Physalaemus, Fitzinger, 1826

Frost, Darrel R., Grant, Taran, Faivovich, Julián, Bain, Raoul H., Haas, Alexander, De Sá, Célio F. B. Haddad Rafael O., Channing, Alan, Wilkinson, Mark, Donnellan, Stephen C., Raxworthy, Christopher J., Campbell, Jonathan A., Blotto, Boris L., Moler, Paul, Drewes, Robert C., Nussbaum, Ronald A., Lynch, John D., Green, David M. & Wheeler, And Ward C., 2006, The Amphibian Tree Of Life, Bulletin of the American Museum of Natural History 2006 (297), pp. 1-291 : 1-291

atelmatobius, Physalaemus [including En­

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE

57

inae, aggregated on the basis of plesiomor­ closely related to at least some component of phy. Within his Telmatobiinae Lynch defined Limnodynastidae ( Lechriodus ) than to other five tribes, each aggregated on a variable ba­ South American ‘‘leptodactylids’’. Another sis of synapomorphy and symplesiomorphy: leptodactylid satellite is Brachycephalidae , a Telmatobiini ( Batrachophrynus , Caudiver­ small monophyletic taxon, likely the sister bera, Telmatobius , and Telmatobufo ); Also­ taxon of Euparkerella ( Leptodactylidae : dini ( Batrachyla , Eupsophus [including Al­ Eleutherodactylinae) based on digit reduction sodes], Hylorina , and Thoropa ); Odonto­ (Izecksohn, 1988; Giaretta and Sawaya, phrynini ( Macrogenioglottus , Odontophry­ 1998). Similarly, Rhinodermatidae (Rhinodnus, and Proceratophrys ); Grypiscini erma) is a small group that is likely also a ( Crossodactylodes , Cycloramphus , and Za­ telmatobiine leptodactylid (Barrio and Rinchaenus); and Eleutherodactylini (Eleuther­ aldi de Chieri, 1971; Lavilla and Cei, 2001), odactylus, Euparkerella , Holoaden , and Is­ differing from them in having partial or comchnocnema, as well as several other genera plete larval development within the male vosubsequently placed in the synonymy of cal sac and, except for Eupsophus , in having Eleutherodactylus ), with Scythrophrys being endotrophic larvae (Formas et al., 1975; Alleft incertae sedis. Subsequently, Heyer tig and McDiarmid, 1999).

(1975) provided a preliminary clustering Laurent (1986) provided the subfamilial (based on the nonphylogenetic monothetic taxonomy we employ for discussion (his arsubset method of Sharrock and Felsenstein, rangement being the formalization of the 1975) of the nominal genera within the fam­ groupings tentatively recommended by Heyily that assumed monophyly of both the fam­ er, 1975). He recognized Ceratophryinae (in ily and the constituent genera (see Farris et the larger sense of including J.D. Lynch’s al., 1982a, for criticism of the approach) in Odontophrynini, transferred from Telmatowhich Heyer identified, but did not recognize biinae), Telmatobiinae (including calyptoceformally, five units that were recognized sub­ phallelines and excluding J.D. Lynch’s sequently (Laurent, 1986) as Ceratophryinae , Eleutherodactylini), Cycloramphinae (as Eleutherodactylinae, Cycloramphinae , Lep­ Grypiscinae, including Grypscini and Elositodactylinae, and Telmatobiinae. J.D. Lynch inae of J.D. Lynch ), Eleutherodactylinae, and (1978b) revised the genera of Telmatobiinae, Leptodactylinae .

