<<

THE LOWER DISSOROPHOID

(), AND THE EVOLUTION OF MODERN

AMPHIBIANS

Trond Sigurdsen

Department of Biology

McGill University, Montreal

November 2009

A thesis submitted to McGill University in partial fulfillment of the requirements

of the degree of Doctor of Philosophy

© Trond Sigurdsen 2009

1 ACKNOWLEDGMENTS

I am deeply grateful to my supervisors Robert L. Carroll and David M. Green for their support, and for revising and correcting the drafts of the individual chapters. Without their guidance, encouragement, and enthusiasm this project would not have been possible. Hans Larsson has also provided invaluable help, comments, and suggestions.

Special thanks go to John R. Bolt, who provided specimens and contributed to

Chapters 1 and 3. I thank Farish Jenkins, Jason Anderson, and Eric Lombard for making additional specimens available. Robert Holmes, Jean-Claude Rage, and

Zbyněk Roček have all provided helpful comments and observations.

Finally, I would like to thank present and past members of the Paleolab at the

Redpath Museum, Montreal, for helping out in various ways. Specifically,

Thomas Alexander Dececchi, Nadia Fröbisch, Luke Harrison, Audrey Heppleston and Erin Maxwell have contributed helpful comments and technical insight.

Funding was provided by NSERC, the Max Stern Recruitment Fellowship

(McGill), the Delise Allison and Alma Mater student travel grants (McGill), and the Society of Vertebrate Student Travel Grant.

2 CONTRIBUTIONS OF AUTHORS

Chapters 1 and 3 were written in collaboration with Dr. John R. Bolt from the

Field Museum of Chicago. The present author decided the general direction of these chapters, studied specimens, conducted the analyses, and wrote the final drafts. Dr. John Bolt made many of the relevant specimens available, provided additional observations, data, and photographs, and corrected drafts.

3 TABLE OF CONTENTS

Acknowledgments 2

Contributions of Authors 3

General abstract / Résumé 7

General introduction 9

CHAPTER 1. The Lower Permian Amphibamid Doleserpeton

(Temnospondyli: Dissorophoidea), the interrelationships of amphibamids, and the origin of modern

Abstract 15

Introduction 16

Systematic paleontology 19

Materials and methods 21

Description 22

Phylogenetic analyses 48

Discussion 52

Bridging text 1 64

CHAPTER 2. The otic region of Doleserpeton (Temnospondyli) and its implications for the evolutionary origin of

Abstract 65

Introduction 66

Materials and methods 68

Description 69

Discussion 88

4 Bridging text 2 98

CHAPTER 3. The lissamphibian and elbow joint, and the origins of modern amphibians

Abstract 99

Introduction 100

Materials and methods 102

Results 104

Discussion 115

Bridging text 3 123

CHAPTER 4. The anuran humerus and the evolution of jumping in early salientians

Abstract 125

Introduction 126

Materials and methods 128

Results 130

Discussion 142

Bridging text 4 145

CHAPTER 5. The origin of modern amphibians: a reevaluation

Abstract 147

Introduction 148

Materials and methods 151

Results 154

Discussion 169

5 CONCLUSIONS 175

BIBLIOGRAPHY 179

APPENDIX 1 199

APPENDIX 2 201

APPENDIX 3 211

APPENDIX 4 216

APPENDIX 5 217

APPENDIX 6 224

APPENDIX 7 229

APPENDIX 8 230

APPENDIX 9 241

APPENDIX 10 246

6 ABSTRACT

The origin and evolution of modern amphibians is still a subject of controversy.

The amphibamid temnospondyl Doleserpeton has often been suggested as a close relative of modern amphibians, but the skeletal morphology of this important

taxon has never been fully described. In this thesis, a study of the skeletal anatomy of Doleserpeton is presented, and the interrelationships of dissorophoids are discussed. The relationships of modern amphibians and are studied using phylogenetic analyses based on both Bayesian inference and parsimony. The of Doleserpeton is preserved in such detail that it allows the first description of the of an amphibamid temnospondyl. It is found to house a posteriorly positioned perilymphatic duct, a feature which is today restricted to amphibians. Evidence for the presence of a -like tympanic annulus is also described. During the studies of the limb skeleton of Doleserpeton, it was realized that important features of the of modern were misinterpreted in previous anatomical descriptions. Comparisons of the of salientians (frogs), caudates (salamanders) and the Lower reveal important shared derived traits that are also found in

Doleserpeton. Humeral evidence also indicates that saltation was an important part of the locomotion of the Lower salientian . Previous phylogenetic analyses are reexamined and numerous problems in previous character coding are revealed. An analysis of the new data collected suggests that modern amphibians may be a monophyletic group closely related to dissorophoid temnospondyls.

7 RÉSUMÉ

L’origine des amphibiens modernes reste un sujet de controverse. Doleserpeton a longtemps été suggéré comme un proche parent des lissamphibiens, mais l’anatomie de ce spécimen important n’avait toujours pas été complètement décrite. La présente thèse porte sur ce spécimen énigmatique en comparant sa morphologie aux amphibiens modernes et fossiles. Ceci nous nous informe sur les liens phylogénétiques entre les amphibiens modernes et ceux datant du

Paléozoique en utilisant des analyses basées sur la parcimonie ainsi que sur l’inférence bayésienne. Les conditions de préservation du crâne de Doleserpeton sont telles qu’elles permettent la première description de l’oreille interne d’un membre de ce groupe. Ceci a permis de trouver la présence de conduits périlymphatiques, un trait généralement réservé aux lissamphibiens. Les structures tympaniques rappellent aussi celles des anoures. Ces nouvelles études comparatives ont permis de corriger certaines informations au sujet de l’anatomie des membres supérieurs des salamandres. Les comparaisons anatomiques entre les anoures, les salamandres et le fossile Eocecilia datant du Jurassique présentent de nombreux caractères communs qui permettent de mieux situer les caractéristiques de Doleserpeton. Les données provenant de l’observation de l’humérus ont permis d’établir que la saltation était présente chez Triadobatrachus. De nouvelles analyses phylogénétiques ont été effectuées afin de rectifier les stades des caractères utilisés. Ces nouvelles analyses suggèrent une descendance commune des amphibiens modernes à partir d’un groupe relié aux temnospondyles dissorophoides.

8 INTRODUCTION

The Permian amphibamid Doleserpeton annectens (Temnospondyli:

Dissorophoidea) is generally regarded as one of the most important taxa for our understanding of the origins of modern amphibians (Bolt, 1968, 1969;

Holmes, 2000; Ruta and Coates, 2007). It was found in the Lower Permian

(Leonardian) fissure fills of the Dolese Brothers limestone quarry near Fort Sill in

Oklahoma. Although this small Permian temnospondyl is a subject worthy of discussion in its own right, the possibility of a close relationship between

Dolseserpeton and modern amphibians provides a unique opportunity to study the evolutionary origins of a major group of tetrapods living today.

Modern amphibians (often united in the taxon ) consist of three orders within Tetrapoda. These orders are , and Apoda (Milner,

1988; Cannatella and Hillis, 2004). Salientia comprise crown-group Anura (frogs and ) and fossil stem salientians (proanurans). Caudata includes Urodeles

(salamanders and ) and stem taxa. Finally, Apoda consists of

() and their fossil relatives. The problem of modern origins is not only due to the lack of plausible sister-groups, but also the fact that the three modern orders are morphologically disparate, making the search for potential “proto-lissamphibians” difficult (Carroll, 2000). Consequently, there are several opposing hypotheses seeking to explain the evolutionary origins of extant amphibians (Vallin and Laurin, 2004; Carroll, 2007; Ruta and Coates, 2007;

Anderson et al., 2008b).

9 A relatively diverse fossil record of modern amphibians first occurs in the

Lower Jurassic (Carroll, 2007). The earliest forms include the virtually modern- looking salientians (Shubin and Jenkins, 1995) from the in Arizona, and Vieraella (Baez and Basso, 1996) from the Lower

Jurassic of Argentina. Among the earliest unequivocal caudates we find

(Ivachnenko, 1978), and Kokartus (reviewed in Milner, 2000) from Kazakhstan and Kirghizstan respectively. The earliest undisputed caecilian is Eocaecilia

(Jenkins et al., 2007), which is the only limbed apodan known to science. Like

Prosalirus, it was found in the Kayenta formation. In addition to these Jurassic forms, undisputed salientian fossils are known from the Lower Triassic.

However, only one specimen is represented by an articulated skeleton; that of

Triadobatrachus massinoti from Madagascar (Rage and Roček, 1989; Roček and

Rage, 2000). Whereas the Lower Jurassic salientian Prosalirus is though to have been a capable jumper (Shubin and Jenkins, 1995; Jenkins and Shubin, 1998), the mode of locomotion in Triadobatrachus is uncertain (Rage and Roček, 1989;

Roček and Rage, 2000).

Almost nothing is known of the evolution of the modern amphibian groups prior to the Triassic (Carroll, 2007). Although a great variety of tetrapods are known from the and Permian, very few show any features that can demonstrably link them to recent forms. Two important groups of Palaeozoic tetrapods are often discussed in connection with the origins of modern amphibians. These are the lepospondyls and the dissorophoid temnospondyls.

10 The lepospondyls are a group of small Paleozoic tetrapods, which include a large number of elongate forms (Carroll et al., 1998). The consist of

Microsauria, , , Adelospondyli, and Aїstopoda. Although the affinities of the group is somewhat uncertain, the lepospondyls are usually placed closer to than to temnospondyls in phylogenies (e.g. Vallin and

Laurin, 2004; Ruta and Coates, 2007).

As noted above, Doleserpeton belongs to the other group of interest; the

Dissorophoidea, which is a group of small to medium-sized temnospondyls (e.g.

Holmes, 2000). They generally have large interpterygoid vacuities on the palate, a well-developed otic notch (squamosal embayment), and a laterally exposed (LEP). Except for larvae and perennibranchiate forms (e.g. Fröbisch et al., 2007), they appear to be more fully terrestrial than other temnospondyls

(Holmes, 2000). Doleserpeton, together with many of the smaller dissorophoids, is generally classified within the Amphibamidae (Clack and Milner, 1993; Daly,

1994; Fröbisch and Reisz, 2008). The interrelationships of this group has been discussed in a number of recent papers (Huttenlocker et al., 2007; Anderson et al.,

2008a; Fröbisch and Reisz, 2008), but the most intriguing question regarding these forms is their putative relationship to some or all modern amphibians (e.g.

Watson, 1940; Carroll, 2007; Ruta and Coates, 2007). The earliest occuring amphibamids include the Upper Carboniferous taxa grandiceps from Mazon Creek, , and Platyrhinops lyelli from Linto, Ohio and Nyřany,

Czech Republic (Gregory, 1950; Clack and Milner, 1993). The latest occuring

11 is stowi from the Lower Triassic Assemblage

Zone of South Africa (Schoch and Milner, 2004).

After the publication of two seminal papers by Parsons and Williams (1962,

1963), the three modern orders of amphibians have frequently been viewed as a

monophyletic group, based mainly on their shared possession of a unique kind of dentition, in which the crown rests on a cylindrical pedicel. The earliest description of Doleserpeton (Bolt, 1969), subsequently linked the modern amphibians to the Temnospondyli (e.g. Milner, 1988; Bolt, 1991; Ruta and

Coates, 2007), partially based on the lissamphibian-like dentition of this form.

The view that all modern amphibians are allied to the Paleozoic temnospondyls is called the Temnospondyl Hypothesis herein.

The use of modern phylogenetic techniques saw the rise of an alternative view, dubbed the Lepospondyl Hypothesis, in which a monophyletic

Lissamphibia was seen as derived from the Paleozoic lepospondyls (Laurin and

Reisz, 1997; Vallin and Laurin, 2004). Finally, Carroll (2007) and Anderson

(2007) have suggested that while frogs and salamanders are closely related to

dissorophoid temnospondyls such as Doleserpeton, caecilians may be closer to

the microsaurs within Lepospondyli, making Lissamphibia a polyphyletic

assemblage. Consequently, this view is called the Polyphyletic Hypothesis.

Although the Temnospondyl Hypothesis (e.g. Ruta and Coates, 2007) is currently

the most widely accepted view, major new studies by Carroll (2009) and

Anderson (2007; Anderson et al., 2008b) have favoured the Polyphyletic

12 Hypothesis. The Lepospondyl Hypothesis continues to be favoured by a small

group of researchers (e.g. Marjanovic and Laurin, 2008, 2009).

Chapter 1 of this thesis includes a phylogenetic analysis shedding new light on

the relationships of the Amphibamidae, and provides a long-awaited full description of the anatomy of Doleserpeton. It also presents a re-analysis of the data matrix of Anderson et al. (2008).

A number of previous studies (e.g. Bolt and Lombard, 1985; Lombard and

Bolt, 1988; Robinson et al., 2005) have suggested that temnospondyls share important features of the middle ear with anurans. Such an important functional complex could provide convincing evidence for the affinities of the groups in question. This was the incentive for Chapter 2, in which the otic capsule and hearing system of Doleserpeton are described in detail.

Compared to the number of discussions of cranial features, there has been a general lack of interest in the postcranial skeleton. In Chapter 3, I discuss some important features of the humeri and elbow joints of modern amphibians and

Doleserpeton, highlighting new features that have never before been discussed in connection with the evolution of modern amphibians. Chapter 4 focuses on the evolution of saltation in anuran locomotion. It builds on observations made during my work on Chapter 3, showing how the morphology of the humeri of the

Triassic salientian Triadobatrachus can be used to draw conclusions about its mode of locomotion.

A number of problems with previous phylogenies were discovered during the work on Chapter 1, and this, together with a need for a new study of modern

13 amphibian origins, was the basis for Chapter 5, in which the individual data

matrices used in the discussion of amphibian relationships are studied, critiqued,

reanalyzed, and combined. (Note: Chapter 1 was finished after Chapters 2, 3, and

4, and draws to a certain extent on information from those chapters. It was moved to the beginning of this thesis in to provide a general introduction to

Doleserpeton).

14 CHAPTER 1.

The Lower Permian Amphibamid Doleserpeton (Temnospondyli:

Dissorophoidea), the interrelationships of amphibamids, and the origin of modern amphibians

ABSTRACT

The Lower Permian amphibamid Doleserpeton annectens (Temnospondyli) is of great importance to our understanding of the origin of modern amphibians. Here, this taxon is described in detail. The pedicellate bicuspid teeth of Doleserpeton are more similar to the dentition of modern amphibians than those of other

Palaeozoic forms, including . The dermal bones of the skull are almost completely smooth, unlike other dissorophoids. The braincase and parts of the postcranial skeleton are more highly ossified than in other amphibamids, and

Doleserpeton is unlikely to represent a juvenile stage. The are relatively short and posteriorly curved, a feature that may be shared with caudates,

Triadobatrachus, and Eocaecilia. The palate shares numerous features with modern amphibians, including the configuration of the interpterygoid vacuities and the structure of the . Previous phylogenies of amphibamids have failed to take into account the similarities between dissorophoids and stem temnospondyls. Our analysis indicates that dissorophoids may be more basal within Temnospondyli than previously thought, and that amphibamid interrelationships are far from resolved. Micropholis is probably not more primitive than the other amphibamids. The evidence for a temnospondyl origin of

15 batrachians is strong, particularly for anurans, and the Lepospondyl Hypothesis is

therefore questioned. Doleserpeton also shares numerous features with caecilians, supporting the theory of the monophyletic origin of lissamphibians from amphibamid temnospondyls.

INTRODUCTION

There is at present no consensus as to whether amphibamids constitute the sister-group of all modern amphibians (e.g. Ruta and Coates, 2007), of parts

thereof (e.g. Carroll, 2007; Anderson, 2007; Anderson et al., 2008b) or of none at

all (e.g. Laurin and Reisz, 1997; Vallin and Laurin, 2004). However, most

molecular studies support the of Lissamphibia relative to extant

amniotes (San Mauro et al, 2004; Zhang et al., 2005; Igawa et al., 2008).

Doleserpeton annectens, an amphibamid dissorophoid from the Lower Permian

of Oklahoma, was first described by Bolt (1968, 1969). It has repeatedly been

suggested as a possible “proto-lissamphibian”, either as a potential to

all Lissamphibia (e.g. Bolt, 1969, 1977, 1991; Ruta and Coates, 2007), or as the

closest Paleozoic relative to one or more of the modern orders (Bolt and Lombard,

1985; Holmes, 2000; Anderson, 2007; Carroll, 2007). One feature that caught

attention was the fact that this small temnospodyl has bicuspid pedicellate teeth

with labial and lingual cuspules, features that are thought to be diagnostic

lissamphibian characteristics (Parsons and Williams, 1962; Bolt, 1969).

Descriptions of the hearing system of Doleserpeton have suggested a close

resemblance to that of modern frogs (Bolt and Lombard, 1985; see also Chapter

16 2). Thus, the phylogenetic position of Doleserpeton reflects directly on the problem of the origins of modern amphibians.

The original description of Doleserpeton emphasized its dentition and vertebral morphology as pointing to lissamphibian affinities (Bolt, 1969). A few details of the skull were described a few later by Bolt (1974). That author went on to characterize several features as indicative of the Doleserpeton specimens being juvenile (Bolt, 1977), but nevertheless stressed its separate status from related forms, such as . Bolt and Lombard (1985) pointed to the otic notch and of Doleserpeton as being remarkably similar to those of modern anurans, particularly in the fact that both anurans and Doleserpeton have a ventral articulating notch on the stapedial footplate, indicating a rocking motion of the stapes rather than the piston-like movement seen in other tetrapods with a stapedial hearing system. However, such a stapes may have a wide distribution within the temnospondyli (Robinson et al., 2005).

During recent reexaminations of the Doleserpeton specimens, several new characters have come to light that may have a bearing on both the interrelationships of amphibamids, and the question of lissamphibian origins.

Here, Doleserpeton annectens (Fig. 1) is described in full in order to clarify and amplify previous descriptions. It is hoped that this work will shed new light on the origins of the modern amphibian orders.

17

FIGURE 1. The reconstructed skeleton of Doleserpeton in lateral view. The approximate position of the fenestra ovalis is shaded.

18

SYSTEMATIC PALEONTOLOGY

TEMNOSPONDYLI Zittel, 1888

DISSOROPHOIDEA Bolt, 1969

APHIBAMIDAE Moodie, 1909

DOLESERPETON ANNECTENS Bolt, 1969

Revised diagnosis

Small dissorophoid temnospondyl with skull length less than two centimeters,

having the following autapomorphies: non-labyrinthine dentition with lissamphibian-like pedicels (sensu Parsons and Williams, 1962), crowns with labial and lingual cuspules, vertebral centra with large and ventrally continuous pleurocentra that sometimes fuse with the neural arch, with a short transverse row of five or six pedicellate teeth, smooth dermal bones of the skull,

ectopterygoid reduced or absent. It shares with Platyrhinops and Gerobatrachus a

humerus that is about three times longer than wide. It also shares with

Platyrhinops the short posterior contact of the palatine with the pterygoid. It shares with Amphibamus and Platyrhinops the bicuspid tooth crown morphology.

19

FIGURE 2. The skull of Doleserpeton in dorsal view. A, FMNH UR1308 (type). B, FMNH UR1296.

20 MATERIALS AND METHODS

Specimens and locality

Doleserpeton annectens (Bolt, 1969) is represented by numerous specimens

from the Lower Permian fissure fills of (Fort Sill) in Oklahoma

(Appendix 1). They were collected from the discarded heaps of clay in the Dolese

Brothers limestone quarry. The matrix is a gray clay, which is sometimes densely

packed with fossil bony material, particularly in matrix from the D-concentrate

(sensu Bolt, 1969). Other pieces of matrix contain fewer Doleserpeton specimens,

but the included fossils, whether pertaining to Doleserpeton or other taxa, are

often very well preserved. The matrix is usually considerably softer than the fossil

specimens, and it is sometimes easily removed with a damp brush. The color of

the specimens varies from deep brown to gray or black. Most specimens consist

of completely disarticulated elements, and no complete skeletons have been found. However, several partially articulated skeletons exist, including the type specimen (FMNH UR1308).

Analyses

The data matrices (Appendices 4 and 6) were run trough a parsimony-based cladistic analysis using PAUP (version 4.0b10), and a Bayesian inference analysis using MrBayes (version 3.1). In PAUP, sequences were added randomly, and the bootstrap analysis was undertaken using 10 000 replicates. The Bayesian analyses were run using Markov chain Monte Carlo sampling (MCMC), and the default settings of MrBayes. Samples were taken every 100 generations, and burn-in set

21 to 25% of the samples. The stopping point was when the average standard

deviation of split-frequencies reached 0.01 (Ronquist et al., 2009).

DESCRIPTION

The of Doleserpeton annectens (from the D-concentrate) consists of a

skull (Fig. 2A) and partial (15 vertebrae) with associated ribs,

and a few elements of the appendicular skeleton. Although no complete skeletons

exist, enough remains to make a probable estimate of body length (snout-to-vent)

of about 55 mm.

The dermal integument of Doleserpeton is not as clearly preserved as in certain

other amphibamids (e.g. Daly, 1994). One specimen (e.g. FMNH UR1309) seems

to have a vague imprint resembling large overlapping ovoid scales. However, this

interpretation is far from certain, and a no other specimens show any signs of the scaly dermal covering of other amphibamids.

A few specimens show elements belonging to the hyobranchial apparatus (Fig.

3B). It appears to consist of a central basihyal, a pair of ceratohyals, and one hypobranchial.

Skull

The most complete are those of FMNH UR1308 (type), UR1296, UR1323 and UR1307 (Figs. 2, 3). One additional skull was examined, but it lacks a specimen number (to be donated to the Field Museum shortly). The skull of the type specimen is approximately 18 mm as preserved, but the skull length is about

22

FIGURE 3. The palate and lower jaw of Doleserpeton. A, FMNH UR1323. B, FMNH UR1307. C, dorsal view of lower jaw, FMNH UR1335 (stereo photo).

23

15 mm when corrected for distortion (Fig. 4). Some incomplete skulls appear to

be somewhat larger, perhaps 18-19 mm in life. The smaller specimen FMNH

UR1323 is 12.3 mm long.

The orbits are larger relative to the skull size than those of most amphibamids

save, perhaps, Eoscopus (Daly, 1994) and Gerobatrachus (Anderson et al.

2008b). Platelets were found in the area of the orbits in several specimens (Fig.

2). Most of these resemble the squareish palpebral (eyelid) plates described in

other amphibamids by Carroll (1964), Daly (1994) and others.

Although slight ridges and pits are present, the external surface texture of the dermal bones is smoother than those of other amphibamids, with none of the external sculpturing seen in other dissorophoids, including the similarly sized

Amphibamus grandiceps. The smooth surface of the external surface of the skull is only broken by low ridges around the orbits, otic notches, and on the .

The premaxilla borders the , nasal and vomer and it forms the anterior rim of the external nares (Figs. 2 and 4A). It has between 20 and 25 teeth. The alary process is well developed, and fits into a notch in the nasal bone, which is floored with a thin osseous sheet, as in Eoscopus (Daly, 1994). Internally, the premaxilla has a horizontal shelf running dorsal to the tooth row, and contacting the vomer. The internal edge of the shelf bends ventrally, forming a low, vertical ridge that runs parallel to the tooth row. This feature continues on to the maxilla.

The premaxilla is pierced by many small foramina, and internally, there is a

24 rounded recess below and slightly medial to the alary process. An inter-rostral

fontanelle was almost certainly present (Fig. 4A).

The maxilla contacts the premaxilla, septomaxilla, lacrimal, palatine, jugal, and

quadratojugal. It is an antero-posteriorly elongated element bearing more than 40

pedicellate teeth. The anterior portion extends dorsally, forming a sheet of bone

(pars facialis) covering parts of the lateral surface of the skull anterior to the

orbits. The external surface is tallest at the contact with the lacrimal (Fig. 4A).

The long posterior process widens somewhat at the jugal, and extends further

posteriorly to the quadratojugal (Fig.1). Like the premaxilla, the maxilla has an

internal shelf, which curves ventrally at the medial (lingual) rim (Fig. 4B). In

FMNH UR1323 the maxilla has been rotated post mortem along is longitudinal

, so that this shelf appears to be directed laterally (labially) (Fig. 3A). The

shelf contacted the palatine and the anterior process of the pterygoid bone.

The septomaxilla is present in the external naris of the holotype (FMNH

UR1308). It is a small, crescent-shaped bone contacting the lacrimal and maxilla,

but seemingly in closer contact with the latter. This is close to the situation seen in

Tersomius (Carroll, 1964).

The lacrimal is roughly L-shaped as in Amphibamus grandiceps, with one arm

bordering the orbit, and the other extending anteriorly to the external naris (Figs.

2A, 4A). Three openings for the nasolacrimal duct are present in the orbit of

FMNH UR1308 (Fig. 2A), but FMNH UR1344 clearly shows a fourth opening.

The duct is enclosed in the bone and can be seen as a low ridge on the ventral side, although it is only partially enclosed in small specimens, such as FMNH

25

FIGURE 4. The skull of Doleserpeton reconstructed. A, dorsal view. B, ventral view. C, posterior view. Based on several specimens.

26 UR1323 (Fig. 3A). It has a single opening anteriorly at the posteromedial part of

the external naris. Ventrally, there is a steep ridge, which is part of the narial

flange (Bolt, 1974), and it is situated close to the contact with the prefrontal.

The nasal is a wide element that contacts the premaxilla, frontal, prefrontal and

lacrimal. It forms almost half of the rim surrounding the external naris. A short,

powerful ridge extends on the internal surface from the corner of the contact with

the prefrontal and lacrimal running towards, but not reaching, the notch for the

alary process. This ridge is the anterior part of the narial flange (Bolt, 1974;

Dilkes, 1990). The nasal is overlapped by the alary process of the premaxilla.

The prefrontal bone contacts the lacrimal, nasal, frontal, and, by way of a ventro-lateral process (the ventral prefrontal process of Bolt, 1974), the palatine.

The part that forms the orbital rim is set off by a low ridge. It does not contact the postfrontal. The prefrontal also forms part of the narial flange ventrally.

The frontal is an elongated bone constituting part of the medial orbital margin.

Shallow anterior and posterior incisures along the lateral margin of this bone accommodate the prefrontal and postfrontal bones. The frontal widens slightly anteriorly, but is generally more elongate than in most other amphibamids.

The postfrontal has an elongated anterior process which fits into a posteromedial groove on the frontal. It overlaps parts of the parietal and supratemporal bones, and forms a flange underlying the postorbital.

The postorbital constitutes the posterior margin of the orbit. This bone overlaps the squamosal and postfrontal, and is overlapped by the jugal. The circumorbital ridge is quite pronounced on this bone.

27 The jugal is antero-posteriorly elongated, and it contacts the maxilla throughout its lateral border. It curves slightly dorsally around mid-length, corresponding to the region where the maxilla widens somewhat posterior to the orbit.

The parietals are relatively narrow anteriorly, but widen towards the level of the posterior edge of the postparietals (Figs. 2 and 4A). The large parietal foramen opens slightly anterior to the midpoint between the anterior and posterior edges of this bone. The parietal underlies the supratemporal, frontal, and postfrontal bones, but it overlaps the .

The postparietal bone is transversely elongated. Laterally, it forms a blunt point, which contacts the tabular. The dorsal side of the postparietal has a transverse ridge, which runs the length of the bone from the medial suture to the tabular.

There is a prominent, blade-like process on the posterior edge, immediately above the exoccipital bone (Fig. 4A, C).

The tabular bone forms the posterolateral corner of the skull roof (Fig. 4A), overlying the posttemporal fossa. Its main body is triangular with a medial point meeting that of the postparietal. Laterally and posterorly, a process extends posteriorly and ventrally to contact the paroccipital process of the opisthiotic. A pit on the antero-ventral surface presumably also contacted the otic capsule.

The supratemporal is a large element constituting most of the postero-lateral part of the skull roof, contacting the postfrontal, parietal, postparietal, tabular, squamosal and postorbital bones. A semilunar flange is present posterolaterally, defining the dorso-medial rim of the otic notch. A flange of the supratemporal underlies the postfrontal bone.

28 The squamosal forms most of the otic notch, but it does not reach the tabular

dorsally. The squamosal portion of the otic notch is smoothly rounded, and it is

wide as in most amphibamids and armored dissorophids. There is a thickened rim along the border of the notch, separating the surface of the notch from the lateral exposure of the bone. The ventral and posterior part overlapping the quadratojugal forms an irregular edge.

The quadratojugal is antero-posteriorly elongated (Figs. 3B, 4A). It is ventrolaterally flattened and only a narrow strip of this bone is visible laterally. It is overlapped broadly by the squamosal, and anteriorly, it is contacted laterally by the maxilla. The quadrate underlies the slightly widened posterior end of the quadratojugal and also borders it medially. This postero-medial part of the quadratojugal has a small notch for the dorsal process of the quadrate.

The quadrate is a robust element constituting the jaw articulation area of the skull. It underlies the quadratojugal, but is exposed dorsally on the medial side of that bone. In this area, the quadrate forms a short dorsal process (Bolt and

Lombard; 1985). The articulating area is saddle-shaped, with a large condyle medially, separated from a smaller lateral condyle by a short trough (Fig. 4C).

Palate

The palate of Doleserpeton is dominated by the large interpterygoid vacuities. It is distinguished from that of other amphibamids by the presence of short rows of pedicellate teeth on the vomer and palatine bones (evidently supplanting the usual fang-and pit pairs). The ectopterygoid is either absent or greatly reduced. A small

29 sliver of bone anterior-lateral to the pterygoid in specimen FMNH UR1296 (Fig.

2B) may represent this element, but this is uncertain. The interpterygoid vacuities

are bordered by the vomer, platine pterygoid and bones. A shagreen

of denticles covers the vomers, the lateral shelf of the pterygoids and a diamond-

shaped field at the base of the cultriform process of the parasphenoid. The latter

bone widens posterior to the basicranial articulation. Tooth-covered platelets

covering the interpterygoid vacuities could not be positively identified in any

specimen, contrary to the situation seen in Pasawioops (Fröbisch and Reisz,

2008).

The vomer of Doleserpeton is easily recognized by its short row of pedicellate teeth located medial to the internal naris. This row consists of five to six teeth set

at the anterior ridge of a roughly rectangular field, which is devoid of the

denticles covering the rest of the vomerine surface. The shape of the vomer is

otherwise typically amphibamid. It is a wide sheet of bone with a lateral process

contacting the palatine. The anterior edge is devoid of denticles, and the same is

true for a shallow intervomerine depression located antero-medially (Fig. 4B).

Some specimens appear to lack this depression, but this is evidently due to

crushing. Close examination of specimens FMNH UR1346 and UR1323 shows

that the floor of the pit was broken and pushed to the level of the rest of the bone.

The palatine bone is quite complex, and has been described in detail by Bolt

(1974). In ventral view, it has three processes: two anteriorly on either side of the

choana, and one posteriorly, overlapping the anterior process of the pterygoid

(Figs. 2B, 3B, 4B). It forms a lateral exposure bordering the lacrimal, maxilla, and

30 jugal bones (Figs. 2A, 4A). Ventrally, the palatine has a raised area with a short

row consisting of six or seven pedicellate teeth.

The pterygoid has three well-developed processes (Figs. 3, 4B). The anterior process reaches the posterior part of the palatine bone. The medial process participates in the basipterygoid articulation with the parasphenoid-basisphenoid- epipterygoid complex. The basicranial articulation takes the form of the medial process of the pterygoid cupping the dorsal surface of the bulbous lateral processes of the parasphenoid-basisphenoid, and the epipterygoid apparently also participates in the dorsal part of this articulation. The quadrate (posterior) process of the pterygoid bends slightly laterally and contacts the quadrate with a flat lateral surface. The pterygoid is somewhat less extensively covered in denticles than that of Amphibamus grandiceps. The denticles cover a well-defined area on the lateral shelf of the main body of the pterygoid. It does not extend far on to the processes as in Amphibamus. On the dorsal surface of the pterygoid, lateral to the basipterygoid articulation, a bony vertical wall extends dorsally and curves on to the anterior part of the quadrate process. This is the lamina ascendens of the pterygoid (Bystrow and Efremov, 1940). It is only preserved fully in FMNH

UR1322.

Braincase

The parasphenoid and basisphenoid are invariably fused (Fig. 5). The basal plate of the parasphenoid is very broad, and it is similar to Amphibamus grandiceps in that it widens considerably posterior to the basicranial articulation.

31 It has a diamond-shaped raised area of denticles at the base of the cultriform process (Fig. 5A). This process is concave dorsally (Fig. 5B), following the shape of the overlying sphenethmoid. The carotid foramina open ventrally between the basicranial process and the medial denticle field.

Dorsally, the basisphenoid forms a thickening of the parasphenoid-basisphenoid complex at the level of the basicranial articulation (Fig. 5D). In this region a sharp dorsal process, the dorsum sellae, juts dorsally and anteriorly. Immediately in front of the sellae, two sulci open at the base of the cultriform process. This appears to be similar to the recesses interpreted as attachment points for the rectus eye muscles, as described in Tersomius (Carroll, 1964). However, in

Doleserpeton, the sulci are walled off laterally and seem to be closely associated with the brain (Fig. 5D), making this interpretation unlikely. They appear to be too thick to be vascular, and seem to end blindly. Judging from dissections of modern amphibians, we tentatively suggest that this structure accommodated the two-lobed infundibulum immediately anterior to the pituitary proper.

The three-dimensional shape of the sphenethmoid is shown in specimens

FMNH UR1350 (Fig. 5E) and UR1322. It can also be seen in specimen FMNH

UR1307. In cross section it is U-shaped, overlying the cultriform process of the parasphenoid. The sphenethmoid has a smooth exterior surface. In FMNH

UR1322 it appears pitted, but this is probably due to distortion of the thin bone, as the surface is not similar on the right and left halves.

The epipterygoid is flattened in FMNH UR1307 (Fig. 3B) but is found in situ in specimen FMNH UR1322. It consists of an ascending process and a broad base.

32

FIGURE 5. Braincase of Doleserpeton. A, cultriform process of parasphenoid, FMNH UR1369, ventral view. B, same in dorsal view. C, parasphenoid in ventral view (catalogued under same specimen number as above). D, same in dorsal view (showing fused basisphenoid). E, sphenethmoid in ventral view, FMNH UR1350.

33 The ascending process curves strongly medially as in (Sawin, 1941) but unlike that of or , and its ossified parts apparently did not reach the dermal skull roof (as in Edops, Romer and Witter, 1942). The medial part of the base sits on top of the basipterygoid articulation immediately anterior to the prootic. The lateral and posterior part of the base curves upwards towards the lateral part of the prootic, and it may have contacted the anterior groove on the crista parotica.

The otic capsule and adjacent parts of the braincase are described in detail in

Chapter 2. The otic capsule consists of the prootic and opisthotic, although these elements appear to be fused in FMNH UR1322. The opisthotic is notable for its

lateral ridges on the paroccipital process, and its ventrolateral ledge resembling

that of anurans.

The exoccpitals are fused to the basioccipital, and sometimes partially fused

dorsally to the opisthotic (Fig. 4C). They form the two and

border the jugular foramen medially. A small foramen, presumably for cranial nerve XII, is located lateral to the occipital condyle. The large internal jugular foramen and other details of the internal braincase are described in Chapter 2

(Sigurdsen, 2008 [Chapter 2]).

Lower jaw

Like the skull proper, the individual bones of the lower jaw lack dermal

sculpturing apart from some low ridges on the angular. In ventral view (Figs. 3,

4B), the jaw bones are arranged as in Eoscopus (Daly, 1994), but the dentary of

34 Doleserpeton is enlarged relative to this form, reaching well past the level of the posterior edge of the post-splenial. The jaw appears to have the same bones found in basal temnospondyls (e.g. Milner and Sequeira, 1994). This includes a set of at least two coronoids, the anterior of which is covered in denticles, whereas the posterior-most coronoid is bare (Fig. 3C). The presence of an anterior coronoid is uncertain. The dentary has a ventral groove for the presplenial, and the latter contacts its fellow at the symphysis, where the dentary is thickened lingually. In this region, the dentary bears a patch of symphyseal teeth lingual to the main tooth row (Fig. 6A). This patch of teeth is sometimes elongated to form a short secondary tooth row on the lower jaw. The symphysial teeth are of the same size as marginal teeth, and appear to be pedicellate. The post-splenial covers much of the anterior process of the prearticular, which is quite long in dorsal view. The slit-like meckelian fenestra is bordered by the angular and the prearticular elements. There is a small foramen close to the posterior edge of the prearticular.

