudinii Trachycephalus jordani Triprion petasatus* Same species Same species Leptodactylus gracilis* Paratelmatobius lutzii* (n 5 4) Telmatobius brevipes Same species* Same species Same species Same species* (n 5 3) Rhaebo haematiticus Same species (n 5 4) Same species Atelopus spumarius* Pedostibes hosii* Dendrobates auratus Oophaga pumilio* Same species* Cardioglossa leucomystax Same species Hyperolius lateralis Same species* Same species* (n 5 1) Same species Same species Same species* Phrynomantis bifasciatus* Microhyla berdmorei Same species Same species* Chaperina fusca* THE EVOLUTION OF LOCOMOTION IN FROGS 153 TABLE 1. (Continued) Sampled taxa Emerson’s (1979, 1982) study species Asterophryninae Cophixalus riparius Ptychadenidae Hildebrandia ornata Rhacophoridae Rhacophorus leucomystax Rhacophorus colletti Chiromantis rufescens Rana (Lithobates) catesbeiana Rana(Lithobates) pipiens Ranidae Species in this study Cophixalus cheesmanae Cophixalus oxyrhinus Same species (n 5 4) Ptychadena anchietae* Rhacophorus viridis Rhacophorus chenfui* Chiromantis xerampelina* Same species (n 5 3) Same species* Amolops hosii* (n 5 3) Meristogenys jerboa* The 54 species used by Emerson (1979, 1982) in her description of the pelvic functional complex in frogs (middle column) are currently assigned to the 34 higher listed taxa in column 1. We resampled Emerson’s taxa (n 5 5–8 per species (Appendix) or as noted, column 3) with some substitutions and added 18 species within these families as well as extinct taxa (y) of known phylogenetic placement for which traits could be scored from literature accounts. Micro-CT material is marked with asterisks. Three species were omitted in families represented by two species. Sources for paleotaxa and additional leiopelmatids are indicated with numbers in column 3. 1, Rage and Roček, 1989; 2, Roček, 2008; 3, Jenkins and Shubin, 1998; 4, Pugener, 2002; 5, Báez and Nicoli, 2004; 6, Gao and Wang, 2001; 7, Ritland, 1955; 8, Stephenson, 1952. Unique character states (indicated by square symbols) were not included in the ML reconstructions. Patterns of trait correspondence across all four traits were then used to identify pelvic systems and compare them to locomotor modes (burrowing, walking, hopper, terrestrial jumping, and arboreal jumping) of Zug (1978), Emerson (1979), and Brown and Crespo (2000). We follow the convention of Zug (1978) and Emerson (1979) who used an eight body-lengths threshold to distinguish short from long-distance leaping and termed them hoppers and jumpers, respectively. Although the hopper/ jumper distinction is a relatively quantitative one, the other general locomotor modes used describe the most common types of locomotion used by each species and are the best comparative data available in the literature. Clearly, there are exceptions to these patterns in some taxa, but the simplification of locomotor states is necessary to begin to understand patterns of evolution in anuran locomotor anatomy. The work was conducted from 2008 to 2010 at Ohio University under approved animal care and use protocols. RESULTS Emerson’s Skeletal Sacral Traits Sacral shape. Micro-CT images illustrating Emerson’s three basic types of frog pelvic systems are shown in Figure 1, and pelvic movements associated with these types are presented in movies in the Supporting Information S1–S6. Sacral diapophyseal expansion was the most obvious skeletal character to score in dorsal view, and it has three states (Emerson, 1982). The fore-aft-slider morph possesses longitudinally expanded and flat-sided butterfly-like diapophyses (Fig. 1A). We will call this the flat-sided morph. These sacra are either completely bony (e.g., Pseudhymenochirus) or have rounded sacral diapophyses upon which large extended cartilaginous edges make up the flat sides that articulate with the ilia (e.g., Triprion and Litoria). Here, a large ilio-ilial ligament extends from the ventrolateral and anterior half of one iliac shaft, dorsally over the back to the other iliac shaft, forming a tunnel in which the sacrum slides over the ilia (Fig. 1A, bottom). Manipulations of this pelvic type show that the ilia appear to be limited to bilateral fore-aft sliding because lateral movements of the pelvis translate through the sacrum to cause bending in the presacral vertebral column (Supporting Information S1). The lateral-bender morph has a bowtie-like sacrum ranging from narrow to widely expanded diapophyses (Fig. 1B). The diapophyses are dorsoventrally flattened with curved bony margins covered with curved cartilaginous caps. The cartilaginous caps extend the lateral edges of the sacrum to produce a larger curved articular surface than the underlying bone and may extend beyond the bony tip anteriorly and posteriorly as well (Fig. 4C,D). This morph has a dorsal iliosacral ligament with a wide attachment on the anterolateral ilium (slightly narrower than the width of the distal diapophysis) that extends medially over the diapophysis and narrows to its attachment on the medial half of the dorsal aspect of the diapophysis (Fig. 1B, bottom). Each iliac shaft can move fore and aft several millimeters to the extent that the ligament allows. Thus, the ilia can slide fore and aft together (Supporting Information S2) or in opposite directions during lateral bending (Fig. 4B; Supporting Information S3). The sagittal-hinge morph has rod-like diapophyses that extend posterolaterally and are usually elevated above the horizontal plane that runs along the tips of the vertebral neural arches (Fig. 1C). In these sacra, the ilia are tightly bound to the diapophyses by short, dorsal iliosacral ligaments incorporated into a capsule by a binding of fascia. In this pelvic type, there is little movement Journal of Morphology 154 S.M. REILLY AND M.E. JORGENSEN around these ligaments except for hinge-like bending as the ilia rotate under the rod-like diapophyses (Supporting Information S4). Here, lateralbending movements translate through the tightly bound iliosacral articulation to bend the presacral vertebral column (Supporting Information S5). Emerson’s iliac ridge trait. Emerson’s second bony trait was the presence of a large dorsal ridge on each iliac shaft that she described as diagnostic of the sagittal-hinge morph (1979, 1982). Our resurvey reveals that, when present, this dorsal ridge is a large to very large ‘‘fin’’ that arises on the dorsal aspect of the ilia just posterior to where the ilia articulate with the sacral diapophyses and extends posteriorly to the ischium (Fig. 2B). The iliac ridge exists in striking contrast to the smooth ilia seen in other pelvic types (Fig. 2A). In the smooth ilia, the ilial cross section has a round or oval shape and looks much like the cortical cross section of a long bone (Fig. 2). In ilia with a dorsal ridge, the ridge extends well above the diameter of the iliac shaft, generally curves medially on its dorsal aspect, and it often increases in height posteriorly (Fig. 2B). The pipids are aquatic and have laterally directed ridges that we scored as unique (&); we did not include these in the character reconstruction. Reinterpreting Emerson’s Traits in Prosalirus and Ascaphus Sacral shapes of fossil and living basal anurans are illustrated in Figure 3A–F. The most basal frogs are represented by the sister taxa Notobatrachus and Prosalirus (Gao and Wang, 2001). Notobatrachus is the best-known Jurassic frog with over 100 specimens, many from the same site with exceptional preservation (Báez and Nicoli, 2004). The sacral diapophyses of Notobatrachus are dorsoventrally flattened (Fig. 3A, from Pugener, 2002), and the well-preserved series of sacra illustrates the range of intraspecific variation in diapophyseal expansion in this species (Fig. 3B, from Báez and Nicoli, 2004). The sacrum of Prosalirus (Fig. 3C) is drawn from one partial fossil found in a group of four disarticulated specimens (Jenkins and Shubin, 1998). The single known Prosalirus diapophysis (Fig. 3C) was described as approximately triangular in cross section but widest in the anteroposterior direction. Although it lacks a key sagittal-hinge trait (iliac ridges), Shubin and Jenkins (1995) interpreted Prosalirus to have a sagittal-hinge pelvis apparently because of the general morphological similarity of its sacrum to a ranid sacrum. Others consider Prosalirus to be most similar to Notobatrachus on the basis of sacral