where he recognized three tribes: Telmato­ ‘‘ CERATOPHRYINAE ’’ (6 GENERA, 41 SPE­ biini ( Alsodes , Atelognathus , Batrachophry­ CIES): Reig (1972) and Estes and Reig (1973) nus, Eupsophus , Hylorina , Insuetophrynus , suggested that the leptodactylid subfamily Limnomedusa , Somuncuria , and Telmato­ Ceratophryinae was ‘‘ancestral’’, in some bius), Calyptocephalellini ( Caudiverbera and sense, to Bufonidae , although others rejected Telmatobufo ), and Batrachylini ( Batrachyla this (e.g., J.D. Lynch , 1971, 1973). Laurent and Thoropa ). The justification for this ar­ (1986), following Heyer (1975), transferred rangement was partially based on character Macrogenioglottus , Odontophrynus , and argumentation, although plausibility of re­ Proceratophrys (J.D. Lynch’s tribe Odontosults was based on subjective notions of phrynini) into this nominal subfamily, with overall similarity and relative character im­ Ceratophrys , Chacophrys , and Lepidobatraportance. A cursory glance at figure 24 (Fai­ chus being placed in Ceratophryini. Haas vovich et al., 2005) shows that several of (2003; fig. 15) presented morphological evithese groups are nonmonophyletic. dence that Ceratophryini and Odontophry­ Burton (1998a) suggested on the basis of nini are not each other’s closest relatives (folhand muscles (although his character polarity lowing J.D. Lynch , 1971), with Odontophrywas not well supported) that the leptodactyl­ nus most closely related to Leptodactylus , id tribe Calyptocephalellini is more closely and the clade Ceratophryini (Lepidobatrarelated to the South African Heleophrynidae chus 1 Ceratophrys ) most closely related to than to other South American leptodactylids. hylids, exluding hemiphractines. Duellman, San Mauro et al. (2005; fig. 17 ) suggested on (2003) treated the two groups as subfamilies the basis of DNA sequence data that Cau­ Odontophryninae and Ceratophryinae , prediverbera ( Calyptocephalellini ) is more sumably following the results of Haas

58 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297 2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE

59

(2003), and this was followed by Dubois to be close to Eleutherodactylini on the basis (2005). Faivovich et al. (2005; fig. 24) also of overall similarity.) Grypiscines and hylofound Ceratophryinae to be polyphyletic. We dines differ in (1) the shape of the transverse sampled exemplars from all nominal cerato­ processes of the posterior presacral vertebrae, phryid genera except Macrogenioglottus , being short in hylodines and not short in grywhich is similar to Odontophrynus (J.D. piscines; (2) the shape of the facial lobe of Lynch , 1971) and karyologically similar to the maxillae (deep in grypiscines, shallow in Proceratophrys (Silva et al., 2003; Odonto­ hylodines); (3) the shape of the nasals (large phrynus not examined in that study) that we and in median contact in grypiscines, small doubt that this will be an important problem. and widely separated in hylodines); and (4) Ceratophryini does have synapomorphies, whether the nasal contacts the frontoparietal for example: (1) transverse processes of an­ (contact in grypiscines, no contact in hyloterior presacral vertebrae widely expanded; dines). We were unable to obtain samples of (2) cranial bones dermosed; and (3) teeth Crossodactylodes , Rupirana , or Zachaenus , fanglike, nonpedicellate (J.D. Lynch , 1971, but we did obtain at least one species of ev­ 1982b), although nominal Odontophrynini ery other nominal genus in the group: Crosdoes not have unambiguously synapomor­ sodactylus schmidti , Cycloramphus boraphies , and the group is united on overall sim­ ceiensis, Hylodes phyllodes , Megaelosia ilarity. All ceratophryids have free­living ex­ goeldii , Paratelmatobius sp. , Scythrophrys otrophic larvae (Altig and McDiarmid, sawayae , and Thoropa miliaris . Denser sam­ 1999). We sampled three species of Cerato­ pling of this particular taxon would have phryini ( Ceratophrys cranwelli , Chacophrys been preferable, but what we obtained will pierotti , and Lepidobatrachus laevis ) and test cycloramphine monophyly and its putathree species of Odontophrynini (Odonto­ tive relationship to Dendrobatidae and will phrynus achalensis , O. americanus , and Pro­ provide an explicit hypothesis of its internal ceratophrys avelinoi). Our sampling of Pro­ phylogenetic structure as the basis of future ceratophrys should have been denser, but this studies.