Dentition

The dentition of Doleserpeton is remarkable for its close resemblance to that of modern amphibians (Bolt, 1969). The tooth base is a long, cylindrical pedicel

(Fig. 6A) and the tooth-crown is apparently not firmly fused to this base. Because of this, the crown often is broken off from the pedicel, as is often seen in fossil lissamphibians. The tooth crown consists of two cusps situated lingually and labially (Fig. 6B). A curved edge runs towards the apex of each cusp. This type of dentition is present in the upper and lower jaws, as well as in short rows on the

35

FIGURE 6. The dentition of Doleserpeton. A, dentary with pedicels in medial (lingual) view (FMNH PR 2823). B, tooth crown from maxilla (FMNH PR 2822). This tooth is preserved in situ, and shows the orientation of the cusps. Also note the ridges associated with the cusps.

36 vomers and palatine bones. The teeth are small, and more than 20 teeth sit on the

premaxilla, and about 40 on the maxilla. The dentary had at least 60 teeth. The

vomerine, palatine, and symphyseal rows each consists of between five and seven

teeth. The patch of pedicellate teeth at the symphysis of the lower jaw resembles

that of some caecilians (e.g. Evans and Sigogneau-Russell, 2001; personal

observations of Typhlonectes), and in at least one specimen this patch was

elongated to constitute a short secondary row of teeth in the lower jaw, as in

Eocaecilia (Jenkins et al., 2007). Smaller denticles cover a raised patch on the

parasphenoid, part of the pterygoid, most of the vomers, and the coronoid

elements of the lower jaw.

Vertebral column

The vertebrae and vertebral column have been described in part by Bolt (1969),

Daly (1994) and Carroll et al. (2004). FMNH UR1317 preserves 22 presacral

vertebrae, but this specimen has a gap in the vertebral column, making the

original number of presacrals somewhat uncertain (Fig. 7A). However, judging

from comparisons to other specimens, it seems safe to assume that the number of

presacrals was between 22 and 25 in life. The most likely number is 24.

The consists of a single-element centrum fused to the neural arch (Fig.

7B). There is a channel for the which is open dorsally, and this channel

divides the centrum into two lateral lobes. Each lobe has an anterior cotyle which

is considerably wider than the corresponding posterior articulating surface. The

neural arch of the atlas consists of two processes that do not meet dorsally. The

37 dorsal part of each process bends sharply posteriorly. At the anterior angle of this bend there is normally a prezygapophysis, similar to that articulating with the proatlas in Eryops (Moulton, 1974). However, this prezygapophysis is sometimes absent, and no proatlas has been found.

The centrum of the axis is poorly preserved, but was presumably similar to those of the dorsal vertebrae. The neural arch of the axis is expanded antero- posteriorly.

The presacral vertebrae are distinctive among temnospondyls in that mature specimens have very large pleurocentra that form nearly the whole body of the vertebral centrum (Fig. 7C), although the pleurocentra apparently remained hollow throughout the life of the . The pleurocentrum fuses with the neural arch in mature specimens. The proportionally smaller and crescent-shaped intercentra always remain separate from other vertebral elements. There is a small facet for the capitulum of the on the intercentrum, and a larger one for the tuberculum on the neural arch. In less mature specimens (judging by size and degree of ), the vertebrae closely resemble those of Eoscopus and

Tersomius (Daly, 1994). In these, the pleuro- and intercentra are more equal in size, and there is no fusion of the pleurocentra to the neural arch. The neural arch of the fourth is shorter than both the preceding and subsequent vertebrae, somewhat resembling Eryops (Moulton 1974). The sacral vertebra has wide transverse processes for the sacral rib (Fig. 10D).

The tail is partially preserved in specimen FMNH UR1319 (Fig. 7D). It is impossible to ascertain the number of caudal vertebrae, but it is certainly more

38

FIGURE 7. Vertebral column of Doleserpeton. A, FMNH UR1317, vertebral column, showing at least 22 presacrals (right side). B, detail of same, showing atlas and axis complex. C, FMNH UR1309, presacral vertebrae (left side). D, FMNH UR1319, caudal vertebrae, left side.

39 than 15. Most likely, Dolseserpeton had a tail similar to that sometimes seen preserved in Amphibamus grandiceps (Godfrey, 1997). Although the anterior caudals had ribs, these are not well-preserved. More posteriorly, the ribs are absent, but well-developed haemal arches appear. In this region of the tail, all the elements of the caudal vertebrae appear to be fully fused. However, the different parts of these vertebrae can still be discerned, and are generally similar to those of

Eryops, as described by Moulton (1974). The intercentra appear to constitute a larger part of the caudals than in the trunk vertebrae.

Ribs

The ribs of Doleserpeton start on the second vertebra, and are generally short and gently curved so that the distal ends are directed posteriorly (Figs. 1, 7A).

This observation is based not only on partly articulated specimens, but also on individual disarticulated elements. The first four ribs grow increasingly robust, but then diminish in size posteriorly after the fifth rib. The robust ribs in the region of the pectoral girdle, that is ribs 2-4, have expanded and flattened distal ends. More posteriorly the ribs come to a thin point. The longest ribs (pectoral region) are slightly shorter than three vertebral centra.

Pectoral girdle

The pectoral girdle of Doleserpeton (Fig. 8) is closely comparable to those of other terrestrial temnospondyls (e.g. Holmes et al. 1998; Pawley and Warren,

2006). The dominant element in the pectoral girdle is the scapulocoracoid (Fig.

40

FIGURE 8. The pectoral girdle of Doleserpeton. A, right scapulocoracoid, FMNH UR1360, lateral view. B, same in medial view (slightly dorso-medial). C, (fragment), FMNH UR1307, ventral view. D, right , FMNH UR1359, anterior view. E, FMNH UR 1358, right (fragment), lateral view.

41 7A, B). In mature specimens, this element is similar to that of Dendrerpeton

(Holmes et al., 1998), being somewhat less ossified dorsally than those of

Dissorophus, or Eryops (Williston, 1910a, b; Pawley and Warren, 2006), but more so than that described for Amphibamus grandiceps (Watson, 1940). The element consists of a vertical broad sheet of bone () dorsally, and a thicker ventro-medial () part. However, these elements are completely fused so that there is no discernable suture. The acetabulum is of the screw-shape found in many early tetrapods (e.g. Holmes, 1977) with the anterior part of the facet facing

posteriorly, whereas the posterior part faces dorso-laterally. Several foramina

pierce the scapulocoracoid, including the glenoid, supraglenoid and coracoid foramina. These follow the same pattern as in Eryops (Pawley and Warren, 2006).

The interclavicle (Fig. 8C) appears to be of the normal diamond-shape, and it

does not to have any long processes. The clavicle is better preserved (Fig. 8D). It

consists of a ventro-medial blade-like process articulating with the interclavicle,

and a dorsal shaft. Only incomplete cleithra have been found (Fig. 8E). However,

its preserved parts show that it consisted of a ventral shaft and a dorsal blade-like

area as in other temnospondyls.

Forelimb

The humeri of Doleserpeton are better preserved than those of any other

amphibamid, and they are more elongate than most of these (Fig. 9A-D), except

for Platyrhinops and Gerobatrachus. The contrast is quite pronounced when

compared to the similar-sized Amphibamus grandiceps specimen YPM794. The

42

FIGURE 9. The forelimb of Doleserpeton. A, right humerus (type, FMNH UR1308), medial view. B, right humerus, lateral view (FMNH UR1361). C, same, flexor view. D, FMNH UR1321, right forelimb, humerus is in flexor view, whereas zeugopodium and autopodium are seen in extensor view (as preserved). E, left , extensor view (FMNH UR1377). F, same in anterior view. G, left in flexor (posterior) view (FMNH UR1378). H, same in extensor view.

43 distal and proximal ends of the humerus are twisted about 90 degrees, or somewhat less. The distal articulation area is often poorly preserved (Fig. 8A, B,

C), but the large capitulum can be seen in specimen FMNH UR1321, and as an imprint in the sediment of the type specimen. The distal end of the bone, including the large capitulum and smooth trochlear region, is similar to that of modern lissamphibians (Chapter 3).

The radius is a relatively simple hourglass-shaped bone (Fig. 9D, E). Although the distal part is often poorly ossified, specimen FMNH UR1321 shows the fan- shaped distal end which resembles that of modern salamanders (Fig. 9D). The radius fits into a well-defined facet on the ulna. The latter bone is a slender version of the ulna of Eryops (Pawley and Warren, 2006). However, the radius appears to have been more closely associated with the ulna in Doleserpeton, as evidenced by a radial facet on the ulna. This may reflect a shared articulation surface, as in modern amphibians (Chapter 3). The proximal articulation and olecranon process are poorly ossified.

The manus has four digits, parts of which are preserved in FMNH UR1321 (Fig.

9D). The phalangeal formula is impossible to determine, due to the disarticulation of the specimens. The distal phalanges on both manus and pes come to a ventrally curved point superficially resembling a claw (Fig. 2A). Several carpal elements are preserved in specimen FMNH UR1321 (Fig. 9D), including the radiale, ulnare, intermedium, and apparently all the distal carpals, which do not form a basale commune. Some of the centralia are apparently fused to form a larger element, but other carpal elements may be missing or unossified.

44

FIGURE 10. The pelvic girdle of Doleserpeton. A, , ventral view (FMNH UR1379). B, same, dorsal view. C, same, lateral view. D, FMNH UR1306, parts of sacral vertebra and rib, right lateral view. E, left sacral rib, FMNH UR1362, anterior view. F, same in posterior view.

45 Pelvic girdle

The pelvis of Doleserpeton (Fig. 10A, B, C) is similar to those of primitive

temnospondyls, such as Dendrerpeton and Balanerpeton (Holmes et al. 1998;

Milner and Sequeira, 1994). The pubis, ilium and ischium are fused, although the

area of the pubis is partially unossified. The dorsal surface of the pubis slants

antero-ventrally, and the obturator foramen pierces through dorsally on this

sloped surface. The dorsal blade of the ilium is directed postero-laterally. Most of

the ischium is a flat sheet of bone posterior to the pubis and acetabulum. The

concave surface of the latter is facing almost directly laterally. The single sacral

rib (Fig. 10D, E, F) is a robust blade curving posteriorly. It has a wide tuberculum

facing the thickened transverse process of the sacral vertebra, and a well-

developed capitulum, which presumably articulated with the sacral intercentrum.

Hind limb

The is an elongate and often well-ossified bone. It is similar to that of

modern salamanders in having a well-developed internal trochanter connected to

the adductor crest, but this morphology is relatively common in other tetrapods as

well (e.g. Sumida, 1997). As with the other limb bones, there is a pronounced

difference between some slightly immature specimens (subadults) and the fully

ossified adults (Fig. 11A, B, C) in that the condyles and complete internal

trochanter are only present in the latter.

The (Fig. 11D, E) is similar to that of Eryops in being relatively stout and with an expanded proximal head (Pawley and Warren, 2006). It has a slight flexor

46

FIGURE 11. A, well-ossified left femur in flexor view (FMNH UR1380). B, right femur, FMNH UR1364, flexor view. C, same, extensor view. D, left tibia, extensor view (FMNH UR1381). E, same in flexor view. F, left in extensor view (FMNH UR1382). G, same in extensor view. H, tibia and pes (FMNH UR 1320).

47 crest, and small facets for the tibiale and intermedium. A similarity to Eryops can

be seen in the fibula as well. This curved bone has a low posterior ridge, and a

well-developed facet for the intermedium.

The pes is incompletely preserved (Fig. 11H) and no set of tarsals have been

found in articulation, although they are ossified in some disarticulated specimens.

In specimen FMNH UR1319, the articulation between the fibula and intermedium

can be seen. The phalangeal count is difficult to determine, other than that the

second digit had two phalanges and that digit three almost certainly had three.

Digit four had at least three. The phalangeal formula of both Amphibamus

grandiceps and Eoscopus is 22343 (Daly, 1994), and this is not unlikely to be the correct formula for Doleserpeton as well.

PHYLOGENETIC ANALYSES

Amphibamid character matrix

We based our phylogenetic analysis on a matrix modified from Fröbisch and

Reisz (2008), which is largely built on data collected by Schoch and Rubidge

(2005). Problematic characters were either recoded or removed (see Appendix 2), and a careful review of the character states of primitive forms such as

Dendrerpeton and Balanerpeton was incorporated (Holmes et al., 1998; Milner and Sequiera, 1994). During our review of previous amphibamid phylogenies, it became apparent that many characters used in the previous data matrices are of dubious value (see Discussion and Appendix 2), and the matrix was therefore cut down considerably. The new matrix (Appendices 3-4) reflects our current

48

FIGURE 12. Phylogenetic analyses of the interrelationships of dissorophoids. A, strict consensus of 11 most parsimonious trees. B, 50% majority rule bootstrap tree (10 000 replicates). C, Bayesian inference tree (1000 000 generations). Numbers indicate support in percentiles.

49 knowledge of amphibamids better than previous data matrices, as the different

amphibamid taxa are very similar in general morphology, and few characters are

known to elucidate the interrelationships within this group. The few characters

that set the individual taxa apart are often unique to individual genera, providing

little phylogenetic information. The previously published, well-resolved

phylogenies (e.g. Schoch and Rubidge, 2005; Huttenlocker et al., 2007; Fröbisch

and Reisz, 2008) are therefore partially based on poor evidence.

The branchiosaurid (e.g. Boy and Sues, 2000) and the primitive temnospondyl Balanerpeton were included in the analysis, and one new character was added (the shape of the otic notch). The phylogenetic analysis using

parsimony resulted in 11 most parsimonious trees, the strict consensus of which is

given in Fig. 12A. This cosensus tree, the bootstrap analysis (Fig. 12B), and the

Bayesian analysis (Fig. 12C) all show a lack of support for most of the internal

relationships of amphibamids.

Re-analysis of data from Anderson et al. (2008)

In order to get an impression of the importance of the new data for modern

amphibian relationships, the matrix by Anderson et al. (2008b) was examined,

adding previously unknown information on Doleserpeton. The phylogenies of

Anderson (2007) and Anderson et al. (2008) are notable for being the first

phylogenetic analyses to support the Polyphyletic Hypothesis.

Several characters that were coded similarly in the stem caecilian Eocaecilia

(Jenkins et al., 2007) and microsaurs were found to be just as easily, or even

50

FIGURE 13. Analyses based on modified character matrix from Anderson et al. (2008). A, strict consensus tree of 91 equally parsimonious trees. B, bootstrap analysis (10 000 replicates). Arrows indicate positions of modern amphibians.

51 preferably, interpreted as temnospondyl characters. The addition of new data and the reinterpretation of several characters (Appendix 5) led to Eocaecilia being classified with frogs and salamanders within Amphibamidae (Fig. 13). See

Appendix 5, Chapter 5 and the discussion below for further details on the re- analysis of the data matrix based on that of Anderson et al. (2008). Chapter 5 also contains bootstrap and Bayesian analyses of this matrix.

DISCUSSION

Compared to modern tetrapods, Doleserpeton resembles terrestrial caudates

(newts and salamanders) in general body proportions, although this may not reflect a direct sister-group relationship with this taxon specifically. In several areas of the anatomy, such as the dentition, limb morphology, and the structure of the ribs, the resemblance encompasses primitive anurans and caecilians as well.

The limb girdles and skull are more heavily ossified than modern caudates, although Doleserpeton is lighter built than many other dissorophoids. Its hearing system was evidently most similar to that of modern anurans (Chapter 2).

The overwhelming majority of tetrapods from the Fort Sill fissures were terrestrial (Sullivan et al., 2000). This, coupled with the well-ossified appendicular skeleton of Doleserpeton, and the lack of any aquatic adaptations in this form (such as lateral lines or reduced tympanic hearing system), indicates that

Doleserpeton was primarily a . However, it is likely that it had an aquatic larval stage, given its close relationship to forms with known larval stages, such as Amphibamus grandiceps, and barnchiosaurids (Milner, 1982). The

52 dentition is of the lissamphibian type (Parsons and Williams, 1962), and

Doleserpeton probably fed on small as many modern amphibians

do.

Comparison of dentition

After Bolt (1969) drew attention to the lissamphibian-like dentition of

Doleserpeton, several papers have described taxa from the Paleozoic with

dentitions approaching the lissamphibian condition (Bolt, 1977, 1979; Anderson

and Reisz, 2003; Schoch and Carroll, 2003; Anderson et al., 2008b). Bolt (1977,

1979) described the dissorophoids Tersomius, Amphibamus and Broiliellus as

having, at least in some specimens, bicuspid teeth. According to these studies,

Amphibamus may also have pedicellate tooth bases. Anderson and Reisz (2003)

described the microsaur Bolterpeton as having weakly bicuspid tooth crowns,

whereas Germain (2008) finds pedicel-like structures in an aїstopod. Davit-Beal

et al. (2007) warn against using the bicuspid condition as an argument for

lissamphibian relationships, but their arguments seem to be built on a

misunderstanding of the plesiomorphic condition (compare Bolt, 1979).

Among all modern lissamphibian orders, at least some forms have bicuspid

pedicellate dentition (Parsons and Williams, 1962; Bolt, 1969; Duellman and

Trueb, 1994). The dentition of Eocaecilia (Jenkins et al., 2007) is remarkably

similar to that of Doleserpeton. Its crowns are clearly bicuspid, with lingual and

labial cusps, and the base is a well-defined cylinder. The crowns in both

Eocaecilia and the form Rubricacaecilia (Evans and Sigogneau-

53 Russell, 2001) often break off before fossilization, indicating a poorly calcified area between the tooth crown and base, as in modern lissamphibians. In all modern amphibians, the long, cylindrical pedicel is situated on the lingual side of a vertical sheet of bone, which forms the outer surface of the tooth-bearing element (e.g dentary or maxilla).

Apart from Doleserpeton, the amphibamid with the most lissamphibian-like dentition appears to be Amphibamus grandiceps (Bolt, 1979). This form has bicuspid crowns, and its tooth bases appear to be semi-cylindrical. However, the teeth, when missing, appear to have broken off below the base, perhaps indicating a lack of a separation between crown and base. Also, the bases are not elongate cylinders as in Doleserpeton and modern lissamphibians. Tersomius specimens appear to have highly variable tooth morphologies (Bolt, 1977), but few specimens show as clear bicuspid crowns as Doleserpeton. Its tooth bases appear to lack a lissamphibian-type pedicel. The same is true for Broilellus (Bolt, 1977).

Gerobatrachus (Anderson et al., 2008b) was described as having monocuspid, pedicellate dentition. The lack of secondary cusps is unexpected in a form which is perceived to be basal to batrachians (frogs and salamanders), although this may be seen as secondary loss. A more serious problem is the interpretation of the tooth base. Anderson et al. (2008b) describes this as being of the pedicellate type, but reexamination of the specimen showed that the teeth of this form lack the cylindrical bases (i.e. pedicels) of lissamphibians. Nor is the line of poor calicification any clearer than in Amphibamus.

54 Schoch and Carroll (2003) described lissamphibian-like pedicels in the branchiosaurid Apateon. Although stouter in shape than in most modern forms, these appear to be similar to modern amphibians, and add informative evidence for the relationship of this group. However, the tooth crowns of branchiosaurids have been described as either monocuspid (Schoch and Milner, 2008), or as having three cusps arranged transversely (Werneburg, 2009).

Bolterpeton (Anderson and Reisz, 2003) is the only microsaur described as having a dentition approaching the lissamphibian condition. Its bases are not pedicellate, but the tooth crowns were interpreted as weakly bicuspid (as in

Broiliellus). The secondary cusp is not clear in the figures however, and it is uncertain if its orientation is similar to lissamphibians. Furthermore the overall shape and surface texture of the ridged crowns appear to be typical for most microsaurs, and quite unlike modern amphibians. A second lepospondyl, an aїstopod of the Phlegetonia, has been suggested to have pedicellate teeth

(Germain, 2008), based on two narrow zones separating the crown and base.

However, the presence of two “zones of weakness” are not typical for modern amphibians (compare Parsons and Williams, 1962), and it is unclear if the base forms an elongate cylinder on the lingual side of the bone, as is the case in modern amphibians and Doleserpeton (Fig. 6).

In conclusion, although some other dissorophoids may rightly be said to approach the condition seen in Doleserpeton, the latter form remains the only

Paleozoic taxon with an unambiguously lissamphibian-like pedicellate bicuspid dentition.

55 Ontogenetic stage of Doleserpeton

Bolt (1979), Holmes (2000), and Marjanovic and Laurin (2008) discussed the

ontogenetic stage of Doleserpeton and considered the possibility that it might

represent a juvenile, postmetamorphic stage. However, new studies of the

specimens reveal a slight ontogenetic range in the available material. Whereas

some specimens show a moderately ossified otic region and braincase as well as

unossified ends of the long bones, some specimens (e.g. FMNH UR 1321 and UR

1322) appear to be “further developed” than most other amphibamid specimens

described, judging from the degree of ossification. For instance, the braincase of

FMNH UR 1322 is more highly ossified than that of any other amphibamid,

including the much larger amphibamid Pasawioops (Fröbisch and Reisz, 2008),

which was found at the same locality. In most amphibamids, the braincase is only

partly preserved. Likewise, Doleserpeton is the only known amphibamid in which

some specimens have well-preserved and highly ossified distal ends of the humeri

(Fig. 9D). A low degree of ossification is not, in itself, proof of the immaturity of

a taxon, since some taxa may become mature without fully ossifying certain

elements. However, the high degree of ossification seen in some Doleserpeton

specimens appears to preclude the immaturity of these specimens.

Amphibamid phylogeny

Doleserpeton is, in many ways, a typical amphibamid. However, it does show a

number of apomorphies, almost all of which point to lissamphibian affinities.

Apart from the typical amphibamid traits, such as the laterally exposed palatine,

56 few of its traits are shared with other dissorophoids, complicating the question of

their interrelationships. Several characters previously thought to be limited to

dissorophoids are possibly plesiomophic within Temnospondyli. For example, the ventral contact between the prefrontal and the palatine below the lacrimal has been described as an unambiguous synapomorphy of dissorophoids (Fröbisch and

Reisz, 2008). However, there is a tongue-like prefrontal process in this region of the primitive temnospondyl Dendrerpeton (Holmes et al., 1998). This process may have contacted the palatine bone. Judging from the available photographs of

Balanerpeton (Milner and Sequeira, 1994), the prefrontal and palatine appear to approach each other ventrally in this form as well.

Several other features, such as the stapes and pelvic girdle, are more similar in primitive temnospondyls and amphibamids than the previous character coding indicates (e.g. Schoch and Rubidge, 2005). In the case of the stapes (Appendix 2, character 32), morphological differences might be informative in a wider analysis of tetrapod groups, but the great similarity between Dendrerpeton, Amphibamus, and Doleserpeton, coupled with a lack of information on this bone in many amphibamid taxa, indicates that this character is uninformative for the interrelationships within dissorophoids.

Another problem was the previous coding of the distal condyles of the amphibamid humerus. These were coded as “small” in size (Schoch and Rubidge,

2005; Anderson et al., 2008a), unlike most other temnospondyls (e.g. Pawley and

Warren, 2006). In fact, the distal end of the amphibamid humerus has never been fully described until recently (Chapter 3), but it has now been shown that the

57 ulnar condyle is absent, whereas the radial condyle (capitulum) is very large (Fig.

9D).

A largely unresolved phylogeny may best represent our current knowledge of

amphibamid interrelationships (Fig. 12). Within this group, there appears to be

strong evidence for a close relationship between Pasawioops, Micropholis and,

perhaps, Tersomius (Fig. 12C). Doleserpeton appears to be closely allied to

Amphibamus and Platyrhinops, but this clade is poorly supported in the bootstrap

and Bayesian analyses (bootstrap clade support 55%). Note also that the branchiosaurid Apateon is included in the amphibamid group. This taxon is sometimes associated with the origin of salamanders (Fröbisch et al., 2007;

Carroll, 2007). Although not supported by the Bayesian analysis, the sister-group relationship seen between armored dissorophids (Dissorophinae and Cacopinae) and Amphibamidae, as presented in the bootstrap analysis, may well be correct

(Fig. 12B, unresolved in the consensus tree Fig. 12A). These two taxa share

features of the interpterygoid vacuities and otic notches not seen in the more

primitive trematopids.

An interesting feature of the present phylogeny is the clade containing

Eryopidae and the stereospondyl . The relationship between these

two taxa may come as a surprise, but it is also supported by a recent analysis by

Ruta (2009). Most of the taxa included here are dissorophoids, and these may well

be basal to most other temnospondyls excepting Dendrepeton and Balanerpeton.

Such a view is supported by numerous symplesiomorphies shared with the latter

58 two taxa, and by the early occurrence of amphibamids in the Westphalian series of the Carboniferous (Carroll, 1964).

Modern amphibians

As mentioned above, Doleserpeton shares numerous traits with modern

amphibians, such as the pedicellate bicuspid dentition (Bolt, 1969), the presence

of a posteriorly positioned perilymphatic duct (Chapter 2), and a generally frog-

like hearing system (Bolt and Lombard, 1985; Sigurdsen, 2008 [Chapter 2]). In

addition, lissamphibians share a number of traits of the humerus and elbow

region, features that are remarkably close to those of Doleserpeton (Chapter 3). It

is at present uncertain whether all of these traits are found in Doleserpeton exclusively, or if they were present in other amphibamids as well. Several parts of the morphology of Doleserpeton are better preserved than in other amphibamids, making the exact position of the individual taxa relative to lissamphibians ambiguous. However, a few such traits appear to be exclusive to Doleserpeton, and are also consistent with affinities to modern amphibians. These include the presence pedicellate bicuspid teeth, short rows of teeth on the vomers and palatines, and the reduction or absence of the ectopterygoid.

Salientia

Strong support for the temnospondyl origin of frogs is found in the structure of the palate, humeri (Chapter 3), the hearing system and braincase (Bolt and

Lombard, 1985; Sigurdsen, 2008, [Chapter 2]), including most features of the

59 inner ear to the structure of the stapes and tympanic annulus. The anuran palate shares with Doleserpeton the large interpterygoid vacuities bordered by the palatine, pterygoid and parasphenoid (Duellman and Trueb, 1994). Individual skull bones, such as the premaxilla and maxilla, also resemble those of

Doleserpeton in overall shape. The primitive contact between the quadratojugal and maxilla (Clack, 2002) is retained in most anurans, as in Doleserpeton, but unlike most lepospondyls.

Caudata

Salamanders often have large interpterygoid vacuities (present in the primitive fossil form Karaurus, modern hynobiids, cryptobranchids, and many other families), although this feature is partially obscured by the opening of the cheek region. The presence of pedicellate bicuspid dentition (Parsons and

Williams, 1962; Duellman and Trueb, 1994), a posteriorly positioned perilymphatic duct (Lombard, 1977; see also Chapter 2), a large capitulum on the humeri (Chapter 3), and short ribs directed posteriorly also support the amphibamid affinities of caudates.

Apoda

Among modern amphibians, the most problematic taxon is Apoda (caecilians), a group which may be allied to microsaurian lepospondyls, and therefore unrelated to batrachians (Carroll, 2007; Anderson et al., 2008b). However, most of the features listed above, including the humeral features, also apply to the primitive

60 caecilian Eocaecilia. Carroll (2007) lists nine derived characters shared by the microsaur and primitive caecilians:

1. Orbital openings small relative to those of comparably sized early tetrapods

2. Jaw articulation well anterior to occipital condyle

3. Absence of fangs on palatal bones

4. Rows of teeth on vomer and palatine, parallel to those of premaxilla and

maxilla

5. Loss of intertemporal and supratemporal bones common to early

temnospondyls

6. Ossification of pleurosphenoid

7. Double occipital condyle, not present in Carboniferous temnospondyls or anthracosaurs

8. Cylindrical centra fused to neural arch early in development, in contrast to labyrinthodonts

9. Greater elongation of vertebral column than in any temnospondyls, anthracosauroids, nectrideans, or early amniotes

With the possible exception of character 5, all of these can be seen as valid synapomorphies. However, characters 3, 4, 7, and 8 are also found in

Doleserpeton, although the vertebral fusion may occur later in development in this form. As for character 5, the intertemporal bone is also absent in

Doleserpeton, but the tabular bone of Eocaecilia may in fact be the supratemporal

(Jenkins et al., 2007), and hence not a shared loss with Rhynchonkos (see below,

61 and Appendix 5, Character 5). Some of the remaining characters are linked to the

elongation of the body (Character 9) and possibly to fossoriality (Character 1),

factors that commonly occur within Tetrapoda.

Many of the features coded by Anderson et al. (2008b) as being shared by

Eocaecilia and microsaurs are probably misinterpretations. For instance, the

palate of Eocaecila has been coded as having numerous microsaurian features, but

the interpterygoid vacuitites of this primitive caecilian are unusually large for a

microsaur, and also larger than most modern caecilians. In a phylogenetic analysis

that places caecilians among dissorophoid temnospondyls, this would emerge as a

reduction from the dissorophoid condiditon, rather than a reversal, as the

Polyphyletic Hypothesis implies. The wide shape of the vomer is also closer to

Doleserpeton than to microsaurs, as it forms a broad sheet anteriorly, rather than a

narrow medially situated structure. The interpterygoid vacuities in Eocaecilia are

bordered by the vomer, palatine, pterygoid and parasphenoid elements, exactly as

in Doleserpeton, but unlike microsaurs, in which the palatine is excluded from

this opening. The close resemblance of the dentition of Eocaecilia and

Doleserpeton adds to this pattern, and it is notable that another primitive

caecilian, Rubricacaecilia (Evans and Sigogneau-Russell, 2001) has a lower jaw with pedicels in almost the exact configuration seen in Doleserpeton (Fig. 6).

As mentioned above, the configuration of the “tabular” bone of Eocaecilia

(Appendix 5, character 5) is actually more similar to the supratemporal of both

Doleserpeton and primitive tetrapods (Clack, 2002). It should be coded as

uncertain (?). Finally, the ribs of Eocaecilia appear to be closer to the short,

62 posteriorly pointing type found in amphibamids, rather than to the longer,

ventrally curved ribs of microsaurs, again contrary to the character coding of

Anderson et al. (2008b).

Phylogeny

The corrections above were entered into the matrix provided by Anderson et al.

(2008b). When the characters in question were reinterpreted in light of the new

data from Doleserpeton, Eocaecilia turned up within a monophyletic taxon

containing Doleserpeton, Gerobatrachus, and batrachians (Fig.

13). Interestingly, Doleserpeton was coupled with Eocaecilia in the consensus tree, although this clade was not supported in the subsequent bootstrap analysis

(Fig. 13B). In this context, it is worth noting some specific points of resemblance shared by Eocaecilia and Doleserpeton, such as the shape of the palatine elements. However, the plesiomorphic condition of this element is uncertain in other modern amphibians. As no convincing synapomorphies connect

Doleserpeton and Eocaecilia to the exclusion of salamanders and frogs, this apparent relationship should be treated with caution.

This study has generally confirmed the importance of Doleserpeton to the discussion of the origins of modern amphibians. Although many questions concerning modern amphibian origins remain, the anatomy of this key form appears to support the hypothesis of a monophyletic origin of the recent amphibian orders from within Temnospondyli.

63 Bridging text 1.

As outlined in Chapter 1, the braincase and otic capsules of Doleserpeton are exceptionally well-preserved. The amount of detail found in these specimens allows for a detailed study of the otic region. This area is of prime importance, because the evolution of hearing is at the centre of the question of lissamphibian origins. Also, the well-preserved otic capsules of Doleserpeton can provide one of the most detailed views of the inner ear of any Paleozoic tetrapod.

Bolt and Lombard (1985) and Robinson et al. (2005) provided evidence for the essentially frog-like condition of the hearing system of temnospondyls. However,

Laurin (e.g. 1998b) has repeatedly expressed doubts as to whether temnospondyls had a functional tympanic hearing system.

Lombard (1977, 1979) and Clack (2002) pointed out essential differences of the inner ears of amniotes and modern amphibians. One important difference is the presence of the amphibian papilla in amphibians, but this feature is unlikely to be preserved in fossil forms. However, there are also fundamental differences in the perilymphatic system of the inner ear of these taxa, particularly in the position of the perilymphatic duct (called the helicotrema in amniotes).

Chapter 2 provides detailed descriptions of the anatomy of the inner ear and some new details of the middle ear of Doleserpeton. It also outlines some important new evidence for the discussion of the origins of modern amphibians.

64 CHAPTER 2.

The otic region of Doleserpeton (Temnospondyli) and its implications for the evolutionary origin of frogs

(Reference: Sigurdsen T, 2008. The otic region of Doleserpeton

(Temnospondyli) and its implications for the evolutionary origins of frogs.

Zoological Journal of the Linnean Society 154: 738-751)

ABSTRACT

There is increasing evidence that the Paleozoic temnospondyls had a frog-like tympanic hearing system. For this reason, the otic region of Doleserpeton is described and compared to modern anurans. The otic capsules are expanded laterally and ventrally relative to other temnospondyls. The opisthotic has a bulbous ventral region resembling the ventrolateral ledge in modern frogs. Two lateral processes are located on the paroccipital process. Comparison to the condition in modern anurans with a tympanic hearing system shows that this may have been the attachment site for the tympanic annulus. Parts of the osseous labyrinth are described. The inner ear shows numerous features resembling the condition found in frogs. These include strong evidence for the presence of a lissamphibian-type perilymphatic duct most closely resembling that of anurans.

This is the first time such a perilymphatic system is described in any Palezoic form. The posterior part of the braincase shows a jugular foramen closely associated with the perilymphatic foramen as in anurans. Although the

65 distribution of these traits among other temnospondyl groups remain little known, the sum of the evidence points to affinities between anurans and temnospondyls, and adds to the evidence for a close relationship between anurans and the Permian amphibamid Doleserpeton.

INTRODUCTION

The origins and interrelationships of the three extant orders of amphibians

(sometimes classified in the Lissamphibia) are still largely unknown and proposed hypotheses as to how they are related to the various Paleozoic forms differ considerably (e.g. Laurin and Reisz, 1997; Ruta et al., 2003; Carroll, 2007). This is partly due to their widely differing morphologies, making direct comparisons of the extant groups problematic. The Lower Permian amphibamid temnospondyl

Doleserpeton has been regarded as a possible relative of modern amphibians ever since it was first described (Bolt, 1969). This was mainly based on its possession of pedicellate bicuspid teeth and nearly monospondylous vertebrae in which the pleurocentra dominate. However, the exact nature of this relationship has been in doubt. In the original description, it was suggested that Doleserpeton may be a sister group to all Lissamphibia. As an alternative hypothesis, Laurin and Reisz

(1997) suggested that the link between modern amphibians and Paleozoic forms is to be found among the lepospondyls rather than the temnospondyls. However, this idea has not been generally accepted, and the origins of lissamphibians are still regarded as controversial. More recently, Carroll (2001, 2007) suggested that

Doleserpeton could be related to anurans specifically, since it has well-developed

66 otic notches as well as large orbits and interpterygoid vacuities, giving it a rather

anuran-like appearance. It should be noted that frog-like otic notches and stapes

appear to be primitive for temnospondyls, as these features are found in early

forms such as Dendrerpeton (Robinson et al., 2005) and Balanerpeton (Milner

and Sequiera 1994). In addition, the large interpterygoid vacuities can also be

found in other temnospondyls, including the branchiosaurids as well as some

small stereospondyls. However, Doleserpeton remains the most frog-like

amphibian known from the Paleozoic, with the possible exception of the

enigmatic form Gerobetrachus (Anderson et al., 2008b).

The increasing evidence for the presence of an essentially anuran hearing

system in temnospondyls (Lombard and Bolt, 1988, Robinson et al., 2005) may, if

confirmed, provide a link to the Paleozoic antecedents of modern anurans in the

form of a complex functional system. In this paper, the otic region of

Doleserpeton is described in detail in order to assess its affinities. This description

adds crucial information on the biology and anatomy of this important Paleozoic

temnospondyl, and some possible phylogenetic implications of the new data are pointed out. However, in order to fully take advantage of these observations, more information is needed from other Paleozoic forms. It is hoped that this study will spur interest in identifying similar traits in other fossil amphibians, adding to our knowledge of their relationships to each other as well as to modern forms.