morphology and several other osteological traits (Báez and Nicoli, 2004; Pugener and Maglia, 2009). As shown in Figure 3, the Prosalirus sacJournal of Morphology rum is similar to those of other basal anuran taxa, and the fossil reconstruction falls well within the range of phenotypic variation known in both its sister taxon Notobatrachus (Fig. 3B) and the most basal living frog Ascaphus (Fig. 3D). In addition, shading in the reconstruction indicates that the Prosalirus diapophyses are not elevated above the horizontal plane as in sagittal-hinge sacra (Fig. 1C). We follow Pugener and Maglia (2009) in scoring the Prosalirus sacrum as the expanded type, and thus, a lateral-bender morph. Our reexamination of the sacra of the most basal living taxa confirms that Ascaphus and Leiopelma also have dorsoventrally flattened diapophyses (Duellman and Trueb, 1994; Gao and Wang, 2001; Pugener, 2002; Fig. 3E,F). Ritland (1955) illustrated the variability in diapophyseal expansion in Ascaphus (Fig. 3D) similar to that in Notobatrachus (Fig. 3B). Ascaphus possesses expanded cartilaginous flanges on the diapophyses illustrating the larger extent of sacral expansion often not revealed by ossified portions of frog sacra (Fig. 4C,D). Leiopelma has flattened diapophyses, as well, but possesses a unique sacrum among frogs in that it is proximally ‘‘Y’’ shaped diapophyses articulate on the midline (Stephenson, 1952; Fig. 3F). Each diapophysis can be moved independently anteroposteriorly relative to the sacral centrum. Both Ascaphus and Leiopelma lack the iliac ridges diagnostic of Emerson’s sagittal-hinge morph (Fig. 3E,F). It is unclear why Emerson scored Ascaphus as having a rod-like diapophysis unless she examined a specimen with an extremely narrow diapophysis or a juvenile. Emerson also based her scoring on the size of the dorsal iliosacral ligament. In our dissections of Ascaphus, we found a long dorsal iliosacral ligament essentially identical to that of other lateral-bender sacra (Fig. 4A right, B) with a broad attachment along the ilium and a sacral attachment extending to the proximal end of the diapophysis (Fig. 4C). In addition, there is a large articular surface on the dorsal face of the ilium where the sacrum slides in the horizontal plane (Fig. 4C, right ilium) unlike any sagittalhinge sacrum. Finally, manipulations of the Ascaphus pelvis (Fig. 4D) show a great deal of fore-aft play in the iliosacral joints, with the dorsal iliosacral ligament limiting the extent of forward and backward motion (bilaterally and unilaterally) in the same way it does in other lateral-bender morphs (compare Fig. 4B–D or Supporting Information S3–S6). It is clear that Ascaphus has the lateral-bender morphology on the basis of expanded sacral diapophyses, smooth ilia, long dorsal iliosacral ligaments, and the horizontal plane of pelvic mobility. The leiopelmatids also have a fundamentally different swimming (Abourachid and Green, 1999; Nauwelaerts and Aerts, 2002) and leaping (Essner et al., 2010) behavior than other frogs. THE EVOLUTION OF LOCOMOTION IN FROGS 155 Fig. 2. Ilial and urostyle morphology in frogs. Frogs either have smooth ilia (A) or large dorsal ridges on them (B). The urostyle of frogs is either smooth (A) or has large dorsal ridges extending along the anterior half or the full length of the bone (B). Some frogs with smooth ilia retain some architectural remnant of the neural arch components in the form of raised lines of bone on the dorsal surface (A). Line drawings illustrate the plesiomorphic ‘‘L’’ shaped sacral diapophyses and independent caudal vertebrae but smooth anuran-like extended ilia of the immediate outgroup to Anura (Triadobatrachus, modified from Rage and Roček, 1986) and the smooth urostyle and smooth ilium of the basal most anurans [the Notobatrachids Notobatrachus and Prosalirus (modified from Pugener, 2002 and Jenkins and Shubin, 1998, respectively)]. Journal of Morphology 156 S.M. REILLY AND M.E. JORGENSEN Fig. 3. Sacral shapes and intraspecific variation in the degree of expansion in basal lateralbender frogs. Line drawings are modified from A: Pugener, 2002; B: Báez and Nicoli, 2004; C: Jenkins and Shubin, 1998; and D: Ritland, 1955. Two New Pelvic System Traits on the Urostyle One of the diagnostic features of anurans is that the adult caudal vertebrae are fused into a single rod-like tail, the urostyle (Gaupp, 1896). In surveying taxa for this analysis, we were able to discern and score two additional traits from the urostyle that are relevant to the design of the pelvic girdle in frogs because many of the muscles extending to the sacrum and ilia arise from this fused remnant of the tail. One is the nature of the urostyle’s articulation with the sacrum and the other concerns the presence and extent of a dorsal ridge on the urostyle. Journal of Morphology Sacro-urostylic articulation. Although Triadobatrachus possessed caudal vertebrae (Fig. 7, &), the earliest frogs (e.g., Notobatrachus, Figs. 2A and 3A) retained only one unfused postsacral vertebra between the sacrum and the urostyle. The shapes of the centra in these taxa indicate that normal intervertebral cartilaginous disks lie on either side of the unfused postsacral vertebra. In higher taxa, all of the vertebrae are fused into the urostyle that articulates with the sacral centrum (Gaupp, 1896; Duellman and Trueb, 1994). In basal frogs, the sacro-urostylic articulation ranges from having a normal intervertebral disk and large concave articular surfaces (Fig. 5A) to THE EVOLUTION OF LOCOMOTION IN FROGS 157 Fig. 4. Lateral-bender morphology and pelvic movements in Ascaphus. The large dorsal iliosacral ligaments (A, right) that limit the lateral-bending movements of the ilia on the sacrum are shown in the lateral-bender morph Rhinella marinus (B, Supporting Information S3). Dorsal ligaments and a flat iliosacral articulation surface (C) and lateral-bending movements (D, Supporting Information S4) are present in Ascaphus revealing that Ascaphus is not a sagittal-hinge morph. Panel A is modified from Emerson (1982). Fig. 5. Sacro-urostylic articulations in frogs. Frogs have either a single articulation surface with a full range of axial movements possible (A, monocondyly), a bicondylar articulation limiting lateral bending of the urostyle (B), or the urostyle is fused to the sacrum (Fig. 1: A, top and bottom panels; B, top panel). Line drawings, top to bottom, are modified from Jenkins and Shubin, 1998 and Pugener, 2002. Journal of Morphology 158 S.M. REILLY AND M.E. JORGENSEN centra forming a ball and socket (Fig. 5A, Barbourula). For the purposes of this study, we consider all urostyles with single articular surfaces as monocondylar (Fig. 5A). Many frogs have developed what has been called a bicondylar sacro-urostylic joint (Fig. 5B; Duellman and Trueb, 1994; Pugener and Maglia, 2009). Here, instead of a single articular surface, the sacral centrum has two side-by-side hemispherical condyles that articulate with a matching pair of concave surfaces on the anterior face of the urostyle. In a third and derived form (discussed below), the urostyle is solidly fused to the sacrum (e.g., Fig. 1A: Pseudhymenochirus; 1B: Scaphiopus). Cases of urostylic fusion that appear in several taxa across the phylogeny from both mono- and bicondylar ancestry are discussed below. The details of variation in monocondyly and the development of the bicondylar sacro-urostylic complex and the origin of the fused condition have recently been described by Pugener and Maglia (2009) but require a broader sampling and comparative analysis. Much has been said about the hypothesized function of the urostyle in locomotion and the contrasting range of movements possible with a monoversus bicondylar urostyle (Green, 1931; Whiting, 1961; Emerson, 1979, 1982; Emerson and De Jongh, 1980; Duellman and Trueb, 1994; Jenkins and Shubin, 1998; Pugener and Maglia, 2009). Currently, the function of the urostyle in different modes of locomotion is not well understood as it has only been examined in one comparative study of three species (Emerson and De Jongh, 1980). In tetrapods, muscles that control tail movement are also widely used in moving the hindlimb. However, in frogs with elongated ilia shifting the acetabula posterior to the tail, the muscles of the tail (urostyle) have become involved in novel mechanisms controlling movements of the pelvis relative