proved a practical impossibility. Duellman (2003) did not accept Laurent’s ‘‘ CYCLORAMPHINAE ’’ (10 GENERA, 79 SPE­ (1986) unification of J.D. Lynch’s Hylodinae CIES): Haas (2003) suggested that this group and Grypiscini and recognized Hylodinae may be closely related to Dendrobatidae , in ( Crossodactylus , Hylodes , and Megaelosia ) part supporting the earlier position of Noble as a different subfamily from Cycloramphi­ (1926) and J.D. Lynch (1973) that the hylo­ nae. Duellman distinguished Hylodinae and dine part of this nominal subfamily (Cros­ Cycloramphinae by T­shaped terminal phasodactylus, Hylodes , and Megaelosia ) is par­ langes in Hylodinae and knoblike terminal aphyletic with respect to Dendrobatidae . Fai­ phalanges in Cycloramphinae ; and glandular vovich et al. (2005; fig. 24) recovered Cros­ pads on the dorsal surface of the digits, absodactylus (their exemplar of this group) as sent in Hylodinae and present in Cycloramthe sister taxon of Dendrobatidae . Laurent phinae. However, neither the particulars of (1986) recognized this subfamily, thus uni­ distribution of these characters in the taxa fying J.D. Lynch’s (1971, 1973) Grypiscini nor the levels of universality of their appli­ and Elosiinae (5 Hylodinae), although the cation as evidence was discussed. Duellman evidentiary basis for uniting these was based (2003) also suggested that Hylodinae and on Heyer’s (1975) results based on monoth­ Cycloramphinae differ in chromosome numetic subsets, not parsimony. (Note that J.D. bers, with 13 pairs in Cycloramphinae and 3 Lynch , 1971, had considered his Grypiscini pairs in Hylodinae. However, Kuramoto four mitochondrial genes (12S, 16S, tRNAVal, cytochrome c) and five nuclear genes (rhodopsin, tyrosinase, RAG­1, seven in absentia, 28S) and analyzed by Direct Optimization in POY under equal cost functions. Gaps were treated as evidence.

60 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

(1990) noted that hylodines in Duellman’s that work in the near future can rectify this sense have 11–13 pairs of chromosomes, and with the recognition of major monophyletic cycloramphines in Duellman’s sense also groups from within Eleutherodactylus . What have 11–13 pairs, so Duellman’s statement is we could sample of the non­ Eleutherodactaken to be an error. tylus eleutherodactyline taxa were Barycho­ ELEUTHERODACTYLINAE (13 GENERA, 782 los ternetzi , Ischnocnema quixensis , and two SPECIES): The only suggested synapomorphy species of Phrynopus . Of Eleutherodactylus of this taxon is direct terrestrial development (sensu lato) we sampled two species of the of large eggs deposited in small clutches North American subgenus Syrrhophus (J.D. Lynch, 1971) . The universality of direct ( Eleutherodactylus marnocki of the E. mardevelopment in this group is based on ex­ nocki group of J.D. Lynch and Duellman, trapolation from the few species for which 1997, and E. nitidus of the E. nitidus group direct development has been observed; the of J.D. Lynch and Duellman, 1997); one speoccurrence of large, unpigmented eggs, and cies of the Antillean subgenus Euhyas because free­living larvae are unknown (see ( Eleutherodactylus planirostris of the E. ricautionary remarks in Thibaudeau and Altig, cordii group of J.D. Lynch and Duellman, 1999). Inasmuch as this taxon contains the 1997); two species of the South American largest vertebrate genus, Eleutherodactylus subgenus Eleutherodactylus ( E. binotatus (ca. 600 species) of which the vast majority and E. juipoca , both of the E. binotatus are not represented by genetic samples, this group of J.D. Lynch , 1978a; see also J.D. taxon will remain inadequately sampled for Lynch and Duellman, 1997); and six species some time. There has never been any com­ of the Middle American subgenus Craugasprehensive phylogenetic study of the rela­ tor 12 ( E. bufoniformis of the E. bufoniformis tionships within the group and the likelihood group of J.D. Lynch , 2000, E. alfredi of the of many (or even most) of the non­ Eleuth­ E. alfredi group of J.D. Lynch , 2000, E. auerodactylus genera being components of gusti of the E. augusti group of J.D. Lynch , Eleutherodactylus is high. Indeed, Ardila­ 2000, E. pluvicanorus of the E. fraudator Robayo (1979) suggested strongly that for group of Köhler, 2000, E. punctariolus and the taxon currently referred to as Eleuthero­ E. cf. ranoides 13 of the E. rugulosus group dactylus (sensu lato) to be rendered mono­ of J.D. Lynch , 2000) and E. rhodopis of the phyletic it would need to include Barycholos , E. rhodopis group of J.D. Lynch , 2000). (For Geobatrachus , Ischnocnema , and Phrynopus expediency, all of these are noted in ‘‘Re­ (and likely Adelophryne , Phyllonastes , Phy­ sults’’ in combination with their subgeneric zelaphryne, Holoaden and Euparkerella , and names; e.g., Eleutherodactylus (Syrrhophus) Brachycephalidae [Izecksohn, 1971; Giaretta marnockii is treated as Syrrhophus marnock­ and Sawaya, 1998; Darst and Cannatella, ii.) As noted earlier, we expect that Eleuth­ 2004; Faivovich et al., 2005] 11). Regardless, erodactylus will be found to be paraphyletic many of the nominal eleutherodactyline gen­ with respect to a number of other eleutherera represent rare and extremely difficult an­ odactyline taxa (e.g., Barycholos , Phrynopus , imals to obtain (e.g., Atopophrynus , Dischi­ and Ischnocnema ) and hope that this selecdodactylus), so our sampling of this partic­ tion will allow some illumination of this. ular taxon is clearly inadequate to address most systematic problems. We could not ob­ 12 Craugastor was recently considered to be a genus tain samples of Adelophryne , Atopophrynus , by Crawford and Smith (2005) and we follow that ar­ Dischidodactylus , Euparkerella (even though rangement, although we refer to Craugastor in this secit was suggested to be closely related to Bra­ tion and elsewhere as part of Eleutherodactylus (sensu lato) for consistency with the immediately relevant litchycephalidae), Geobatrachus , Holoaden , erature. Phyllonastes , or Phyzelaphryne . We hope 13 We report this species as Craugastor cf. ranoides , because we discovered late in this project that the vouch­ 11 Dubois (2005) noted that if Brachycephalidae is a er specimen was lost. However, the identification in the synonym of Eleutherodactylinae, as suggested by the re­ associated field notes was ‘‘ Eleutherodactylus rugulo­