67 MATERIALS AND METHODS

The Doleserpeton specimens all come from the Lower Permian (possibly

Leonardian) fissure fills of the Dolese Quarry, Oklahoma (near Richards Spur), and mostly from the D-concentrate (see Bolt, 1969). The specimens belong to the

Field Museum of Chicago, but are currently on loan to the Redpath Museum,

McGill University, Montreal. The material varies from completely disarticulated

elements to partly articulated skeletons. Many specimens were already prepared,

whereas others needed some additional preparation. The matrix is relatively soft,

and individual elements could be cleaned using soft brushes and water. In some

cases, old glue had to be removed using acetone. The attribution of the specimens

of the current study to the genus Doleserpeton is regarded as secure, as the

disarticulated elements were closely comparable to bones found in articulated

specimens such as the holotype. Information on the cranial anatomy of frogs,

salamanders and caecilians was obtained from the literature, as well as dissections

of Bufo americanus and Rana catesbeiana. In addition, cleared and stained specimens of Leiopelma archeyi, serial sections of the skull of Ascaphus truei and skeletal specimens of Necturus spp, Cryptobranchus alleganiensis and

Schistometum thomense were available for comparative studies.

68 DESCRIPTION

Otic region of extant amphibians

The otic region of modern amphibians has been described in some detail by

Gaupp (1904), Lombard (1977), Wever (1985), Mason (2007) and others, and

only features of importance to the present study will be discussed here. Among the extant amphibians, frogs alone have a tympanic hearing system as well as an operculum. Salamanders lack the tympanum, but retain the operculum, whereas

caecilians apparently lack both (Mason, 2007). A problem in assessing the

plesiomorphic condition in the ear region of frogs is the reduction of the

tympanum and related structures in the most primitive known members of the

Anura, notably the genera Leiopelma and Ascaphus. However, the primitive

salientian Triadobatrachus has a well-developed otic notch (Roček and Rage,

2000), which presumably supported a tympanum. In modern anurans with a

functional tympanic hearing system, the tympanum is supported by a

cartilaginous ring called the tympanic annulus. Ontogenetically, this structure first appears as a proliferation of the cells of the quadrate cartilage (Gaupp, 1904; Bolt and Lombard, 1985). It later migrates posteriorly and dorsally to its functional position beside the otic capsule. During , it is first sickle-shaped, before finally attaining a round or oval shape in the adult (Lombard and Bolt, 1985). An often overlooked feature of the annulus is the fact that it is attached not only to the squamosal and quadratojugal, but also to the otic capsule by means of a pair of short processes on the crista parotica (Fig. 1 A, B). The fusion of the upper rim of the annulus to the crista parotica is described by several authors (Gaupp, 1904; de

69 Beer, 1937; Sedra and Michael, 1959; Swanepoel, 1966). In most studies of recently metamorphosed frogs, the lateral end of the crista parotica is unossified, and this area is in synchondrotic continuity with the annulus. However, in more highly ossified specimens of Rana, the area of contact takes on the appearance of two ridges or processes where the thickened inner and outer rims of the annulus meets the crista parotica (Fig. 1A, B). These processes are usually cartilaginous, but may be partly ossified in some cases. In the adult anuran skull, the upper and posteriormost of these processes is attached to the outer thickened rim of the annulus, posterior to the area where the latter attaches to the squamosal. The other process, which is situated more ventrally and anteriorly, is attached to the inner rim (Fig.1 A, B). This latter process is situated close to the ascending process of the extrastapes.

That the condition described above is primitive may be questioned because of the situation in discoglossids. In this , the annnulus does not appear to be attached to the otic capsule as in most other anurans with a functional tympanum.

However, since the whole auditory system is frequently reduced in anurans (e.g.

Mason, 2007), the situation in discoglossids may be seen as partial reduction of the hearing apparatus. At least one of the processes is present on the crista parotica of Discoglossus (Van Zyl, 1950). It is situated close to the annulus, which looks unfinished dorsally. The annulus may vary from complete to sickle- shaped within the same species, and the taxonomic value of the reduced condition is therefore doubtful (see Sedra and Michael, 1959).

70

FIGURE 1. The otic region of Rana catesbeiana. A, right posterior part of braincase, ventral view. B, right side of the braincase, posterior view (based on several specimens).

71 The otic capsule of frogs is generally described as consisting of two parts, the

prootic and exoccipital, which normally fuse in the adult. The shape of the

capsule is unlike those of other lissamphibians not only in that it has a well- developed paroccipital process (crista parotica), but also a ventrolateral ledge

located ventral to the fenestra vestibuli (Figs. 1 and 2). This basic shape of the

otic capsule is remarkably consistent among the different anuran families,

including the primitive genera such as Leiopelma and Ascaphus (Stephenson,

1951), although the ventrolateral ledge appears to be somewhat less developed in the latter genus. The advanced pipids, such as the genus Pipa, are among the greatest exceptions with its highly derived skull structure, resulting in the main body of the otic capsule looking dorsoventrally compressed and expanded laterally. However fossil pipids, such as the unnamed specimen from the Upper

Cretaceous In Beceten locality (Fig. 2B), show that the pipid otic capsule primitively followed the same pattern as other anurans (Baez and Rage, 1998). In modern anurans the bulbous ventrolateral ledge is filled with the posterior part of the perilymphatic cistern which is connected to the operculum and the stapes footplate, although sometimes the lateral part of the cistern is set off as a lateral chamber (Wever, 1985).

The flattened nature of the sole specimen of Triadobatrachus makes it difficult to assess the exact shape of the otic capsule in early salientians. However, there is good reason to think that it had a ventrolateral ledge of the opisthotic, judging from the available evidence. First, the parasphenoid bone is almost identical to that of modern anurans. This includes the wide posterior part of the bone

72

FIGURE 2. The occiputs of extant and fossil amphibians. Not to scale. A, Conraua goliath (from Starck, 1979). B, unidentified pipid anuran from the Upper Cretaceous In Beceten locality (from Baez and Rage, 1998). C, Cryptobranchus alleganiensis (from Carroll and Holmes, 1980). D, Hynobius naevius (from Carroll and Holmes, 1980). E, Hypogeophis rostratus (from Carroll and Currie, 1975). F, Eocaecilia micropodia (from Jenkins et al., 2007).

73 supporting the otic capsules and posterior part of the braincase, and which

underlies the ventrolateral ledge ventrally in modern frogs as well as in

Doleserpeton. Also, there is a bulbous area on the otic capsules situated

immediately laterally to the occipital condyles. This structure is seen on both

sides of the specimen, and should probably be interpreted as a partly ossified

ventrolateral ledge (observations based on cast). This is probably the element that

was interpreted as an operculum by Estes and Reig (1973) but correctly reinterpreted as parts of the opisthotics by Rage and Roček (1989).

The braincase and otic capsules of salamanders are quite variable in external structure, and the interpretation here is based mainly on the primitive hynobiids and cryptobranchids (Figs 2C and D). The skulls of salamanders differ considerably from frogs in the otic region of the cranium, at least partly because of their lack of a tympanic hearing system. Since the stapes is not connected to a tympanum, it usually articulates with the squamosal or quadrate bones, leaving the operculum as the main functional ear ossicle (Duellman and Trueb, 1986;

Wever, 1985). In hynobiids and cryptobranchids the posterior portion of the otic capsule is less expanded laterally than in frogs, lacking the prominent crista parotica and ventrolateral ledge, although some other salamander families have retained, or re-evolved, a crista parotica. This feature is normally situated somewhat more anteriorly than in frogs. The occiput differs from that of frogs in that the occipital condyles are set further back, the whole occiput slanting postero- ventrally. Judging from the work of Lombard (1977) the inner ear of salamanders appears to be generally similar to frogs, except in lacking the superior

74

FIGURE 3. The inner ear of lissamphibians. A, Ascaphus truei, dorsal view, left side. B, Ascaphus truei, ventral view, right side. C, Hynobius nigricens, dorsal view, left side. D, Caecilia occidentalis, dorsal view, left side. Not to scale. From Lombard (1977).

75 perilymphatic space and in having a more ventral origin of the perilymphatic duct

(Fig. 3C).

In caecilians the otic capsule, exoccipital and parasphenoid elements are fused to form the os basale. The otic capsule is an evenly rounded structure in posterior view (Figs. 2E and F) and, as in primitive salamanders, the crista parotica and ventrolateral ledge are lacking. In extant forms, the otic capsule appears to be slightly expanded ventrally compared to that of anurans and urodeles, but this is not evident in the Upper Jurassic caecilian Eocaecilia (Fig. 2F). Whether or not

caecilians are derived from an animal with a Doleserpeton-type of otic capsule

must remain unresolved until further studies are undertaken.

The inner ear of lissamphibians has been described in detail by Gaupp (1904),

Lombard (1977), Wever (1985) and others (Fig. 3). The otic labyrinth resembles

that of the sarcopterygians (e.g. Platt et al., 2004) in that the lagena is a relatively

small and simple structure, and the semicircular canals are generally short relative

to the utriculus and more evenly curved than in most amniotes. In addition to the

basilar papilla, which is a sensory papilla located in the recessus basilaris close to

the lagena, there is the amphibian papilla in a recess located dorsomedially in the

posterior part of the sacculus. There are two basic kinds of perilymphatic systems

in the inner ear of extant tetrapods (Lombard and Bolt, 1979; Lombard, 1980). In

amniotes a duct, called the helicotrema, passes in front of the lagena connecting

the perilymphatic cistern (scala vestibuli) to the perilymphatic sac (scala

tympani), whereas in lissamphibians a similar duct, the perilymphatic duct or

periotic canal, is positioned posterior to the lagena (Figs. 3 and 4). These

76

FIGURE 4. The perilymphatic systems of Latimeria and tetrapods. Schematic figures of otic capsules oriented in such a way that anterior is to the right and lateral is towards the top of the drawings. Modified from Lombard (1980) and Fritzsch (1992).

77 conditions are generally considered to be independently evolved from a primitive sarcopterygian state (Lombard and Bolt, 1979; Fritzsch, 1992; Clack, 2002).

In anurans, the lateral origin of the perilymphatic duct is the superior perilymphatic space, which is a dorsal extension of the perilymphatic cistern (Fig.

3A). This space seems to be typical for the anuran inner ear, and it has been described in Ascaphus (Lombard, 1977), Bufo (Amano, 1931; Ecke, 1934), and

Rana (Gaupp, 1904). The descriptions by Lombard (1977) and Fritzsch and

Wake (1988) of salamanders and caecilians show a very different structure in this area. The perilymphatic duct exits the cistern more ventrally in salamanders before passing posterior to the lagena (Fig. 3C). In caecilians, the perilymphatic duct exits the cistern dorsally as in frogs, but apparently without a well-developed superior perilymphatic space (Fig. 3D). The perilymphatic cistern of salamanders and caecilians appears to be expanded relative to that of amniotes. However, it is usually not oriented as far laterally as in anurans, in which the expanded cistern fills the ventrolateral ledge. However, some salamanders, such as the primitive

Hynobius (Fig. 2D) give the impression of having a slight ventrolateral ledge.

Taken as a whole, the entire otic region of lissamphibians seems at first to present more differences than similarities between the extant orders. However, two important features are shared between these groups: the amphibian papilla and the posteriorly located perilymphatic duct.

78 Doleserpeton

The distinct morphology and often highly ossified condition of many of the bony elements, including the inner ear, confirms Holmes (2000) suggestion that

Doleserpeton does not constitute a juvenile stage of other known dissorophoids.

The otic capsules are large relative to those of other small temnospondyls, such as

Micropholis (Schoch and Rubidge, 2005), and they constitute a dominant part of

the occipital surface (Fig. 5). In disarticulated specimens, the otic elements are

commonly found as separate prootics and opisthotics. However, in specimen

FMNH UR1322 (Fig. 5) the two bones are fused without a discernable suture. In specimen FMNH UR1333 (Fig. 8) the exoccipital-basioccipital complex is fused to the opisthotic. The shape of the fenestra vestibuli is hard to study in detail,

since it is damaged or partially obscured by other elements in the articulated specimens. It is bordered by the opisthotic, prootic and, perhaps, the parasphenoid bones. It is unclear whether the parasphenoid constitutes its ventral edge, but it seems more likely that this area consisted of a cartilaginous part of the otic capsule proper. It appears to resemble the situation found in Eryops or Kamacops

(Schoch, 1999a, b), but the fenestra vestibuli of Doleserpeton is apparently larger relative to the otic capsule size.

Prootic

In anterior view (Fig. 6A), the prootic is similar to that of Dendrerpeton as described by Robinson et al. (2005). However, it lacks the groove for the internal jugular vein, the anterior surface of the bone being smoothly convex. This

79

FIGURE 5. Braincase and otic region of Doleserpeton. Specimen FMNH UR1322. A, occipital view. B, right side in ventral view.

80 convexity reflects the position of the anterior semicircular canal and its ampulla.

The prootic has two raised articulating surfaces dorsally, separated by the anterior

extension of the posttemporal fossa. The dorso-lateral articulating surface has the appearance of a curved oblique groove on the dorsal and anterior surface of the crista parotica. The anterior portion of this groove contacted the posterior edge of the epipterygoid, or a cartilaginous extension of this bone. The dorso-medial articulation is a short elevated area, which probably contacted the ventral surface of the . The ventral area of the prootic contacts the parasphenoid- basisphenoid complex immediately above the basipterygoid articulation. A prootic was separated from the matrix of specimen FMNH UR1314, and this specimen reveals a great deal of detail of the anterior osseous labyrinth (Fig. 6B).

The plane of view cuts through the otic capsule immediately posterior to the level where the lateral semicircular canal joins its ampulla. The lateral and posterior semicircular canals are shown, as well as the ampulla of the lateral canal. The utriculus is separated ventrally from the compartment of the sacculus by a crescentic osseous wall, which only just fails to separate the two compartments completely. The overall configuration at this plane of view conforms very closely to the situation found in anurans when viewing a transverse section through the otic capsule at the level of the ampulla for the lateral semicircular canal (e.g.

Gaupp, 1904, his Fig. 159; personal observation of sections through the skull of

Ascaphus). Some of the anterior pockets of the perilymphatic system would

presumably have been situated in this region of the otic capsule situated dorsally

in the space of the osseous labyrinth containing the utriculus, as well as an

81 anterior diverticulum of the perilymphatic cistern bordering the sacculus (Gaupp,

1904). However, no trace of an anteriorly situated perilymphatic duct

(helicotrema) could be found in this part of the osseous labyrinth.

The ventrolateral groove in the prootic is in the position of the branch of the

vena capitis lateralis (or “primitive head vein”) as seen in Ascaphus (Pusey,

1943). A more dorsal and medial branch of this vein was seen preserved as a result of calcification in the Tersomius specimen described by Carroll (1964).

Opisthotic

The opisthotic (Figs. 5 and 7) is a dominant element of the skull as seen in occipital view. Three bony areas project from the main body of this bone. These are the paroccipital process located dorsolaterally, the medial ramus situated above the exoccipital and foramen magnum, as well as the ventrolateral ledge, which is a bulbous area located ventral to the fenestra vestibuli. The medial ramus forms the roof of the foramen magnum, and it is fused with its fellow along the midline in specimen FMNH UR1322. The shape of the otic capsule of

Doleserpeton, as compared to other temnospondyls, is distinct because of the prominent ventrolateral ledge (pseudobasal process of Wever, 1985). It is not fully ossified ventrally, but the position of the posteriormost parts of the parasphenoid may indicate its shape in life. The jugular foramen is visible between the opisthotic and exoccipital elements (Figs. 5A and 7A).

In the well-preserved and mature specimen FMNH UR1322 the paroccipital process has two short lateral processes; one is located at the posterodorsal apex of

82

FIGURE 6. Right prootic of Doleserpeton. Specimen FMNH UR1314. A, anterior view; B, posterior view. Sharp edges are drawn in solid lines, whereas impressions and concavities of the bone are stippled.

83 the paroccipital process, whereas the other is situated slightly more ventrally and

anteriorly (Fig. 5A, B). The dorsal ridge contacted the tabular bone, and both

ridges may have been attached to a tympanic annulus (see discussion below). In

posterior view, there is a convex area at the centre of the bone, immediately

lateral and dorsal to the articulation with the exoccipital. This corresponds to the

position of the posterior semicircular canal and its ampulla.

Internally, the best preserved opisthotics (e.g. specimen FMNH UR1317) show

several details of the posterior osseous labyrinth (Fig. 7B). The plane of view cuts

through the posterior and lateral semicircular canals twice, due to the curve of

these canals at the posterior end of the membranous labyrinth. Anterior to the

posteriormost portions of the semicircular canals, and positioned immediately

medial to the fenestra vestibuli, is the perilymphatic cistern, part of the posterior

wall of which has been preserved. A portion of the perilymphatic cistern ascends dorsally and medially to form the superior perilymphatic space. There is also a small indentation in the bone located medially and ventrally to the perilymphatic

cistern. This probably marks the position of the recess for the basilar papilla.

The opisthotics show a slight ontogenetic range, in that the smaller specimens

show little or no preservation of the lateral ridges, less developed ventrolateral

ledge, and medial rami of the opisthototic that are not fused along the midline.

Thus, specimen FMNH UR1317 (Fig. 7) is apparently less mature than the only

slightly larger FMNH UR1322 (Fig. 5). This may also reflect interspecific

variation, although part of the difference may be due to damage to FMNH

UR1317.

84

FIGURE 7. Right opisthotic of Doleserpeton. Specimen FMNH UR1317. A, posterior view; B, anterior view. Sharp edges are drawn in solid lines, whereas impressions and concavities of the bone are stippled.

85

Exoccipital and basioccipital

The exoccipitals are fused to the basioccipital such that it is impossible to tell

where one ends and the other begins. In mature specimens this complex is also

partially fused dorsally to the medial ramus of the opisthotic, as in Edops (Romer

and Witter, 1942).

The exoccipital and basioccipital bones form the lateral and ventral walls of the foramen magnum (Figs. 5 and 8). Dorsally and laterally, the exoccipitals contact the opisthotics. The external jugular foramen opens between the two bones. The exoccipitals also form the two occipital condyles, which are slanted obliquely postero-laterally. A tiny foramen, probably for cranial nerve XII, pierces through the bone internally half way between the foramen magnum and the fissura metotica, and opens externally immediately laterally to the condyle (Figs. 5 and

8).

In Doleserpeton, there are two or three foramina in the region of the fissura metotica as seen in medial view (Fig. 8). The postero-dorsal foramen, called the metotic foramen, is equivalent to a similar foramen found in Edops (Romer and

Witter, 1942), Tersomius (Carroll 1964) as well as some amniotes (e.g. Rieppel,

1980), and it probably accommodated cranial nerves X and IX. Ventrally and anteriorly to this, there is a large foramen only rarely preserved in other temnospondyls. It is very similar to the anuran jugular foramen in shape and

position, but it seems to be subdivided so that the antero-ventral portion could be

regarded as a separate foramen, the dividing wall possibly being finished in

86

FIGURE 8. Posterior part of the braincase of Doleserpeton in medial view. Specimen FMNH UR1333. Part of the anterior edge of the jugular foramen has been reinforced with glue.

87 cartilage in life. This ventral part of the jugular foramen corresponds closely to the region where the perilymphatic duct communicates with the jugular foramen in modern anurans. It is therefore interpreted as the perilymphatic foramen. In the medial wall of the otic capsule, there are traces of the posterior acoustic foramen which would have been located immediately anterior to the perilymphatic foramen, but most of this region of the skull is missing.

DISCUSSION

Allowing for the reduction and loss of several dermal elements in the anuran skull, the overall resemblance of the occiput of Doleserpeton to that of frogs is striking (Figs. 1B, 2A and 5). In anurans, the often trough-like area of the dorsal part of the otic capsule between its dorso-lateral tip and the top of the frontoparietal may be a remnant of the posttemporal fossa of temnospondyls, although this feature is less clear in some species, depending on the orientation of the paroccipital process. The posttemporal fossa of early tetrapods is thought to have been a point of insertion for the epaxial musculature (e.g. Romer and Witter,

1942; Clack, 1998). In the modern frogs, the longissimus dorsi muscle has two heads inserting at the back of the skull (Gaupp, 1904; personal observation). The medial head inserts dorsolateral to the foramen magnum, in a depression at the posteriormost part of the frontoparietals. The other head inserts slightly more laterally at a prominent ridge on the otic capsule (Fig. 1A, B). The latter ridge is absent in Doleserpeton, and the edges of the posttemporal fossa presumably provided an adequate insertion for the muscle. The fact that the anuran ridge is

88 situated close to where the posttemporal fossa is found in temnospondyls suggests that the ridge may have evolved in order to accommodate the epaxial musculature in frogs when the fossa was reduced. In anurans, the dorsal aspect of the otic capsule may have increased importance as an attachment site for jaw muscles relative to the situation seen in temnospondyls (Carroll and Holmes, 1980).

In large temnospondyls, and particularly the stereospondyls, the exoccipital is a more dominant part of the occiput than the situation seen in Doleserpeton. It is often fused to, and sometimes covers, most of the posterior aspect of the otic capsule, resulting in a more laterally directed jugular foramen. Although the otic capsule of smaller forms can be expected to be larger relative to body size, this is clearly not the whole explanation for the frog-like shape of the otic elements in

Doleserpeton, which is to a large extent defined by the presence of a ventrolateral ledge of the opisthotic. This feature is similar in Doleserpeton and anurans not only in shape and position, but also in ossification pattern. The lower edge of the ledge is slow to ossify in both, often leaving an unossified strip between the otic capsule and parasphenoid. As pointed out above, the ventrolateral ledge seems to be present universally in fossil and extant frogs. Since Holmes (2000) suggested that many of the lissamphibian traits of Doleserpeton could be due to its small size, it is perhaps significant that the similarities of the otic capsules between

Doleserpeton and anurans persist even in some very large modern anurans such as

Rana catesbeiana (Fig. 1A, B), and Conraua goliath (Fig. 2A). Although the otic capsules of other small amphibamids such as Amphibamus are poorly known, related forms such as Micropholis (Schoch and Rubidge, 2005) and Tersomius

89 (Carroll, 1964) as well as other temnospondyls such as Acheloma (personal observation) and Perryella (Carlson, 1987) all seem to lack the well-developed ventrolateral ledge that makes the posterior shape of the otic capsules of

Doleserpeton and anurans so distinctive. It is also worth noting that the lateral expansion of the otic capsule and the ventrolateral ledge are not found in descriptions of lepospondyls (e.g. Carroll et al., 1998) nor early amniotes (e.g.

Clark and Carroll, 1973) both of which are frequently very small. The absence of the ventrolateral ledge in other small Paleozoic tetrapods, as well as its presence in very large modern frogs, makes it unlikely that this feature is related solely to the small size of the animal, and suggests rather that it may represent a synapomorphy of Doleserpeton and anurans. However, it should be mentioned that the ventral aspect of the opisthotic is frequently poorly ossified in temnospondyls, and therefore a cartilaginous ledge could have been present in other members of this group, as suggested by the position of the parasphenoid in some forms, such as the primitive Dendrepeton (Robinson et al., 2005).

Therefore, a ventrolateral ledge may have been present in primitive temnospondyls, but probably in less developed form than in Doleserpeton. In very large temnospondyls, such as Edops and Eryops, this incipient ventrolateral ledge may have become indistinct due to the smaller relative size of the otic region and the larger exoccipitals that cover most of the posterior aspect. Thus, there seems to be a complex relationship between size and morphology of the otic capsule within the temnospondyli. However, the important point to consider is that the ventrolateral ledge of the opisthotic is apparently never as pronounced within

90 other members of this group as it is in Doleserpeton. It is interesting to note that

Robinson et al. (2005), basing their observations on stapes morphology, suggest that Doleserpeton had an expanded perilymphatic cistern relative to the primitive condition seen in Dendrerpeton. This conclusion is confirmed here.

As for a functional interpretation of the presence of well-developed crista parotica and ventrolateral ledge, it seems that a clue may lie in the fact that although both are present in most families of frogs, they seem to be somewhat reduced in forms that lack a functional tympanic hearing system such as Ascaphus and Pipa. It is possible that the expanded ventrolateral ledge is necessary in order to have a functional operculum as well as a stapedial hearing system. The fact that both are incorporated in the fenestra vestibuli, which is connected to the perilymphatic cistern, may call for an enlargement of the latter, which fills most of the ventrolateral ledge. If this is the case, then it is likely that the well- developed ventrolateral ledge of Doleserpeton points to the presence of an operculum. The presence of an operculum in this genus is made all the more likely when one considers the common ancestry of frogs and salamanders, which probably dates back to the Permo-Carboniferous dissorophoids (e.g. Carroll,

2007), a group that includes the amphibamids. Since both frogs and salamanders have an operculum, then this feature was very likely present in their last common ancestor. If a dissorophoid origin of salamanders and frogs is accepted (Carroll,

2007; Ruta and Coates, 2007), we would expect that the operculum should have been present in at least some dissorophoids. However, no trace of a lissamphibian-type operculum was found in the otic region of Doleserpeton. This

91 is perhaps not surprising, as it is usually cartilaginous in anurans. If an operculum was indeed present in Doleserpeton (also suggested as a possibility by Bolt and

Lombard, 1985), it would presumably have been smaller than that of most modern frogs, judging from the shape of the opisthotic. It would be particularly interesting to know if this feature is present in the branchiosaurid dissorophoids. This group has sometimes been suggested as a close relative of caudates (Carroll, 2007), and it shares some unique ontogenetic features with salamanders (Fröbisch et al.,

2007).

The osseous labyrinth of Doleserpeton is very similar to that of anurans. This includes the relative positioning of the utriculus, sacculus and semicircular canals, and the presence of a crescentic wall between the utriculus and sacculus. In some respects, Doleserpeton resembles immature anuran specimens more than adults. A good example is the way the internal aspect of the prootic seems to be more comparable to the sections made from immature anurans (Gaupp, 1904), than to the more mature specimens sectioned by Wever (1985). However, the inner ear of

Doleserpeton is more highly ossified than most adult anurans and temnospondyls, and so cannot be seen to represent a juvenile stage.

The helicotrema of amniotes is generally visible in sections through the otic capsule immediately posterior to where the horizontal canal connects to its ampulla (e.g. De Burlet, 1934; Hamilton, 1964). The lack of a helicotrema in the prootic of Doleserpeton, and the generally frog-like configuration of the osseous labyrinth, suggests that it had a lissamphibian type of perilymphatic system.

Stronger evidence is found in the best preserved opisthotic (Fig. 7), which reveals

92 the presence of a superior perilymphatic space (sensu Wever, 1985). It is a clear indication of the presence of a posterior perilymphatic duct of the lissamphibian type, as the superior perilymphatic space constitutes the area where the perilymphatic duct exits the cistern. Indeed, in the works of Ecke (1934) and

Amano (1931) the superior perilymphatic space is viewed as part of the perilymphatic duct proper. As pointed out above, salamanders and caecilians show a different structure in this area. The presence of the superior perilymphatic space is thus a strong indication that Doleserpeton had a lissamphibian-type perilymphatic duct most closely resembling that of frogs.

In modern anurans the perilymphatic foramen is closely associated with the internal jugular foramen, which is part of the embryonic fissura metotica. The latter also accommodates a branch of the jugular vein, as well as the vagus (X) and glossopharyngeal (IX) nerves (Starck, 1979). It is bordered posteriorly by the contact with the occipital arch (roughly equivalent to the exoccipital-basioccipital complex in Doleserpeton). In modern frogs, the medial view of the braincase reveals the superior and inferior perilymphatic foramina, which usually open into the antero-ventral part of the enlarged jugular foramen. However, this region of the skull of adult frogs is highly variable, even at the individual level (Case,

1979). The fissura metotica is subdivided by a septum in early ontogeny, separating the jugular foramen proper from the perilymphatic foramina (de Beer,

1937; Case, 1979). Later in development, this septum usually disappears, but it is retained in some forms, such as Leiopelma (Stephenson, 1951). The close association of the perilymphatic and jugular foramina in Doleserpeton may

93 therefore be another trait that it shares with anurans. However, our lack of knowledge of this trait in other Paleozoic forms limits the ability to judge this feature. This region is often poorly ossified in other temnospondyls, but the shape of the posterior border of this unossified area in the braincase of Edops, as described by Romer and Witter (1942), indicate that it marks the posterior edge of a jugular foramen roughly similar to that of Doleserpeton. The position of the perilymphatic foramen in this form cannot be judged with any certainty, as is also the case in Eryops (Sawin, 1941). In salamanders, the jugular foramen is usually

slit-like and widely spaced from the more anteriorly positioned perilymphatic

foramen (Francis, 1934; Rose, 2003). The situation in caecilians appears to be

similar to that of salamanders (de Beer, 1937; Brand, 1956). Amniotes have a slit-

like jugular foramen, often situated near the perilymphatic foramen (Starck, 1979;

Rieppel, 1980). The region of the internal jugular foramen is poorly ossified in the

microsaur (Romer, 1969). The situation seen in Doleserpeton is again

more like that of frogs than any of the other extant groups of tetrapods in that it

has an enlarged jugular foramen closely associated with the perilymphatic duct.

Furthermore, the position of the perilymphatic foramen supports the above

suggestion that Doleserpeton had a frog-like perilymphatic system in general.

Bolt and Lombard (1985) pointed to important evidence for the attachment of

the tympanum within dissorophoids. However, they overlooked the additional

evidence that the structure of the otic capsules can provide in this regard. There is

a close correspondence between the lateral processes of the anuran otic capsule

and the ossified lateral processes found in Doleserpeton (Figs. 1 and 5). Although

94 the uppermost of the two processes contacted the tabular, this does not exclude the possibility that it also served as attachment for the annulus. It may have resembled the way the dorsal ridge is overlapped by the squamosal and serving as attachment for the annulus in Chiromantis (Swanepoel, 1966). The presence of attachment sites for the annulus adds to the growing evidence that temnospondyls had a frog-like tympanic hearing system.

Cannatella (2006) points out that the study of the distribution of a tympanic hearing system among extant frogs would lead to the assumption that the system evolved within Anura, due to the lack of this system in primitive frogs such as

Ascaphus and Leiopelma. However, the fossil evidence indicates that this hearing

system is ancient, dating back to primitive temnospondyls such as Dendrerpeton

(Robinson et al., 2005). The new evidence from Doleserpeton shows that by the

Lower Permian, the tympanic hearing system of temnospondyls had reached

essentially the level of modern frogs.

Features of the otic region shared by Doleserpeton and anurans now include the

stapedial morphology (Lombard and Bolt, 1985), the well-developed otic notch,

the tympanic annulus attached to the crista parotica (paroccipital process), the

expanded perilymphatic cistern, and the posteriorly situated perilymphatic duct

originating from a superior perilymphatic space (Table 1). Of these, the otic notch

and stapedial morphology is also shared with primitive temnospondyls such as

Dendrerpeton (Robinson et al., 2005). The similarities of the otic notch make it likely that the annulus was attached in much the same way in other dissorophoids as in Doleserpeton (Lombard and Bolt, 1985). However, whether the annulus was

95 directly connected to the crista parotica in other temnospondyls remains to be

seen. The ventrolateral ledge and associated perilymphatic cistern seems to be

closer to the anuran state in Doleserpeton relative to the situation seen in

Dendrerpeton and other temnospondyls, whereas the configuration of the

perilymphatic duct is entirely unknown in other Paleozoic forms. However, when

compared to extant forms, the inner ear labyrinth and posteriorly positioned

perilymphatic duct of Doleserpeton correspond to those of lissamphibians, and

among the latter it clearly resembles anurans most of all in having a superior

perilymphatic space as well as perilymphatic and jugular foramina closely

associated with each other. Add to this the other features known to be shared by these two groups (reviewed by Carroll, 2007) and the case for a close relationship between anurans and dissorophoid temnospondyls, and Doleserpeton specifically,

becomes very strong.

The gradual appearance of the anuran hearing system within the temnospondyls

seems to have involved a combination of features that can be linked to the

evolution of smaller body size, perhaps as a result of , as well as the

appearance of new features. However, more extensive descriptions of both fossil

and extant taxa are needed in order to fully comprehend the evolutionary steps

leading to the hearing systems of modern frogs, salamanders, and caecilians.

96 TAXON Ventral hinge Annulus on Ridges on Ventrolateral Perilymphatic Enlarged jug. Amphibian

on stapes squamosal crista ledge duct foramen papilla

footplate (developed from parotica associated with

quadr. cartilage) per. foramen Salientia Present Present present present posterior present present

Caudata Absent Absent absent intermediate posterior Absent present

Apoda Absent Absent absent absent posterior Absent present

Amniota Absent Absent absent absent anterior Absent absent

Doleserpeton Present Present present present posterior Present ?

Dendrerpeton Present Present ? intermediate ? ? ?

Table 1. Some important traits of the otic region of amphibians and amniotes. In some taxa, the ventrolateral part of the otic capsule shows a slight ledge, termed “intermediate” here. The tympanic annulus is inferred from the presence of a squamosal notch and dorsal process of the quadrate (Bolt and Lombard, 1985).

97 Bridging text 2.

The potential of the otic capsules and hearing system for providing evidence for

the evolutionary affinities of modern amphibians has been mentioned by a

number of authors (Bolt and Lombard, 1985; Clack, 2002; Robinson et al., 2005).

However, there has been far less focus on the postcranial system, despite its

obvious functional importance. Milner (1988) and Carroll (2007) discuss some

important postcranial features in light of the Temnospondyl Hypothesis and

Polyphyletic Hypothesis respectively. It is notable that the limb bones have

received little attention in this discussion. Part of the reason for this is the highly

specialized nature of the limb morphology of the modern orders, as anurans have adopted a jumping locomotion, whereas caecilians lack limbs. It would therefore seem that any effort to find morphological similarities would be in vain. However,

with the description of the limbed stem caecilian Eocaecilia (Jenkins et al., 2007),

and the new data available from Doleserpeton (Chapter 1), a comparison of limb

structures is possible. Chapter 3 focuses on the forelimb, and specifically the

humerus and elbow joint of amphibians.

98 CHAPTER 3.

The lissamphibian humerus and elbow joint, and the origins of modern amphibians

(Reference: Sigurdsen T, Bolt J. 2009. The Lissamphibian Humerus and Elbow

Joint, and the Origins of Modern Amphibians. Journal of Morphology [in press])

ABSTRACT

The origins and evolution of the three major of modern amphibians are still a source of controversy, and no general consensus exists as to their relationship to the various known Paleozoic taxa. This may indicate that additional character complexes should be studied in order to resolve their phylogenetic relationship.

The salamander elbow joint has been fundamentally misinterpreted in previous morphological descriptions. In caudates and anurans, both the radius and ulna

(fused in anurans) articulate with the characteristically large capitulum (radial condyle), although part of the ulnar articulating surface fits into to the smooth trochlear region. The salamander “ulnar condyle” of previous descriptions, is in fact the entepicondyle. The condition seen in batrachians (i.e., salamanders and frogs) may be a lissamphibian synapomorphy, because the elbow region of the primitive fossil caecilian Eocaecilia resembles those of frogs and salamanders. In addition to the large and bulbous capitulum, all lissamphibian humeri lack an entepicondylar foramen, and possess a distally pointing entepicondyle, a low and

99 rounded ectepicondyle, and an elongated shaft. These characters are identified in

key fossil forms to assess the support for the different hypotheses proposed for the

evolutionary origins of lissamphibians. Temnospondyli is the only group of early

tetrapods that shows a progressive evolution of lissamphibian traits in the humerus and elbow joint. Furthermore, among Paleozoic taxa, the dissorophoid temnospondyl Doleserpeton annectens is the only taxon that has the full set of humeral features shared by all lissamphibians. These results add support for the theory of a monophyletic origin of lissamphibians from dissorophoid temnospondyls.

INTRODUCTION

Contemporary phylogenies of modern amphibians vary considerably with respect to their relationship to both living amniotes and fossil forms (reviews in

Schoch and Milner, 2004; Anderson, 2008). Lissamphibians have been proposed as a monophyletic group with the fossil amphibamid dissorophoid Doleserpeton as closest sister taxon (Bolt, 1969; Ruta and Coates, 2007), or as a monophyletic group allied to the Paleozoic lepospondyls (Laurin and Reisz, 1997). Furthermore, lissamphibians have been suggested to be a polyphyletic assemblage, with frogs and salamanders closely related to dissorophoids, and caecilians constituting a sister group to the microsaurian lepospondyl genus Rhynchonkos (Carroll, 2007;

Anderson, 2007). Molecular data tend to confirm the monophyly of lissamphibians relative to amniotes and to show that batrachians (frogs and salamanders) form a monophyletic group, with caecilians as their closest sister

100 taxon (e.g., San Mauro et al., 2004; Igawa et al., 2008; review in Anderson,

2008).

A number of traits have been proposed as synapomorphies linking lissamphibians with various Paleozoic groups (e.g., Parsons and Williams, 1962;

Schoch and Milner, 2004; Caroll, 2007). However, the structure of the

appendicular skeleton is rarely discussed in this context, other than in connection

with the evolution of jumping in modern frogs (e.g., Jenkins and Shubin, 1998).