to the vertebral column during locomotion (Emerson, 1979, 1982). Functionally, there may be different biomechanical implications for a monocondylar joint with the freedom to move three dimensionally compared to the bicondylar joint that appears to be uniaxial with movement restricted to axial flexion-extension. Accordingly, the bicondylar joint has been considered the best joint for jumping locomotion because it is hypothesized to be involved in allowing the body to straighten out during take off (Emerson, 1979, 1982; Emerson and De Jongh, 1980; Jenkins and Shubin, 1998; Kargo et al., 2002; Pugener and Maglia, 2009). Although the relationship between jumping and bicondyly (and monocondyly and lateral bending) has been proposed, the relationship between the state of the sacro-urostylic articulation and Emerson’s pelvic morphs has not been quantified. Therefore, we added a scoring of mono- versus bicondylar sacrourostylic articulation for comparison to the other pelvic traits and variation across locomotor modes. Journal of Morphology Urostyle dorsal ridge. In all frogs, the body of the urostyle is more or less cylindrical but differs in the presence and nature of a ridge on its dorsal surface. In the survey of our sample taxa, we found and scored three different types of urostyles. Some frogs have smooth urostyles with full-length circular cross sections essentially like a limb bone (Fig. 2A). In some of these, a small increase in dorsal thickness is observed as a single (Fig. 2: Bombina; Fig. 3F, Leiopelma) or double (Fig. 2: Myobatrachus; Fig. 3E, Ascaphus) line of thicker bone anteriorly that extends about a third of the way posteriorly on the urostyle. When present on these otherwise smooth urostyles, this feature appears to be a remnant of the neural arch components. The other two types of urostyles we scored have a distinct, elevated dorsal ridge several times thicker than the cortical thickness of the shaft of the urostyle (Fig. 2) and similar to the iliac ridges. These ridges are larger anteriorly and generally expand into a dorsal knob, which extends over the sacro-urostylic joint (Fig. 2B). We distinguished two forms of urostyle ridges by length, one with the ridge extending back only along the anterior half of the urostyle (Fig. 2: Triprion and Hemisus) and others with a fulllength urostyle ridge (Fig. 2: Lithobates). The pipids have a short ‘‘T’’ shaped ridge on the urostyle that we scored as unique (&) and did not include in the character reconstruction. The ridges on ilia have been proposed to function in providing larger muscle attachment surfaces (Přikryl et al., 2009). These would allow for larger muscles of the ilia to control the movements of the pelvis relative to the body and the limbs relative to the pelvis (Emerson and De Jongh, 1980; Duellman and Trueb, 1994; Přikryl et al., 2009). We propose that the ridges on the urostyle provide the same function of providing larger muscle attachment sites. Thus, the presence of ridges can provide insights into the movements and function of the urostyle in relation to the body and the limbs. In addition, as discussed later, the correspondence of iliac and/or urostyle ridges with anatomical features allowing (monocondyly) or limiting urostylic movements (fusion, bicondyly) can provide insights into the functional role of the urostyle during locomotion in different frog pelvic morphs. Patterns of Pelvic Trait Evolution in Frogs Evolutionary patterns of Emerson’s traits are shown in Figure 6. With the corrected states for Prosalirus and Ascaphus presented above, it becomes clear that the basal condition for frogs is expanded sacral diapophyses (Fig. 6, left) and smooth iliac shafts (Fig. 6, right). This is not only the basal condition but the general condition across frogs based on high probabilities of expanded diapophyses at the base of the Neobatra- THE EVOLUTION OF LOCOMOTION IN FROGS 159 Fig. 6. Patterns of evolution in Emerson’s osteological pelvic traits. Maximum-likelihood ancestral character state reconstructions for the evolution of sacral expansion (left) and the presence of iliac ridges (right) are mapped on the phylogeny for our sample taxa-based topology of Gao and Wang, 2001; Frost et al., 2006; Van der Meijden et al., 2007; and Hedges et al., 2008. Character states scored for the terminal sample taxa are illustrated on the figure and in Figures 1–3. Pie diagram shading indicates the maximum likelihood of alternative character states at each ancestral node on the phylogeny. Probabilities of basal states are indicated for key nodes showing the high likelihood of the retention of expanded sacra and smooth ilia well into the Neobatrachia. The Triadobatrachus illustration is modified from Rage and Roček, 1986. chia (93%) and the bases of the Hyloides (90%) and Ranoides (82%). Flat-sided diapophyses appear independently in six terminal taxa, all involving unique transitions from expanded sacral diapophyses (Pipidae, Bombinatoridae, Megophryidae, Hylidae, and two microhylids). However, only in the Pipidae, they are completely bony as opposed to having flat-sided cartilaginous caps. Rod-like diapophyses appear independently in four different terminal taxa. Judging from the high probabilities for expanded diapophyses at preceding ancestral nodes (90, 88, and 82% for Hyloides, Arthroleptidae, and Natatanura, respectively), each case exhibits a saltatorial shift to rod-like sacral diapophyses. Smooth iliac shafts are also basal and general for the Anura. Iliac ridges appear in the Megophryidae and in several taxa within the Neobatrachia (Eleutherodactylidae, some Leptodactyliformes, some Afrobatrachia, and Natatanura, Fig. Journal of Morphology 160 S.M. REILLY AND M.E. JORGENSEN 6, right). Within the Neobatrachia, iliac ridges appear with rod-like sacral diapophyses in the Eleutherodactylidae, Leptodactylidae, and Natatanura but appear to arise earlier than rod-like sacra in the Afrobatrachia. The Hylidae and Microhylidae retain smooth iliac shafts. Only in the Megophryidae are flat-sided sacra associated with iliac ridges. The presence of iliac ridges in some taxa with flat-sided (Megophryidae) and expanded sacra (Hemisotidae) shows that iliac ridges are not always diagnostic of the sagittalhinge morph as proposed by Emerson (1979, 1982). Evolutionary patterns of urostylic traits are presented in Figure 7. A monocondylar sacro-urostylic articulation is basal for frogs, with a 62% or better probability in basal nodes until the Neobatrachia. Among terminal basal taxa, bicondyly appears in two of the Xenoanura and in Alytidae. Basal pipids are monocondylar (Nevo, 1968), suggesting that terminal pipids shifted to fusion from monocondyly. Bicondyly appears at a high probability (96%) at the base of the Neobatrachia (it is lost to fusion in a few bufonids, breviceptids, and microhylids within Gastrophryninae and Phrynomantinae; Pugener and Maglia, 2009). Like the pipids, some of the Anomoceola (Scaphiopodidae and some pelobatids and megophryids; Pugener and Maglia, 2009) develop fused urostyles from monocondylar articulations. Smooth urostyles are found in all basal frogs except the Megophryidae. Urostyle ridges arise repeatedly within the Neobatrachia. In the Hyloides, except for the Cycloramphidae and Myobatrachidae, there appears to be a widespread transition to possession of urostylic ridges, with half-length ridges appearing in many taxa and the two sagittal-hinge taxa independently developing full ridges. In the Ranoides, there are three basic groups: 1) the Afrobatrachia are dominated by half-length ridges (with the one sagittal-hinge morph developing a full ridge), 2) the microhylids primarily retain smooth urostyles, and 3) full ridges appear in the Natatanura. DISCUSSION A summary of the four pelvic traits used in this analysis is presented on the sacral diapophysealshape character tree (Fig. 8). Sacro-urostylic articulation state is simply scored on a few branches, as bicondyly only appears in three basal terminal taxa (rhinophrynids, pipids, and alytids) and in the Neobatrachia. Primary locomotor states for each family are based on Zug’s (1978), Emerson’s (1979), and Brown and Crespo’s (2000) scoring of locomotor modes [dedicated AQuatic (Pipidae), Burrowing habits, Walking, Hopping (less than eight body lengths per leap), and Jumping (more than eight body lengths per leap) with environment (Terrestrial and