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE

61

Nevertheless, we are aware that this tiny dense enough to evaluate well the likely parfraction of the species diversity of Eleuth­ aphyly of this taxon with respect to others, erodactylus is insufficient to fully resolve the such as Adenomera (Heyer, 1998) , being rephylogeny of this massive taxon and that the stricted to only two of the five nominal spevalue of the results will be in highlighting cies groups. Leptodactylines vary from havoutstanding problems and providing a basis ing endotrophic larvae, facultatively endotrofor future, more densely sampled studies. phic larvae ( Adenomera ) to having exotroph­ LEPTODACTYLINAE (12 GENERA, 159 SPE­ ic, free­living larvae ( Edalorhina, CIES ): Monophyly of this group is supported Engystomops , Eupemphix , Leptodactylus , by the possession of foam nests (except in Lithodytes , Physalaemus , Pleurodema , Pseu­ Limnomedusa [Langone, 1994] and Pseudo­ dopaludicola, Vanzolinius ; Altig and Mcpaludicola [Barrio, 1954], and in some spe­ Diarmid, 1999).

cies of Pleurodema [Duellman and Veloso ‘‘ TELMATOBIINAE ’’ (11 GENERA, 98 SPE­ M., 1977]) and the presence of a bony ster­ CIES): Telmatobiinae is a similarity grouping num (rather than the cartilaginous sternum of of mostly austral South American frogs. As other leptodactylids; J.D. Lynch , 1971). currently employed, contents of this subfam­ However, Haas (2003; fig. 15) sampled three ily stem from Laurent’s (1986) formalization species of Leptodactylinae ( Physalaemus bi­ of Heyer’s (1975) informal grouping. Telligonigerus, Leptodactylus latinasus , and matobiines are currently arranged in three Pleurodema kriegi ) for mostly larval mor­ tribes (J.D. Lynch , 1971, 1978b; Burton, phology and found the group to be para­ or 1998a): Telmatobiini ( Alsodes , Atelognathus , polyphyletic with respect to Odontophrynus , Eupsophus , Hylorina , Insuetophrynus , So­ and with Physalaemus 14 and Pleurodema muncuria , and Telmatobius ); Batrachylini forming, respectively, more exclusive out­ ( Batrachyla and Thoropa ); and Calyptocegroups of Haas’ hylodines and dendrobatids. phalellini ( Batrachophrynus , Caudiverbera, In Darst and Cannnatella’s (2004) phyloge­ and Telmatobufo ). All telmatobiines have netic analysis of mtDNA ( fig. 22), their lep­ aquatic, exotrophic larvae except Eupsophus , todactyline exemplars are monophyletic in which has endotrophic larvae (Altig and the maximum­likelihood analysis of mtDNA, McDiarmid, 1999), and Thoropa , which is but polyphyletic in the parsimony analysis. semiterrestrial (Bokermann, 1965; Wassersug In Faivovich et al.’s (2005; fig. 24) parsi­ and Heyer, 1983; Haddad and Prado, 2005). mony analysis of multiple mtDNA and Batrachylini (in J.D. Lynch’s sense of innuDNA loci, exemplars of most genera of cluding Thoropa ) is diagnosed by having ter­ Leptodactylinae obtained as monophyletic, restrial eggs and aquatic to semiterrestrial with the exception of Limnomedusa . There­ larvae and T­shaped terminal phalanges. fore, the monophyly of Leptodactylinae is an Laurent (1986) did not (apparently) accept open question. We could not obtain samples J.D. Lynch’s (1978b) transferral of Thoropa of Hydrolaetare (or the recently resurrected into Batrachylini, and retained Thoropa in Eupemphix and Engystomops ), but we sam­ Cycloramphinae following Heyer (1975).