The new and intriguing amphibamid Gerobatrachus presents some evidence for

the evolution of the amphibian postcranial skeleton (Anderson et al., 2008b), but

many of its features are poorly preserved.

Previous descriptions of the forelimbs of modern amphibians give the

impression of completely disparate anatomies, with few, if any, features being

shared by frogs, salamanders, and caecilians. When we dissected the elbow joints

of modern amphibians, we realized that the anatomy and function of the forelimb

of modern salamanders have been misinterpreted by previous workers. Modern

interpretations of salamander anatomy usually are based on the descriptions of

Francis (1934). Although his figures are generally accurate, his interpretation of

the elbow area is misleading. Francis’ description is at odds with available

descriptions of anurans (e.g., Gaupp, 1904; Ritland, 1955; Maglia et al., 2007),

even though anurans and caudates have many forelimb features in common. This

has led to incorrect interpretations, making the anatomy of the forelimb of

anurans seem more divergent from that of salamanders than it really is (e.g.,

Duellman and Trueb, 1994; Báez and Basso, 1996). In contrast, the brief

101 descriptions by Miner (1925) and Evans (1946) are fairly accurate, but both lack detail and fail to point out the uniqueness of the lissamphibian elbow joint. Here, we describe the unique features of the lissamphibian elbow joint, including the fossil limbed caecilian Eocaecilia micropodia. We then compare the relevant fossil forms to modern amphibians, adding much-needed information to the discussion of lissamphibian origins. Although there is no current consensus as to the relationships and origins of extant amphibians, the term “Lissamphibia” is used here to denote modern amphibians, irrespective of the question of monophyly.

MATERIALS AND METHODS

Modern and fossil tetrapod taxa were sampled as broadly as possible. Within

Lissamphibia, basal and derived taxa were examined based on both traditional

(Duellman and Trueb, 1994) and recent (Frost et al., 2007) phylogenies. Other

tetrapods were sampled with emphasis on stem taxa. The Lower Triassic stem-

anuran Triadobatrachus massinoti was studied with the kind help of Jean-Claude

Rage (Muséum National d'Histoire Naturelle, Paris). Primitive fossil

lissamphibians, most notably Eocaecilia micropodia and Prosalirus bitis, as well

as numerous Paleozoic fossils representing most of the major tetrapod groups,

were available for study at the Museum of Comparative Zoology, the Field

Museum of Natural History, and the Redpath Museum. For a full list of

specimens, see Appendix 7.

102 In the following descriptions, as in most descriptions of modern tetrapods, the humeri are assumed to be directed posteriorly and more or less parallel to the axis of the body for purposes of description. Thus, the capitulum (= radial condyle, capitellum or humeral ball) is situated on the ventral and lateral side of the distal end of the humerus.

Measurements were obtained of the humeri from 19 modern amphibians

(representing 15 families), and 30 non-lissamphibian tetrapods (Appendix 7).

Some measurements were taken from published descriptions. The humeral length

(L), distal width (W), and transverse capitulum size (C) were measured for all specimens. Most humeri were measured with calipers, but the smallest specimens were measured using outline drawings made with a camera lucida on a microscope, adjusting for the magnification. All measurements were log- transformed before the analysis, which was undertaken using PAST

(Paleontological Statistics) version 1.90. In order to test the hypothesis that the capitulum size is larger in lissamphibians than in other tetrapods, we ran an analysis of covariance (one-way ANCOVA). The capitulum size of batrachians and amniotes was analyzed, adjusting for the covariance of the distal humeral width. This is justifiable because the capitulum is located on the distal end of the humerus. We also ran an analysis adjusting for the humeral length. Finally, we analyzed the variables separating the lissamphibian humeri from other taxa, using a discriminant analysis first for batrachians versus amniotes, and then on batrachians versus all “non-batrachians” of this study.

103 Institutional Abbreviations

CAS, California Academy of Sciences, San Francisco, California; FMNH, Field

Museum of Natural History, Chicago, Illinois; MCZ, Museum of Comparative

Zoology, Harvard, Massachusetts; MNA, Museum of Northern Arizona,

Flagstaff, Arizona; RM, Redpath Museum, Montreal, Quebec, Canada.

RESULTS

Humerus and Elbow Joint of Salamanders

The salamander humerus is slender (e.g., Ambystoma) to moderately robust

(e.g., Andrias), and no foramina pierce it. As in many other tetrapods, the proximal and distal ends are expanded and twisted about 90º relative to each other. The rounded humeral head is more or less confluent with the deltopectoral crest (= crista ventralis humeri). There is a dorsal process on the extensor surface of the humeral shaft, located at about one third of the length of the bone from the proximal end. This is the insertion point of the subscapularis muscle. The usually cartilaginous distal end has two prominent protrusions (Figs. 1A, B, 2A).

According to Francis (1934), the most medial of these is the ulnar condyle.

However, it is better described as the entepicondyle because of its function, as well as its position, which reflect that of the entepicondyle described in other tetrapods (Romer, 1956; Coates, 1996; Pawley and Warren, 2006). The condyle in question is situated medially on the distal end of the bone, and it is directed distally (Figs. 1A, 2A). Although its lateral edge forms part of the trochlear region

(described below), the condyle itself does not function as the articulation site of

104

FIGURE 1. The elbow joint of salamanders (some features are removed in the drawings for clarity). A: Andrias davidianus, dorsal view. B: Ambystoma tigrinum, lateral view. In both cases, the articular capsule has been opened and the joint surfaces slightly separated. Note that the main articulating surface of the ulna faces the capitulum (radial condyle).

105 the ulna. Rather, it is the attachment site for tendons of the flexor musculature of the carpus, as are the entepicondyles of other tetrapods. The capitulum (radial condyle) is large and hemispherical (ratio of distal end width to capitulum width about 0.5; Appendix 7), and it lies immediately lateral to the trochlear region. It extends distally and ventrally from the main body of the humerus, and serves as the articulation site of both the radius and the ulna (Fig. 1A, B). The smooth concavity between the capitulum and entepicondyle is the trochlea (Fig. 2A), which articulates with a rounded facet on the ulna. Unlike the trochlear regions of most amniotes, it is rather indistinct and lacks convexities. Thus, there is no ulnar condyle proper, contrary to the description by Francis (1934). The ectepicondyle

(epicondylus lateralis), is the attachment site for the extensor muscles of the lower arm and . It is a low, rounded ridge situated lateral and slightly proximal to the capitulum. The ulna has a partially cartilaginous olecranon, which is the attachment site of the anconaeus musculature (= triceps; Walthall and Ashley-

Ross, 2006). The proximal ulnar articulation facet consists of a large concave surface laterally, and a smaller rounded facet situated more medially (Fig. 2B).

The large concavity articulates entirely with the capitulum, whereas the rounded edge fits into the trochlear region. This structure has been confirmed in all salamander species studied.

Salientians (Anurans and Proanurans)

The anuran humerus differs from that of salamanders in having a longer deltopectoral crest (Fig. 2C). The shaft is usually slightly sigmoidally curved. The

106

FIGURE 2. Tetrapod forelimb elements. All humeri are from the right side and seen in ventral (flexor) view. A: Ambystoma tigrinum, humerus. B: Ambystoma tigrinum, zeugopodium and autopodium in anterior (radial) view. C: Rana catesbeiana, humerus. D: Rana catesbeiana, radio-ulna in anterior (radial) view . E: Eocaecilia micropodia, humerus (based on MCZ 9163 and 9169). F: Lacerta sp. humerus.

107 proximal and distal ends are often more highly ossified than in salamanders. The

humerus is strikingly similar to that of salamanders in having a large capitulum, a

distally directed entepicondyle (epicondylus ulnaris of Gaupp, 1904), as well as a

low, rounded ectepicondyle. In large, highly ossified species, such as Rana

catesbeiana, the entepicondyle has a short medial process for the insertion of the

flexor carpi musculature (Fig. 2C).

The fused radio-ulna of frogs (Fig. 2D) articulates with the capitulum (the humeral ball of Maglia et al. 2007), which is even larger than that of most salamanders. Part of the ulnar component of the radio-ulna reaches the trochlear

region between the capitulum and entepicondyle. This medial ulnar edge is rounded to fit into the trochlear region of the humerus, but the major articulation facet of the ulnar part of the radio-ulna faces the capitulum. Apart from the fusion of the radio-ulna, this is exactly the situation seen in salamanders.

The earliest remains of crown-group anurans are found in the Jurassic. The fossils of such otherwise informative forms as Vieraella and Notobtrachus have no preserved capitulum (Roček, 2000; Báez and Nicoli, 2004). Furthermore, the humeri of Prosalirus bitis are only partly ossified along the outer edge of the capitulum (Shubin and Jenkins, 1995). However, the preserved parts of the latter taxon are complete enough to reconstruct the size of this condyle, which was of comparable size to that of modern anurans (Appendix 7).

The Triassic proanuran Triadobatrachus massinoti was recently redescribed by

Roček and Rage (2000). The humeral shaft of this form has a curvature and deltopectoral crest similar to those of most modern anurans. The distal and

108 proximal ends are incomplete, owing to lack of ossification. However, a large gap

distally indicates the presence of a capitulum which may have been similar in size

to that of modern anurans. This feature is confirmed in the presumably related

Polish form polonicus (Evans and Borsuk-Bialynicka, 1998), in

which a large capitulum is preserved in an otherwise fragmentary humerus.

Eocaecilia

Although modern caecilians lack limbs, important new evidence is available

from the Lower Jurassic form Eocaecilia micropodia (Jenkins et al., 2007). The

limb bones of this form are reduced in relative size, but well preserved. The

humerus is elongated, and lacks foramina (Fig. 2E). The entepicondyle is offset

slightly medially, but it comes to a sharp point distally as in modern frogs. The

trochlear area is smoothly concave as in salamanders, and the capitulum is large

and bulbous. The condylar size relative to the width of the bone is somewhat

smaller than in frogs, but resembles those of some hynobiid and dicamptodontid

salamanders examined (Appendix 7). The ectepicondyle is a low, rounded ridge

similar to those of anurans. The ulna has been described in detail by Jenkins et al.

(2007). As in salamanders, the proximal ulnar articulating area has two facets, the larger of which faces the radial side (i.e. facing the capitulum). Overall, the humerus and elbow joint exhibit the structure seen in batrachians, showing all the traits shared by salamanders and frogs. Unlike anurans, the humerus has a dorsal process similar to that of some salamanders (Jenkins et al., 2007). The limb bones

109 of Eocaecilia are highly ossified, despite being reduced in size. This may indicate that the limbs were still being used actively in locomotion.

Other Fossil Forms

Among the known Paleozoic forms, only the dissorophoids have a humeral morphology and an elbow joint resembling those of lissamphibians. A relatively

large capitulum seems to be typical of dissorophoid and eryopid temnospondyls

(Yates and Warren, 2000), although this trait may be present in some primitive

temnospondyls as well, judging by the unossified area of the captiulum of Edops

(unpublished data). A good example of the large temnospondyl capitulum can be

seen in the well-studied form Eryops, in which the capitulum dominates the distal

end of the humerus (Pawley and Warren, 2006). Stronger similarities to modern

amphibians can be found among the dissorophoids (Fig. 3A, B). Whereas the

humeri of trematopid dissorophoids such as Acheloma are remarkably similar to

those of Eryops (Olson, 1941), the humeri of appears smoother in

that it lacks the supinator process (DeMar, 1968). Although the bone of the latter

form is relatively short and robust, the distal end is, in fact, remarkably similar to

that of modern lissamphibians (Fig. 3B). This includes the distally directed

entepicondyle and rounded ectepicondyle, as well as the smooth concavity of the

trochlear region. The capitulum of Dissorophus is large relative to those of most

other early tetrapods, including Eryops, although its relative size is not as large as

those of most modern amphibians (Appendix 7). The medial process of the

entepicondyle resembles that of modern anurans (Figs. 2C, 3B). The described

110

FIGURE 3. Forelimb and humerus of dissorophoids. A: Doleserpeton annectens FMNH UR1321. B: Dissorophus multicinctus MCZ 4176. Illustrations are slightly restored based on other specimens.

111 humeri of Cacops (Williston, 1910a) resemble more poorly preserved specimens of Dissorophus, in which the distal end of the bone is incompletely ossified.

In amphibamid dissorophoids, the humeri are generally more slender than in other temnospondyls, perhaps partly due to their small size. This is most pronounced in the Lower Permian form Doleserpeton annectens. The humerus of this taxon is a relatively smooth bone lacking an entepicondylar foramen and supinator process. The capitulum is preserved as a mold in the matrix surrounding the distal end of the bone in the type specimen (FMNH UR1308). In a few specimens, such as FMNH UR1321, the large capitulum is preserved (Fig. 3A). It is a large, hemispherical structure situated next to a smoothly concave trochlear region. As in Dissorophus and modern amphibians, the entepicondyle is distally directed, and the ectepicondyle is low and rounded. Uniquely among Paleozoic tetrapods, the humerus of Doleserpeton possesses all the characters shared by anurans, salamanders, and Eocaecilia. In addition, the radius and ulna of

Doleserpeton resemble those of modern salamanders. Although the proximal parts are only partly preserved in these bones, the articulating surfaces of the ulna appear to be oriented similarly to those of salamanders.

Morphometric comparisons

Batrachians (frogs and salamanders) and amniotes are usually regarded as monophyletic groups (Frost et al., 2006; Carroll, 2007; Ruta and Coates, 2007; but see review by Anderson, 2008), and are therefore used for comparison to the other taxa in this study. The reduced major axis regression lines of the variables C

112

FIGURE 4. Analysis of the humeral data. A: The relationship between the distal humeral width to the size of the capitulum. Major axis regression lines are given for batrachians (black line) and amniotes (gray line). Black dot, batrachian; cross, ; gray box, lepospondyl; light gray diamond, temnospondyl; circle, stem tetrapod. B: Distribution of frequencies along the discriminant axis (defined in the text). Black columns: batrachians (frogs and salamanders), gray columns: “non- batrachian tetrapods”. The overlap of the columns is due to Eocaecilia and Doleserpeton.

113 and W for amniotes and batrachians are given in Fig. 4A. The ANCOVA gave adjusted means of log C as 0.52 for batrachians (N = 18) and 0.26 (N = 16) for amniotes (F = 90.15, p < 0.0001), demonstrating the larger relative size of the capitulum in anurans and caudates compared to amniotes. Adjusting for humeral length gave similar results, but with somewhat less extreme difference in adjusted means (0.49 and 0.29 respectively, p < 0.0001). The regression lines of the other groups did not pass the bootstrap test, but plotting the data points onto the regression lines (Fig. 4A) confirms the observation that Eocaecilia and

Doleserpeton both closely approximate the batrachian condition. A possible objection to using ANCOVA is that the slopes of the regression lines of amniotes and batrachians are significantly different (p < 0.01). However, a similar pattern is seen when the size of the capitulum relative to the size of the bone are compared in lissamphibians and other tetrapods. Thus, the ratio of capitulum to humeral width is significantly larger in lissamphibians than in other tetrapods (mean ratio of capitulum size to distal humeral width for lissamphibians: 0.54, others: 0.33.

Student’s t test gives t = 8.99, p < 0.0001).

After dividing the taxa into two groups consisting of batrachians and “non- batrachians” respectively, the discriminant analysis (Fig. 4b) applied to the log- transformed measurements identified 95.9% of the taxa as being in the correct group (p < 0.001). The discriminant function is defined by: v = (21.90 log C) +

(5.93 log L) – (27.72 log W). The analysis identified Eocaecilia and Doleserpeton

(included for purposes of the analysis among the “non-batrachians”) as belonging to the same group as batrachians.

114 DISCUSSION

The Lissamphibian Humerus and Elbow Joint

Our results indicate that the specialized elbow region described above may be a

shared trait of modern amphibians. The elbow joint consists of a large capitulum

articulating with both radius and ulna, although the latter bone also has a rounded facet articulating with the smooth and relatively featureless trochlear region. A low ectepicondyle and a distally directed entepicondyle are also present. Other traits of the humerus that are shared by these forms include the lack of humeral foramina and the elongation of the humeral shaft.

The elbow joint approximates a ball-and-socket joint rather than a hinge. Such a joint allows for considerable movement, the opposite of the situation seen in some primitive fossil tetrapods, such as , in which elbow movement was restricted (Jenkins, 1973). Manipulations of freshly dissected specimens indicate that the nature of the lissamphibian elbow joint may allow some fore-and-aft movement of the radius and ulna relative to the humerus, but this movement is

restricted by the medial edge of the olecranon fitting into the trochlear region. The

joint is clearly flexible enough to allow for the highly divergent morphologies and

modes of locomotion seen in anurans, caudates and primitive apodans.

This forelimb morphology contrasts markedly with the ancestral tetrapod

condition, in which the humerus is a short bone with an entepicondylar foramen, a

small capitulum and a separate ulnar condyle (Holmes, 1980; Carroll and Holmes,

2007). The entepicondyle of Acanthostega is large and medially (posteriorly)

directed, although it also has a distally directed edge, making this trait somewhat

115 uncertain in polarity (Coates, 1996). Within the Amniota, a distinct shaft evolved early, but stem amniotes still retain the small capitulum and the entepicondylar foramen (Reisz, 1980, 1981; Sumida, 1997). In modern , there is often a true ulnar condyle between the capitulum and the entepicondyle (Fig. 2F). In some amniotes, the capitulum forms part of the lateral edge of the trochlear region, and a facet of the ulnar articulating surface faces this medial area of the capitulum (Holmes, 1977; Jenkins, 1973). Others, such as many modern , might be said to have a continuous joint surface for the radius and ulna in the form of a wide trochlear-shaped area (Starck, 1979). However, none has an elbow joint in which the major articulating surfaces of both the radius and ulna face a large, bulbous capitulum. The morphology of the lissamphibian humerus and elbow joint thus seems to be unique among modern tetrapods.

Data analysis

Because of the extremely fragmentary nature of many of the fossil specimens used in this study, it is hard to assess the effects of body mass on the humeral features discussed. However, several features seem to indicate that body size is not a determining factor for the morphology of the elbow region. For example, the humeral morphology and elbow region of the smallest caudates resemble those of the giant Chinese salamander (Andrias davidianus), which is the largest modern amphibian. Conversely, the approximately similar-sized, but unrelated, Paleozoic forms Eryops and had strikingly different elbow regions, with the temnospondyl Eryops having a larger capitulum perhaps approaching the

116 lissamphibian condition. Warren and Snell (1991) pointed out the lack of size-

related variation among temnospondyl humeri, and that humeral morphological

differences were likely to reflect phylogenetic relationships. Our observations of

various temnospondyl humeri support this view. Also, the lissamphibian-like

humerus of the amphibamid temnospondyl Doleserpeton is not paralleled by any similar-sized amniote as far as is known. Nor is it apparently paralleled by microsaurs or other lepospondyls, although the often incompletely preserved fossils and highly variable morphologies of lepospondyls make this group difficult to interpret at present.

Both the ANCOVA and the discriminant function analysis confirm the distinctiveness of the batrachian humerus relative to that of amniotes, as well as to other tetrapods. Furthermore, the discriminant function analysis classified the humeri of Eocaecilia and Doleserpeton as belonging to the batrachian group. It

should be noted that the latter analysis was not robust when changes were

introduced by removing parts of the data. However, Eocaecilia and Doleserpeton

also appear closer to the batrachian regression line than any other tetrapods (Fig.

4A), confirming the close resemblance of the humeri of amphibamids, batrachians

and early caecilians.

The Origin of Lissamphibia

Apart from adding to, and correcting, our knowledge on the forelimb

morphology of amphibians, our results also provide valuable phylogenetic

information. Although it would be unwise to propose a novel phylogeny based on

117 one new character complex, it is informative to compare the distribution of the

traits considered here in the light of previously published phylogenies. A wide variety of fossil forms was examined, including lepospondyls, temnospondyls and

early tetrapods. A problematic feature of many of the relevant fossil forms, such

as many lepospondyls and amphibamids, is the lack of ossified condyles.

However, the position and, with less certainty, the size of the capitulum

sometimes can be inferred from the presence of a gap where the cartilaginous

condyle was located in life.

In recent phylogenies, temnospondyl dissorophoids, as well as lepospondyl

microsaurs and lysorophians, have been suggested as possible sister-groups to

one, two, or all modern lissamphibians (Laurin and Reisz, 1997; Vallin and

Laurin, 2004; Ruta and Coates, 2007; Carroll, 2007; Anderson et al., 2008b).

Some lepospondyls share certain features of the humerus with modern amphibians, such as elongation of the shaft and loss of the entepicondylar foramen, but these are relatively common in other tetrapods as well.

Among the lepospondyls, the microsaur has been described as

having a humerus fitting the lissamphibian descriptions above (Gregory et al.,

1956). However, the limb bones assigned to this taxon appear to be identical to

elements that are associated with the type specimen of the temnospondyl

Doleserpeton (Bolt, 1969). Because these taxa occur in the same locality (Fort

Sill, Oklahoma), the limb elements assigned to Cardiocephalus by Gregory et al.

(1956) should be regarded as belonging to Doleserpeton until further studies of

the former are undertaken. Specimen MCZ 3692, which was attributed to

118 Cardiocephalus by Carroll and Gaskill (1978) and Carroll et al. (1998) is a

somewhat more robust bone with a smaller capitulum than that of Doleserpeton,

and may have been correctly identified as belonging to Cardiocephalus. Data for this specimen are given in Appendix 7.

In the polyphyletic origins hypothesis (Carroll, 2007; Anderson, 2007;

Anderson et al., 2008b) salamanders and frogs are proposed to be closely related to branchiosaurids and amphibamids, whereas the microsaur genus Rhynchonkos is suggested as a sister taxon to caecilians. This is supported by several traits of the skull and vertebral column (Carroll, 2007), but contradicted by evidence from dentition (Parsons and Williams, 1962; Bolt, 1969) and the inner ear (Chapter 2).

The humerus of Rhynchonkos shares the lack of an entepicondylar foramen and the elongated shaft with lissamphibians, giving some credence to the hypotheses of a relationship between this form and caecilians. The distal end is poorly ossified, but the capitulum appears to have been medially placed, judging from the figures given by Carroll and Gaskill (1978). The structure of the distal end of the bone might have resembled that of the related microsaur Trihecaton, including a medially placed capitulum, as well as epicondyles that differ from those of lissamphibians (Carroll and Gaskill, 1978). The data presented here fit well with the temnospondyl origins of salamanders and frogs, but they also point to some remarkable similarities among Eocaecilia, dissorophoid temnospondyls, and batrachians. Because the humerus of Eocaecilia shares the features found in batrachians, and no microsaur has been shown to have the full set of

119 lissamphibian features, the suggested microsaurian origin of caecilians is questioned here.

The lysorophians were proposed as immediate sister-group to lissamphibians by

Laurin and Reisz (1997). Sadly, the distal ends of the humeri of these forms are poorly preserved. However, the preserved parts of the lysorophian humeri constitute a short hourglass-shaped bone, with distal and proximal ends situated approximately in the same plane (Wellstead, 1991). These features are strikingly different from the lissamphibian humeri described above. Of the well-preserved lepospondyl remains studied or described in the literature, none has the full set of lissamphibian characters. Some nectrideans do have a large capitulum overlapping that of modern amphibians in size (Carroll et al., 1998). However, this character is combined with features that are distinctively different from lissamphibians, such as the retention of an entepicondylar foramen, and a short humeral shaft. These features make the humeri of the nectridean Scincosaurus appear more like those of the possible stem amniote Westlothiania (Smithson et al., 1994) than those of lissamphibians. From the above evidence, which is admittedly sparse in the case of lepospondyls, it seems unlikely that a lissamphibian type of forelimb occurred within Lepospondyli.

In the phylogeny by Ruta and Coates (2007), the amphibamid Doleserpeton is proposed as the closest sister-group to all lissamphibians (including

Albanerpetontidae), with Amphibamus as the sister taxon to the Doleserpeton-

Lissamphibia clade. When applying the known morphologies of the humerus and

elbow of the relevant taxa to this phylogeny, a gradual evolution of the

120 lissamphibian traits within the temnospondyls appears. Thus, the entepicondylar foramen disappears in primitive temnospondyls. The capitulum of euskelian temnospondyls (such as Eryops) is enlarged and the entepicondyle points distally.

In Dissorophus, the ectepicondyle takes on the shape of a low, rounded ridge and the supinator process is absent. Finally, in Doleserpeton, the capitulum is larger still and the bone is elongated, essentially reaching the lissamphibian condition.

Thus, the new data fit well with the phylogeny of Ruta and Coates (2007).

However, it should be noted that the frequently poorly ossified distal and proximal ends and the preaxial limb ossification of salamanders may fit the branchiosaurid dissorophoids better than the closely related amphibamids

(Fröbisch et al., 2007). Because these traits do not occur in anurans or in amphibamids, Carroll (2007) may be right in allying salamanders to branchiosaurids and frogs to amphibamids. The enigmatic albanerpetontids also share several humeral features with lissamphibians, including the lack of foramina and possession of a large, rounded capitulum (McGowan, 2002). These characters support the presumed lissamphibian affinities of Albanerpetontidae.

A potentially informative pattern is emerging when reviewing the evidence for lissamphibian origins, namely that although some studies find the Lepospondyl

Hypothesis to be more parsimonious (Laurin and Reisz, 1997; Vallin and Laurin,

2004; but see Ruta and Coates, 2007), no known uniquely derived features of lissamphibians are shared with any lepospondyl, contrary to the situation for temnospondyls. Although there is always the possibility that convergence has taken place, this seems to be unlikely in the case of lissamphibian limb

121 morphology because of the widely differing modes of locomotion seen in these

animals. For instance, although Eocaecilia retains its limbs, it shows clear signs

of limb reduction and elongation of the body (Jenkins et al., 2007), whereas frogs

found at the same location were already highly specialized for saltation (Jenkins and Shubin, 1998). The monophyletic dissorophoid origins hypothesis (e.g., Ruta

and Coates, 2007) fits the present data better than the other theories. If this

hypothesis is correct, then the structures of the humerus and elbow joint as well as

several previously described features such as pedicellate bicuspid teeth (Parsons

and Williams, 1962; Bolt, 1969) and the posterior perilymphatic duct (Sigurdsen,

2008 [Chapter 2]) all would be uniquely derived traits shared by lissamphibians

and their closest relatives among the dissorophoids. We suggest that this is the

most likely explanation for the origins and relationships of lissamphibians at

present.

122 Bridging text 3.

The studies of the amphibian humeri in Chapter 3 described an interesting new

character complex, which is informative for the question of the relationships of

modern amphibian orders. However, anurans, salamanders, and caecilians have

very different modes of locomotion, and it is to be expected that this affects the

morphology of their appendicular skeletons. In particular, the jumping locomotion of anurans is very different from any other tetrapods, and it is generally agreed that this has affected a great number of features of their postcranial anatomy (e.g.

Jenkins and Shubin, 1998). One open question is when exactly saltation became the dominant mode of locomotion, and the form Triadobatrachus is especially interesting this regard, as it is generally regarded to be the earliest well- preserved salientian (Roček and Rage, 2000). However, the interpretation of its anatomy has proven to be equivocal, and its mode of locomotion is still uncertain.

In both Triadobatrachus and Prosalirus, the hindlimb morphology was used as the main piece of evidence for the mode of locomotion, although Jenkins and

Shubin (1998) briefly mentioned the importance of the forelimb.

In Chapter 3, the differences in humeral morphology between anurans and other tetrapods is explored, and the humeri of Triadobatrachus are described in detail to see if there is additional evidence for the locomotion of this enigmatic form.

123

124 CHAPTER 4.

The anuran humerus and the evolution of jumping in early

salientians.

ABSTRACT

The Lower Triassic amphibian Triadobatrachus massinoti is universally regarded

as a stem anuran, but its form of locomotion is still controversial. It is not known

whether it was capable of jumping, as the postcranial features studied so far have

provided equivocal evidence in this regard. New detailed studies of the humeri of

Triadobatrachus show potentially informative features that are shared with modern anurans, but not with salamanders and other tetrapods. In modern

anurans, the morphology of the deltopectoral crest (crista ventralis) is modified

relative to that of other amphibians. The shape of the deltopectoral crest of

Triadobatrachus shows a lengthening of the crest, and a lateral deflection of is

ventral edge creating a slight concavity on the lateral face of the humerus. These

features are very uncommon in most tetrapod groups with a sprawling stance, but

typical of modern jumping anurans, in which the anterior chest musculature is

enlarged. This indicates that the importance of the anterior parts of the chest

musculature, most notably the deltoid, had increased relative to that of the

posterior (abdominal) parts of the pectoralis muscle. This may be an adaptation

for resisting the forces of impact when landing during saltation, although

observations of the most plesiomorphic modern anuran genera (Leiopelma and

Ascaphus), make this conclusion somewhat uncertain.

125 INTRODUCTION

In previous studies of the evolution of jumping in anurans, the anatomy of the

hind limbs and pelvic region has naturally received much attention (e.g. Shubin and Jenkins 1995; Jenkins and Shubin 1998; Wang et al. 2007; Přikryl et al.,

2009). When these data have been compared to fossil forms, the evolution of jumping seems to have been fully realized by the Lower Jurassic (Jenkins and

Shubin, 1998). The earlier salientians (proanurans) from the Lower Triassic remain equivocal in this regard (Rage and Roček, 1986; Evans and Borsuk-

Bialynicka, 1998). Although they do show some features that may be associated with saltation, such as anteriorly directed iliac blades, the relatively short hind limbs and lack of fusion of limb bones have left the manner of locomotion an open question (e.g. Roček and Rage, 2000).

There can be little doubt that the forelimbs play a vital part in anuran saltation.

Several authors stress their importance as shock absorbers in frogs (e.g. Zug,

1972; Liem et al., 2001; Nauwelaerts and Aerts, 2006). The forces of impact absorbed by the forelimbs and pectoral girdle have been measured to be three times higher than the take-off forces, due to the relatively abrupt deceleration phase when landing (Nauwelaerts and Aerts, 2006). However, how this relates to the morphology of the forelimbs has scarcely been mentioned in the literature. A few studies discuss the importance of the morphology of the pectoral girdle

(Emerson, 1984, 1988), but even in these studies, the anatomy of the forelimbs is rarely discussed. In several recent anatomical descriptions of anurans, the humeri were among the few bones not figured at all (e.g. Pugener and Maglia, 1997;

126 Banbury and Maglia, 2006). Here, I discuss features of the osteology and myology of the forelimbs that are likely to be related to jumping, and show how these features may throw light on our knowledge of the locomotion of extinct forms, such as the enigmatic early salientian Triadobatrachus (Fig. 1) from the

Lower Triassic of Madagascar.

Although it is not stated unambiguously, some studies seem to indicate that the

frog-like specializations seen in the pelvic structure of Triadobatrachus massinoti

could be considered a preadaptation, i.e. an adaptation originally serving a

function differing from its current use (e.g. Rage and Roček, 1989; Roček and

Rage, 2000). If this were the case, then other parts of the anatomy are unlikely to show clear trends towards saltatory adaptations. Furthermore, the evolution of saltation in anurans is now thought to be independent from that of swimming

(Abourachid and Green, 1999; Přikryl et al., 2009), although it may have been linked to escaping predators by jumping towards water (Gans and Parsons, 1966).

If saltation evolved primarily as a means of escaping predation by jumping from land to nearby water, then the forelimbs of the earliest salientians would presumably not be adapted for impact absorption.

Emerson (1978) and Wells (2007) divide anurans roughly into strong jumpers

(e.g. Rana), hoppers (e.g. Bufo, and most Pelobatidae), walkers and runners (e.g.

Kaloula and Kassina), and fully aquatic swimmers (e.g. Pipa and Xenopus).

These modes of locomotion are overlapping, and combined with various degrees of aquatic habits and/or arboreality and terrestriality. The results of Gomes et al.

(2009) show that phylogeny and ecology both can have an impact on anuran

127 postcranial morphology, and these features can sometimes be hard to tease apart.

In this study, the humeral features of Triadobatrachus are compared to both the presumed ancestral state (amphibamids or lepospondyls) and modern anurans in order to assess the most likely mode of locomotion in this enigmatic Triassic taxon.

MATERIALS AND METHODS

Triadobatrachus massinoti (MNHN, No. MAE 126, Fig. 1.) was studied in detail using high-definition casts. Dr. Jean-Claude Rage (Muséum National d'Histoire Naturelle, Paris), who has access to the original specimen, kindly provided photographs and confirmed my observations. In addition, fossil and extant tetrapod specimens were examined at the Museum of Comparative

Zoology, Harvard, Field Museum of Natural History, Chicago, and the Redpath

Museum, McGill University, Montreal. Most notable among these taxa are the primitive caecilian Eocaecilia micropodia (MCZ 9163), and the possible proto- lissamphibians Amphibamus grandiceps (casts) from the Lower Carboniferous of

Mazon Creek, Illinois, and Doleserpeton annectens (for list of specimens, see

Appendix 1) from the Lower Permian of Oklahoma. Primitive amniotes and microsaurs were also studied for comparison.

Ethanol preserved specimens representing Ranidae (Rana pipiens, RM 2812, and Rana catesbeiana, RM 2785), Bufonidae (Bufo americanus, RM 4999),

Ascaphidae (Ascaphus truei, RM 4430), Leiopelmatidae (Leiopelma hochstetteri,

RM 2215), (Xenopus laevis RM 2230), Pelobatidae (Scaphiopus

128

FIGURE 1. The Lower Triassic salientian Triadobatrachus massinoti from Madagascar, natural mold (MNHN, No. MAE 126). A, dorsal view. B, ventral view.

129 holbrookii, RM 2425) and the salamander families Hynobiidae (Hynobius

nigrescens, MCZ 22513), Cryptobranchidae (Andrias davidianus, FMNH

166872) and Ambystomatidae (Ambystoma tigrinum, RM 2161) were dissected to

compare the musculature and osteology of the thorax and forelimbs. In addition,

dry skeletons of various caudates, anurans and amniotes were compared to the

lissamphibian specimens. Unfortunately, no specimens of the “running frogs”

Kassina sp. were available for study.

Institutional Abbreviations

FMNH, Field Museum of Natural History, Chicago, Illinois; MCZ, Museum of

Comparative Zoology, Harvard, Massachusetts; MNHN, Muséum National d'Histoire Naturelle, Paris; RM, Redpath Museum, Montreal, Quebec, Canada.

RESULTS

Tetrapod humeri vary greatly in form and function. For the purposes of this paper, the humeri of taxa retaining the primitive sprawling locomotion were compared to those of salientians, whereas highly derived tetrapods (e.g. mammals and birds) were generally ignored.

Salamanders (Caudata)

Modern amphibian humeri are described in Chapter 3, and only a few points of

interest will be repeated here. Salamander humeri consist of a well-developed

shaft (in contrast with the earliest amphibians e.g. Carroll and Holmes, 2007),

130

.

FIGURE 2. The right humeri of modern amphibians compared. A, Ambystoma tigrinum in ventral (flexor) and medial (posterior) views. B, Rana catesbeiana in ventral and medial views. Thin lines indicate the level of the schematic section shown below. Note the differing orientation of the ventral edge of the deltopectoral crest.

131 with usually cartilaginous ends (Figs. 2A). In general, salamanders share with anurans and the primitive fossil caecilian Eocaecilia a large capitulum (radial condyle or capitellum), a distally directed entepicondyle, and the absence of foramina (Chapter 3). The humeral shaft of caudates is usually straight or only slightly curved. As in Eocaecilia, but unlike frogs, there is often a dorsal process

(crista dorsalis), which is the insertion point of the subscapularis muscle.

However, this process is absent in some caudates, such as Cryptobranchus

(Reese, 1906) and Necturus. As in most tetrapods with a sprawling posture, the ventral edge of the deltopectoral crest (crista ventralis) curves medially and ventrally, creating a ventro-medial concavity near the proximal end of the bone

(Fig. 2A). This feature is almost universally present in tetrapods with a sprawling posture, the major exception being anurans (see below).

Frogs (Anura)

The distal end of the anuran humerus is very similar to that of salamanders

(Figs. 2 and 3). It is dominated by the capitulum, which is usually even larger than that of caudates. The capitulum articulates with both the radial and ulnar parts of the fused radioulna (Chapter 3). The proximal head of the humerus is rounded, except for the anterior concavity formed by the deltopectoral crest (see below).