Arboreal)]. The scoring of the locomotor mode of Ascaphus and Leiopelma as Journal of Morphology walker/hoppers is from Essner et al. (2010). The paleotaxa are assumed to use the basal walker/ hopper mode evident in all extant basal taxa given their widespread possession of lateral-bender pelvic traits. We have coded the jumping taxa in bold and the letter F after the taxon names indicating cases of taxa in which all (F) or some (f) members have a fused urostyle (Fig. 8). Pelvic Designs in Relation to Locomotor Mode in Basal Anurans The walker/hopper mode is both basal and general for the anura. Based on the four skeletal characters, the early design of the frog pelvic region is highly conserved in the basal Anura. Below the neobatrachians, all six major frog lineages possess expanded diapophyses, smooth ilia, and smooth urostyles. In all terrestrial basal taxa, smooth ilia are correlated with smooth urostyles (except in the megophryids). In terms of locomotion, all basal taxa use walking/hopping locomotion (except the aquatic pipids). Thus, Emerson’s lateral-bender morph (with monocondyly and smooth urostyles) and the walker/hopper locomotor mode is clearly the basal condition for the Anura. Walker/hopper-lateralbender morphs with expanded diapophyses and smooth ilia and urostyles (***: Fig. 8) are also widespread within the Neobatrachia (Myobatrachidae, Cycloramphidae, and two microhylids) although they differ in that they are bicondylar. The remaining neobatrachian walker/hoppers exhibit two other pelvic systems in which they retain expanded sacral diapophyses, but add a half urostyle ridge (Limnodynastidae, Bufonidae, and Microhylidae: Microhylinae) or a half urostyle ridge and iliac ridges (Ceratophryidae, Leiuperidae, Hemisotidae, and some hyperoliids). Are bicondyly and sacro-urostylic fusion innovations for burrowing? Patterns of sacrourostylic articulation provide a basis to hypothesize the function of the urostyle in lateral-bender morphs and the relationship of urostyle movement to burrowing. The most basal extant lineage (Leiopelmatidae) is monocondylar and is not known to burrow in the soil. However, many of the burrowing taxa exhibit a strong tendency to develop ways to limit the lateral movement of the urostyle, thus directing muscle activity to moving the ilia and limbs and not the urostyle. First, the xenoanuran, alytid, and neobatrachian burrowers have bicondylar sacro-urostylic articulations. Second, the Anomoceola (Scaphiopodidae and many pelobatids and megophryids) have fused urostyles. Finally, some bufonids and some microhylids (within Gastrophyninae and Phrynomantinae) add fusion to bicondyly. Thus, in our taxonomic sample, all of the 14 burrowing taxa have (or have some members with) limited urostylic mobility in the lateral direction. This correlation appears to be evidence THE EVOLUTION OF LOCOMOTION IN FROGS 161 Fig. 7. Patterns of evolution in two new osteological pelvic traits. Character states for the sacro-urostylic articulation (left) and urostyle ridges (right) scored for the sample taxa are illustrated on the figure and in Figures 2 and 5. Maximum likelihood ancestral character state reconstructions (shading in nodes) indicate the probability of alternative states for each ancestral node on the same phylogeny as Figure 6. Probabilities of basal states are indicated for key nodes showing shifts to bicondyly within the Xenoanura, the Alytidae, and the Neobatrachia and the high likelihood of the retention of smooth urostyles well into the Neobatrachia. The Triadobatrachus illustration is modified from Rage and Roček, 1986. that urostylic fusion or bicondyly (or both) may provide an extended posterior axial anchor from which the ilia can move relative to the body with digging limb strokes. In addition, as mentioned above, megophryids and many of the neobatrachian burrowers have added ridges on the ilia and/ or urostyle, which would provide additional surface area for muscles to produce larger and more directly aligned lateral-bending movements in digging. We posit that laterally rigid urostyles may facilitate independent and stronger lateral movements of the limbs relative to the axial column in digging. From the wide ranging correspondence of burrowing and traits that would limit lateral movement of the urostyle, we propose that bicondyly may have first appeared as an adaptation to burrowing/lateral-bending (and walking) locomotion. Thus, bicondyly may have been an exaptation to sagittal-hinge jumping which does not appear until well into the Neobatrachia. Journal of Morphology 162 S.M. REILLY AND M.E. JORGENSEN Fig. 8. Evolution of pelvic girdle designs in frogs in relation to locomotor modes. Patterns of sacral and pelvic traits (from Figs. 6 and 7) are summarized in relation to primary locomotor modes for each family from Emerson, 1979 and Zug, 1978. Bicondyly is indicated with black bars. Locomotor modes are: AQ, dedicated aquatic (Pipidae); B, burrowing habits, W, walking, H, hopping (less than eight body lengths per leap). Jumping forms (more than eight body lengths per leap) are divided by environment into terrestrial jumpers (JT) and arboreal jumpers (J A). Note the generality of the basal walker/hopper traits across the phylogeny, the convergence in traits in terrestrial jumpers (JT), and divergence in arboreal jumping systems (J A) in the Neobatrachia. Radiation in Flat-Sided Morphs in the Anura: Emerson’s Type 1 Sacrum Is Not One Type There are several independent transitions to flat-sided diapophyses (Emerson’s ‘‘Type I’’ sacra) among frogs. The pipids are the most derived group of frogs being dedicated aquatic swimmers Journal of Morphology and the only group with massive, flat, bony-sided diapophyses fusing with a urostyle. Pipids have many other osteological (e.g., unique lateral iliac flanges: & in Figs. 6 and 8) and myological (e.g., shoulder muscles extending to the thighs) novelties associated with their fully aquatic habits and THE EVOLUTION OF LOCOMOTION IN FROGS extreme fore-aft sliding pelvic system (Whiting, 1961; Van Dijk, 2002). The remaining flat-sided morphs are terrestrial and have rounded bony diapophyses capped with flat-sided cartilages (Fig. 1A). If not carefully evaluated, the flat shape of these cartilaginous caps would be scored as ‘‘expanded’’ and thus lateralbender morphs. Among the taxa with cartilaginous flat-sided sacra, there appear to be three basic pelvic designs. First, three widely separated taxa (Bombinatoridae and two microhylid lineages) have cartilaginous flat-sided sacra but are basal in that they lack iliac or urostylic ridges. Essentially, these appear to be lateral-bender morphs with their cartilaginous caps formed into flat sides. Given their otherwise basic similarity to lateralbender morphs, it is not surprising that they are walker/hoppers. Second, perhaps the most unique pelvic design is among the basal frogs in the Megophryidae. From among its externally and ecologically similar lateral-bender-morph sister taxa in the Anomoceola, the megophryids have shifted to flat-sided sacra (similar to bombinatorids) with half urostylic ridges and iliac ridges. Some are burrowers (and have fused urostyles; Pugener and Maglia, 2009) and the family is unique among frogs in having intervertebral disks with ossified centers (Frost et al., 2006). This family, like many other taxa, uses forest floor and streamside habitats. However, it contains twice as many species (140) as all of the rest of the extant lineages below the Neobatrachia combined (70). It remains to be seen why this speciose walker/hopper family is so derived in pelvic and axial design and how its morphological novelties relate to its great success in Southeast Asia. Finally, the last flat-sided form is the family Hylidae, which are climber/arboreal jumpers and have added urostylic ridges to the system (discussed below with other arboreal forms). The Radiation of Jumping Locomotion in the Neobatrachia Figure 8 shows that a majority (14/25) of our neobatrachian sample taxa represent a range of different kinds of walker/hoppers with the basal lateral-bender pelvis, some having a half urostyle ridge. In terms of jumping frogs, there are clear patterns of convergence in terrestrial jumpers and divergence in pelvic traits in arboreal jumpers, which appear well within the Neobatrachia. Pelvic convergence in terrestrial jumpers. One of the