pled at least one species of each of the other Calyptocephalellini was most recently disnominal leptodactyline genera: Adenomera cussed and diagnosed by Burton (1998a) on hylaedactyla, Edalorhina perezi , Leptodac­ the basis of hand musculature, but the chartylus fuscus , L. ocellatus , Limnomedusa ma­ acter argumentation was essentially that of croglossa, Lithodytes lineatus , Physalaemus overall similarity, not synapomorphy. Forgracilis, Pleurodema brachyops , Pseudopa­ mas and Espinoza (1975) provided karyologludicola falcipes , and Vanzolinius discodac­ ical evidence for the monophyly of Calyptylus). Our sampling of Leptodactylus is not tocephallelini (although they did not address Batrachophrynus ). Cei (1970) suggested on 14 Nascimento et al. (2005) recently partitioned Phys­ the basis of immunology that Calyptocealaemus into Physalaemus , Eupemphix , and Engysto­ phalellini is phylogenetically distant from mops on the basis of phenetic comparisons. Unfortu­ the

nately, the historical reality of these taxa will remain leptodactylids, being closer to Heleophryniarguable until a phylogenetic analysis is performed on dae than to any South American leptodactylthis group. id group. J.D. Lynch (1978b) suggested 62 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 297

following to be synapomorphies of Calyp­ the basis of mtDNA evidence that Hemitocephalellini (composed of solely Caudiv­ phractinae is polyphyletic, with Cryptobatraerbera and Telmatobufo ): (1) occipital artery chus closest to direct­developing eleutheroenclosed in a bony canal; and (2) very broad dactylines, and Gastrotheca imbedded in anpterygoid process of the premaxilla. In ad­ other group of leptodactylids. Similarly, in dition, (1) a very long cultriform process of the analysis by Faivovich et al. (2005; fig. the parasphenoid; and (2) presence of a me­ 24) of multiple mtDNA and nuDNA loci, dial process on the pars palatina of the pre­ hemiphractines do not appear as monophymaxilla are osteological characters suggested letic. They recovered one clade composed of by J.D. Lynch possibly to unite Batracho­ Gastrotheca and Flectonotus , one clade comphrynus with Caudiverbera and Telmatobu­ posed of Stefania and Cryptobatrachus , and fo. they found Hemiphractus to form a clade

We sampled representatives of two of the with the few included exemplars of Eleuthgenera of Calyptocephallelini ( Caudiverbera erodactylinae and Brachycephalidae . Further, caudiverbera and Telmatobufo venustus ). We inasmuch as the sole noncontingent synapocould not sample Batrachophrynus , which morphy of nominal Hemiphractinae , bellwas considered a calyptocephalelline by shaped larval gills, has not been surveyed Burton (1998a), and in some of the clado­ widely in direct­developing leptodactylids, grams presented by J.D. Lynch (1978b) Ba­ we consider the morphological evidence for trachyophrynus was considered to form the the monophyly of the hemiphractines to be sister taxon of his Calyptocephalellini , so its questionable. absence from our analysis is unfortunate. Faivovich et al. (2005; fig. 24) transferred