In anurans, the humeral shaft often has a more pronounced curvature than most other sprawling tetrapods, including salamanders. In most of the specimens

examined, the bone bends ventrally towards the distal end, and dorsally towards the proximal head, giving the shaft as a whole a curvature resembling a shallow,

132

FIGURE 3. The humeri of modern anurans. A, Ascaphus truei, ventral and posterior views. B, Leiopelma hochstetteri, male in ventral and posterior views and partly ossified female humerus in ventral view (cartilage removed). C, Scaphiopus holbrookii in ventral and posterior views. D, Xenopus laevis, in ventral and posterior views. Schematic sections shown below as in Fig. 2.

133 elongated letter S (Fig. 2B). However, the degree of curvature varies quite

considerably, even within species, making this trait hard to interpret.

The deltopectoral crest (crista ventralis) serves as insertion for the deltoideus and pectoralis muscles, as well as for several smaller muscles (see below). In anurans, this crest continues further distally than in salamanders and most other tetrapods. The deltopectoral crest of the primitive genus Leiopelma tapers down and disappears gradually towards the midpoint of the shaft (Fig. 3B). In strong jumpers such as Rana catesbeiana, the deltopectoral crest is a prominent elongated ridge that extends for almost half the length of the bone, at which point it ends abruptly (Fig. 2B). Proximally, it continues on to the head of the humerus, sometimes being interrupted by a slight break between the main part of the ridge and the head of the humerus (Fig. 2B). In most frogs, the deltopectoral crest curves laterally (anteriorly), forming a concavity on its lateral face (Figs. 2B,

3A,B,C). This lateral concavity is a striking difference from the situation seen in salamanders, the early caecilian Eocaecilia, as well as lizards and most Paleozoic tetrapods, in which the deltopectoral crest is normally concave medio-ventrally. In anurans, the lateral concavity is most prominent in strong jumpers such as Rana catesbeiana (Fig. 2B) but it is also clearly present in Leiopelma hochstetteri (Fig.

3B) and Scaphiopus holbrookii (Fig. 3C). On the other hand, it is weakly developed in the semiaquatic Ascaphus truei (Fig. 3A) and the terrestrial hopper/walker Bufo americanus.

The most atypical anuran humeri are those of the specialized aquatic pipids. In

Xenopus laevis, the deltopectoral crest is relatively short. It is almost straight in

134 cross-section and the proximal area is best described as biconcave (Fig. 3D). In this species a prominent boss for the insertion of the pectoralis musclulature is located close to the posterior edge of the deltopectoral crest. Immediately lateral to this boss, there is a partially covered groove running from the deltopectoral crest down part of the ventral surface of the shaft. This groove accommodates the long tendon extending from the coraco-radials muscle to the radio-ulna.

Chest Musculature of Amphibians

The musculature which is most important for the present discussion originates on the trunk and inserts on the forelimb (Fig. 4). These include the latissmus dorsi, coraco-brachialis, dorsalis scapulae, supracoracoideus, deltoideus and the pectoralis musculature. A brief summary can be found in Duellman and Trueb

(1986), but the present description is based on Gaupp (1904) as well as personal observations.

The latissimus dorsi is a relatively thin, triangular muscle, which originates from the connective tissue sheet comprising the fascia dorsalis, extends over the back edge of suprascapula, joins the dorsalis scapulae muscle and inserts on the crista ventralis (deltopectoral crest) of the humerus. The latissimus dorsi and dorsalis scapulae muscles are both very similar in salamanders and frogs.

In anurans, the coraco-brachialis longus muscle originates from the medial and posterior end of coracoid and inserts ventrally on the humerus, close to the distal end of the deltopectoral crest. There are two slips of the muscle. The superficial slip originates from posterior part of the lateral half of the coracoid to insert at the

135

FIGURE 4. The chest musculature of modern amphibians. A, the American bullfrog Rana catesbeiana (Ranidae). B, the salamander Ambystoma tigrinum (Ambystomatidae).

136 base of the proximal end of crista ventralis. The deep slip (profunda) originates a

little more dorsally on the coracoid and also on parts of the cartilage that connects

the scapula and coracoid elements. It inserts medially close to the head of the

humerus. The coraco-brachialis muscles of salamanders, as described by Miner

(1925) are very similar to those of frogs, although the longus slip of the muscle inserts further distally in salamanders than in frogs (Francis, 1934).

The deltoid muscle (m. procoraco-humeralis of Francis, 1934) of salamanders is

relatively simple (Fig. 4B). It originates from the procoracoid cartilage of the

pectoral girdle and inserts on the deltopectoral crest of the humerus (Francis,

1934). In contrast, the deltoid of anurans is a complex and massive muscle (Fig.

4A). Gaupp divided the musculus deltoideus into three parts (pars episternalis,

clavicularis and scapularis) with separate origins and insertions. In fact, the

muscle is a more unified mass with two or three origins (heads) and two regions

of insertion. Nevertheless, Gaupp’s classical description is still largely accurate,

and the major divisions given by that work is used here, although the fibers do not

always follow the patterns described by Gaupp. The scapularis part of the deltoid

is by far the most robust part and the most readily apparent in ventral view (Fig.

4A). It originates from the anteriorly projecting area of contact between the

scapula and clavicle (i.e. the acromion) and from the interior of the scapula. It

inserts on the deltopectoral crest of the humerus, as noted by Gaupp (1904).

However, fibers from this slip also join the pars episternalis and continue on to

insert more distally on the humerus, along a low ridge running from the

deltopectoral crest to the entepicondyle. The pars episternalis originates from

137 episternum (omosternum) and inserts on the distal end of the humerus, above the

entepicondyle (ulnar epicondyle) of the humerus. This slip of the anuran deltoid is

partially covered by the coraco-radialis muscle. Finally, the pars clavicularis of the deltoid originates from the lateral end of clavicle. It inserts inside the lateral concavity of the deltopectoral crest of the humerus. In most frogs, the deltoid muscle is considerably thicker than the surrounding musculature (Fig. 4A), and it appears to be the most powerful of the muscles originating on the trunk and inserting on the forelimb. The exception is again the aquatic pipids, in which the deltoid is thinner than in most other anurans.

The pectoralis muscle of anurans also has three parts: pars epicoracoidea, sternalis and abdominalis. Pars epicoracoidea originates from the epicoracoid cartilage (midline of pectoral girdle) and from the medial end of the coracoid, to insert on the posterior side of the crista ventralis of the humerus. It appears to be homologous to the supracoracoideus muscle described in salamanders (the latter has sometimes been interpreted as a slip of the pectoralis, see Francis, 1934). Pars sternalis originates from the sternum (both cartilaginous and osseous parts) and inserts on the posterior base of the crista ventralis. The pars abdominalis is the largest element, originating from the septa of the rectus muscle in the abdomen and inserting on the posterior edge of the crista ventralis. This posterior part of the pectoralis covers a large area of the abdomen, but it is very thin, being slightly transparent even in large ranid specimens (Fig. 4A). This is in clear contrast with the thicker pectoralis musculature found in salamanders, in which the posterior slips of this muscle are more powerful than the deltoid (Fig. 4B).

138 Although not directly connected to the humerus, the coraco-radialis muscle is also large in anurans (Fig. 4A). When observed in ventral view, this muscle is partially covered by the anterior slip of the pectoralis muscle. It originates from the episternum (osseous and cartilaginous), the epicoracoid cartilage, and the medial parts of the coracoid and clavicle. The muscle tapers to a point laterally and continues as a long tendon which enters a canal through part of the deltoid muscle and inserts on the proximal radial end of the radioulna, close to the articulation with the humerus. In salamanders, the supracoracoideus muscle ( = pectoralis pars epicoracoidea) covers the deeper coraco-radials, which is hard to discern from the supracoracoideus (Francis, 1934), and is therefore presumably a relatively small muscle.

The Humeri of Early Fossil Salientians

The Lower Triassic Malagasy salientian Triadobatrachus massinoti has been described by Piveteau (1937), Rage and Roček (1989) and Roček and Rage

(2000), with additional comments by Estes and Reig (1973). These descriptions gave little information on the humeri, other than that they resemble modern anurans (Piveteau, 1937). The bone is about 17.5 mm long as preserved. The humeral shaft of Triadobatrachus is sigmoidally curved, much like most modern frogs (Fig. 5). There is no dorsal process. The deltopectoral crest tapers off distally until it disappears at not quite mid-length of the shaft (preserved length of crest approximately 8 mm). There is a slightly raised area on the medial side of the bone, near the ventral apex of the deltopectoral crest. This was presumably the

139

FIGURE 5. The humeri of the Lower Triassic salientian Triadobatrachus massinoti. A, original specimen (top) and cast (bottom) in ventral view showing the forelimbs. B, dorsal view of same. C, humerus, ventral (flexor) view, with schematic section below. D, medial (posterior) view. E, dorsal (extensor) view. The line drawings are based on both sides of the specimen, but reoriented to correspond to the right side.

140 insertion point of the pectoralis musculature as in modern pipids. However, the

ventral edge of the deltopectoral crest is deflected laterally, so that the proximal

part of the humerus is concave on the lateral side, as in true jumping anurans. The

distal and proximal ends of the bone are poorly preserved, due to the lack of

ossification of these parts. There is a large rounded gap at the distal end of the

bone, located lateral to the entepicondyle (Fig. 5C). The size and shape of this gap

indicates that Tridobatrachus had a large unossified capitulum, and this feature is

confirmed in the similar Polish taxon Czatkobatrachus polonicus (see below). The

gap continues on to the dorsal side of the bone, again resembling the condition in

modern anurans, in which part of the cartilaginous capitulum extends dorsally into

a trough-like depression. The preserved parts of the medial condyle also indicate a distally directed entepicondyle as in anurans. In general, Piveteau’s (1937) observation that the humeri of Tridobatrachus are much like those of modern

anurans is confirmed here.

The humerus of the Triassic salientian Czatkobatrachus polonicus is only

known from fragments (Evans and Borsuk-Bialynicka, 1998). However, it

preserves some parts not seen in the similar and better known Triadobatrachus.

The large capitulum is better shown in Czatkobatrachus (Evans and Borsuk-

Bialynicka, 1998). However, the general similarities of these two forms, as well

as the large gap in the distal end of the humerus of Triadobatrachus, make it

likely that these typically batrachian features were present in both Triassic taxa.

141 DISCUSSION

Although the humeri of frogs, salamanders and Eocaecilia are surprisingly similar in the elbow region (Chapter 3), salientians (frogs and basal salientians) appear to differ from all other tetrapods in the muscle attachments of the humeri.

In this regard, the humeri of Eocaecilia, salamanders, and lizards are functionally equivalent in having a deltopectoral crest which is deflected medially and ventrally, creating a medial concavity proximally. This condition is also present in temnospondyls, such as Dissorophus and Doleserpeton (Chapter 3). Overall, the humeri of salamanders are remarkably similar to those of the Lower Jurassic

Caecilian Eocaecilia (Jenkins et al., 2007) and the possible “proto-lissamphibian”

Doleserpeton, an amphibamid from the Lower Permian. Although lepospondyl limb elements often are poorly preserved, the medial concavity of the deltopectoral crest is confirmed in the microsaurs Pantylus, Cardiocephalus

(Carroll and Gaskill, 1978) and Euryodus (FMNH PR 983). The similarity to possible lissamphibian sistergroups may indicate that the appendicular morphology of modern salamanders is close to the primitive lissamphibian condition (Chapter 3).

The humeral morphology found in most sprawling tetrapods outside Salientia ensures a strong edge for the attachment of the pectoralis musculature, which is important in pushing the animal forwards during locomotion. The most striking differences between an anuran humerus and that of salamanders are thus associated with the deltopectoral crest. The fact that the proximal area of the humerus of most salientians is concave laterally rather than medially presumably

142 reflects a shift in the importance and strength of the muscles attached to the

deltopectoral crest. The deltoid muscle inserts on the proximal portion of the

humerus in salamanders, whereas the different slips of this muscle in frogs insert

not only proximally, but also further distally, one slip inserting almost at the distal

end of the bone (Gaupp, 1904). The lateral concavity of the anuran humerus

serves as attachment site for the powerful deltoid musculature of frogs. This

muscle is in an ideal position to resist the forces of impact when the frog is in the

landing phase of a jump. Contraction of the deltoid pulls the humerus forward,

and rotates it so that the radio-ulna and manus are directed anteriorly (Gaupp,

1904; Duellman and Trueb, 1994). During this phase, the long axis of the body of

the frog is at a sharp angle to the substrate, the forelimbs being stretched

anteriorly and ventrally to resist the impact (Nauwelaerts and Aerts, 2006). The

fusion of the radio-ulna strengthens the forelimb osteology against the force of

impact. Even in poorer jumpers (e.g., Bufo, Scaphiopus), the movement of the

forelimb is very different from that of crawling animals such as salamanders and,

presumably, primitive tetrapods. Thus, the morphology of the humeri probably reflects the differences in muscle insertion and muscle action on this element, which in turn depends on the mode of locomotion. New observations of Ascaphus

(Essner, Suffian and Reilley, unpublished data), show a quite different mode of

jumping behavior. In this species, the forelimbs are not extended forward during

the landing phase, the animal landing on its belly or head. This may be an

adaptation to the streams in which it lives, but it also poses a problem in assessing

the primitive condition of modern anurans.

143 The extended insertion of the deltoid of frogs may serve to counteract the force

acting on the distal end of the humerus when landing. Since the force applied

further from the fulcrum (i.e. the glenoid) is larger, the insertion of the deltoid muscle may have shifted distally to counteract the landing force. The deltopectoral crest of Triadobatrachus is slightly extended distally relative to that of salamanders but less so than in the strongest jumpers among modern frogs.

When compared to extant anurans, and keeping in mind the missing cartilaginous parts, the humerus of Triadobatrachus appears to resemble that of the primitive genus Leiopelma most. The Leiopelma species are best characterized as hoppers rather than strong jumpers, although they are capable of longer jumps when agitated (Abourachid and Green, 1999). Like Ascaphus, Leiopelma appears to rarely land directly on its forelimbs, although they differ in that Leiopelma streches the forlimbs anteriorly when jumping. The primitive condition of modern anurans is therefore somewhat uncertain.The slight anterior concavity of the proximal head of the humerus of Triadobatrachus (Fig. 5) shows that the anterior chest musculature including the deltoid was probably more important in locomotion than was the posterior parts of the pectoralis muscle. Although the anatomy of Triadobatrachus precludes the possibility of this form being as strong a jumper as some modern forms, the attachment of relatively strong deltoid musculature, unusual in other tetrapods, may suggest that hopping or jumping on dry land was an important form of locomotion for Triadobatrachus. However, further studies are needed in order to assess the primtive jumping behavior of anurans, and the importance of the forelimb in salientian locomotion.

144 Bridging text 4.

Whereas the anatomy of the Triassic form Triadobatrachus clearly belongs within the salientian stem, and thus within crown-group (assuming a sister-group relationship between Urodela and Anura), the position of

Doleserpeton relative to modern amphibians is still controversial (e.g. Holmes,

2000; Vallin and Laurin, 2004; Carroll, 2007; Ruta and Coates, 2007).

Conversely, there is a general consensus as to the relationship of Doleserpeton relative to other Paleozoic tetrapods. Its position within the dissorophoid temnospondyls has never been contested (e.g. Daly, 1994; Holmes, 2000; Vallin and Laurin, 2004). The new information on Doleserpeton presented in earlier chapters could therefore change the position of dissorophoids (and the

Temnospondyli in general) relative to modern amphibians. Furthermore, the review of the data matrix of Anderson et al. (2008b) in Chapter 1 revealed many problems in the character coding, affecting the our conclusions regarding lissamphibian origins. In Chapter 5, the data matrices of Vallin and Laurin (2004),

Ruta and Coates (2007) as well as Anderson et al. (2008b) are discussed and reanalyzed. Also, a supermatrix is built using previous data sets in order to increase the amount of evidence in the analysis of amphibian relationships.

145

146 CHAPTER 5.

The origin of modern amphibians: a reevaluation

ABSTRACT

There are currently three competing hypotheses seeking to explain the

evolutionary origins of modern amphibians. The Lepospondyl Hypothesis holds that modern amphibians all derive from the Paleozoic lepospondyls, Lysorophia being the closest sister-group to all lissamphibians. The Temnospondyl

Hypothesis suggests that modern amphibians are most closely related to the dissorophoid temnospondyls such as Doleserpeton. Finally, the Polyphyletic

Hypothesis holds that the modern amphibian orders have separate evolutionary origins from among different groups of Paleozoic tetrapods. The latter hypothesis suggests that batrachians (frogs and salamanders) are related to the dissorophoid temnospondyls, whereas caecilians are derived microsaur lepospondyls. Here, the character matrices used in the three different hypotheses are critiqued. New data on Doleserpeton are introduced, along with new observations of Eocaecilia. A

supermatrix is built based on data from all three hypotheses, and analyzed using

both Bayesian and parsimony-based software. The unmodified supermatrix yielded inconclusive results. However, when the matrices are corrected according to the new observations, the phylogeny based on the resulting supermatrix supports the Temnospondyl Hypothesis of lissamphibian origins.

147 INTRODUCTION

The evolutionary origins of the modern amphibian groups is still the subject of

contention (review in Anderson, 2008). Modern amphibians (sometimes united in

the taxon Lissamphibia) consist of Salientia (frogs and their fossil relatives),

Caudata (salamanders and newts), and Gymnophiona (caecilians), with Anura,

Urodela, and Apoda constituting their respective crown groups (Milner, 1988;

Cannatella and Hillis, 2004). Three competing hypotheses have been forwarded to explain the relationships of modern amphibians. After the publication of several important studies of the 1960s (Parsons and Williams, 1962, 1963; Bolt, 1969), modern amphibians have often been regarded as a monophyletic group allied to temnospondyls (e.g. Milner, 1988; Trueb and Cloutier, 1991; Ruta and Coates,

2007). For instance, Ruta and Coates (2007) proposed that the three modern

orders together constitute the sister-taxon of the dissorophoid temnospondyl

Doleserpeton. However, Vallin and Laurin (2004), building on the work of Laurin

and Reisz (1997), suggested that modern amphibians are derived from the

Paleozoic tetrapods belonging to the Lepospondyli. This is a diverse group of

animals consisting of microsaurs, nectrideans, adelospondyls, lysorophians and

aїstopods. In the phylogeny of Vallin and Laurin (2004) the lysorophians (Fig.

1A, B, C) were seen as the closest sister taxon to modern amphibians, whereas the

microsaurian taxa Brachystelechidae and Rhynchonkos formed consecutive stem groups. In contrast, Carroll (2007), Anderson (2007), and Anderson et al. (2008b), suggested that modern amphibians are a polyphyletic assemblage. According to these authors, batrachians (anurans and caudates) form a monophyletic group,

148

FIGURE 1. Some fossil taxa of importance to the discussion of the origin of modern amphibians. A-C, the lysorophian lepospondyl Brachydectes (from Wellstead, 1991). D-F, the microsaurian lepospondyl Rhynchonkos (from Carroll and Gaskill, 1978). G-I, the amphibamid temnospondyl Doleserpeton (from Chapter 1). A, D, G, full reconstruction. B, E, H, skull, dorsal view. C, F, I, skull, ventral view. Not to scale.

149

which is closely related to dissorophoid temnospondyls, whereas caecilians are

more closely related to the microsaur Rhynchonkos (Fig. 1D, E, F). Although

Carroll (2007) provided an insightful discussion of the polyphyletic hypothesis, his main conclusions were not based on a phylogenetic analysis. As the present study focuses on data matrices used in such analyses, Anderson et al. (2008b) is used here as the basis for a discussion of the Polyphyletic Hypothesis.

In the present study, I seek to explore the strengths and weaknesses of these hypotheses. Schoch and Milner (2004) discussed previous work and hypotheses in

this field. However, they included only a superficial appraisal of the relevant data

matrices, and no attempts were made to modifiy and reanalyze the data. More

recently, Marjanovic and Laurin (2009) critiqued the matrix provided by

Anderson et al. (2008), and concluded that the modified matrix supports the

Lepospondyl Hypothesis. Here, I am including a more complete discussion of the

character coding of the previous data matrices, and I also provide a reanalysis

based on these matrices. Emphasis is placed on the critique of existing data, rather

than introducing yet another data set to this discussion. However, it is useful to

discuss the problem of lissamphibian origins in light of a supermatrix based on all the current hypotheses. Although details of the interrelationships of modern amphibian taxa is not the primary concern of this study, it is worth noting that most molecular studies favor a sister-group relationship between frogs and salamanders relative to caecilians (San Mauro et al., 2005; Zhang et al., 2005;

Frost et al., 2006; Hugall et al., 2007). These same studies also tend to favor the

150 monophyly of modern amphibians relative to modern amniotes (for a critique of

the molecular studies, see Anderson, 2008).

MATERIALS AND METHODS

Fossil specimens belonging to all the pertinent taxa were studied and compared.

These taxa include Lysorophia, (e.g. Rhynchonkos and

Cardiocephalus), Dissorophoidea (e.g. Amphibamus, Apateon, Doleserpeton, and

Gerobatrachus) and early lissamphibians (e.g. Eocaecilia, Prosalirus, and casts

and photographs of Triadobatrachus). Doleserpeton was of special interest,

because of the numerous well-preserved specimens available, and because this

form has repeatedly been the focus of the Temnospondyl and Polyphyletic

Hypotheses since it was first described (Bolt, 1969). Skeletal material and cleared

and stained specimens of the three modern amphibian orders were extensively

used as a basis for comparison.

The data matrices of Vallin and Laurin (2004) and Anderson et al. (2008b) were

obtained from the literature (online in the case of Anderson et al., 2008b). That of

Ruta and Coates (2007) was obtained by personal correspondence. A supermatrix

(e.g. de Queiroz and Gatesy, 2006) was based on the matrices of the above

authors. This was done by restricting the number of taxa to those covered by all

the three studies. In some cases, a combination of related taxa had to be used. For instance, the matrix of Ruta and Coates (2008) includes fossil forms only, whereas Vallin and Laurin (2004) and Anderson et al. (2008) also used recent data. In these cases, taxa were chosen with emphasis on minimizing the amount of

151

FIGURE 2. The Jurassic limbed caecilian Eocaecilia. A, full reconstruction. B, skull in dorsal view. C, skull in ventral view. D, skull in lateral view. From Jenkins et al. (2007).

152 missing data. For instance, the taxon Anura in the supermatrix had to be constructed using a combination of data from the taxa Discoglossidae,

Notobatrachus and “frogs” (the latter from Anderson et al., 2008). For Caudata,

Hynobiidae, Valdotriton, and “salamanders” were used.

This approach to constructing a supermatrix maximizes the number of characters in the data, but limits the analysis to include only key taxa. Excluding taxa that were not covered by all the matrices avoids the potential problems resulting from the use of excessive amounts of missing data (but see de Queiroz and Gatesy, 2006). Both the concept of parsimony and Bayesian inference have proven their usefulness to science (e.g. Gauch, 2003) and phylogenetic analyses

(e.g. Ronquist et al., 2009). Therefore, the matrices were analyzed using both

MrBayes (version 3.1.2) and PAUP (version 4.0 b10). Repetitive characters were excluded using the “exclude” command in both programs. Both modified and unmodified versions of the supermatrix were analyzed. For the modified analyses, new data were entered for Doleserpeton where appropriate, and corrections were made whenever the character coding was found to be incorrect or misleading

(Appendices 5, 8, 9). However, no new characters were added to this analysis.

The resulting data matrix contains 719 characters, but 130 of these were excluded to avoid repetition of characters. The reduction in the number of taxa also increased the number of uninformative characters, leaving only 400 that were phylogeny-informative. All characters were treated as unweighted and unordered, and the bootstrap analyses were set to 10 000 repetitions. The Bayesian analyses were run using Markov chain Monte Carlo sampling (MCMC), and the default

153 settings of MrBayes. Samples were taken every 100 generations, and burn-in set

to 25% of the samples. Generations were added until the average standard

deviation of split frequencies reached below 0.01 (but see discussion, below).

RESULTS

Lepospondyl Hypothesis

This hypothesis is originally based on the work of Laurin and Reisz (1997),

whose data were reworked by Laurin (1998a). The resulting matrix was enlarged by Laurin (1998b), Laurin and Reisz (1999) and finally Vallin and Laurin (2004).

Although not exclusively concerned with the origins of modern amphibians, the most controversial claim of these studies is that Lissamphiba constitutes a monophyletic taxon within Lepospondyli. Temnospondyls are seen as a side- branch splitting off early from the lineages leading to modern amniotes and lissamphibians. Table 1 summarizes the most important features that are shared between lepospondyls and modern amphibians (but not with dissorophoids) according to this data matrix.

Schoch and Milner (2004) noted a number of problems with this hypothesis.

Most characters supporting the lepospondyl-Lissamphibia clade are shared absences of individual skeletal elements. Very few “positive” characters, such as homologous features of similar structure and function, support this grouping.

Furthermore, the resulting phylogeny implies at least eight character reversals within the Lepospondyli-Lissamphibia clade (Schoch and Milner, 2004). Some of

154

FIGURE 3. Analysis of modified matrix from Vallin and Laurin (2004). Numbers indicate node support (posterior probabilities in the Bayesian analysis). A, bootstrap analysis (PAUP). B, analysis of MrBayes, 3 000 000 generations. Arrows indicate the position of clades containing modern amphibians.

155 these character reversals seem counter-intuitive, such as the reappearance of the circumorbital bones in Eocaecilia (Fig. 2).

Many of the characters in the matrix of Vallin and Laurin (2004) are concerned with the presence or absence of skeletal features. Compared to the work of Ruta and Coates (2007), relatively few characters cover the shapes of the elements in question. Furthermore, some of the shape-based characters that are present were coded in a way that may be misleading (Appendix 8). For instance, character 5, concerning the shape of the orbits, is coded similarly for lysorophians and lissamphibians (except Eocaecilia), but differing from that of temnospondyls. The authors justify this by the fact that the orbits in these forms are open posteriorly

(character state 2). However, this “open” condition reflects the fact that a number of dermal bones are absent in both lissamphibians and lysorophians, features that are covered by other characters in the same matrix. Furthermore, this condition does not result in a general similarity in orbit shape, the orbits of modern salamanders and frogs usually being considerably wider relative to the skull width, than those of lysorophians. Alternatively, it can be argued that the wide orbits of amphibamid temnospondyls are more similar in shape to modern batrachians, despite the fact that many more circumorbital elements are present in temnospondyls.

The temporal emargination (otic notch, character 31) was also coded in a way that may be questioned. In this case, any resemblance between temnospondyls and anurans (both sharing a large otic notch of the squamosal) was erased by introducing different character states depending on whether or not the tabular and

156

Characters favoring Lysorophia-Lissamphibia clade

Ch. nr. Character state 5 Orbit shape open posteriorly (not Eocaecilia) 27 Postfrontal absent (not Eocaecilia) 28 Postorbital absent (not Eocaecilia) 30 Jugal absent (not Eocaecilia) 84 Postsplenial absent 98 Vertebrae with one central element only (not Eocaecilia) 134 Humeral pectoral tubercle indistinct (“?” in Eocaecilia)

Characters favoring Microsaur-Lysorophia-Lissamphibia clade

Ch. nr. Character state 3 Lack of dermal surface sculpturing 6 Jaw joint position anterior to or level with the occiput 13 Parietal-squamosal contact present 15 Absence of pineal foramen (not in Triadobatrachus) 17 Supratemporal absent 51 Pterygoid denticles absent (not Eocaecilia) 58 Parasphenoid denticles absent (not Eocaecilia) 61 Post-temporal fenestra small or absent 78 Stapedial foramen absent (not in caecilians and Karaurus) 80, 81, 82 Individual coronoids absent 87 Parasymphyseal fangs absent 105 Atlantal odontoid present (not Salientia) 122 Clavicle without ventral plate (Anura only)

Table 1. Characters favoring the Lepospondyl Hypothesis according to the data of Laurin (1998) and Vallin and Laurin (2004). See the text for discussion.

157 supratemporal bones form part of the notch. These bones are primitively present in temnospondyls, but absent in anurans, a fact that is covered elsewhere in the matrix. The similarity of the otic notches of temnospondyls and anurans was discussed by Bolt and Lombard (1985; see also Chapter 2).

Other problems of the character coding may be harder to resolve. Examples include the individual coronoid and splenial elements of the lower jaw. In the original matrix, each of these is coded as a separate character, and lysorophians thus end up with four mandibular characters supporting their lissamphibian affinities. This is in spite of the fact that the of the medial bony element of the lower jaw of lissamphibians is unclear (it is probably the result of a fusion of several elements), and this element is variously called the prearticular, angulosplenial, pseudoangular (Duellman and Trueb, 1994) or even the coronoid

(Reese, 1906). The situation seen in lysorophians is that the meckelian canal is covered by the single splenial and a prearticular (Wellstead, 1991). The lower jaw of lysorophians is thus coded as having many characters in common with lissamphibians, in spite of the obvious differences in jaw shape, including the general curvature of the jaw, width of the ramus, length of the tooth bearing ramus, or the height of the coronoid process. Such characters are not covered by this matrix.

Other problem characters, such as the jaw joint position in anurans (character

6), and the post-temporal fossa being coded as “small or absent” in modern amphibians (character 61), are noted in Appendix 8.

158

Modified analysis

Using PAUP, the analysis of Vallin and Laurin’s (2004) corrected matrix yields a consensus tree that is not significantly different from the original topography.

Thus, simply adding more information on Doleserpeton and correcting some of the data provided on the other taxa did not exclude lissamphibians from their position as sister taxon of the Lepospondyli. This illustrates the robustness of this matrix, a fact that has been pointed out by its authors (Laurin, 1998b). However, it

could be argued that some of this robustness results from unjustified repetition of

characters, such as the mandibular characters discussed above. When these repetitive characters were united to form single multistate characters, and the above corrections were introduced, the resulting 50% majority rule bootstrap tree collapsed almost completely (Fig. 3A). Lissamphibians still formed a monophyletic taxon, but the Lissamphibia, Temnospondyli and Lepospondyli constituted an unresolved polytomy. Although the strict consensus tree still

favored the lepospondyl hypothesis, the apparent robustness of the matrix had

disappeared. The Bayesian analysis favored the monophyletic Lepospondyl

Hypothesis (Fig. 2B). However, this result was only achieved after increasing the number and temperature of the hot chains in MrBayes. Using the standard settings resulted in the average standard deviation of split frequencies never reaching the recommended value of 0.01, a situation which probably reflects the presence of

major local optima, or islands (see discussion below).

159 Interestingly, in a test run in which only Eocaecilia was included among modern amphibians in the modified matrix of Vallin and Laurin (2004), this taxon was included within the Amphibamidae.

Polyphyletic Hypothesis

The data matrix provided by Anderson et al. (2008b) is lepospondyl-focused, in the sense that many more taxa are included of the relatively poorly understood

Lepospondyli (29 taxa) than of the species-rich Temnospondyli (15 taxa). Several closely related microsaur species have been coded, sometimes within the same genus. In contrast, a species-rich temnospondyl family like is only represented by one taxon. This is a surprising choice, particularly considering the fact that Anderson et al. (2008b) are concerned primarily with the description of Gerobatrachus, which is a temnospondyl.

Marjanovic and Laurin (2009) recently critiqued the character matrix of

Anderson et al. (2008b). Their reanalysis of this matrix resulted in a tree supporting the Lepospondyl Hypothesis. They commented that the argument for the presence of a basale commune in Gerobatrachus was unconvinicing, and that the pedicelly of the teeth in this form is uncertain. These arguments are justified, and personal observations by the present author have cast doubt on the identity of the putative basale commune. As noted in Chapter 1, the description of pedicellate teeth in Gerobatrachus was almost certainly a misinterpretation. However, for the analysis of the modified matrix given here (Fig. 4) fewer character changes were made than those introduced by Marjanovic and Laurin (2009). The coding of the

160

FIGURE 4. Analysis of modified matrix from Anderson et al. (2008). A, bootstrap analysis. B, Bayesian analysis, 1000 000 generations. Numbers indicate node support. Arrows indicate clades containing modern amphibian orders. See Chapter 1 for the strict consensus tree obtained using PAUP.

161 basale commune was kept unchanged, but that of the dentition was changed to

uncertain (?) for a more conservative approach.

The character matrix of Anderson et al. (2008b) represents a fairly complete

picture of the overall anatomy of the taxa in question, including both

presence/absence data as well as the shape of the elements in question. Elements

of the lower jaw have been coded as single characters representing groups of

related bones (e.g. coronoids and splenials), rather than one character per element

as in Vallin and Laurin (2004). Unfortunately, little discussion of the individual

characters is included in this work, and the character descriptions are written in a

shorthand manner (Appendix 5), sometimes making the interpretation of the

character coding difficult.

A relatively large number of problems was found in the character coding of

Anderson et al. (2008b). Some of these were noted in Chapter 1, but are repeated here for clarity. The problematic features include characters covering palatal elements (particularly the vomers), dentition, and ribs. In general, the ribs and palatal elements were coded as if the morphology of these elements supports the

Polyphyletic Hypothesis. However, detailed comparisons of microsaurs (e.g.

Carroll and Gaskill, 1978), temnospondyls (e.g. Holmes, 2000; see also Chapter

1), and Eocaecilia (Jenkins et al., 2007), show the opposite to be the case. For instance, Anderson et al. (2008b) coded the ribs of Eocaecilia as being similar to

those of microsaurs (see Appendix 5, character 179). However, microsaurian ribs

tend to be similar to those of amniotes in that they are longer and more ventrally

curved (Fig. 1D) than those of modern amphibians. The ribs of Eocaecilia are

162 similar to modern salamanders and Doleserpeton in being short and posteriorly curved (Fig. 2A). Further, Anderson et al. (2008b) interpret the interpterygoid vacuities of Eocaecilia as being closer to the microsaurian condition (character

115, coded as “narrow”, state 0), by dividing the vacuity morphologies into three groups based on size (narrow, wide, and closed). However, it could be argued that the vomers, palatines and interpterygoid vacuities of Eocaecilia are all closer to the condition found in Doleserpeton than to that of microsaurs (Figs. 1C, F, I, and

2C). Specifically, the interpterygoid vacuities of Eocaecilia and Doleserpeton are bordered by the same palatal elements, including the palatine bone, whereas the latter bone is never found bordering the vacuities of microsaurs. It could also be objected that the vacuities of Eocaecilia are too large to be described as “narrow”

(Fig. 2C).

Modified analysis

A strict consensus tree of the modified matrix of Anderson et al. (2008b) was provided in Chapter 1. This phylogeny places Doleserpeton as the sister taxon of

Eocaecilia. However, most of the majority rule bootstrap tree is collapsed (Fig.

4A), and the clade including Doleserpeton, Eocaecilia and modern amphibians is weakly supported. An analysis of the same data in MrBayes gives a somewhat different picture. The Bayesian analysis (Fig. 4B) rearranged the tree so that

Eocaecilia is the sister group of batrachians and Gerobatrachus, and there is strong support for the position of Doleserpeton basal to modern amphibians. In

163 general, the temnospondyl affinities of modern amphibians are supported in all the analyses of the modified matrix of Anderson et al. (2008b).

Temnospondyl Hypothesis

The analysis of Ruta and Coates (2007) builds on that of Ruta et al. (2003).

Encompassing 102 taxa and 339 characters, this matrix is the largest of the three

hypotheses discussed here. Characters in Ruta and Coates (2007) are frequently

more fully described than either Anderson et al. (2008b) or Laurin (1998b; Vallin

and Laurin, 2004). The data also include shape characters more often than the

other matrices. An effort has clearly been made to cover the whole anatomy of the

taxa in question, and while this is in many ways a sound procedure, it has led to

the inclusion of some characters of dubious value. For instance, the coding of

sutural characters (characters 44, 45, 60, 66, and 71) covering differences in

degrees of interdigitation may be superfluous, because such characters are

frequently obscured by differences of ontogenetic stage (e.g. Liem et al., 2001),

and possibly also fossil preservation. However, it was found that the exclusion of

these characters made little difference for the final phylogenies. Character 245,

dealing with the presence of a “subterminal glenoid”, is also highly dependent on

the degree of ossification (personal observations) and was excluded from the

present analyses (see Appendix 9). The data follow Vallin and Laurin (2004) in

coding each coronoid and splenial separately. In the context of the present data,

this seems to be less out of place, due to the great detail in which virtually every

feature is described in this matrix. These characters (i.e. loss of elements) tend to

164 favor the Lepospondyl Hypothesis, and have been kept in the modified analysis so as to not balance the matrix needlessly in favor of the temnospondyl hypothesis.