clearest patterns in this study is the strong correspondence of one set of skeletal pelvic traits with terrestrial jumping. All of the terrestrial jumpers have rod-like diapophyses and ridges on the ilia and urostyle (Fig. 8, lll and JT). This pattern independently appears twice in the Hyloides (eleutherodactylids and leptodactylids) 163 and twice in the Ranoides (arthroleptids and Natatanura). All of these taxa (some previously included together in the same taxa) resemble and appear to jump like the ranid frogs most often studied as the model sagittal-hinge jumping frog. However, based on the patterns of trait change and large shifts in probabilities on the individual trait trees (Figs. 6 and 7), it is clear that each of these distantly related taxa have independently converged on the sagittal-hinge jumping system. Although this suite of characters describes the sagittal-hinge morph and provides strong support for the relationship between the sagittal-hinge morph and jumping, it remains to be seen how similar the four independently evolved terrestrial jumping taxa are in the details of myology, motor pattern, kinematics, and function. It seems probable that tightly bound iliosacral joints and large dorsal ridges (and muscles) on pelvic elements are related to the increase in jumping ability in these frogs. However, relative limb and body size appear to play a role as well, as most of these taxa are large frogs with relatively larger hindlimb muscles (Emerson, 1978; Marsh, 1994; Choi and Park, 1996; Choi et al., 2003). Functionally, the relative contributions to jumping of potential sagittal pelvic movements versus limb extension per se have not been quantified in frogs. Nonetheless, we propose that the sagittal-hinge system may be an adaptation to jumping in large terrestrial frogs. Because mass increases as a cube, heavier frogs may need the sagittal-hinge to raise the heavier front end during take off and they may need a good pelvic hinge to help fold up their longer hindlimbs for landing. However, comparisons of the launch, aerial, and landing kinematic timing patterns in Ascaphus, Bombina, and Lithobates showed no difference in aerial duration despite the fact that Lithobates jumped significantly farther (Essner et al., 2010). Thus, the larger sagittalhinge frog must rotate its longer/heavier limb segments into landing position in the same amount of time as the small frogs. Thus, we propose an alternative hypothesis that sagittal-hinge may actually function in facilitating limb placement during landing in relatively larger frogs. One other family (Dendrobatidae) has members converging on terrestrial jumping. The family Dendrobatidae as we have sampled is based on Emerson’s sample of the subfamily Dendrobatinae, which have the basal pelvis (Fig. 8) and are walker/hoppers (Zug, 1978). However, Zug (1978) showed that the more basal dendrobatids (Colostethinae) are good jumpers, and our X-ray data (not included in the present analysis) show them to have rod-like sacra, iliac ridges, and a half urostyle ridge similar to terrestrial jumpers. Thus, the basal dendrobatids (Colostethinae) appear to be a fifth group to independently develop terrestrial jumping traits and habits. Accordingly, the more Journal of Morphology 164 S.M. REILLY AND M.E. JORGENSEN derived poison-dart frogs (Dendrobatinae) have reverted to basal walker/hoppers (in design and performance) in concert with small body size, visually smaller hindlimbs, toxicity, and aposematic coloration, all of which appear to relate to the loss of the need for long-distance predatory escape. Systematically, the Dendrobatidae is one of the less resolved branches, having been historically aligned with the Bufonidae, Ranidae, Leptodactylidae, and Arthroleptidae (Grant et al., 2006). The presence of terrestrial jumping basally does not support a close relationship of the Dendrobatidae with the lateral-bender-morph Bufonidae. Pelvic divergence in arboreal jumpers: Four kinds of tree frogs. In contrast to the convergences seen in terrestrial jumpers, several pelvic types have moved into the trees. Three arboreal jumpers have arisen within the Ranoides, two with expanded sacral diapophyses and one with rod-like diapophyses. First, among the lateralbender morphs of the Microhylidae, members of the subfamily Asterophryninae have become arboreal with the basal pelvic system traits (expanded sacra, no iliac or urostyle ridges) except they are bicondylar. This large radiation of small- to medium-sized Southeast Asian frogs has added intercalary cartilages and toe pads to the lateral-bender morph and moved into the arboreal niche. Second, the Hyperoliidae has a few walker/hopper taxa (e.g., Kassina) but is dominated by small- to medium-sized reed and tree frogs. This family has added iliac and half urostyle ridges to the expanded sacrum. The third type of tree frog has arisen from among the sagittal-hinge ancestry in the Natatanura. The Rhacophoridae have developed intercalary cartilages and toe pads for climbing and appear to have taken the sagittal-hinge system into the trees. The same pattern also appears in another Natatanuran family not in our study (Mantellidae) and in some eleutherodactylid taxa, although they lack intercalary cartilages. The fourth and dominant types of tree frogs are in the Hylidae. This family, with toe pads and intercalary cartilages in the digits, is the most speciose frog family comprising nearly a sixth of all known frogs. The hylids have the cartilaginous flat-sided sacral diapophyses, and no iliac ridges, but possess half ridges on the urostyle. As posited above, the flat-sided cartilages may be a simple adaptation of the basal lateral-bender morph because the bony diapophyses of hylids appear the same as lateral-bender morphs and one has to look carefully for the flat-sided cartilaginous caps to score them as a flat-sided morph. Interestingly, one small group of hylids (Pseudacris and sister taxa) has transitioned back to rounded cartilaginous diapophyseal caps and more terrestrial habits, but they retain the urostyle ridge and jumping ability. The discovery of four basic arboreal pelvic designs has some general implications in relation Journal of Morphology to tree frog evolution. Although the sagittal-hinge is tightly correlated to terrestrial jumping, the system works well in the trees as evidenced by the huge radiation of rhacophorid tree frogs (300 sps). However, more basal pelvic designs are found in the other three tree frog radiations that comprise a quarter of all frogs. The lateral-bender morph (Microhylidae: Asterophryninae and Hyperoliidae) and flat-sided (Hylidae) tree frogs, in general, are smaller and have relatively longer, more gracile hindlimbs than the sagittal-hinge frogs (Zug, 1972; Emerson, 1978) and can attain greater jumping distances (Zug, 1978). In fact, the best jumpers measured to date are from the Hylidae (Litoria: Pelodryadinae and Pseudacris and Acris: Hylinae) in which eight species have been shown to average from 20 to 26 body lengths per jump compared to the high of 19.6 body lengths (Platymantis papuensis) measured in the best of 14 Ranoid jumpers (Zug, 1978). Thus, expanded or flat-sided sacral diapophyses (5the absence of a sagittal-hinge system) can serve as a successful platform for long-distance jumping. The most widely accepted explanation for differences in jumping performance in frogs is the length of the hindlimb (Zug, 1972, 1978; Dobrowolska, 1973; Emerson, 1978, 1985, 1988; Choi and Park, 1996; Choi et al., 2003; Gomes et al., 2009), which reflects the time and distance through which jumping forces act (Gray, 1968). However, like the contribution of sagittal-hinge extension versus limb extension in terrestrial jumpers, the contribution of limb extension and fore-aft sliding to arboreal jumping has not been demonstrated. Because most tree frogs are able to jump far without a sagittal-hinge and iliac ridges, adaptations to limb extension performance alone may be the primary driving locomotor mechanism in most tree frogs. The contrast in the presence and absence of iliac ridges in the terrestrial (sagittal-hinge) versus arboreal (lateral-bender and flat-sided morphs) tree frogs, respectively, suggests that there are fundamental differences in the amount of muscle mass and perhaps primary directions of action in controlling the relative movements of ilia, urostyle, and