‘‘Telmatobiini’’ of J.D. Lynch (1978b) is ‘‘Hemiphractinae’’ out of a reformulated Hyexplicitly paraphyletic with respect to Batra­ lidae and into ‘‘Leptodactylidae’’ on the bachylini and as such has no diagnosis other ses that continued inclusion in Hylidae than that of the inclusive clade ‘‘Telmatobi­ would render Hylidae polyphyletic; its nomini’’ 1 Batrachylini: (1) presence of an outer inal inclusion in ‘‘Leptodactylidae’’ did no metatarsal tubercle (dubiously synapomorph­ violence to a taxon already united solely by ic), and (2) reduction of imbrication on the plesiomorphy and geography; and placing it neural arches of the vertebrae. Among spe­ incertae sedis within Hyloidea was to suggest cies of ‘‘Telmatobiini’’ we sampled Alsodes its possible placement outside of the ‘‘ lepgargola, Atelognathus patagonicus , Eupso­ todactylid’’ region of the overall tree, which phus calcaratus , Hylorina sylvatica , Telma­ it is not. ‘‘Hemiphractinae’’ is a grouping of tobius jahuira, T. cf. simonsi , and T. sp. Of South American frogs united by (1) brooding Batrachylini, we sampled Batrachyla lepto­ of eggs on the female’s back, generally withpus. On this basis we provide a weak test of in a dorsal depression or well­developed telmatobiine relationships with regard to Ba­ pouch; (2) possession in the developing lartrachylini. We were unable to sample any vae of bell­shaped gills (Noble, 1927); and member of Insuetophrynus or Somuncuria . (3) presence of a broad m. abductor brevis

‘‘ HEMIPHRACTINAE ’’ (5 GENERA, 84 SPE­ plantae hallucis (Burton, 2004). Larvae may CIES): Mendelson et al. (2000) provided a be exotrophic and endotrophic among specladogram of Hemiphractinae but assumed cies of Gastrotheca and Flectonotus , and enits monophyly and its hylid affinities, as had dotrophic alone in Cryptobatrachus, Hemiall authors since Duellman and Gray (1983) phractus, and Stefania . Based on Faivovich and Duellman and Hoogmoed (1984). Haas et al.’s (2005) topology ( fig. 24), claw­ (2003) suggested ( fig. 15), on the basis of shaped terminal phalanges and presence of morphological data, that his examplar of intercalary cartilages between the ultimate Hemiphractinae , Gastrotheca , was far from and penultimate phalanges must be considother hylids and imbedded within a heter­ ered either convergent with those found in eogeneous group of leptodactylids and ran­ Hylidae or plesiomorphically retained in hy­

oids. Darst and Cannatella (2004), who ex­ lids (and lost in intervening lineages), while amined one exemplar species each of Gas­ the proximal head of metacarpal II not betrotheca and Cryptobatrachus , suggested on tween prepollex and distal prepollex, and the

2006 FROST ET AL.: AMPHIBIAN TREE OF LIFE

63

larval spiracle sinistral and ventrolateral acters that are possible synapomorphies. Two (Duellman, 2001) are convergent with those species are currently recognized, Rhinoderin the Phyllomedusinae . Our sampling of ma darwinii and R. rufum . We sampled R. Gastrotheca is not dense enough to allow for darwinii . the detection of the paraphyly suggested by DENDROBATIDAE (CA. 11 GENERA, 241 SPE­ Mendelson et al. (2000). Our sampling pre­ CIES): The monophyly of Dendrobatidae has cludes evaluation of paraphyly of any of the been upheld consistently (e.g., Myers and nominal genera. Nevertheless, we did sample Ford, 1986; Ford, 1993; Haas, 1995; Clough at least one species per genus, which allows and Summers, 2000; Vences et al., 2000b), us to test the monophyly of the hemiphrac­ but different datasets place Dendrobatidae at tines based on more extensive outgroup sam­ various extremes within the neobatrachian pling. Our sampled taxa are Cryptobatrachus clade. It is either nested deeply within hysp., Flectonotus sp. , Gastrotheca fissipes , G. loids and arguably related to cycloramphine cf. marsupiata, Hemiphractus helioi , and leptodactylids (Noble, 1926, 1931; J.D. Stefania evansi . Lynch , 1971, 1973; Burton, 1998a; Haas,