Some characters are more accurately coded by Ruta and Coates (2007) compared to the other studies. For instance, the presence or absence of the prefrontal of Triadobatrachus has been correctly coded as uncertain (see Roček and Rage, 2000) rather than present, as in Vallin and Laurin (2004).

Modified analysis

The modified matrix yielded a consensus tree that is virtually identical to that given by Ruta and Coates (2007). As with the matrix of Vallin and Laurin (2004), the Bayesian analysis of the modified matrix encountered problems in that the average standard deviation of split frequencies stabilized at values significantly higher than the recommended value of 0.01. This indicates that the two simultaneous runs (a standard feature of MrBayes) are trapped on two different islands, and unable to converge. This continued to be the case even after increasing the temperature of the heated chains, and running the analysis for 10

000 000 generations.

Many taxa were poorly supported by the bootstrap analysis (Fig. 5), as is the case with the original matrix (Ruta et al., 2003). A conservative reading of the bootstrap analysis suggests that the relationships between many early tetrapod taxa are still poorly understood. However, the temnospondyl affinities of all modern amphibians are strongly supported in the bootstrap tree.

165

Figure 5. Simplified bootstrap tree based on modified matrix of Ruta and Coates (2007). Numbers indicate node support. The position of modern amphibians (and Albanerpeton) is indicated with an arrow. All nodes of less than 50% support are collapsed. Note that all lissamphibians are positioned within Dissorophoidea, and that the “Lepospondyli” are completely collapsed.

166 Supermatrix

With a combined matrix of 719 characters, the number of repetitive characters

was surprisingly low (130), demonstrating the differing emphasis of the studies

discussed here. Unfortunately, Gerobatrachus had to be excluded from the matrix, as it is only covered by Anderson et al. (2008b). However, this may also

be justified due to the many problems with the description of this form (personal

observations; see also Marjanovic and Laurin, 2009).

The results of the analyses of the unmodified supermatrix are inconclusive, as

the bootstrap analysis yields the Lepospondyl Hypothesis (Fig. 6A), whereas

Bayesian analysis supports the Polyphyletic Hypothesis (Fig. 6B). The

monophyletic Temnospondyl Hypothesis is therefore not supported by the

combination of the original data matrices. As noted above, several mistakes and

omissions were found in the character coding of the included data matrices

(Appendices 5, 8, 9), and a separate set of analyses of a supermatrix based on the

modified matrices is therefore warranted.

The analyses of the modified supermatrix (Appendix 10) strongly support the

monophyletic Temnospondyl Hypothesis (Fig. 7A, B, C). Whereas the Bayesian

analysis and strict consensus trees both suggest a sister-group relationship

between lissamphibians and Doleserpeton, the majority rule bootstrap tree shows

this relationship to be relatively weak (Fig. 7A), and leaves some room for doubt

as to whether lissamphibians are monophyletic within the Dissorophoidea

(bootstrap support value: 52%).

167

Figure 6. Phylogenetic analysis based on unmodified characters from all three hypotheses. A, Bootstrap analysis of PAUP. B, Bayesian inference tree of MrBayes, 1000 000 generations. Repetitive characters were excluded. Numbers indicate node support (PAUP), or posterior probabilities (MrBayes). Arrows indicate modern amphibian taxa.

168 DISCUSSION

The reanalysis of the data matrices from the literature yielded variable results, and the analyses of the unmodified supermatrix were likewise inconclusive. This highlights the problems of modern amphibian relationships, as different parts of the anatomy of these animals appear to give incompatible phylogenetic signals.

This is also clearly illustrated by a tendency of the analyses in MrBayes to be trapped in local phylogenetic optima. This phenomenon was detected by noting the average standard deviation of split frequencies provided by the Bayesian analysis (Ronquist et al., 2009), which sometimes failed to reach a value below the recommended limit of 0.01, indicating that the two simultaneous runs had reached two different phylogenetic islands. The problem could sometimes be avoided by increasing the hot chain temperature, but this failed in the case of the modified matrix of Ruta and Coates (2007). The clade support values (posterior probabilities) of MrBayes were frequently higher than the bootstrap values, sometimes strongly supporting clades that were collapsed in the bootstrap analyses (e.g. Fig. 4A, B).

Characters

Complete lists of the problems in the character coding of the individual matrices are given in Appendices 5, 8, and 9. Although the Lepospondyl Hypothesis appears to be supported by a number of characters, many of these are here interpreted as problematic or incorrect (Appendix 8). Also notable is the lack of characters shared between Eocaecilia and lepospondyls, particularly lysorophians

169 (Table 1). This is surprising, given that caecilians are regarded as the most lepospondyl-like among modern amphibians (Carroll, 2007). Some of the character states noted in Table 1, such as the lack of dermal sculpturing, are actually also present in Doleserpeton (Chapter 1). Finally, note that none of the characters presented in the Lepospondyl Hypothesis, whether correctly coded or not, are unique among tetrapods. Rather, most of them also occur within several groups of amniotes. This is in stark contrast to many of the characters shared by

Doleserpeton and modern amphibians. For instance, the pedicellate dentition

(Chapter 1) and large humeral capitulum (Chapter 3) are rare or non-existent among lepospondyls and amniotes. The presence of the palatine bone in the rim of the interpterygoid vacuity also appears to be uncommon outside modern amphibians and the Amphibamidae. Finally, a number of characters are shared between most lepospondyls and amniotes, but not with modern amphibians (Table

2). These include the shape of the vomers, the length and curvature of the ribs, and the presence of the jugal between the maxilla and quadratojugal elements.

This is a problem for both the Lepospondyl and the Polyphyletic hypotheses, as modern amphibians still retain the primitive condition (exemplified by

Acanthostega) of these characters. Some of these features are noted as synapomorphies of modern amphibians by Milner (1988), although they actually seem to be symplesiomorphies. However, this does not mean that these characters are uninformative for the discussion of the origins of modern amphibians, as the morphology of the ribs, vomers, and cheek would all have to be reversed for modern amphibians to be nested within Lepospondyli. The rib morphology may

170 be especially informative, because of its importance to the mode of respiration.

Thus, the resemblance of most lepospondyl ribs to those of amniotes may indicate

that this group relied upon an aspiration pump system for respiration, as opposed to the buccal pump system of modern amphibians (Liem et al., 2001).

Amphibian phylogeny

The three studies discussed here differ considerably in relative robustness, as demonstrated by the bootstrap analyses. Whereas all the modified bootstrap analyses show strong tendencies towards clade collapse, only the analysis of

Anderson et al. (2008b) changed with regard to which theory was supported after

introducing corrections. In both the Bayesian analysis and the bootstrap analysis,

the modified matrix of Anderson et al. (2008b) subsequently supported the

monophyletic temnospondyl hypothesis, albeit only weakly in the case of the

bootstrap analysis (bootstrap support value 50%).

The analyses of the unmodified supermatrix highlight the lack of consensus of

the different data sets, resulting in variable results depending on the software used

for analysis (Fig. 6). This may result from the data being obtained from three very

different studies, each being slightly biased towards a specific solution. The

supermatrix analyses may thus reflect the different ways of coding counteracting

each other. The fact that this part of the analysis did not show any support for the

Temnospondyl Hypothesis may also point to a relative lack of shared derived

characters supporting this view, compared to the many shared absence characters

supporting the Lepospondyl Hypothesis.

171

Figure 7. Phylogenetic analyses based on modified matrices from all three hypotheses. A, bootstrap analysis (PAUP). B, Bayesian inference tree (MrBayes), 1000 000 generations. C, strict consensus of two equally parsimonious trees (PAUP). Repetitive characters were excluded. Numbers indicate node support (PAUP), or posterior probabilities (MrBayes). Arrows point to modern amphibians.

172 The modified analyses, which introduce new information on Doleserpeton, and correct inconsistencies according to the views of the present author, yields significant support for the Temnospondyl Hypothesis (Fig. 7). It may be argued that the modifications made to the matrices introduce a new bias on top of those of the original data sets. This is a valid argument, but it should be noted that many of the changes introduced to the matrices consisted of simply adding information where there previously were gaps in the data. Furthermore, the comparative studies of the palate, ribs, limbs and dentition suggest that the Temnospondyl

Hypothesis is likely to be correct. It is hoped that this study will spur interest in critical examination of previous phylogenetic analyses, perhaps leading to a consensus in the question of lissamphibian origins.

173

TAXON Posterior Vomers Palatine Humeral Rib length, curvature maxilla wide bordering capitulum size contact anteriorly interpterygoid vacuity Acanthostega Quadratojugal Yes No Small Short, straight/posteriorly curved

Amniota Jugal No No Small Long, ventrally curved Microsauria Jugal No No Small/medium Long, ventrally curved

Lysorophia Palatine No - ? Long, ventrally curved

Doleserpeton Quadratojugal Yes Yes Large Short, straight/posteriorly curved

Salientia Quadratojugal Yes Yes Large Short, straight/posteriorly curved

Caudata - Yes - Large Short, straight/posteriorly curved Eocaecilia Quadratojugal Yes Yes Large Short, straight/posteriorly curved

Table 2. Characters that are frequently overlooked or misinterpreted in phylogenetic analyses of amphibians. Acanthostega is used as an example of a primitive tetrapod (Clack, 2002). Note that characters of the cheek (jugal), ribs and vomers support the affinities of amniotes and lepospondyls, whereas humeral and palatal characters support the amphibamid-lissamphibian relationship, in addition to better-known characters such as dental pedicelly. Dashes indicate inapplicable characters.

174 CONCLUSIONS

Doleserpeton annectens is, in many ways, a typical amphibamid dissorophoid.

The position of this species within the Temnospondyli is therefore

uncontroversial. Its possession of features that point to affinities to frogs,

salamanders, and caecilians suggests a link between temnospondyls, which is the largest group of Paleozoic tetrapods, and modern amphibians. However, the interrelationships of the Temnospondyli and Dissorophoidea are not as well understood as some previous studies suggest. The previous data matrices

generally ignored similarities between dissorophoids and primitive

temnospondyls, thus making the dissorophoids seem more derived than they

really are (see discussion in Chapter 1). Dissorophoids are here suggested to have

split off early from the temnospondyl stem. This is based on the phylogenetic

analysis presented in Chapter 1, the resemblance of many dissorophoids to the

most primitive temnospondyls, and the early occurrence of Amphibamus in the

Westphalian D (Carboniferous).

The otic region and inner ear of Doleserpeton are closely comparable to modern

anurans (Chapter 2). The shape of the otic capsule, the orientation of the stapes

and the otic notch, and the presence of ridges that may have served as attachment

for an annulus, all suggest a hearing system that is closely comparable to modern anurans, even when ignoring observations that are generally unavailable from other early tetrapods (such as details of the inner ear). This confirms observations made by other authors (e.g. Bolt and Lombard, 1985; Robinson et al., 2005), who

175 have concluded that comparisons of tetrapod hearing systems demonstrate the

temnospondyl affinities of anurans. Salamanders and caecilians lack a tympanic

hearing system, but their inner ears contain an amphibian papilla and a posteriorly

positioned perilymphatic duct, as in anurans. The fact that such a duct was almost

certainly present in Doleserpeton (Chapter 2) strengthens the connection between

this form and modern amphibians.

Although some postcranial features have been discussed in the previous

literature on lissamphibian evolution (e.g. Milner, 1988; Carroll et al., 2004;

Carroll and Holmes, 2007), these characters have generally been given less attention than cranial features. Chapter 3 illustrates that the humeri and elbow joints of modern salamanders and frogs are more similar than previous descriptions suggest. Furthermore, the close resemblance of the humeri and elbow

morphology of batrachians, Eocaecilia, and Doleserpeton indicates that this may

be another important character complex linking all modern amphibians to temnospondyls. The forelimbs of modern anurans appear to be specialized for

absorbing the impact of jumping (Chapter 4). This feature is already apparent in

the Triassic form Triadobatrachus, possibly indicating that saltatory locomotion

was part of the locomotion of the most basal salientian known to science.

As noted above, there are various views of the relationships between modern

amphibians and Paleozoic tetrapods, and this thesis has generally confirmed the

observations of the Temnospondyl Hypothesis (e.g. Trueb and Cloutier, 1991;

Ruta and Coates, 2007). The Lepospondyl Hypothesis of Laurin and Reisz (1997)

and Vallin and Laurin (2004) shows that a number of features are shared between

176 lepospondyls and lissamphibians. However, the morphology of the characters in

question are rarely discussed in the papers supporting the Lepospondyl

Hypothesis, the authors often choosing to critique the characters supporting

opposing views (e.g. Laurin, 1998b). A review of the character matrix used by

Vallin and Laurin (2004) indicates that there are few, if any, character complexes

that convincingly link lissamphibians to the Lepospondyli (Chapter 5). This does

not mean that phylogenetic software, such as PAUP or MrBayes, connected these

forms based on nonexistent evidence. Some features are indeed shared between

lepospondyls and modern amphibians, but these frequently occur in other major

groups as well. A case in point is the loss of skull elements, which is a commonly

occurring phenomenon in nearly all branches of the tetrapod phylogeny. For

instance, modern mammals and lizards have far fewer skull elements than their relatives in the Carboniferous.

Phylogenetic software can only be expected to provide plausible when the data entered have been selected with care. Thus, part of the controversy

seems to be rooted in character coding. It has been shown that introducing

additions and corrections to the matrix can easily topple the phylogeny of

Anderson et al. (2008b), as exemplified in Chapter 5 and by Marjanovic and

Laurin (2009). However, this is not the case with the matrix of Vallin and Laurin

(2004).

The matrices of Laurin and Reisz (1997) and Vallin and Laurin (2004) differ

from that of Ruta and Coates (2007) in having relatively few shape-based

characters, and parts of the resulting phylogenies are therefore based mostly on

177 shared absences, as pointed out by Schoch and Milner (2004). The inclusion of a

variety of characters from several sources, as in the supermatrix presented in

Chapter 5, may help to counteract some of these individual preferences of character coding. Correcting the matrix and inserting new data on Doleserpeton

yielded the Temnospondyl Hypothesis as the most likely scenario. Although the

bootstrap analysis did not suggest a specific dissorophoid taxon as the

lissamphibian sister group, Doleserpeton is tentatively proposed as the sister-

taxon of modern amphibians based on the consensus tree and Bayesian analysis.

This is supported by features of the palate, otic region, and postcranial skeleton of

Doleserpeton, and it is therefore the preferred hypothesis of this study.

178 BIBLIOGRAPHY

Abourachid A, Green DM. 1999. Origins of the frog-kick? Alternate-leg

swimming in primitive frogs, families Leipelmatidae and Ascaphidae. Journal

of Herpetology 33: 657–663.

Amano KW. 1931. The labyrinth of Bufo vulgaris japonicus . Especially,

upon the morphologic study of the perilymphatic space, in connection with the

membranous labyrinth and auditory capsule. Journal of Morphology and

Physiology 51: 207–242.

Anderson JS. 2007. Incorporating ontogeny into the matrix: a phylogenetic

evaluation of developmental evidence for the origin of modern amphibians. In:

Anderson JS, Sues H-D, eds. Major transitions in vertebrate evolution.

Bloomington, IN: Indiana University Press, 182–212.

Anderson JS. 2008. Focal review: the origin(s) of modern amphibians.

Evolutionary Biology 35: 231–247.

Anderson JS, Henrici AC, Sumida SS, Martens T, Berman DS. 2008a.

Georgenthalia clavinasica, a new genus and species of dissorophoid

temnospondyl from the early Permian of Germany, and the relationships of the

family amphibamidae. Journal of Vertebrate Paleontology 28: 61–75.

Anderson JS, Reisz RR. 2003. A new microsaur (Tetrapoda: Lepospondyli)

from the Lower Permian of Richards Spur (Fort Sill), Oklahoma. Canadian

Journal of Earth Sciences 40: 499–505.

179 Anderson JS, Reisz RR, Scott D, Fröbisch NB, Sumida SS. 2008b. A stem

batrachian from the Early Permian of and the origin of frogs and

salamanders. Nature 453: 515–518.

Banbury B, Maglia AM. 2006. Skeletal development of the Mexican spadefoot,

Spea multiplicata (Anura: Pelobatidae). Journal of Morphology 267:803–821.

Báez AM, Basso NG. 1996. The earliest known frogs of the Jurassic of South

America: review and cladistic appraisal of their relationships. Münchner

Geowissenschaftliche Abhandlungen Reihe A (Geologie, Palaeontologie)

30:131–158.

Báez AM, Nicoli L. 2004. A new look at an old frog: the Jurassic

Reig from Patagonia. Ameghiniana 41: 257–270.

Báez AM, Rage J. 1998. Pipid frogs from the Upper Cretaceous of In Beceten,

Niger. Palaeontology 41: 669-691.

Bolt JR. 1968. The osteology and relationships of Doleserpeton annectens, a new

rhachitomous amphibian from the Lower Permian of Oklahoma. Ph. D.

Dissertation, University of Chicago.

Bolt JR. 1969. Lissamphibian origins: possible protolissamphibian from the

Lower Permian of Oklahoma. Science 166: 888–891.

Bolt JR. 1974. Evolution and functional interpretation of some suture patterns in

Paleozoic labyrinthodont amphibians and other lower tetrapods. Journal of

Paleontology 48: 434–458.

Bolt JR. 1977. Dissorophoid relationships and ontogeny, and the origin of the

Lissamphibia. Journal of Paleontology 51: 235–249.

180 Bolt JR. 1979. Amphibamus grandiceps as a juvenile dissorophoid: evidence and

implications. In: Nitecki MH, ed. Mazon Creek Fossils. New York, NY:

Academic Press, 529–563.

Bolt JR. 1991. Lissamphibian origins. In: Schultze HP, Trueb L, eds. Origins of

the Higher Groups of Tetrapods: Controversy and Consensus. Ithaca: Comstock

Publishing Associates, 194–222.

Bolt JR, Lombard RE. 1985. Evolution of the amphibian tympanic ear and the

origin of frogs. Biological Journal of the Linnean Society 24: 83-99.

Boy JA, Sues H-D. 2000. Branchiosaurs: larvae, and

heterochrony in temnospondyls and seymouriamorphs. In: Heatwole H, Carroll

RL, eds. Amphibian biology, volume 4: Paleontology. Chipping Norton, NSW:

Surrey Beatty and Sons, 1150-1197.

Brand DJ. 1956. The cranial morphology of Scolecomorphus uluguruensis

(Barbour and Loveridge). Annals of the Universty of Stellenbosch 32A: 1-25.

Bystrow AP, Efremov JA. 1940. sushkini Efr., a labyrinthodont

from the Eotriassic of Sharzhenga River. Trudy Paleozoologicheskogo Instituta

Akademia Nauk SSSR 10: 1–152. [in Russian].

Cannatella DC. 2006. An integrative phylogeny of Amphibia. In: Narins PM,

Feng AS, Fay RR, Popper AN. Eds. Hearing and Sound Communication in

Amphibians. New York, Springer Verlag, 12–43.

Cannatella DC, Hillis DM. 2004. Amphibians: Leading a life of slime. In:

Cracraft J, Donoghue MJ. Eds. Assembling the tree of life. New York, Oxford

University Press, 430–450.

181 Carlson KJ. 1987. Perryella, a new temnospondylous amphibian from the Lower

Permian of Oklahoma. Journal of Paleontology 61: 135–147.

Carroll RL. 1964. Early evolution of the dissorophid amphibians. Bulletin of the

Museum of Natural History Harvard University 131: 161–250.

Carroll RL. 2000. The lissamphibian enigma. In: Heatwole H, Carroll RL, eds.

Amphibian biology, volume 4: Paleontology. Chipping Norton, NSW: Surrey

Beatty and Sons, 1270-1273.

Carroll RL. 2001. The origin and early radiation of terrestrial vertebrates.

Journal of Paleontology 75: 1202–1213.

Carroll RL. 2007. The Palaeozoic ancestry of salamanders, frogs and caecilians.

Zoological Journal of the Linnean Society 150 (Suppl. 1): 1–140.

Carroll RL. 2009. The rise of amphibians: 365 million years of evolution.

Baltimore: Johns Hopkins University Press.

Carroll RL, Boisvert C, Bolt JR, Green DM, Philip N, Rolian C, Schoch RR,

Tarenko A. 2004. Changing patterns of ontogeny from osteolepiform

through Permian tetrapods as a guide to the early evolution of land vertebrates.

In: Arratia G, Wilson MVH, Cloutier R, eds. Recent advances in the origin and

early radiation of vertebrates. München, DE: Verlag Dr. Friedrich Pfeil, 321-

343

Carroll RL, Bossy KA, Milner AC, Andrews SM, Wellstead CF. 1998.

Encyclopaedia of paleoherpetology, part 1, Lepospondyli. München, DE:

Verlag Dr. Friedrich Pfeil. 216pp.

182 Carroll RL, Gaskill P. 1978. The order Microsauria. Memoirs of the American

Philosophical Society 126: 1–211.

Carroll RL, Holmes R. 1980. The skull and jaw musculature as guides to the

ancestry of salamanders. Zoological Journal of the Linnean Society 68: 1–40.

Carroll RL, Holmes RB. 2007. Evolution of the appendicular skeleton of

amphibians. In: Hall BK, ed. Fins into limbs. Chicago: University of Chicago

Press, 185–224.

Case SM. 1979. Observations on some cranial foramina in the Ranidae. Copeia 2:

346–348.

Clack JA. 1998. The neurocranium of Acanthostega gunnari Jarvik and the

evolution of the otic region in tetrapods. Zoological Journal of the Linnean

Society 122: 61–97.

Clack JA. 2002. Gaining Ground. The Origin and Evolution of Tetrapods.

Bloomington and Indianapolis: Indiana University Press.

Clack JA, Milner AR. 1993. Platyrhinops from the Upper Carboniferous of

Linton and Nyrany and the family Amphibamidae (Amphibia: Temnospondyli).

In: Heidtke U, ed. New research on Permo-Carboniferous faunas. Bad

Dürkheim, Pollichia-Buch 29, 185-191.

Clark J, Carroll RL. 1973. Romeriid from the Lower Permian. Bulletin

of the Museum of Comparative Zoology 144: 353–407.

Coates MI. 1996. The tetrapod Acanthostega gunnari Jarvik:

postcranial anatomy, basal tetrapod interrelationships and patterns of skeletal

183 evolution. Transactions of the Royal Society of Edinburgh: Earth Sciences 87:

363–421.

Colbert EH. 1952. A pseudosuchian from Arizona. Bulletin of the

American Museum of Natural History 99: 561–592.

Daly E. 1994. The Amphibamidae (Amphibia: Temnospondyli), with a

description of a new genus from the Upper of Kansas. University

of Kansas Miscellaneous Publications 85:1-59.

Davit-Beal T, Chisaka H, Delgado S, Sire JY. 2007. Amphibian teeth: current

knowledge, unanswered questions, and some directions for future research.

Biological reviews 82: 49–81.

De Beer GR. 1937. The development of the vertebrate skull. London, Oxford

University Press.

De Burlet HM. 1934. Vergleichende anatomie des stato-akustischen organs. A)

Die innere Ohrspäre; B) Die mittlere Ohrspäre. In: Bolk L, Göppert E, Kallius

E, Lubosch W. Eds. Handbuch der vergleichenden Anatomie der Wirbeltiere,

Vol. 2. Berlin: Urban und Schwarzenberg, 1293-1432.

De Queiroz A, Gatesy J. 2006. The supermatrix approach to systematics. Trends

in ecology and evolution 22: 34–41

DeMar R. 1968. The Permian labyrinthodont amphibian Dissorophus

multicinctus, and adaptations and phylogeny of the family .

Journal of Paleontology 42: 1210–1242.

184 Dilkes DW. 1990. A new trematopsid amphibian (Temnospondyli:

Dissorophoidea) from the Lower Permian of Texas. Journal of Paleontology 10:

222–243.

Duellman WE, Trueb L. 1986. Biology of amphibians. London, Johns Hopkins

University Press Ltd.

Duellman WE, Trueb L. 1994. Biology of amphibians, 2nd ed. Baltimore: Johns

Hopkins University Press.

Ecke H. 1934. Anatomische und histologische Untersuchungen am Labyrinth der

Erdkrote. (Bufo vulgaris Laur.). Zeitschrift für Morphologie und Ökologie der

Tiere 29: 79–113.

Emerson SB. 1978. and jumping in frogs: helping the twain to meet.

Evolution 32: 551–564.

Emerson SB. 1984. Morphological variation in frog pectoral girdles: testing

alternatives to a traditional adaptive explanation. Evolution 38: 376–388.

Emerson SB. 1988. Testing for historical patterns of change: a case study with

frog pectoral girdles. Paleobiology 14:174–186.

Estes R, Reig OA. 1973. The early fossil record of frogs: a review of the

evidence. In: Vial JL. Ed. Evolutionary biology of the anurans. Columbia:

University of Missouri Press, 11-63.

Evans FG. 1946. The anatomy and function of the foreleg in salamander

locomotion. Anatomical Record 95: 257–281.

Evans SE, Borsuk-Bialynicka M. 1998. A stem-group frog from the Early

Triassic of Poland. Acta Palaeontologica Polonica 43: 573–580.

185 Evans SE, Sigogneau-Russell D. 2001. A stem-group caecilian (Lissamphibia:

Gymnophiona) from the Lower Cretaceous of . Palaeontology

44: 259–273.

Francis ETB. 1934. The anatomy of the salamander. Ithaca, New York: Society

for the Study of Amphibians and Reptiles.

Fritzsch B. 1992. The water-to-land transition: Evolution of the tetrapod basilar

papilla, middle ear, and auditory nuclei. In: Webster DB, Fay RR, Popper AN.

Eds. The evolutionary biology of hearing. New York, Springer Verlag 351-375.

Fritzsch B, Wake MH. 1988. The inner ear of gymnophione amphibians and its

nerve supply: a comparative study of regressive events in a complex sensory

system (Amphibia, Gymnophiona). Zoomorphology 108: 201-217.

Fröbisch NB, Carroll RL, Schoch RR. 2007. Limb ossification in the Paleozoic

branchiosaurid Apateon (Temnospondyli) and the early evolution of preaxial

dominance in tetrapod limb development. Evolution and Development 9: 69–75.

Fröbisch NB, Reisz RR. 2008. A new Lower Permian amphibamid

(Dissorophoidea, Temnospondyli) from the fissure fill deposits near Richards

Spur, Oklahoma. Journal of Vertebrate Paleontology 28: 1015–1030.

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. Bulletin of the American

Museum of Natural History 297: 1–370.

186 Gans C, Parsons TS. 1966. On the origin of the jumping mechanism in frogs.

Evolution 20: 92-99.

Gauch H. 2003. Scientific method in practice. Cambridge University Press.

Gaupp E. 1904. A. Ecker and R. Wiedersheim's Anatomie des Frosches. Vols. 1-

2. Braunschweig, Friedrich Vieweg und Sohn.

Germain D. 2008. A new phlegethontiid specimen (Lepospondyli, Aisopoda)

from the Late Carboniferous of Montceau-les-Mines (Saône-et-Loire, France).

Geodiversitas 30: 669–680.

Godfrey SJ. 1997. Tetrapods (amphibians and reptiles). In: Shabica CW, Hay

AA, eds. Richardson’s guide to the fossil fauna of Mazon Creek. Illinois:

Northeastern Illinois University, Chicago, 256–269.

Gomes FR, Rezende EL, Grizante MB, Navas CA. 2009. The evolution of

jumping performance in anurans: morphological correlates and ecological

implications. Journal of Evolutionary Biology 22: 1088–1097.

Gregory JT. 1950. Tetrapods of the Pennsylvanian nodules from Mazon Creek,

Illinois. American Journal of Science 248: 833–873.

Gregory JT, Peabody FE, Price LI. 1956. Revision of the Gymnarthridae,

American Permian Microsaurs. Bulletin of the Peabody Museum of Natural

History 10:1–77.

Gubin YM. 1980. New Permian dissorophids of the Ural Forelands.

Paleontologicheskiy Zhurnal 1980: 82–90 [In Russian].

Hamilton DW. 1964. The inner ear of lizards. Journal of Morphology 115: 255–

271.

187 Holmes R. 1977.The osteology and musculature of the pectoral limb of small

captorhinids. Journal of Morphology 152: 101-140.

Holmes R. 1980. scheelei and the early evolution of the

labyrinthodont pectoral limb. In: Panchen AL, ed. The terrestrial environment

and the origin of land vertebrates. London, EN: Academic Press, 351–376.

Holmes RR. 2000. Paleozoic temnospondyls. In: Heatwole H, Carroll RL, eds.

Amphibian biology, volume 4: Paleontology. Chipping Norton, NSW: Surrey

Beatty and Sons, 1081-1120.

Holmes RB, Carroll RL, Reisz RR. 1998. The first articulated skeleton of

Dendrerpeton acadianum (Temnospondyli, Dendrerpetontidae) from the Lower

Pennsylvanian locality of , Nova Scotia, and a review of its

relationships. Journal of Vertebrate Paleontology 18: 64–79.

Hugall AF, Foster R, Lee MSY. 2007. Calibration choice, rate smoothing, and

the pattern of tetrapod diversification according to the long nuclear gene RAG-

1. Systematic Biology 50: 543–563.

Huttenlocker AK, Pardo JD, Small BJ. 2007. Plemmyradytes shintoni, gen. et

sp. nov., an Early Permian amphibamid (Temnospondyli: Dissorophoidea) from

the Eskridge Formation, Nebraska. Journal of Vertebrate Paleontology 27:

316–328.

Igawa T, Kurabayashi A, Usuki C, Fujii T, Sumida M. 2008. Complete

mitochondrial genomes of three neobatrachian anurans: A case study of

divergence time estimation using different data and calibration settings. Gene

407: 116–129.

188 Ivachnenko MF. 1978. Urodelans from the Triassic and Jurassic of Soviet

Central Asia. Paleontologicheskiy Zhurnal 1978: 84–89. [In Russian]

Jenkins FA. 1973. The functional anatomy and evolution of the mammalian

humero-ulnar articulation. American Journal of Anatomy 137: 281–298.

Jenkins FA, Shubin NH. 1998. Prosalirus bitis and the anuran caudopelvic

mechanism. Journal of Vertebrate Paleontology 18: 495–510.

Jenkins FA, Walsh DM, Carroll RL. 2007. Anatomy of Eocaecilia micropodia,

a limbed gymnophionan of the . Bulletin of the Museum of

Comparative Zoology Harvard 158: 1-81.

Laurin M. 1998a. The importance of global parsimony and historical bias in

understanding tetrapod evolution. Part I. Systematics, middle ear evolution and

jaw suspension. Annales des Sciences Naturelles, Zoologie, 13e Serie 19: 1-42.

Laurin M. 1998b. The importance of global parsimony and historical bias in

understanding tetrapod evolution. Part II. Vertebral centrum, costal ventilation,

and paedomorphosis. Annales des Sciences Naturelles, Zoologie, 13e Serie 19:

99-114.

Laurin M, Reisz R. 1997. A new perspective on tetrapod phylogeny. In: Sumida

SS, Martin KL, eds. Amniote origins. New York, Academic Press, 9-59.

Laurin M, Reisz RR. 1999. A new study of Solenodonsaurus janenschi, and a

reconsideration of amniote origins and stegocephalian evolution. Canadian

Journal of Earth Sciences 36: 1239–1255.

Liem KF, Bemis WE, Walker WF, Grande L. 2001. Functional anatomy of the

vertebrates. Third edition. Belmont: Brooks/Cole - Thomson Learning.

189 Lombard RE. 1977. Contributions to vertebrate evolution 2: Comparative

morphology of the inner ear in salamanders (Caudata: Amphibia). New York,

S.Karger.

Lombard RE. 1980. The structure of the amphibian auditory periphery: a unique

experiment in terrestrial hearing. In: Popper AN, Fay RR, eds. Comparative

studies of hearing in vertebrates. New York, Springer-Verlag. 121-138.

Lombard RE, Bolt JR. 1979. Evolution of the tetrapod ear: an analysis and

reinterpretation. Biological Journal of the Linnean Society 11: 19-76.

Lombard RE, Bolt JR. 1988. Evolution of the stapes in Paleozoic tetrapods. In:

Fritzsch B, Ryan MJ, Wilczynski w, Hetherington TE, Walkowiak W, eds.

Evolution of the amphibian auditory system. New York, Wiley & Sons, 37-67.

Maglia AM, Pugener LA, Mueller JM. 2007. Skeletal morphology and

postmetamorphic ontogeny of Acris crepitans (Anura: Hylidae): a case of

miniaturization in frogs. Journal of Morphology 268: 194–223.

Marjanovic D, Laurin M. 2008. A reevaluation of the evidence supporting an

unorthodox hypthesis on the origin of extant amphibians. Contributions to

Zoology 77: 149–199.

Marjanovic D, Laurin M. 2009. The origins(s) of modern amphibians: a

commentary. Evolutionary Biology 36: 336–338.

Mason MJ. 2007. Pathways for sound transmission to the inner ear in

amphibians. In: Narins PM, Feng AS, Fay RR, Popper AN. Eds. Hearing and

Sound Communication in Amphibians. New York, Springer Verlag, 147–183.

190 McGowan GJ. 2002. Albanerpetontid amphibians from the Lower Cretaceous of

Spain and Italy: a description and reconsideration of their systematics.

Zoological Journal of the Linnean Society 135: 1–32.

Milner AR. 1982. Small temnospondyl amphibians from the middle

Pennsylvanian of Illinois. Palaeontology 25: 635–664.

Milner AR. 1988. The relationships and origin of living amphibians. In: Benton

MJ, ed. The phylogeny and classification of the tetrapods, Volume 1:

Amphibians, Reptiles, Birds. Oxford: Clarendon Press, 59-102.

Milner AR. 2000. and Caudata and Albanerpetontidae. In:

Heatwole H, Carroll RL, eds. Amphibian biology, volume 4: Paleontology.

Chipping Norton, NSW: Surrey Beatty and Sons, 1412-1444.

Milner AR, Sequeira SEK. 1994. The temnospondyl amphibians from the

Visean of East Kirkton, , . Transactions of the Royal

Society of Edinburgh: Earth Sciences 84: 331–361.

Miner RW. 1925. The pectoral limb of Eryops and other primitive tetrapods.

Bulletin, American Museum of Natural History 51: 145–312.

Moulton JM. 1974. A description of the vertebral column of Eryops based on the

notes and drawings of A. S. Romer. Breviora, Museum of Comparative Zoology

428: 1–44.

Nauwelaerts S, Aerts P. 2006. Take-off and landing forces in jumping frogs.

Journal of Experimental Biology 209: 66–77.

Olson EC. 1941. The family Trematopsidae. Journal of Geology 49: 149–176.

191 Parsons TS, Williams EE. 1962. The teeth of Amphibia and their relation to

amphibian phylogeny. Journal of Morphology 110: 375–389.

Parsons TS, Williams EE. 1963. The relationships of the modern Amphibia: a

re-examination. Quarterly Review of Biology 38: 26–53.

Pawley K, Warren A. 2006. The appendicular skeleton of Eryops megacephalus

Cope, 1877 (Temnospondyli: ) from the Lower Permian of North

America. Journal of Paleontology 80: 561–580.

Piveteau J. 1937. Paléontologie de Madagascar XXIII. Un amphibien du Trias

inférieur. Essai sur l’origine et l'évolution des amphibians anoures. Annales de

Paléontologie 26: 135–177.

Platt C, Jørgensen JM, Popper AN. 2004. The inner ear of the

Protopterus. Journal of Comparative Neurology 471: 277-288.

Přikryl T, Aerts P, Havelková P, Herrel A, Roček Z. 2009. Pelvic and thigh

musculature in frogs (Anura) and origin of anuran jumping locomotion. Journal

of Anatomy 214: 100–139.

Pugener LA, Maglia AM. 1997. Osteology and skeletal development of

Discoglossus sardus (Anura: Discoglossidae). Journal of Morphology 231:

267–285.

Pusey HK. 1943. On the head of the Liopelmid frog, Ascaphus truei. I. The

chondrocranium, jaws, arches, and muscles of a partly-grown larva. Quarterly

Journal of Microscopical Science 84: 105–185.

192 Rage J, Roček Z. 1989. Redescription of Triadobatrachus massinoti (Piveteau,

1936) an anuran amphibian from the early Triassic. Palaeontographica Abt. A

206: 1–16.

Reese AM. 1906. Anatomy of Cryptobranchus allegheniensis. The American

Naturalist 40: 287–326.

Reisz RR. 1980. A protorothyridid captorinomorph reptile from the lower

Permian of Oklahoma. Life Sciences Contributions, Royal Ontario Museum

121: 1–16.