axial column in these different tree frog morphs. In addition, the relative amount of hindlimb musculature, muscle contractile properties, and stored energy mechanisms may differ among the sagittal-hinge and tree frog morphs (Emerson, 1978; Marsh, 1994; Choi and Park, 1996; Choi et al., 2003; Roberts and Marsh, 2003). The arboreal hyperoliids illuminate another interesting twist: they have the same expanded pelvic system that has independently appeared in three other walker/hopper families (Hemisotidae, Leiuperidae, and Ceratophryidae) with ridges added to the urostyle and ilia. This suggests that something is functionally different about walker/hoppers that have added ridges and the hyperoliids appear to THE EVOLUTION OF LOCOMOTION IN FROGS have adapted this system to arboreal jumping without compromising walking and hopping. The rhacophorids and hyperoliids of tropical Africa are widely accepted to be the ecological analogs to the tropical American hylids. However, the pelvic systems show these to be three very different kinds of tree frogs. Similarly, the rhacophorids and hyperoliids, which comprise a majority of the African tree frog taxa, are fundamentally different in pelvic design, which may be related to differences in more savanna versus forest arboreality in these taxa. In terms of systematics, in the past the hyperoliids have been considered to be closely related to the rhacophorids (Frost et al., 2006), but their major pelvic differences add to support for their separation within the Ranoides. Finally, the Rhacophoridae and Hylidae contain the two primary radiations of gliding tree frogs (Duellman, 1970; Emerson and Koehl, 1990). Studies of gliding in frogs have shown that gliders from these families are convergent in body morphology and behavior related to aerodynamics during flight (Emerson and Koehl, 1990; Emerson et al., 1990; McCay, 2001). However, these studies have not considered the influence of jumping per se (i.e., take-off velocity) in attaining gliding dynamics when jumping from trees or that these gliding families have converged on gliding locomotion from divergent jumping systems. The hylids appear to have adapted a more basal pelvic system to arboreal jumping with long legs and gracile bodies, whereas the rhacophorids appear to have adapted the terrestrial sagittal-hinge jumping system for use in arboreal jumping. Here again, understanding the relative contribution of limb extension and pelvic movements to jumping in these two forms is necessary to evaluate the influence of pelvic (if any) and limb design on launching performance. Conclusions and Insights From Our New Hypothesis for the Evolution of Frog Locomotion We propose that the patterns revealed by the pelvic traits and the general locomotor patterns presented in Figure 8 represent a new starting point for the understanding of the evolution of locomotion in frogs. In resampling Emerson’s taxa, we have covered only 24 of 49 extant anuran families but these families represent three-fourths of the 6,0001 frog species and all of the major radiations. We point out that assigning family-level states to both the morphological and locomotor traits on the basis of a few genera has its limits. However, we have better accounted for intraspecific variation with better intraspecific sample sizes and improved visualization of the anatomy, and there is evidence of substantial family-level consistency in these traits (we examined familial 165 patterns in more taxa than reported here). The addition of information on more taxa (or use of a different phylogeny) will not significantly change the major pelvic patterns that have emerged in this study, they will only add to them. We feel confident that as the first formal phylogenetic mapping of these characters in frogs, important new patterns and insights are revealed and several general observations can be made. Although frogs can be categorized into flat-sided, round-sided, and rod-like sacral shapes, the most general observation is that there is a wider radiation in the details of overall pelvic design in anurans indicating at least seven pelvic types (Fig. 8) rather than the original three morphs proposed by Emerson. The dedicated aquatic frogs, which do not appear to use lateral bending, are the only frogs identified so far that have ossified flat-sided diapophyses. The walker/hopper condition is both basal and generally conserved across the phylogeny well into the Neobatrachia. Burrowing appears to have had a canalizing effect on caudal rigidity via bicondyly and/or fusion at the sacrourostylic joint. Furthermore, because bicondyly evolved well before either arboreal or terrestrial jumping, it does not appear to have evolved for jumping but to provide axial rigidity to lateralbending behaviors. Long-distance jumping and jumping traits do not appear until well within the Neobatrachia. Thus, the ‘‘sagittal-hinge jumping is primitive’’ hypothesis of Shubin and Jenkins (1995) lacks support. Instead, the terrestrial jumpers have come to the same design independently in widely distant lineages suggesting a strong functional canalizing effect on the evolution of terrestrial jumping in frogs that tend to have larger body- and relative hindlimb muscle masses (Marsh, 1994; Choi and Park, 1996). Because many other frogs can jump as far or farther than sagittal-hinge frogs, one has to question the utility of the sagittal-hinge in jumping and look more to comparisons of other anatomical, ecological, and physiological jumping components. Interestingly, recent comparisons of jumping in basal and ranid frogs have shown that the kinematics of limb movements during the aerial phase and landing are significantly different, suggesting that folding the body and protracting the legs for landing may be the primary functional role of the sagittal-hinge (Essner et al., 2010). Arboreality, on the other hand, has been accomplished in four different ways: 1) with no change from the basal condition; 2) with the addition of urostylic and iliac ridges; 3) with cartilaginous flat-sided sacra; and 4) from sagittal-hinge lineages. This suggests that transitions to climbing adaptations (most obviously, smaller mass and the addition of toe pads), rather than pelvic design, may drive the move into arboreality. However, relatively longer legs appear to contribute to arborJournal of Morphology 166 S.M. REILLY AND M.E. JORGENSEN eality in both extending the reach in climbing limb movements and in increasing arboreal jumping ability (Marsh, 1994; Gomes et al., 2009). Arboreality has only arisen in bicondylar forms, suggesting that the benefits of stiffening the axial column, which we proposed for burrowing and walking, may facilitate vertical climbing as well. The Hylidae are the only group to take flat-sided sacra into the trees. Their appearance in the hylids and their maintenance in nearly 900 species suggest that cartilaginous straight-sided diapophyses are somehow significant to the success of this largest arboreal radiation. However, we need to know how the urostyle and the flat-sided sacra function in the hylid pelvis and whether they are more important for climbing up or jumping down. The generality of the basal hopping frog system, multiple convergences on the terrestrial jumping sagittal-hinge system, and pelvic divergences of tree frogs show a diversity of influences of pelvic design on frog locomotion. Still, there are other functional aspects of frog locomotion that need to be better understood. First, kinematic analyses of sacroiliac and urostyle movements during jumping and landing are needed across the array of pelvic types to begin to understand the function of the pelvis in frogs and how it differs in walker/hoppers, terrestrial jumpers, and arboreal jumpers. Second, comparative studies of the musculature of the pelvic region are needed in light of differences of sacral shape, iliac and urostyle ridges, and mono- versus bicondyly. Third, a broader study of limb design and function in relation to limb and body size is needed across frogs with different pelvic designs. Hindlimb length has been shown to scale with jumping performance in both allometric (Zug, 1972; Emerson, 1985) and phylogenetically corrected (Gomes et al., 2009) comparisons in frogs. Other important features of the limb (takeoff velocity and hindlimb muscle mass, architecture, contractile properties, and enzyme