BRACHYCEPHALIDAE (1 GENUS, 8 SPECIES): 2003; Faivovich et al., 2005); the sister group This tiny group of diminutive south­ to of Telmatobius (Vences et al., 2003b) ; or southeastern Brazilian species are united by closely related to Hylinae (Darst and Can­ (1) the absence through fusion of a distin­ natella, 2004). Alternatively, they have been guishable sternum; (2) digital reduction (pos­ suggested to be deeply imbedded within ransibly homologous with that in Euparkerella oids, usually considered close to arthrolep­ and Phyllonastes in Eleutherodactylinae); tids or petropedetids (Griffiths, 1959b, 1963; and (3) complete ossification of the epicor­ Duellman and Trueb, 1986; Ford, 1993; Ford acoid cartilages with coracoids and clavicles and Cannatella, 1993; Grant et al., 1997). (Ford and Cannatella, 1993; Kaplan, 2002). Rigorous evaluation of the support for these Brachycephalidae was suggested to be im­ contradictory hypotheses is required. bedded within Eleutherodacylinae (Izeck­ Ford and Cannatella (1993; fig. 14 ) prosohn, 1971; Giaretta and Sawaya, 1998), vided the following as synapomorphies of which also shows direct development. Fur­ Dendrobatidae : (1) retroarticular process prether, Darst and Cannatella (2004; fig. 22) sent on the mandible; (2) conformation of the provided molecular data to link this taxon to superficial slip of the m. depressor mandib­ Eleutherodactylinae, but continued its rec­ ulae; and (3) cephalic amplexus. They menognition despite the demonstrable paraphyly tioned other features suggested by other authat its recognition requires. Although there thors but that were suspect for one reason or are several named and unnamed species in another. Haas (2003; fig. 15) considered the the genus, the monophyly of the group is not following to be synapomorphies that nest in question (Kaplan, 2002), and we sampled Dendrobatidae within hylodine leptodactylthe type species, Brachycephalus ephippium , ids: (1) guiding behavior observed during for this study. courtship; and (2) T­ or Y­shaped terminal

RHINODERMATIDAE (1 GENUS, 2 SPECIES): As phalanges. Like most other frogs, most dennoted earlier, the Chilean Rhinodermatidae is drobatids have aquatic free­living tadpoles a likely satellite of a paraphyletic ‘‘Lepto­ (with some endotrophy in Colostethus ), aldactylidae’’; it is like them in having pro­ though the parental­care behavior of carrying coelous and holochordal vertebrae. J.D. tadpoles to water on the back of one of the Lynch (1973) conjectured that Rhinoderma­ parents appears to be synapomorphic (Altig tidae is the sister taxon of Bufonidae , where­ and McDiarmid, 1999), although among as Lavilla and Cei (2001) suggested that Rhi­ New World anurans it also occurs in Cyclornoderma is within the poorly­defined ‘‘Tel­ amphus stejnegeri (Heyer and Crombie, matobiinae’’ (‘‘Leptodactylidae’’). The only 1979). notable synapomorphy of Rhinodermatidae Taxon sampling was designed to provide

Figures

Fig. 17. Tree of amphibians provided by San Mauro et al. (2005). This tree reflects a maximumlikelihood analysis of 1,368 bp of the nuclear protein­coding gene RAG­1, assuming the GTR 1 G 1 I substitution model (as suggested by ModelTest v. 3.6; Posada and Crandall, 1998). Sequence alignment was made manually with only one gap excluded from analysis.

Fig. 14. Narrative tree of relevant anuran taxa by Ford and Cannatella (1993). A branch subtending Hylidae 1 Pseudidae in the original figure is collapsed per errata distributed with reprint. An asterisk was used by these authors to denote a metataxon, and quotation marks to denote nonmonophyly.