Reisz RR. 1981. A diapsid reptile from the Pennsylvanian of Kansas. Special

Publications of the Museum of Natural History, University of Kansas 7: 1–74.

Rieppel O. 1980. The perilymphatic system of the skull of Typhlops and

Acrochordus, with comments on the origin of . Journal of Herpetology

14: 105-108.

Ritland RM. 1955. Studies on the post-cranial morphology of Ascaphus truei.

Journal of Morphology 97: 215–282.

Robinson J, Ahlberg PE, Koentges G. 2005. The braincase and middle ear

region of Dendrerpeton acadianum (Tetrapoda: Temnospondyli).Zoological

Journal of the Linnean Society 143: 577-597.

Roček Z. 2000. Mesozoic anurans. In: Heatwole H, Carroll RL, eds. Amphibian

Biology, Volume 4: Palaeontology. Chipping Norton: Surrey Beatty & Sons

PTY Limited, 1295–1331.

193 Roček Z, Rage JC. 2000. Proanuran stages (Triadobatrachus, Czatkobatrachus).

In: Heatwole H, Carroll RL, eds. Amphibian Biology, Volume 4: Palaeontology.

Chipping Norton, NSW: Surrey Beatty & Sons, 1282–1294.

Romer AS. 1956. Osteology of the reptiles. Chicago: University of Chicago

Press.

Romer AS. 1969. The cranial anatomy of the Permian amphibian Pantylus.

Breviora, Museum of Comparative Zoology 314: 1–37.

Romer AS, Witter RV. 1942. Edops, a primitive rhachitomous amphibian from

the Texas Red Beds. Journal of Geology 50: 925–960.

Ronquist F, van der Mark P, Huelsenbeck JP. 2009. Bayesian phylogenetic

analysis using MrBayes. In: Lemey P, Salemi M, Vandamme A-M, eds. The

phylogenetic handbook. New York: Cambridge University Press, 210–266.

Rose CS. 2003. The developmental morphology of salamander skulls. In:

Heathole H, Davies M, eds. Amphibian Biology, Volume 5. Osteology. Chipping

Norton, NSW: Surrey Beatty and Sons Pty. Ltd., 1686-1783.

Ruta M. 2009. Patterns of morphological evolution in major groups of Palaeozoic

Temnospondyli (Amphibia: Tetrapoda). Special Papers in Palaeontology 81:

91-120.

Ruta M, Coates MI. 2007. Dates, nodes and character conflict: addressing the

lissamphibian origin problem. Journal of Systematic Palaeontology 5: 69-122.

Ruta M, Coates MI, Quicke DLJ. 2003. Early tetrapod relationships revisited.

Biological Reviews 78: 251-345.

194 San Mauro D, Gower DJ, Oommen OV, Wilkinson M, Zardoya R. 2004.

Phylogeny of caecilian amphibians (Gymnophiona) based on complete

mitochondrial genomes and nuclear RAG1. Molecular and

Evolution 33: 413–427.

San Mauro D, Vences M, Alcobendas M, Zardoya R, Meyer A. 2005. Initial

diversification of living amphibians predated the breakup of Pangea. The

American Naturalist 165: 590–599.

Sawin HJ. 1941. The cranial anatomy of Eryops megacephalus. Bulletin of the

Museum of Comparative Zoology, Harvard 88: 407–463.

Schoch RR. 1999a. Studies on braincases of early tetrapods: structure,

morphological diversity, and phylogeny – 1. Trimerorachis and and other

primitive temnospondyls. Neues Jahrbuch für Geologie und Paläontologie,

Abhandlungen 213: 233–259.

Schoch RR. 1999b. Studies on braincases of early tetrapods: structure,

morphological diversity, and phylogeny – 2. Dissorophoids, eryopids, and

stereospondyls. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen

213: 289–312.

Schoch RR, Carroll RL. 2003. Ontogenetic evidence for the Paleozoic ancestry

of salamanders. Evolution and Development 5: 314–324.

Schoch RR, Milner AR. 2004. Structure and implications of theories on the

origin of lissamphibiansin. In: Arratia G, Wilson MVH, Cloutier R, eds. Recent

advances in the origin and early radiation of vertebrates. Munich: Verlag Dr.

Friedrich Pfeil, 345–377.

195 Schoch RR, Rubidge BS. 2005. The amphibamid Micropholis from the

Lystrosaurus Assemblage Zone of South Africa. Journal of Vertebrate

Paleontology 25: 502–522.

Sedra SN, Michael ML. 1959. The ontogenesis of the sound conducting

apparatus of the Egyptian , Bufo regularis Reuss, with a review of this

apparatus in Salientia. Journal of Morphology 104: 359–375.

Shubin NH, Jenkins FA. 1995. An early Jurassic jumping frog. Nature 377: 49–

52.

Sigurdsen T. 2008. The otic region of Doleserpeton (Temnospondyli) and its

implications for the evolutionary origin of frogs. Zoological Journal of the

Linnean Society 154: 738–751. (Chapter 2)

Sigurdsen T, Bolt J. 2009. The Lissamphibian Humerus and Elbow Joint, and the

Origins of Modern Amphibians. Journal of Morphology [in press] (Chapter 3)

Smithson TR, Carroll RL, Panchen AL, Andrews SM. 1994.

lizziae from the Viséan of East Kirkton, West Lothian, Scotland, and the

amniote stem. Transactions of the Royal Society of Edinburgh: Earth Sciences

84: 383–412.

Starck D. 1979. Vergleichende Anatomie der Wirbeltiere, Band 2, das

Skeletsystem. New York, Springer-Verlag.

Stephenson EM. 1951. The anatomy of the head of the New Zealand frog,

Leiopelma. Transactions of the Royal Society of London 27: 255–305.

196 Sullivan C, Reisz RR, May WJ. 2000. Large dissorophoid skeletal elements

from the Lower Permian Richards Spur fissures, Oklahoma, and their

paleoecological implications. Journal of Vertebrate Paleontology 20: 456–461.

Sumida SS. 1997. Locomotor features of taxa spanning the origin of amniotes.

In: Sumida SS, Martin KLM, eds. Amniote origins: completing the transition to

land. New York: Academic Press, 353–398.

Swanepoel JH. 1966. Contributions to the cranial morphology of Chiromantis

xerampelina. Annale Universiteit van Stellenbosch, series A 41: 655–684.

Trueb L, Cloutier R. 1991. A phylogenetic investigation of the inter- and

intrarelationships of the Lissamphibia (Amphibia: Temnospondyli). In: Schultze

HP, Trueb L, eds. Origins of the Higher Groups of Tetrapods: Controversy and

Consensus. Ithaca: Comstock Publishing Associates, 223–313.

Vallin G, Laurin M. 2004. Cranial morphology and affinities of ,

and a reappraisal of the phylogeny and lifestyle of the first amphibians. Journal

of Vertebrate Paleontology 24: 56–72.

Van Zyl JHM. 1950. Die beskywende en vergelykende anatomie van die skedel

van Discoglossus pictus (Gravenhorst). Annale van die Universiteit van

Stellenbosch 26: 3–26.

Walthall JC, Ashley-Ross MA. 2006. Postcranial myology of the California

Newt, Taricha torosa. Anatomical Record 288A: 46–57.

Wang Y, Jones MEH, Evans SE. 2007. A juvenile anuran from the Lower

Cretaceous Jiufotang Formation, Liaoning, China. Cretaceous Research 28:

235–244.

197 Warren AA, Snell N. 1991. The postcranial skeleton of Mesozoic temnospondyl

amphibians: a review. Alcheringa 15: 43–64.

Watson DMS. 1940. The origin of frogs. Transactions of the Royal Society of

Edinburgh 60: 195–231.

Wells KD. 2007. The ecology and behavior of amphibians. Chicago: University

of Chicago Press.

Wellstead CF. 1991. Taxonomic revision of the Lysorophia, Permo-

Carboniferous Lepospondyl amphibians. Bulletin, American Museum of Natural

History 209: 1–90.

Wever EG. 1985. The amphibian ear. Princeton, Princeton University Press.

Williston SW. 1910a. Cacops, Desmospondylus; new genera of Permian

vertebrates. Geological Society of America, Bulletin 21: 249–284.

Williston SW. 1910b. Dissorophus Cope. Journal of Geology 18: 526–536.

Yates AM, Warren AA. 2000. The phylogeny of the ‘higher’ temnospondyls

(Vertebrata: Choanata) and its implications for the monophyly and origins of the

Stereospondyli. Zoological Journal of the Linnean Society 128: 77–121.

Zhang P, Zhou H, Chen Y-Q, Liu Y-F, Qu L-H. 2005. Mitogenomic

perspectives on the origin and phylogeny of living amphibians. Systematic

Biology 54: 391-400.

Zug GR. 1972. Anuran locomotion: structure and function. I. Preliminary

observations on relation between jumping and osteometrics of appendicular and

postaxial skeleton. Copeia 1972: 613-622.

198 APPENDIX 1

Doleserpeton annectens specimens used in this study.

FMNH = Field Museum of Natural History, Chicago

Specimen number Description

FMNH PR 2823 Left dentary FMNH PR 2822 Right maxilla FMNH UR1296 Skull (dorsal) and partial postcranial skeleton FMNH UR1297 Quadrate FMNH UR1299 Premaxilla FMNH UR1300 Two nasals FMNH UR1301 Prefrontal FMNH UR1302 Frontal FMNH UR1303 Three parietals FMNH UR1304 Three postparietals FMNH UR1305 Two tabulars FMNH UR1306 Various elements incl. , skull elements FMNH UR1307 Skull and postcranial elements (ventral view) FMNH UR1308 Holotype, skull and partial postcranial skeleton FMNH UR1309 Vertebrae FMNH UR1310 Jugal FMNH UR1311 Two squamosals FMNH UR1314 Various elements, right prootic, postfrontal FMNH UR1316 Two exoccipitals FMNH UR1317 Vertebral column, opisthotic (diff. individual) FMNH UR1319 Postcranial elements, including tail FMNH UR1320 Various elements,incl. hind limb, skull elements FMNH UR1321 Right forelimb, skull elements incl. Prootic FMNH UR1322 Partial skull, well-preserved braincase FMNH UR1323 Skull preserved in ventral view FMNH UR1333 Left side of braincase FMNH UR1335 Lower jaw, posterior part FMNH UR1344 Lacrimal FMNH UR1345 Septomaxilla FMNH UR1346 Two vomers FMNH UR1350 Sphenethmoid FMNH UR1351 Stapes (2 specimens) FMNH UR1353 Lower jaw FMNH UR1357 Disarticulated ribs FMNH UR1358 Two cleithra FMNH UR1359 Right clavicle FMNH UR1360 Scapulocoracoid FMNH UR1361 Humerus

199 Specimen number Description

FMNH UR1362 Left sacral rib FMNH UR1367 Disarticulated phalanges FMNH UR1367 Right premaxilla FMNH UR1368 Right premaxilla FMNH UR1369 Two parasphenoids FMNH UR1364 Right femur FMNH UR1370 Left palatine bone FMNH UR1371 Left maxilla FMNH UR1372 Lower jaw, anterior part FMNH UR1377 Left radius FMNH UR1378 Left ulna FMNH UR1379 Right side of pelvis FMNH UR1380 Left femur FMNH UR1381 Left tibia FMNH UR1382 Left fibula

200 APPENDIX 2

Comments and changes to the character list of Fröbisch and Reisz (2008)

The following is a list of changes to the data matrix of Fröbisch and Reisz (2008), which was based on that of Schoch and Rubidge (2005). The character numbers and descriptions (in bold) are taken from Fröbisch and Reisz (2008), followed by comments by the present authors.

Character 2. Dorsal quadrate process: quadrate having smooth posterodorsal side in plesiomorphic state (0); quadrate with prominent dorsoposterior outgrowth, the quadrate process (1). The dorsal process of the quadrate is present (1) in Dendrerpeton (Milner, 1980; Holmes et al., 1998; Robinson et al., 2005).

Character 3. Vomerine depression: ventral surface of vomers flat and element divided into anterior and posterior portion by transverse ridges that may or may not bear transverse tooth row (0); single unpaired depression in anterior portion of vomers that may or may not house an opening (1). Vomerine depressions occur sporadically in different temnospondyls, and sometimes individual differences occur. Some of the variation may be due to specimens having been flattened due to compression, erasing the vomerine depression. The transverse ridges mentioned seem to actually mark the posterior border of this depression in Eryops. Contrary to previous character matrices, some specimens of both Amphibamus (YPM 794) and Doleserpeton (UR 1346) show a depression devoid of denticles in the same area as in Tersomius. Bolt (1977) proposed that Doleserpeton was different from Tersomius in this regard, but new observations show this not to be the case. The value of this trait is therefore questioned. This character has been removed.

Character 6. Parasphenoid basal plate: basal plate roughly quadrangular, as long as wide (0); basal plate much shorter than wide, reaching about half the width (1). The difference in relative basal plate width between Broiliellus and Doleserpeton is actually surprisingly small, but the character seems to be otherwise sound. Dissorophinae coded polymorphic (01).

Character 9. Palatine fangs: palatine with one fang pair in its anterior third (0); palatine fangs replaced by patch of small, equally sized teeth (1). The teeth on the palatine bone of Amphibamus are larger than the marginal dentition (e.g. specimen FMNH UC2000). Amphibamus grandiceps neotype specimen YPM 794 has fangs on the palatine. Coded (01).

Character 10. Ectopterygoid fangs: ectopterygoid with one fang pair at about mid level (0); ectopterygoid without fangs (1). The ectopterygoid is absent or greatly reduced in Doleserpeton, and the absence or presence of ectopterygoid fangs should be coded as inapplicable (-).

201

Character 12. Pterygoid flange: palatine ramus of pterygoid merging continuously into basipterygoid ramus (0); palatine ramus broadening abruptly to form transverse flange (1). It is not clear how the medial basipterygoid flange and articulation of Dendrepeton is any different (or a less abrupt broadening of the bone) than, say, Tersomius. Nor is Balanerpeton (Milner and Sequeira, 1993) very different from dissorophoids in this respect. Stereospondyls may have a more gradual transition, but this is probably not a primitive trait, contrary to the previous coding. Removed.

Character 13. Pterygoid-palatine: pterygoid (palatine ramus) reaching and suturing to palatine (0); pterygoid contacting only ectopterygoid, or, when latter absent, maxilla (1). This character is less than clear in many of the taxa in question, except for the most primitive forms. The primitive condition is to have a long suture between palatine and pterygoid, excluding the palatine from the margin of the interpterygoid vacuity. The pterygoid seems to reach the palatine in nearly all the forms in question. In Doleserpeton, the palatine meets the pterygoid posteriorly ventral view, although part of the palatine bone actually overlaps the pterygoid dorsally. If there was an ectopterygoid, it would have been reduced to smaller size than that of Gerobatrachus. This feature is unclear in Amphibamus, but it appears to be similar to Doleserpeton. New coding: Long medio-lateral contact between palatine and pterygoid (0); contact between the two bones reduced or absent (1).

Character 15. Interpterygoid vacuity: interpterygoid vacuity roundish or oval in outline (0); interpterygoid vacuity greatly expanded laterally at mid-level (1). Although trematopids sometimes appear to have smaller interpterygoid vacuities than other dissorophoids (Olson, 1941), others such as Phonerpeton (Dilkes, 1990) have very large openings. However, there is a general difference in shape of the vacuities, with those of trematopids being elongated (0) while those of armored dissorophids and amphibamids are wider, with the palatine bone usually constituting part of the rim of the vacuity (1). Part of the difference seems to be related to the extent of the pterygoid and thickness of palatine bone. It may therefore not be a completely independent character, but has been retained here. Kamacops (Gubin, 1980) has relatively small vacuities, but this condition is clearly secondarily derived, judging from the shape and location of the openings. Plemmyradytes has been coded as other amphibamids (1).

Character 16. Choana: choana narrow, forming elongate oval with parallel, parasagittal lateral and medial margins (0); choana expanded anteromedially (1). The outlines of the choanae of Sclerocephalus hardly differ from those of Doleserpeton. Dendrepeton is different because of the presence of tusks lateral to the choanae. This character has been removed.

202 Character 17. Pedicellate teeth: adult dentition of marginal jaw and palatal elements consisting of conical, fully mineralized teeth with labyrinthine infolding of enamel and dentine (0); jaw and/or palatal teeth—in larger, presumably adult stages—subdivided by unmineralized zone forming separate tooth crown (1). This character is more problematic than it may seem. The teeth of Doleserpeton are widely accepted as being pedicellate, as it has a clear pedicel consisting of a cylindrical, tube-like base. This base is separated from the crown by a line of weakness, where the crown often breaks off (for a picture of Doleserpeton pedicels, see Bolt, 1991). This condition is very similar to that seen in salamanders, frogs, and Eocaecilia (Parsons and Williams, 1962; Jenkins et al., 2007). Among other amphibamids, the most similar type of dentition is seen in Amphibamus, in which the base is raised from the surrounding bone, and the crown morphology resembles that of Doleserpeton. However, the base is not an elongate cylinder as in Doleserpeton, nor do the teeth usually break off at the level immediately beneath the crown (observations based on neotype of A. grandiceps, YPM 794). These factors seem to point to a condition other than true pedicelly. The published figures of the bases of Gerobatrachus (Anderson et al. 2008) also do not to show such cylindrical bases, nor does the line of weakness described in this form appear to be as clear-cut as in lissamphibians and Doleserpeton. Thus, Doleserpeton remains the only amphibamid in which the dentition is clearly pedicellate, although a similar dentition may also be present in Apateon (Carroll, 2007). Amphibamus, Gerobatrachus, and Apateon have been recoded (?).

Character 18. Biscuspid teeth: larval and adult teeth conical or needle shaped with single tip (monocuspid) (0); jaw and/or palatal teeth in larger, presumably adult, stages bicuspid (1). The dentition of some Tersomius specimens may be intermediate and are highly variable (Bolt, 1979). It has been coded as polymorphic (01).

Character 21. Prefrontal process: prefrontal forming simple suture with lacrimal laterally (0); prefrontal underplating lacrimal widely by means of ventral prefrontal process contacting palatine (1). The ventral contact between the prefrontal and the palatine below the lacrimal has been described as an unambiguous synapomorphy of dissorophoids (Fröbisch and Reisz, 2008). However, according to Holmes et al. (1998) there is a tongue-like prefrontal process in this region of the primitive temnospondyl Dendrerpeton (recoded [?]). This process may have contacted the palatine bone, although this is uncertain because of the distortion of the skull. Judging from the available photographs of Balanerpeton (Milner and Sequeira, 1994), the prefrontal and palatine appear to approach each other ventrally in this form as well (recoded [?]).

Character 23. Tabular-squamosal: tabular and squamosal widely separated by supratemporal (0); squamosal meeting tabular, excluding supratemporal from otic notch (1). The degree of ossification in the region of the otic notch affects several characters, including the development of a contact between the

203 squamosal and tabular (because of the development of the supratympanic flange), which, in turn, results in a narrower opening of the notch as seen in lateral view. This feature also seems to be linked to the posterior closing of the notch. The fact that the fully closed condition may reflect a closer relationship between Dissorophidae and Ecolsonia is not reflected. Characters 23 and 61 have therefore been rephrased. Character 23, new coding: Smooth inner surface of otic notch not extending dorsally within notch, or only forming a narrow edge on supratemporal (0). Inner rim of otic notch continuing dorsally to form extended supratympanic flange, including a tabular-squamosal contact (1).

Character 25. Squamosal-supratemporal: suture between squamosal and supratemporal nearly as long as supratemporal itself (0); foreshortened squamosal-supratemporal suture reaching only one third or less length of supratemporal (1). The suture of the supratemporal and squamosal is long (0) in Trematopidae (Dilkes, 1990), unlike previous coding. It is short (1) in Balanerpeton.

Character 26. Supratympanic flange (semilunar flange in Schoch and Rubidge 2005, terminology following Bolt (1974)): squamosal continuously ornamented around margin of otic notch (0); squamosal having dorsally exposed and ornamented area (supratympanic flange) stepping abruptly into steeply aligned, poorly ornamented portion (1). The supratympanic flange of the squamosal is (in general) a non-ornamented dorsal part of the otic notch including parts of the tabular supratemporal and squamosal bones. It is hardly represented on the squamosal in Doleserpeton and Amphibamus. More so in Tersomius and Eoscopus, but not at all in Plemmyradytes (Huttenlocker et al., 2007). This region of the squamosal is the same as that reaching the tabular below the supratemporal in character 23, and it is hard to justify including it as a separate character. Removed.

Character 27. Supratemporal length: supratemporal and parietal elongated, having similar length, reaching more than double the length of all other bones in posterior skull table (0); these elements posteriorly abbreviated, supratemporal markedly shorter than parietal (1). The supratemporals of Doleserpeton and Amphibamus are not markedly shorter relative to the parietal than those of Dendrerpeton and Balanerpeton. This character is related to, and has many of the same problems as, the “abbreviated skull table”. The real difference in skull shape between Balanerpeton and Dendrerpeton on one hand and Doleserpeton and Amphibamus on the other seems to be a larger otic notch and orbit in the latter forms, constricting the postorbital bar. However, this character (the large notch) is also found in other dissorophoids, although often in a more elongate form rather than the rounded notch found in amphibamids. Balanerpeton also has large orbits, but a more restricted otic notch. Removed.

Character 28. Semilunar flange (=supratemporal flange of Schoch and Rubidge, 2005): supratemporal without ventral projection into otic notch (0);

204 supratemporal forming marked ventral flange participating in medial bordering of otic notch (1). If the semilunar flange is defined as in Bolt (1974), then this character should constitute a smooth flange of the supratemporal in the upper part of the otic notch. This flange is roofed by the main part of the bone, which is sculpted in trematopids and armoured dissorophids. The best way to find the semilunar flange in amphibamids is to follow the edge of the otic notch from the squamosal on to the supratemporal. Anything below this edge on the latter bone is the semilunar flange if and only if it constitutes part of the dorsal and medial surface of the otic notch. There is a definite semilunar flange in the Doleserpeton specimen FMNH UR1322. It is a smooth flange in Doleserpeton, but this form lacks prominent sculpting of the dermal bones in general. A similar edge curving into the notch is found in Amphibamus (YPM 794) and Platyrhinops (BMNH R2670), but in these it is sculpted. The situation in Amphibamus appears to be similar to Dendrerpeton, and perhaps, Balanerpeton. The supratemporal forms a flat roof, rather than a curved edge, in Plemmyradytes (Huttenlocker et al. 2007). However, this region of the skull is frequently poorly ossified, as pointed out by the latter authors. Amphibamus and Platyrhinops recoded (?).

Character 30. Skull width: moderately wide skull with jugals, postorbitals, and medial skull roofing series usually longer than wide (0); skull table and cheek overall broadened, most elements being as wide as long or wider (1). Several amphibamids may be dimorphic (sexual dimorphism?), as was noted by Schoch and Rubidge (2005), and this may include Doleserpeton. The primitive condition is also not as clear as previous authors claim, as Balanerpeton has a wide skull. Removed.

Character 31. Palpebral : ossifications in orbit restricted to sclerotic ring (0); numerous palpebral ossicles at medial margin of sclerotic ring (1). Contrary to the previous character coding, Carroll (1964) described palpebral ossifications in several armored dissorophids. Balanerpeton may have had poorly ossified palpebral ossificatins (Milner and Sequeira 1994), but this form is best coded as primitive, as it appears not have had the well-ossified palpebral cup of dissorophoids.

Character 32. Stapes: stapes with pronounced dorsodistal curvature directed towards dorsally located otic notch (0); stapes abbreviated without dorsodistal curvature, directed laterally towards vertically aligned otic notch (1). The stapes of Doleserpeton is actually very similar to that of Dendrerpeton (Bolt and Lombard, 1985; Robinson et al., 2005). There is little “dorsodistal curvature towards dorsally located otic notch” in either of these. The same may be true for Balanerpeton (Milner and Sequeira, 1994) and Dissorophus (DeMar, 1968). In all of these, the stapes is located so that the stapedial shaft is directed (not curved) dorsodistally, but the shaft curves slightly antero-distally. Removed.

Character 34. Rib length: moderately elongate thoracic ribs in adult stage, may or may not bear uncinate processes and usually curved distoventrally

205 (0); thoracic ribs without distal curvature, but may bear uncinate processes (1); thoracic ribs very short, simple rods (2). Temnospondyl ribs tend to be shorter and more posteriorly curved than those of amniotes and microsaurs (when similar-sized animals are compared), in which the distal end of the rib points ventrally. This tendency appears to be particularly strong in small forms. This may be an important character for working out lissamphibian relationships (compare, for instance, Eocaecilia). However, the polarity of rib characters is not at all clear within temnospondyls. Describing the ribs of Doleserpeton as short simple rods is misleading (see description above). The fourth or fifth ribs are the largest ribs in Doleserpeton, Eoscopus and, apparently, in Balanerpeton (Milner and Sequeira, 1994). This may have been the case in Amphibamus grandiceps as well, but the available specimens are unclear in this region. In Dendrerpeton, larger ribs continue further posteriorly (16th rib). The ribs in all these taxa are curved postero-ventrally. Removed.

Character 35. Anterior rib morphology: anterior thoracic ribs usually larger and more solid than following ones, having massive proximal and distal ends (0); anterior thoracic ribs not longer than succeeding ones (1). Same problem as above. Removed.

Character 36. Humerus shaft: short and broad (0); substantially longer and more like rod than blade (2). This humeral character has been recoded. Humeral elongation and torsion: Extremely short shaft or no shaft present, little or no torsion between distal and proximal ends (0); relatively short shaft present, approximately 90 degrees torsion (1); long shaft (humerus approximately three times longer than wide), approximately 90 degrees torsion (2). Dendrerpeton and Balanerpeton both have character state 0. Amphibamids generally have state 2 although the shaft in Amphibamus specimen YPM 794 is so short as to warrant a ploymorphic coding in this taxon (12). Tersomius was described as having 45 degrees torsion by Daly (1994), but the humerus she based this observation on clearly has a broken shaft. The humeri of Tersomius otherwise resemble those of other amphibamids. Micromelerpetontids appear to have surprisingly primitive- looking humeri (Boy, 2002) these are shorter and appear to have less torsion than in branchiosaurids, but because of the lack of good descriptions the latter trait is uncertain. They have been coded as most other dissorophoids (1).

Character 37. Humerus supinator: supinator process arising from humerus in adults (0); absence of supinator throughout ontogeny (1). The “unfinished supinator process” reported by Daly (1994) in Tersomius appears to be too close to the distal articulating area to be a supinator process. Compare trematopids or Ecolsonia (Berman et al., 1985). Dendrerpeton and Balanerpeton lack the supinator process, but this may be due to incomplete ossification (the humeri of both appear to be partly cartilaginous), and these were coded as unknown (?).

Character 38. Humerus head: irrespective of length and slenderness of shaft, humeral condyles generally massive and widened (0); in adult amphibamids

206 and branchiosaurids, condyles minute and usually poorly ossified, giving only slightly broadened proximal head region having reduced deltopectoral buttress rather than crest (1). Far from being “minute”, the radial condyle (capitulum) of Doleserpeton dominates the distal end of the humerus to an even greater degree than in primitive temnospondyls (Chapter 3). Having a large capitulum appears to be more typical for temnospondyls than any other Paleozoic group (sometimes evidenced as a large gap, as in Balanerpeton and Edops), but this character is even more pronounced in Dissorophoids, and in particular, Doleserpeton. However, this character is missing from most amphibamids, and is useless for assessing relationships within the group. Removed.

Character 39. Interclavicle: rhomboidal, at least two times longer than wide (0); with abbreviated anterior and posterior ends and as wide as long (1); much shorter than wide, without anterior process (2). The shape of the interclavicle is unknown for many amphibamids. It is uncertain for Doleserpeton as well, although some parts of the bone remain. However, from what remains, the shape appears to be consistent within amphibamids. Unchanged.

Character 40. Ilium: with thin dorsal shaft, much higher than wide (0); dorsally abbreviated, with dorsoposterior process (1); very short, only as high as wide (2). The ilia of Doleserpeton, Eoscopus and Amphibamus all appear to be much more similar to Dendrerpeton and Balanerpeton than the impression given in the previous character coding. In fact, the pelvis of Doleserpeton is very similar to that of Dendrerpeton, with ilia having a long posteriorly directed bladelike process (0), although the latter is a bit shorter in Doleserpeton. Micropholis and Plemmyradytes appear to share a more robust-looking base of the iliac blade (Huttenlocker et al. 2007) (2). Furthermore, those of trematopids (Olson, 1941) and dissorophoids (Williston, 1910) are virtually indistinguishable, and similar to Eryops (Pawley and Warren, 2006), in being short and wide (1) although in the latter form the ilium leans slightly anteriorly. New coding: shape of iliac blade: long posterior process (0); short and wide dorsal process (1); pointing posteriorly, but with a robust base (2).

Character 41. Cleithrum head: aligned along anterior rim of scapula (0); posterodorsally enlarged, wrapping around scapula dorsally (1). The cleithrum is probably similarly extensive in both trematopids and armored dissorophoids (compare Olson, 1941 and Williston, 1910), but this character may be wholly dependent on the amount of ossification of the pectoral girdle and manner of preservation of the cleithrum . The authors are of the opinion that this character may be suspect, due to frequently poorly preserved cleithra. Removed.

Character 42. Cleithrum size: cleithrum with large dorsal head much wider than shaft (0); cleithrum forming simple rod, without any head (1). Due to the problems with the cleithrum often being poorly preserved, this character, which overlaps with the previous one, has been removed.

207

Character 43. Scapula: forming low or moderately high element, depending on degree of ossification, about two times longer than wide (0); dorsally much extended, being three or four times longer than wide (1). Unlike the previous coding, Eryops has a very high scapular part of the scapulocoracoid (1).

Character 44. Presacral vertebral count: 24-25 or more (0); 20 or fewer (1). Doleserpeton’s presacral vertebral count is about 22-24. State 1 should presumably be 23 or fewer. Doleserpeton has been coded {01}.

Character 47. Tail: bony tail long, at least length of trunk, often exceeding length of trunk and skull (0); bony tail foreshortened, markedly shorter than trunk (1). The length of the tail is unknown in Dendrerpeton and Balanerpeton. However, the former appears to have had a short tail (Holmes et al., 1998).

Character 48. Postfrontal shape: broadly quadrangular (0); falciform or sickleshaped (1) (Anderson et al., 2008b). The falciform postfrontal shape seems to rely on the presence of a thin anterior process of the postfrontal, which is not a new structure, but a thinning of the anterior parts of the bone. It is present in Balanerpeton, making the validity of the derived state of this character somewhat questionable (unchanged).

Character 49. Otic notch to orbit: far (0); medium (1); near (2) (Anderson et al., 2008a). This character is somewhat vague. Micropholis appears to vary from having an otic notch smaller than Dendrepeton to having a notch larger than latter. In some anurans (e.g. Rana catesbeiana), there are sexual differences in tympanum size. We suggest that the different morphs of Micropholis may reflect sexual dimorphism, if not species-level differences. The primitive state is not at all typical of armored dissorophids, several of which have otic notches situated close to the orbit (e.g. the notch of Broiliellus is closer to the orbit than that of Tersomius, the same is true for Dissorophus and, to a less degree, Cacops. Notch to orbit length by skull table width: Dendrerpeton 0.54, Balanerpeton 0.6, Eryps 0.44, Phonerpeton 032, Doleserpeton 0.27, Broiliellus 0.22, Ecolsonia 0.32, Micromelerpeton 0.35, Micropholis 0.54, Pasawioops 0.36, Tersomius 0.33). It thus appears that the orbit-to-notch length is smaller than 2/5 of the skull table width in most dissorophoids (but not in Micropholis), which may account for the enlarged otic notch in the group. Little else can be discerned from this. The character appears to be variable in branchiosaurids. Character rephrasing: Distance from posterior edge of orbit to notch (anterior edge) more than 2/5 of skull table width (distance from posterior edge of orbit to posterior edge of skull roof) (0); less than 2/5 of skull table width (1).

Character 50. Pointed posterior projection on tabular: (0) absent, (1) present (Anderson et al., 2008a). Contrary to the previous coding, Doleserpeton does have a well-developed posterior process on the tabular bone, similar to those

208 described in Eoscopus and Gerobatrachus. It is set off from the rest of the bone, being located somewhat ventral to the skull roof proper. This may or may not be homologous to the more knob-like process described in Georgenthalia, but if this is the case, then it is also present in Tersomius. It is suspected that this character was present in many more taxa, but that it may have broken off easily, as is the case in some Doleserpeton specimens. The character was therefore removed.

Character 51. Skull shape in dorsal view: triangular (0); rounded (1) (Anderson et al., 2008a). The derived state of this character (i.e. rounded shape) is clearly present in the primitive taxon Balanerpeton (1).

Character 52. Alary processes: not present, small (0); not extending farther posteriorly than one half of naris length (1); long, extending to or beyond posterior border of external naris (2). Alary processes of the premaxilla are present in both Dendrerpeton and Balanerpeton so the absence of this character is probably not primitive. The alary processes of Eoscopus and Pasawioops are well developed, but somewhat different in that it far more robust and slightly shorter in Pasawioops than in Eoscopus. The shared coding of these is tentatively retained here, but the others are coded as primitive, or in the case of Micropholis, as polymorphic. New coding: alary process short or absent (0) robust (1).

Character 53. Curvature of marginal teeth: teeth straight, conical (0); recurved posteriorly (1) (modified from Yates and Warren, 2000, and Huttenlocker et al., 2007). In most amphibamids, the teeth curve slightly lingually (Bolt, 1969, 1979; Daly, 1994). There appears to be some intermediary forms between fully recurved and simple, slightly inwards-curving teeth. Tersomius is coded (01). Trematopids certainly have posteriorly recurved teeth (1), unlike the previous coding.

Character 54. Carotid foramina: piercing medial region of basal plate of parasphenoid (0); carotid foramina not piercing basal plate (1). The carotid foramina are sometimes hard to find on the ventral surface of the basal plate, and can only be seen in the best preserved specimens of Doleserpeton. They may actually be present in all amphibamids. The basal plate of Pasawioops seems to have a poorly preserved basal plate (Frobisch and Reisz 2008). However, the previous coding is kept unchanged until further studies of Micropholis and Pasawioops are undertaken.

Character 55. Posterior portion of skull table: elongated (0); abbreviated (1). The skull tables of many amphibamids (e.g. Micropholis) are less abbreviated than that of Balanerpeton and some are more comparable to Dendrerpeton. Eryops also has a short skull table (Sawin, 1941). In addition to Balanerpeton, the skull table is most clearly abbreviated in Platyrhinops (Amphibamidae) and Conjunctio (Dissorophidae). Removed.

209 Character 56. Meckelian foramen: located posteriorly (0); located anteriorly and bounded by postsplenial and prearticular (1) (Yates and Warren, 2000). This character appears to be incorrectly described. The foramen is located posteriorly and not bounded by the postsplenial in Tersomius (Carroll 1964) and Doleserpeton, but it IS bounded by this bone in Dendrerpeton (Holmes et al. 1998).

Character 59. Squamosal shape: higher than long, crescent-shaped (0); longer than high (1). The squamosal of Dendrerpeton is, in fact, longer than high, making the polarity of this character uncertain. Removed.

Character 60. Knobby exostoses ornamenting skull roof: absent (0); present (1). This character was coded as being present in Dissorophinae only, and so is uninformative for interrelationships. Removed.

Character 61. Otic notch: open (0); narrow and closed or partially closed (1). This formulation gives a nearly identical distribution to character 23, because they partly described the same features (a narrow otic notch results from the presence of a supratympanic flange). It also overlooked the fact that a fully closed notch (found in Ecolsonia and Cacopinae) may reflect a relationship. Character rephrased. Otic notch: open posteriorly (0); closed posteriorly (1). See also changes to character 23 above.

Character 62. Contour of occiput in dorsal view: relatively straight (0); concave (1). The contour of the occiput seen in dorsal view is variable within apparently well-defined taxa, such as Micromelerpetontidae (Boy and Sues 2000 p. 1168). There is a slight biconcavity in Dendrerpeton, Edops, and Balanerpeton, making the primitive condition uncertain (or intermediate). It also appears that there is little real difference in this character between, say, Doleserpeton and Amphibamus grandiceps, and the character may partially be due to compression taking place at different angles during fossilization. Removed.

210 APPENDIX 3

Amphibamid phylogeny. New character list.

The previous numbering of characters in Fröbisch and Reisz (2008) is given in parantheses:

Character 1. (1) Laterally exposed palatine: palatine overplated by jugal and lacrimal with no dorsal exposure (0); palatine wedging between lacrimal and jugal to make contribution to skull roof and orbital margin (1).