activity) have also been shown to vary among small taxonomic samples of hopping frogs, tree frogs, and ranids (Calow and Alexander, 1973; Marsh, 1994; Choi and Park, 1996; Lutz and Rome; 1996; Olson and Marsh, 1998; Gillis and Biewener, 2000; Choi et al., 2003; Roberts and Marsh, 2003). However, these studies have not consistently related limb features to relative hindlimb muscle mass and body mass (Gomes et al., 2009). We need analyses that will help us understand the differences in the legs of a toad, a bull frog, and different tree frogs in relation to body mass, launch velocity, and jump distance. Finally, we made two observations in relation to some of Emerson’s traits. First, we could not find consistent evidence of ligament scars for the dorsal iliosacral ligaments of frogs. There is usually a smooth surface at the insertion site of the dorsal iliosacral ligaments. Second, the ilio-ilial ligament Journal of Morphology she used as characteristic of the fore-aft-slider morph is present in all frogs we have examined, albeit in a very thin state. Emerson posited that this ligament was a thickening of the lumbodorsal fascia forming a cuff-like band that binds the ilia to the sacrum (Emerson and De Jongh, 1980). This fascial cuff appears to be present in all frogs. Thus, the loss of the dorsal iliosacral ligament, rather than the presence of the ilio-ilial ligament, may be the definitive ligamentous trait for the fore-aft slider morph. In sum, the early frogs did ‘‘leap forward, high over it.’’ However, our analysis suggests that the initial leap the frog bauplan afforded was a performance increase to hopping locomotion. This may have been a significant increase ‘‘high over’’ the leaping distances of generalized ancestral and other tetrapods but it was not a leap to sagittalhinge jumping. The generalized frog is a walker that can hop better than most vertebrates with some proclivity toward burrowing. Later, from a myriad of neobatrachian walker/hopper frogs, many lineages adopted striking patterns of convergence to increase jumping distance on land and others followed divergent patterns to climb and jump in the trees. ACKNOWLEDGMENTS The authors thank Sarah Gutzwiler, Brian Folt, and Ashley Micelli for assistance with radiographic data collection and management. Joe Eastman provided radiographic, film development resources and advice. Jeff Brown and the OUCOM Communication Department provided digital scanning facilities for radiographs. They are grateful to the following museums and collections staff for X-ray assistance and loan material (Appendix): Linda Trueb, Rafe Brown, and Andrew Campbell in the Department of Herpetology at the University of Kansas; Alan Resetar at the Field Museum; Jim McGuire and Carol Spencer at the Museum of Vertebrate Zoology; Robert Drewes and Jens Vindum at the California Academy of Sciences; and Jonathan Losos and José Rosado at the Museum of Comparative Zoology. Larry Witmer and Ryan Ridgely of the OU micro-CT facility provided assistance with CT scanning. Finally, they thank Rick Essner for many valuable discussions and insights into basal and crown frogs and how they differ in locomotion. LITERATURE CITED Abourachid A, Green DM. 1999. Origins of the frog-kick? Alternate-leg swimming in primitive frogs, families Leiopelmatidae and Ascaphidae. J Herpetol 33:657–663. Báez AM, Basso NG. 1996. The earliest known frogs of the Jurassic of South America: Review and cladistic appraisal of their relationship. Münchner Geowiss Abh A 30:131–158. THE EVOLUTION OF LOCOMOTION IN FROGS Báez AM, Nicoli L. 2004. A new look at an old frog: The Jurassic Notobatrachus Rieg from Patagonia. Ameghiniana 41:257– 270. Brown LE, Crespo EG. 2000. The burrowing behavior of the midwife toads Alytes cisternasii and Alytes obstetricans (Anura, Discoglossidae). Alytes 17:101–113. Calow LJ, Alexander RMcn. 1973. A mechanical analysis of the hind leg of a frog (Rana temporaria). J Zool Lond 171:293– 321. Choi I, Park K. 1996. Variations in take-off velocity of anuran amphibians: Relation to morphology, muscle contractile function and enzyme activity. Comp Biochem Physiol A 113:393– 400. Choi I, Shim JH, Ricklefs RE. 2003. Morphometric relationships of take-off speed in anuran amphibians. J Exp Biol A 299:99– 102. Dobrowolska H. 1973. Body part proportions in relation to mode of locomotion in anurans. Zool Polon 23:59–108. Duellman WE. 1970. The hylid frogs of Middle America. Monogr Mus Nat Hist Univ Kansas 1:1–753. Duellman WE, Trueb L. 1994. The Biology of Amphibians. Baltimore, MD: Johns Hopkins University Press. Emerson SB. 1978. Allometry and jumping in frogs: Helping the twain to meet. Evolution 32:551–564. Emerson SB. 1979. The iliosacral articulation in frogs: Form and function. Biol J Linn Soc 11:153–168. Emerson SB. 1982. Frog postcranial morphology: Identification of a functional complex. Copeia 3:603–613. Emerson SB. 1985. Jumping and leaping. In: Hildebrand M, Bramble DM, Leim KF, Wake DB, editors. Functional Vertebrate Anatomy. Cambridge, MA: Belknap Press. pp 58–72. Emerson SB. 1988. Convergence and morphological constraint in frogs: Variation in postcranial morphology. Fieldiana 43:1– 19. Emerson SB, De Jongh HJ. 1980. Muscle activity at the iliosacral articulation in frogs. J Morphol 166:129–144. Emerson SB, Koehl MAR. 1990. The interaction of behavioral and morphological change in the evolution of a novel locomotor type: ‘‘Flying’’ frogs. Evolution 44:1931–1946. Emerson SB, Travis J, Koehl MAR. 1990. Functional complexes, and additivity in performance: A test case with ‘flying’ frogs. Evolution 44:2153–2157. Essner RL, Suffian DJ, Bishop PJ, Reilly SM. 2010. Landing in basal frogs: Evidence of saltational patterns in the evolution of anuran locomotion. Naturwissenschaften DOI: 10.1007/ s00114-010-0697-4. Estes R, Reig OA. 1973. The early fossil record of frogs: A review of the evidence. In: Vial JL, editor. Evolutionary Biology of the Anurans: Contemporary Research on Major Problems. Columbia, MO: University of Missouri Press. pp 11–63. Evans SE, Borsuk-Bialynicka M. 1998. A stem-group frog from the early Triassic of Poland. Acto Palaento Pol 43:573–580. Frost DR, Grant T, Faivovich J, Bain RH, Haas A, Haddad CFB, de Sa’ RO, Channing A, Wilkinson M, Donnellan SC, Raxworthy CJ, Campbell JA, Blotto BL, Moler P, Drewes RC, Nussbaum RA, Lynch JD, Green DM, Wheeler WC. 2006. The amphibian tree of life. Bull Am Mus Nat Hist 297:1–370. Gao K-Q, Wang Y. 2001. Mesozoic anurans from Lianoning Province, China, and phylogenetic relationships of archeobatrachian anuran clades. J Vert Paleontol 21:460–473. Gao K-Q, Chen S. 2004. A new frog (Amphibia: Anura) from the lower Cretaceous of western Liaoning, China. Cretac Res 25:761–769. Gaupp E. 1896. Anatomie des Frosches. I. Lehre vom Skelet und vom Muskelsystem. Braunschweig: Friedrich Vieweg und Sohn. Gillis GB, Biewener AA. 2000. Hindlimb extensor muscle function during jumping and swimming in the toad (Bufo marinus). J Exp Biol 203:3547–3563. Gomes FR, Rezende EL, Grizante MB, Navas CA. 2009. The evolution of jumping performance in anurans: Morphological correlates and ecological implications. J Evol Biol 22:1088– 1097. 167 Grant T, Frost DR, Caldwell JP, Gagliardo R, Haddad CFB, Kok PJR, Means BD, Noonan BP, Schargel W, Wheeler WC. 2006. Phylogenetic systematics of dart-poison frogs and their relatives (Anura: Athesphatanura: Dendrobatidae). Bull Am Mus Nat Hist 299:1–262. Gray J. 1968. Animal Locomotion. London: William Cloes and Sons, Ltd. 479 p. Green TL. 1931. On the pelvis of the Anura: A study in adaptation and recapitulation. Proc Zool Soc Lond 1931:1259–1291. Hedges SB, Duellman WE, Heinicke MP. 2008. New World direct-developing frogs (Anura: Terrarana): Molecular phylogeny, classification, biogeography, and conservation. Zootaxa 1737:1–182. Jenkins FA, Shubin NH. 1998. Prosalirus bitis and the anuran caudopelvic mechanism. J Vert Paleontol 18:495–510. Kargo WJ, Nelson F, Rome C. 2002. Jumping in frogs: Assessing the design of the skeletal system by an anatomically realistic modeling and forward dynamic simulation. J Exp Biol 205:1683–1702. Lutz GJ, Rome LC. 1996. Muscle function during jumping in frogs. I. Sarcomere length, change, EMG pattern, and jumping performance. Am J Physiol 271:C563–C570. Maddison WP, Maddison DR. 2009. Mesquite: A Modular System for Evolutionary Analysis, Version 2.72. Available at:http://mesquiteproject.org. Marsh RL. 1994. Jumping ability of anuran amphibians. Adv Vet Sci Comp Med 38:51–111. McCay MG. 2001. Aerodynamic stability and maneuverability of the gliding frog Polypedates dennysi. J Exp Biol 204:2817– 2826. Nauwelaerts S, Aerts P. 2002. Two distinct gait types in swimming frogs. J Zool Lond 258:183–188. Nevo E. 1968. Pipid frogs from the early Cretaceous of Israel and pipid evolution. Bull Mus Comp Zool Harvard 136:255– 318. Olson JM, Marsh RL. 1998. Activation patterns and length changes in hindlimb muscles of the bullfrog Rana catesbiana during jumping. J Exp Biol 201:2763–2777. Přikryl T, Aerts P, Havelková P, Herrel A, Roček Z. 2009. Pelvic and thigh musculature in frogs (Anura) and origin of anuran jumping locomotion. J Anat 214:100–139. Pugener LA. 2002. The vertebral column and spinal nerves of Anurans [Dissertation]. Lawrence, KS: University of Kansas.480p. Pugener LA, Maglia AM. 2009. Developmental evolution of the anuran sacro-urostylic complex. SA J Herpetol 4:193–209. Rage JC, Roček Z. 1986. Triadobatrachus revisited. In: Roček Z, editor. Studies in Herpetology. Prague: Charles University. pp 255–258. Rage JC, Roček Z. 1989. Redescription of Triadobatrachus massinoti (Piveteau, 1936) an anuran amphibian from the early Triassic. Palaeontographica A 206:1–16. Ritland RM. 1955. Studies on the postcranial morphology of Ascaphus truei. II. Myology. J Morphol 97:215–282. Roberts TJ, Marsh RL. 2003. Probing the limits of muscle powered accelerations: Lessons from jumping bullfrogs. J Exp Biol 206:2567–2580. Roček Z. 2008. The late Cretaceous frog Gobiates from Central Asia: Its evolutionary status and possible phylogenetic relationships. Cretac Res 29:577–591. Roček Z, Nessov LA. 1993. Cretaceous anurans from Central Asia. Palaeontographica A 226:1–54. Roelants K, Bossuyt F. 2005. Archeobatrachian paraphyly and pangean diversification of crown group frogs. Syst Biol 54:111–126. Sanchiz FB. 1986. A reconstruction of the Anuran locomotor synthetotype. In: Roček Z, editor. Studies in Herpetology. Prague: Charles University. pp 111–115. Schluter D, Price T, Mooers AO, Ludwig D. 1997. Likelihood of ancestor states in adaptive radiation. Evolution 51:1699– 1711. Shubin NH, Jenkins FA. 1995. An early Jurassic jumping frog. Nature 377:49–52. Journal of Morphology 168 S.M. REILLY AND M.E. JORGENSEN Stephenson EW. 1952. The vertebral column and appendicular skeleton of Leiopelma hochstetteri Fitzinger. Trans R Soc New Zeal 79:601–613. Twain M. 1867. The Celebrated Jumping Frog of Calaveras County. New York: C.H. Webb. 198 p. Van der Meijden A, Vences M, Hoegg S, Biostel R, Channing A, Meyer A. 2007. Nuclear phylogeny of narrow-mouthed toads (Family Microhylidae) and a discussion of competing hypotheses concerning their biogeographical origins. Mol Phylogenet Evol 44:1017–1030. Van Dijk DE. 2002. Longitudinal sliding articulations in pipid frogs. S Afr J Sci 98:555–556. Whiting HP. 1961. Pelvic girdle in amphibian locomotion. Symp Zool Soc Lond 5:43–57. Zug GR. 1972. Anuran locomotion: Structure and function. I. Preliminary observations on relation between jumping and osteometrics of appendicular and post-axial skeleton. Copeia 4:613–624. Zug GR. 1978. Anuran Locomotion—Structure and function. II. Jumping performance of semiaquatic, terrestrial, and arboreal frogs. Smiths Contr Zool 276:1–31. APPENDIX Material examined. Specimens are from the following collections: CAS, California Academy of Sciences; FMNH, Field Museum of Natural History; KU, Natural History Museum, University of Kansas; MCZ, Museum of Comparative Zoology, Harvard University; MEJ, Michael E. Jorgensen personal collection; MVZ, Museum of Vertebrate Zoology, University of California, Berkeley. Asterisks indicate micro-CT specimens. Alytidae: Alytes obstetricans, MVZ148850, 148851, 148867, 148874, 148875; Discoglossus pictus, MVZ162442, 162446, 162449, 162451, 162456. Arthroleptidae: Cardioglossa leucomystax, MCZ46624, 46627, 46629, 46631, 46623. Bombinatoridae: Bombina maxima, FMNH232977, 49522, 49520, 49519, 232978; Bombina orientalis, MEJ*. Bufonidae: Rhinella marinus, MEJ; MCZ135098, 135099, 135097, 110976, 110979; Anaxyrus americanus, MEJ*; Rhaebo haematiticus, KU32601, 32602, 32604, 32606, 42688, 102130; Epidalea calamita, CAS27311, 87109, 98080, CAS-SUA21701; Atelopus spumarius, KU211676*, 211677, 211680, 211682, 212530; Melanophryniscus stelzneri, KU93181, 179536, 197288, 197289, 197290; Pedostibes hosii, FMNH148055, 148058*, 148057, 148050, 148053. Ceratophryidae (Telmatobiinae): Telmatobius brevipes, KU212436, 212440, 212415, 212445, 212439. Cycloramphidae: Rhinoderma darwinii, KU161544, 161546, 161540, 161538, 161539, 161533*. Dendrobatidae: Dendrobates auratus, KU33091, 36052, 36053, 36057, 36350; Oophaga pumilio, MEJ*; KU33038, 33042, 33043, 33057, 65295. Eleutherodactilidae: Eleutherodactylus coqui, KU180536, 180503, 180635, 180517, 180525, 180533. Hemisotidae: Hemisus marmoratus, CAS160298, 160299, 160300, 160301, 160336; MEJ*. Hylidae (Hylinae): Hyla cinerea, MEJ; Hyla versicolor, MEJ; Smilisca baudinii, KU156962, 156961, 296164, 296160, 296165, 78466; Journal of Morphology Trachycephalus jordani, KU146598, 202747, 164482, 132461, 146591, 164489; Triprion petasatus, KU296254, 71448, 71446, 296252*, 296292, 296289; Pseudacris triseriata, KU39267, 39260, 224624, 39291, 39302, 224625. Hylidae (Phyllomedusinae): Agalychnis callidryas, KU86497, 86480, 86481, 86488, 86499, 86502; Pachymedusa dachnicolor, KU57921, 57923, 57924, 57927, 57920, MEJ*; Phyllomedusa sauvagii, MEJ*. Hylidae (Pelodryadinae): Litoria nasuta, KU136338, 133653, 133655, 133656, 133657; CAS78621*, 78616, 78594. Hyperoliidae: Kassina senegalensis, CAS146397, 146401, 151196, 151197, 151198; Hyperolius lateralis, CAS180144, 180145, 180146, 180147, 180149. Leiopelmatidae: Ascaphus montanus, CAS175151, 175159*, 175153; Ascaphus truei, MVZ190814*, 190815, 190816, 190817, 190818; CAS187926, 179037, 187927; Leiopelma hochstetteri, CAS6709, 6707; FMNH195489*, 58912, 58210, 58212. Leiuperidae: Engystomops pustulosus, MCZ9253, 9252, 9263, 9256, 1401. Leptodactylidae: Leptodactylus pentadactylus, KU100354, 25714, 65707, 30407, 65709, 33165; Leptodactylus gracilis, MEJ*; Paratelmatobius lutzii, KU92977, 92978, 92980*; MCZ64345. Limnodynastidae: Notaden bennetti, FMNH97662, 97661, 97660, 97657, 97659. Megophryidae: Megophrys nasuta, FMNH156560, 156557, 156556, 148300, 148299; Ophryophryne microstoma, MVZ223701*, 223706, 223709, 223715, 223719. Microhylidae (Asterophryninae): Cophixalus cheesmanae, MCZ81607, 98838, 98839, 98848, 98888. Microhylidae (Microhylinae): Microhyla berdmorei, MCZ132422, 132478, 132479, 132480, 132481; Kaloula pulchra, KU296869, 312761, 296846, 222673, 222672; Glyphoglossus molossus, MCZ23413, 23414, 4863, 4864; KU312757*; Chaperina fusca, KU309555, 309548, 311389, 309549*, 309545. Microhylidae (Phrynomerinae): Phrynomantis bifasciatus, KU207509, 195791, 195792, 173022, 173024; MEJ*. Microhylidae (Gastrophryninae): Gastrophryne carolinensis, KU154170, 155022, 154167, 154169, 154168; Hypopachus variolosus, MCZ26390, 26394, 26532, 26533, 21312; Dasypops schirchi, KU93249, 93250*; MCZ58409, 52170, 58410. Microhylidae (Kalophryninae): Kalophrynus pleurostigma, KU301847, 309650, 301848, 309970*, 309971. Microhylidae: Myersiella subnigra, KU93262*. Myobatrachidae: Pseudophryne bibronii, MCZ87311, 84403, 83712, 84405, 84398; Myobatrachus gouldii, MCZ58816, 18364, 18363, 18362; KU125359*. Pelobatidae: Pelobates cultripes, CAS156259, 156260, 156261, 156262. Pipidae: Xenopus laevis, MEJ; Pipa pipa, MCZ56277, 56278, 57341, 85574, 85575, 85576; Pseudhymenochirus merlini, KU207823, 207820, 207826, 207825*, 207817. Ptychadenidae: Ptychadena anchietae, KU196060, 196062*, 196067, 196058, 196061; Hildebrandia ornata, CAS154657, 154658, 202702, 202703. Ranidae: Lithobates pipiens, MEJ*; Lithobates catesbeianus, MEJ; Amolops hosii, KU155672, 155667*, 155668; Meristogenys jerboa, KU155612*, 155604, 155607, 155606, 155610. Rhacophoridae: Rhacophorus viridis, CAS23754, 23755, 23765, 23769, 23772; Rhacophorus chenfui, KU311851, 311853, 311852*, 311846, 311849; Chiromantis xerampelina, KU195913, 195919, 195922*, 195917, 195915. Rhinophrynidae: Rhinophrynus dorsalis, KU86643, 86649, 86639, 86642, 86644*, 86645, 86646. Scaphiopodidae: Scaphiopus hammondii, MEJ*; MCZ44381, 44390, 44388, 44391, 44380.
© Copyright 2026 Paperzz