Character 2. (2) Dorsal quadrate process: quadrate having smooth posterodorsal side in plesiomorphic state (0); quadrate with prominent dorsoposterior outgrowth, the quadrate process (1).

Character 3. (4) Parasphenoid dentition: basal plate of parasphenoid bearing shagreen of small teeth (denticles) anteromedially (0); plate entirely smooth (1).

Character 4. (5) Parasphenoid denticle field: parasphenoid denticle field well established, with triangular outline and with apex reaching onto base of cultriform process (0); denticle field greatly expanded anteriorly to cover most of the cultriform process (1).

Character 5. (6) Parasphenoid basal plate: basal plate roughly quadrangular, as long as wide (0); basal plate much shorter than wide, reaching about half the width (1).

Character 6. (7) Vomerine denticle field: vomer covered with more or less dense shagreen of teeth in addition to obligatory fang pair (0); shagreen confined to juvenile stages and/or absent throughout ontogeny (1).

Character 7. (8) Vomerine fangs: vomer lacking fangs in its medial portion, outside lateral tooth arcade, but having smaller accessory teeth in that region (0); vomer with additional fang pairs posterior to midvomerine depression (1).

Character 8. (9) Palatine fangs: palatine with one fang pair in its anterior third (0); palatine fangs replaced by patch of small, equally sized teeth (1).

Character 9. (10) Ectopterygoid fangs: ectopterygoid with one fang pair at about mid level (0); ectopterygoid without fangs (1).

Character 10. (11) Pterygoid-vomer: retention of suture between pterygoid (palatine ramus) and vomer (0); pterygoid contacting only posteriormost portion of palatine and lacking suture with vomer (1).

211 Character 11. (13) Pterygoid-palatine: long medio-lateral contact between palatine and pterygoid (0); contact between the two bones reduced or absent (1).

Character 12. (14) Palatine, ectopterygoid: palatine and ectopterygoid much wider than maxilla (0); palatine and ectopterygoid reduced to narrow struts not wider than adjoining maxilla (1).

Character 13. (15) Interpterygoid vacuity: interpterygoid vacuity roundish or oval in outline (0); interpterygoid vacuity greatly expanded laterally at mid-level (1).

Character 14. (17) Pedicellate teeth: adult dentition of marginal jaw and palatal elements consisting of conical, fully mineralized teeth with labyrinthine infolding of enamel and dentine (0); jaw and/or palatal teeth—in larger, presumably adult stages—subdivided by unmineralized zone forming separate tooth crown (1).

Character 15. (18) Biscuspid teeth: larval and adult teeth conical or needle shaped with single tip (monocuspid) (0); jaw and/or palatal teeth in larger, presumably adult, stages bicuspid (1).

Character 16. (19) Labyrinthodont teeth: jaw and palatal teeth labyrinthodont at least in adult stages (0); these teeth never labyrinthodont at any stage of development (1).

Character 17. (20) Narial flange: ventral (inner) side of prefrontal, lacrimal, and nasal smooth (0); inner side of these bones forming complicated bar-like structure (narial flange), permitting contact with antorbital bar (1).

Character 18. (21) Prefrontal process: prefrontal forming simple suture with lacrimal laterally (0); prefrontal underplating lacrimal widely by means of ventral prefrontal process contacting palatine (1).

Character 19. (22) Tabular size: tabular narrower than postparietal, but reaching almost same size as latter (0); tabular minute and laterally constricted by unique enlargement of otic notch (1).

Character 20. (23) Ossification around tic notch: smooth inner surface of otic notch not extending dorsally within notch, or only forming a narrow edge on supratemporal (0). Inner rim of otic notch continuing dorsally to form extended supratympanic flange, including a tabular-squamosal contact (1).

Character 21. (24) Postparietal length: postparietal forming transversely rectangular or quadrangular element (0); postparietal abbreviated and reduced to narrow, poorly ornamented strut at posterior margin of skull table (1).

212 Character 22. (25) Squamosal-supratemporal: suture between squamosal and supratemporal nearly as long as supratemporal itself (0); foreshortened squamosal-supratemporal suture reaching only one third or less length of supratemporal (1).

Character 23. (28) Semilunar flange (supratemporal flange of Schoch and Rubidge, 2005): supratemporal without ventral projection into otic notch (0); supratemporal forming marked ventral flange participating in medial bordering of otic notch (1).

Character 24. (29) Prefrontal-postfrontal: prefrontal and postfrontal firmly sutured, excluding frontal from orbital margin (0); both elements separated by frontal, at least dorsally (1).

Character 25. (31) Palpebral ossifications: ossifications in orbit restricted to sclerotic ring (0); numerous palpebral ossicles at medial margin of sclerotic ring (1).

Character 26. (33) Pleurocentra: temnospondyl pleurocentra primitively paired, dorsally separated elements failing to reach ventral level of adjunct intercentrum (classical rhachitomous condition) (0); pleurocentra approaching each other on ventral side (1); pleurocentra fused ventrally to form large element that has become main bearer of vertebral column, and intercentra much smaller than in plesiomorphic condition (2).

Character 27. (36) Humeral elongation and torsion: Extremely short shaft or no shaft present, little or no torsion between distal and proximal ends (0); relatively short shaft present, approximately 90 degrees torsion (1); long shaft (humerus approximately three times longer than wide), approximately 90 degrees torsion (2).

Character 28. (37) Humerus supinator: supinator process arising from humerus in adults (0); absence of supinator throughout ontogeny (1).

Character 29. (39) Interclavicle: rhomboidal, at least two times longer than wide (0); with abbreviated anterior and posterior ends and as wide as long (1); much shorter than wide, without anterior process (2).

Character 30. (40) Shape of iliac blade: long posterior process (0); short and wide dorsal process (1); pointing posteriorly, but with a robust base (2).

Character 31. (43) Scapula: forming low or moderately high element, depending on degree of ossification, about two times longer than wide (0); dorsally much extended, being three or four times longer than wide (1).

213 Character 32. (44) Presacral vertebral count: 24 or more (0); 23 or fewer (1).

Character 33. (45) Ventral scutes: spindle shaped ventral scutes covering the belly region (0); ventral scutes absent (1).

Character 34. (46) Carapace: dorsal scutes either absent or smallish, oval or round, loosely set (0); dorsal scutes heavily ossified, forming transversely broadened making up strongly ornamented carapace (1).

Character 35. (47) Tail: bony tail long, at least length of trunk, often exceeding length of trunk and skull (0); bony tail foreshortened, markedly shorter than trunk (1).

Character 36. (48) Postfrontal shape: broadly quadrangular (0); falciform or sickleshaped (1) (Anderson et al., 2008b).

Character 37. (49) Otic notch to orbit: distance from posterior edge of orbit to notch (anterior edge) more than 2/5 of skull table width (distance from posterior edge of orbit to posterior edge of skull roof) (0); less than 2/5 of skull table width (1).

Character 38. (51) Skull shape in dorsal view: triangular (0); rounded (1).

Character 39. (52) Alary processes: not present, small (0); not extending farther posteriorly than one half of naris length (1); long, extending to or beyond posterior border of external naris (2) (modified from Huttenlocker et al., 2007).

Character 40. (53) Curvature of marginal teeth: teeth straight, conical (0); recurved posteriorly (1) (modified from Yates and Warren, 2000, and Huttenlocker et al., 2007).

Character 41. (54) Carotid foramina: piercing medial region of basal plate of parasphenoid (0); carotid foramina not piercing basal plate (1) (new character).

Character 42. (56) Meckelian foramen: located posteriorly (0); located anteriorly and bounded by postsplenial and prearticular (1) (Yates and Warren, 2000).

Character 43. (57) External nares: relatively small (0); enlarged, approximating onethird to one-half diameter of orbit or greater (1); posterior extension of external naris present (2) (from Huttenlocker et al., 2007; for Platyrhinops, Amphibamus, and Doleserpeton, coding changed from 0 to 1).

214

Character 44. (58) Lengthening of maxilla: relatively long throughout ontogeny (0); lengthening greatly posteriorly at metamorphosis (1).

Character 45. (61) Otic notch: open posteriorly (0); closed posteriorly (1).

Character 46. Outline of the otic notch: elongated, coming to a rounded point antero-dorsoally (0); evenly rounded anterior edge of large otic notch making a crescent-shaped facet on the squamosal (1). The derived condition occurs in all amphibamids except Plemmyradytes.

Character 47. Entepicondylar foramen present (0); absent (1).

Character 48. Intertemporal: present (0); absent (1).

Character 49. Ulnar condyle: present (0); absent (1). There is a separate condyle for the articulation of the ulna with the humerus in Eryops and trematopids, and this feature appears to be present in some stereospondyl humeri. In contrast, well- preserved humeri of Dissorophidae and Doleserpeton clearly lack this feature, as the radial condyle appears to articulate with both the radius and ulna in these forms (Sigurdsen and Bolt, in press). The distal ends of the humeri are poorly preserved in Balanerpeton and Dendrerpeton. The feature can only be seen in well-preserved specimens.

Character 50. Width of inner surface of otic notch: smooth inner surface notch narrow, leaving a broad strip of sculpted bone laterally and ventrally in the area of the quadratojugal (0); inner surface of otic notch expanded, lateral strip of external bone narrow (1). The derived condition is found in amphibamids and armored dissorophids.

215 APPENDIX 4

Amphibamid phylogeny, new character matrix

Dimensions: number of taxa = 20 number of characters =50;

Balanerpeton 0?000 00000 00000 0??00 01000 00?00 0000? 10100 0(01)0?0 000?0 Dendrerpeton 01000 00000 00000 00?00 00000 00?00 0000? 00000 000?0 000?0 Sclerocephalus 00000 10000 00000 00000 00000 01000 00000 00001 01000 01100 00000 1(01)000 00000 00000 00000 01001 10000 00000 01000 01100 Micromelerpeton 1?000 (01)(01)001 00000 0?100 10111 01(01)10 00000 11000 01010 011?0 Amphibamus 11011 00(01)11 ?11?1 1?110 11?01 2(12)110 01001 11100 01110 111?1 Platyrhinops 110(01)0 0(01)011 11101 1?110 11?01 12110 00101 11100 011?0 111?1 Doleserpeton 11001 001-1 11111 11110 11111 22110 0{01}?01 11100 011?0 11111 Eoscopus 11000 (01)(01)001 01100 01110 11111 11110 00101 11110 ?10?0 111?1 Tersomius 11000 11001 ?110(01) 0110(01) 00111 111?? ????? 0110(01) 011?0 111?1 Ecolsonia 11000 10000 00000 01101 00110 01011 1?001 01001 012?1 011?0 Trematopidae 11000 00000 00000 01101 00110 01011 10000 01001 012?0 01100 Dissorophinae 1110(01) 00001 0010(01) 01101 01111 01121 11011 01000 011?0 01111 Cacopinae 11100 00001 00100 01101 0111? 01121 11011 01000 011?1 01111 Micropholis 1101(01) 1(01)001 (01)1100 0110(01) 00011 11112 01101 00(01)(01)1111?0 111?1 Pasawioops 11010 11001 01100 0?100 00111 ????? ????? 01011 111?0 111?1 Plemmyradytes 1???? ?10?1 1?100 111?0 ?1?11 ?11?2 ????? 11101 ??1?0 011?1 Georgenthalia 11001 01??1 ?1100 ??100 11011 ????? ????? 011?? ??2?0 111?1 Gerobatrachus ??1?1 10101 111?0 1??1? 0??11 ?2??? ????? 011?? ??2?0 ?11?? Apateon 0?101 10001 011?0 1??00 100(01)0 ?21(12)? 01(01)01 1(01)1(01)0 0?110 011?0

216 APPENDIX 5

Comments and changes to the character matrix of Anderson et al. (2008).

The character numbering and descriptions (in bold) are listed exactly as found in the on-line appendix of Anderson et al. (2008), with character states separated by a space. These are followed by comments by the present author.

Character 2. 'Skull:trunk' / '>=0.45' '0.30-0.45' '0.20-0.29' '<0.20', 3 Skull_proportions / longer_than_wide wider_than_long. The skull of Doleserpeton is about 40% of the trunk length (1).

Character 3. Skull_proportions / longer_than_wide wider_than_long. Some anurans have skulls that are longer than wide (e.g. Leptodactylus bolivianus and Rana pipiens, see Trueb, 1973), although their skulls are wider than most tetrapods.

Character 5. Supratemporal / present absent. There is a bone bordered by the parietal, postparietal and squamosal elements in Eocaecilia, similar to the supratemporal bone in Amphibamidae and primitive tetrapods (Clack, 2002). Such an element should not be coded as absent in Eocaecilia. The tabular bone in microsaurs appears to be a fusion of the tabular and supratemporal of primitive tetrapods, judging from its bordering bones. Eocaecilia has been coded “?”.

Character 6. ST_exposure_on_occiput / absent present. This character is also linked to the supratemporal (ch. 5), and should be coded similarly (?).

Character 8. 'T-PF' / absent present. This character seems to be related to a simple fusion of the supratemporal and tabular bones mentioned above (5).

Character 9. Postfrontal_shape / broadly_quadrangular falciform. The postfrontal of Amphibamus is closer to being falciform (1) than quadrangular.

Character 10. 'Squamosal-Tabular' / absent present fused. Squamosal tabular contact. This character is related to a simple fusion of the supratemporal and tabular bones. There may be a fusion of the tabular and squamosal in Eocaecilia.

Character 26. Alary_processes_of_premax / absent present. Eocaecilia was described as having a flange of the premaxilla fitting into a slot in the nasal bone. This is more similar to the presence of an alary process found in Doleserpeton than the lack of any process in Rhynchonkos. Eocaecilia has been recoded (1).

Character 28. Septomaxilla / ossified unossified. The septomaxilla is ossified (0) in Rhynchonkos, Dolserpeton and Tersomius.

217 Character 32. Dorsal_exposure_of_premax / broad narrow none. It is not clear how the dorsal exposure of the premaxilla differs in Doleserpeton and Eocaecilia. Both set to 0.

Character 33. Dorsal_shape_of_skull / triangular diamond rounded. Many anuran skulls are more triangular than rounded (e.g. many Rana, and discoglossus species). However as Amphibamus is regarded as having a rounded skull in the previous coding, then Doleserpeton and Tersomius should also be coded as such (recoded 2).

Character 34. Posterior_skull_margin / concave straight convex undulating. This character is far more problematic than it may seem, due to intraspecific variation and distortion. However, no changes were made.

Character 44. Raised_orbital_rim / absent present. Doleserpeton has raised orbital rims.

Character 52. 'Parietal-tabular contact' / absent present. This character again relates to the problem of the identity of the “tabular” of Eocaecilia. This form recoded (?).

Character 55. Postparietals / moderate large. This is a problematic character. There seems to be no reason to base two characters on an apparent resemblance between the postparietals of Eocaecilia and microsaurs. The postparietal of Eocaecilia (coded 1) is actually smaller than that of Acanthostega, which was coded ‘moderate’. The bone is wide in Doleserpeton, long in Eocaecilia, wide AND long in Rhynchonkos. Recoded (0) in Eocaecilia.

Character 57. Postparietal_length / 'large,_quadrangular' 'abbreviated_anteroposteriorly,_elongate_lateral_rectangle'. See above.

Character 60. Posterolateral_projection_from_lateral_margin _of_tabular_above _squamosal_embayment / absent present. A similar projection is present (1) in Doleserpeton.

Character 65. Dermal_sculpturing / circular_pits shallow_ridges_and_grooves little_to_none. Doleserpeton has little to no dermal sculpturing (2).

Character 87. Jaw_articulation / posterior_to_occiput even_with_occiput anterior_to_occiput. Eocaecilia is actually intermediate (1) in this character. It is notable that the jaw articulation of this early form is posterior to that of modern forms, whereas Rhynchonkos is at the same level as modern caecilians.

Character 84 Occipital_condyle / concave convex. This character is similar in Doleserpton, Eocaecilia and Rhynchonkos. The former recoded (1).

218

Character 94. Vomer_teeth / single_pit_pairs Multiple_in_rows Multiple_random. Doleserpton has multiple teeth in rows (1).

Character 96. Intervomerine_rostral_fenestration / absent present. A slight depression is present in Doleserpeton. Two depressions appear to be present in Eocaecilia. The area in question is often poorly ossified in anurans, but sometimes a partly ossified depression of the vomer occurs in this region of the vomer (e.g. mature Rana catesbeiana). However, it is regrettable that the character matrix does not reflect the shape of the vomer, which is similar in Doleserpeton and Eocaecilia, when compared to microsaurs. Unchanged.

Character 99. Tooth_pedicely / absent present. Doleserpeton has far more lissamphibian-like pedicels than those of Gerobatrachus. The latter should probably not be described as pedicels, and has been recoded “?”.

Character 101. Denticles_on_palatines / present absent. Doleserpeton has no denticles on the palatines, but it does have a row of pedicellate teeth on this bone, as in Eocaecilia. Both were recoded (1).

Character 105. Parasphenoid_basal_plate / 'roughly_quadrangular,_basipterygoid_articulations_narrowly_spaced' 'rectangular_laterally,_anteroposteriorly_narow,_basipterygoid_articulation s_distant'. We fail to see how Eocaecilia is closer to Rhynchonkos in this character. It could equally easily be said that Rhynchonkos and Doleserpeton are more similar in this trait, as they both have more prominent basipterygoid articulation areas. The shape of the basal plate is similar in all. Eocaecilia recoded (?).

Character 103. Palatal_teeth / larger_than_marginals equal_to_marginals smaller_than_marginals. The palatal teeth of Doleserpeton are the same size as the marginal teeth (previously coded polymorphic). Recoded (1).

Character 108. Stapes / perforated_stem imperforate_stem no_stem. Doleserpeton has a stapedial foramen (0).

Character 111. Dorsal_process_of_stapes / absent present. The process is absent (0) in Doleserpeton and Amphibamus.

Character 115. Interpterygoid_vaccuities / 'narrow_("closed")' wide fused_at_midline. The vacuities of Eocaecilia are surrounded by the same bones as those of dissorophoids. The vacuities of modern caecilians are often rounded in shape, and are dissimilar from microsaurs in that the palatines border them. Eocaecilia coded 1.

219 Character 118. 'Pterygoid-palatine suture' / present absent. This contact is similar in Eocaecilia and Doleserpeton in that the contact is at the posterior edge of the palatine, whereas in Rhynchonkos there is a suture between these elements lateral to the palatine. To indicate that the condition in Doleserpeton and Eocaecilia does not really apply, these have been recoded “?”.

Character 120. Lat_process_of_pt_into_posttemp / absent present. This character was not understood. Unchanged.

Character 121. Ectopterygoid / 'present_with_fang-pit_pair' 'present_lacking_fang-pit_pair' absent. Doleserpeton, Eocaecilia, frogs and salamanders should all be coded ‘absent’ (2).

Character 130. Meckelian_fossae / 2_or_more 1 0. Doleserpeton has one Meckelian fossa. All salamanders examined by the present authors had one meckelian fossa in the lower jaw, contrary to the previous coding (recoded 1).

Character 131. 'Ventral border of Meckel''s foss' / splenal angular. In Doleserpeton, the ventral border of the fossa consists of the angular.

Character 132. Retroarticular_process / absent present. Carroll (2000, 2007) described the retroarticular process of Rhynchonkos. This is far shorter than in caecilians, and is similar to projection of the lower jaw sometimes seen in salamanders (e.g. Reese 1906). A slight protrusion beyond the articulation area is common in many tetrapods, and the process of Rhynchonkos is only slightly longer than that of Doleserpeton, making this character problematic. Salamanders recoded (01).

Character 134. Articulation_to_tooth_row / above equal below. It is not clear how this character is different in Rhynchonkos and Doleserpeton. They should both be coded “equal“(1).

Character 139. Symphysis / dentary_and_splenal dentary_alone. The figures of Carroll (2000, 2007) show a splenial bone in the symphyseal region of Rhynchonkos. Recoded (0).

Character 140. Jaw_sculpture / present absent. There is hardly any sculpturing of the lower jaw in Doleserpeton. Recoded (1).

Character 145. Number_of_presacrals / '25-35' '20-24' '>35' '<20'. Salamanders were previously coded 3, but can have anywhere from 10 to 60 trunk vertebrae (Duellman and Trueb, 1994). Conservatively recoded (13).

Character 168. Atlas_Anterior_centrum / same_size_as_posterior laterally_expanded. The anterior part of the atlas centrum is expanded (1) in Doleserpeton.

220

Character 174. Proatlantes / present absent. The proatlas is probably absent (1) in Doleserpeton.

Character 175. Second_cervical_arch / expanded_to_more_posterior equal_to_more_posterior shorter_than_more_posterior. The second arch is expanded (0) in Doleserpeton.

Character 177. Cervical_rib_distal_shape / spatulate pointed. The first rib is more or less spatulate (1) in Doleserpeton.

Character 179. Ribs / elongated_and_sometimes_curved straight 'short,_simple_rod'. The ribs of Doleserpeton are short compared to microsaurs and even to Dendrerpeton, but those of Eocaecilia are shorter still, contrary to the previous coding. The posterior curvature of the ribs is also closer to the temnospondyl condition, contrary to the situation seen in microsaurs, in which the ribs curve sharply ventrally (as in amniotes). The old description of lissamphibian ribs as ‘short straight rods’ is slightly misleading, but the fact remains that their ribs are very different from those of microsaurs and amniotes. Eocaecilia recoded 2.

Character 182. Sacral_parapophysis / on_centrum on_transverse_process. The capitulum (parapophysis) of the sacral rib almost certainly articulated with the centrum (0) in Doleserpeton.

Character 186. Interclavicle / diamond_shaped 't-shaped'. The interclavicle is diamond shaped (0) in Doleserpeton.

Character 187. Interclavicle_anterior_plate / broad narrow. Broad (0) in Doleserpeton.

Character 188. 'Interclavicle shape-diamond' / broad_diamond narrow_diamond. The interclavicle is broad (0) in Doleserpeton.

Character 189. Interclavicle_anterior_fimbrati / present absent. Present (0) in Doleserpeton.

Character 190. Interclavicle_sculpture / present absent. Absent (1) in Doleserpeton.

Character 194. Cleithrum / rounded_or_pointed_dorsally 't-_or_y-shaped'. Rounded (0) in Doleserpeton.

Character 196. Supraglenoid_foramen / present absent. The foramen is present (0) in Doleserpeton.

221 Character 197. Number_coracoid_foramina / 0 1 2. Doleserpeton has 1 coracoid foramen (1).

Character 198. Scapulocoracoid_occification / both scapula_only absent. Both are ossified (0) in Doleserpeton.

Character 199. Entepicondylar_foramen / present absent. Absent (1) in Doleserpeton.

Character 200. Tortion (sic)_in_humerus / absent less_than_80_degrees more_than_80_degrees. More than 80 degrees in Doleserpeton, as well as in some frogs (incl. Triadobatrachus) and salamanders. Doleserpeton recoded (2); frogs and salamanders polymorphic (12).

Character 201. Deltapectoral_crest / weak intermediate prominant (sic). The deltopectoral crest of Doleserpeton is similar to that of salamanders (although the trait is somewhat variable in the latter). Frogs were coded as intermediate in the original matrix, although the crest is, in fact, longer in this group than in salamanders (Chapter 4). Doleserpeton coded 2, frogs coded 3.

Character 202. Supinator_process / absent present. This process is absent (0) in Doleserpeton.

Character 203. Humerus_length / 'long_(>4_trunk_centra)' short. The humerus is long (0) in Doleserpeton. It is sometimes shortened in salamanders, which is recoded (01).

Character 204. 'Radius:humerus' / '>=0.7' '0.5-0.7' '<0.5'. The radius is slightly less than half the humeral length in Doleserpeton (coded 2), this is sometimes the case in salamanders as well (recoded polymorphic [12]).

Character 205. Olecranon_process / unossified ossified. The olecranon of Doleserpeton is poorly ossified (0). It is sometimes well ossified in salamanders (01) such as Ambystoma (pers. obs.).

Character 206. Carpals / fully_or_partially_ossified unossified. Carpals are ossified (0) in Doleserpeton.

Character 208. 'Number digits-manus' / 5_or_more 4 3. Doleserpeton has 4 (1).

Character 209. Pelvis / fused sutured poorly_ossified. Sutured (1) in Doleserpeton.

Character 211. Illiac_blade / 2_dorsal_processes narrowly_bifurcate single_blade. Single blade in Doleserpeton (2).

222

Character 212. 'Internal trochanter-articulatio' (sic)/ disctinct continuous. This character is somewhat unclear. We interpret this as the trochanter being set off from the proximal head of the femur in the primitive condition. If so, both Doleserpeton and salamanders should be coded (0).

Character 213. Femoral_shaft / robust slender. The femoral shaft of Doleserpeton is as slender as that of salamanders. Recoded (1).

Character 214. Femur / long short. The exact length is not indicated in the character coding, but the length of the bone in Doleserpeton (about 4 centra long) is clearly longer than in Eocaecilia, and similar to modern salamanders. Recoded (0).

Character 215. Tarsals / ossified unossified. Doleserpeton specimen FMNH UR1320 preserves some ossified tarsal elements. Recoded (0).

Character 219. 'Number of digits, pes' / 5_or_more 4_or_less. Doleserpeton has five digits (0).

223 APPENDIX 6

Modified data matrix of Anderson et al. (2008):

Dimensions: number of taxa = 54 number of characters = 219;

Acanthostega 000100000100000?000000000010000000000000000000000100000000010--000000000000000-0000-- 00000000000000000000000000000000000000000000000- 010000000000?0000000000000000(01)0000000000000000000000000000000000000000000000000100000 0

Proterogyrinus 01000000(0 1)00{0 1}001?00010000000?000000000000000000000001000000010--000000000010000- 0020000000000????000?1000000?0??000000000000000000?00-00{0 1 2}001000000?00000000200001?0100000000000000?000000000010000000200100?1010010111000100

Tuditanus 31011-110111001100010000?0000000000001--000100000001?000?00-0--110000?00011000- 011??0?10??00?200000000?0000200000?000000??01001002-0-21???0010000?012-001021---- ??1000112202011010001021111-?100001--2022001000110001000010

Asaphestera 22011-110111001000010000?00100000300010-0001000010010000000-0--{0 1}10000001010000- 01210002001000???01000?0??00????001010000?000?0100????20????110000??12?00102???????10001122 ??011??000???1110-?0????102002100?1???200?10?????

Hapsidoparion 3?011-010101001001--0000?0001000010001--0000000010010000110-0--120001000011000- 01011002001001-0000011010000101010?0100000000110112-0- 20???110000????0??0?02????????00011212001?01??0????????????????????????00??202????????

Saxonerpeton 31011-110101001000010100?0011000010001--0001000000010000100-0--020001000011000-0110?0?201- 001-00000000-?000??????10100000000110{0 1}1??1020????1100001011-001021---- ??10001121?2011001000?21111-1110001??0020002100110201000100

Pantylus 10011-11011100100001000000000101011001--0000000000010110000-0--101000000222010- 010100011020002000000011100010011?00000001000100011(0 1)10202010000001?01101010200000011000012202011101000?01111- 1010002010022002100110211000110

Cardiocephalus_sternbergi 3?011-110111001110010000?0000101021001--0001000000010(0 1)10000-0-- 12110001022211??1100000300100010011011121000100010000000000001?0012-0- 20202111000?????????0?????????????????????????????????????????????????????????????????

Cardiocephalus_peabodyi 32011-110111001100010000?0001101121001--0001000010010(0 1)10000-0-- 121000?10222111011000003???0???????0??????0??0???????0????0?0111?1??0- 20????110002??01?00102???????100001220?0110000010????????1000?0-1121012?0??202010?????

Euryodus_primus 2?011-110111001100000000?0001102111001--0001000000010110000-0--??1000010222010- 1100001200100010000000121000?0???0?0000000000111112-11203-- 1?1000???02?00102???????1000????????????00???1111-1?????1??0(0 1)2101?100?102010?0???

224 Euryodus_dalyae 2?011-110111001100010000?0001102011001--0001000000010010000-0--121000010222010- 010000020010?0???00000021000000010?0000000000111112-0-203-- 111000???02?00102???????10000122000100???0?0??????????????????????1????0?????????

Pelodosotis 13011-110111001100110000?0001000031001--0000000000010110010-0--12101-010012010- 01210002?0100????000??0210001021100010000?000{0 1}0111111020?021100002??01?10102???????10001121?10110010010?0111-11100011- 10220021???00201000???

Micraroter 13011-110111001010110000?0000000030001--00010000100101(0 1)0010-0--121100010112010- 0101001200100020000000020000001010?000000000010101??1020???1100000?001010102000010010000 12200011000002?????????1000?0-1???0??????102110?????

Rhynchonkos 33011-110111001000010000?0010102121001--0001000010010010?00-0--121000010012000- 0101101310100010010011011000000011001000{0 1}0000100001110101010110002??02?00102???????10000122- 2011210001?3??????????0????12201210011020100010?

Microbrachis 32011-100111011000010000?0010000010001--0000000011110000100-0--010000000(0 1)11000- 010?00?201-001-00000000- 100000000010100000000101001110000021001002101100010200000011010112121011000000020110- 00010000-102001100?21011110??00

Batropetes 32011-10111100100???010000012102130001--00000000011120---00-0--12101-?100120011011110- 3???00????100110- 110010000000?0??0??001??11??101????1110003?0120001020010000100011220?011000000001110- 11???0102002201210?10020110?110

Utaherpeton 310????00??100111???0000?00?20000?0001--0000000?01???0???0?-0--12010??001?1000-01?????{1 2}???0?1???000?0????0?????????????????00??10??0-21????110?00?01200000200000??(0 1)0?01?????????00001?00-000(0 1)???????-102201011?11?1?101?—0

Stegotretus 3?011-1101110011000{0 1}0000?0001101010001--000100000001001000000--111001010222010- 0101000210200020001000011000?????0000000010001?001110-00?0(0 1)0?10?0???11?00102???????10001121????????00???1111-??????011012200210??2?00100????

Sauropleura_scalaris 210100000100001110001000?0012000001101--0000010101010000000-0--1101(0 1)0001012000- 011??0?0001001???01011120000?????000?00001100000100010003-- 1000010?01200011201010?01010101222111210000010-0010000000??0100110000100211100?10

Ptyonius 110101000101001100010000?0012000011001--0000010101010000000-0--110000000012000- 011??0?000100010010001120000?????011000000100001?1??1101???1000021?112100(0 1)1201021001010102222211211000020-0000000000???10001001?12021111?—1

Urocordylus 22010??0??000011000??????????0?00?1001--???00100?1???????0?-0--1100???010??000- 01???0?00010?????000?1?2?0?0?????0000?0???1000?0?0??1100???100??21?1121001120111(0 1)10101010122221?210000030-000?000000??010201111??1021011?—0

225 Keraterpeton_galvani 32011-000111001000010000?00120000000?1--00000000?1110010000-101010000?00022000- 010?11?20010?1???0000111??0??????0020??????0100010111111???0000013?01200011201100001010101 222111210000030-000?010012???10001011?12021111?—0

Batrachiderpeton ??011-000111001000010000?0011000000011--0000000001110010000-110010100100012000- 011?11?20010001000000111?001?????0020000011{0 1}10001011111120(1 2)0000013?012000(0 1)12011000010101112221122?0?00?10-000?010012?????????????20?????????

Diceratosaurus ?1011-000111001010010000?0012000000011--00000000011100100000100010100?00????00- 010?11?2???0???????0??????01?????00001?????000?11011102?2011000013?012001012010000010101012 22111210000030-000?010012--210001011?12021111?—0

Diplocaulus_magnicornis 01011-000111000-1001110100011100000011--00000010011100100000111010000100012000-010011{0 1}300100010010011110001?????001110101010020010(0 1)0- 202010000013?01200101201000001011111222112200100130000010???12??200001101?120??111?--0

Diploceraspis 0?111-00011-1---0001110100012000000001--00000010011100100000111010000100012000- 01001103001001?0010011110001?????00111010001010110111020201000002??012001012010100?101111 22221122???0?1?0-00010???12????????????????????????

Scincosaurus 33011-110111001000010000?00120000{1 3}0011--00000000000120---0000--010100000012000- 01??11?301-001?00100101- 0001?????00010000100?0???????????01?000011?01210001201111001000101222?11210000?20- 000?000002??0022001000010201000?10

Brachydectes 33011----1110011-1--000000012000-20001--00001---0-1101110-000--120100000012000- 012100031110001001001111100?10100100100001001100111110203-- 1100002?011000102000010100101021200112100000311000100000010011001(0 1)11?12021110??00

Adelogyrinus 2?011-10021-1----0000000?00110000300000-00000010?111000000{0 1}10--0101???000?0000- 01000012?1-0?1???0?0001-0000??????00??0?0000000?00??100(0 1)???100000??010?00102????????0001020220112??00???0-00000???0???2????????-????????????

Oestocephalus 230100-0010100101011000000012101020001--00001---0-00001001000-- 120100100010000-01200000001--????100?101?0-0????000000000100000001110-003--1110012?012- 000021----00111010122200121-01--0------00------

Phlegethontia 23011---0-1-1------110000012000021001---0001---1---21---01-0--120101-00022000-012000-2???-- ????100??1?0--020000000?0-?010000??2-2-0-0?3--11100121012-000021----00111010{1 2}22200121-01-- 0------00------

Greererpeton 02010000010001100000000000012(0 1)000300010000000000010000(0 1)000000--000000000000000- 000000?000(0 1)000000000110(0 1)00000000000000000000000000100- 001010010002?00000000000000??10100000?0000211100000-0100000001110011011110100201010—0

Seymouria_baylorensis 0100000000010010000000000000000001000010000000000001000000000--000000000012010- 000000000000000000000001010011??0102000000000000000(0 1){0 1}-200020001001?000000(0 1)020000100100000000001100000100110001001001020022101000000111100?00

226 0101010001010010000100000000000000100100000000000101100010000--121000001022010- 0001000111-00120001000010000?0??0000100000000000101(0 1)0- 001020110000?000001002000000010000000000100000010011000??0100?0?0022101010010111000110

Balanerpeton 000000001011001000000000?101000000000000000000011000000000000- -100000000000000-0000?000000000000000000000000100000110001000000100000-0001- 1010001?000000000000000?00000000?000020010000000000000000??0022001100110211000100

Dendrerpeton 000000000011001000000000?110000000000000000000010000000000000--100000000000000- 000000?000000000000000000000010000?110000000000001110- 00???1010001?00000000000001??10000000?00?0?00100000000000000000100220011001002?00?0??0

Branchiosauridae 311100001010001210000000?100100001000020???0000(0 1)1000000010010--110100000010000- 0000???(0 1)00(0 1)00001000{0 1}0001?100??????1110010010????????0- ???????020010000?00?0000?0?0000000?????0?1200200?00-000??000????210000101?120211010—0

Micromelerpetontidae (2 3)11100001010001010000100?1{0 1}000000(0 1)000000???000011001000010010--010000000000000- 00000???00(0 1)10001(0 1)0000000?0(0 1)0?0????1110011000?????????????????010000000?00?0000?0?00000?0?????0?1210000?00- 000??000????210001101?120211010—0

Tersomius 2?010000{0 1}0110010{0 1}000011011100000010000(0 1)0001000011000001000000--110{0 1}0000000000(0 1)00(0 1)0001000001000100000000?0001?00?001100110100000000(0 1)0- 010021010?0??????????1????????????????????????????????????????????????????????????0???

Ecolsonia 0?11000001110011100001001100201000000001121000000000001000000-- 10000000001{0 1}010-000000?000010001?0000000?000?0??00?11000100000000?0(0 1)0- 0?00200?000??00010000000001011000????????????100?????????0000?0101?11??0?0?100110?0???

Acheloma 0?010000011100??000001001?1020100000000111100?000000001000000-- 100000?00011010-0000???000010001??00???0?0?0?0?????11?00?0000?0100??0- 01???00?000??00000000000000?01000??????????001???????????000????01???????0??0?????0???

Eryops 01010000?1?1000-000000?001??0000000100?0000000000000000000000--100000000000010- 000001?0000000?0??00???0?001?0????01100010000?00????0???????0?0?01?00000000000000001000???? ??????0000??00-000?000000??0???????0?10?1???0?000

Doleserpeton 3?01000010110010000001001111000021000020000100011000000010010- -12000000000000100100111000110010?0010101?10001?0?00110211210000000110- 110020110001?00(12)?0000200?0??110000112220102002000???000010100?010122002000110201000??0

Gerobatrachus 3111000010110012100001101??000?02100002000000???0??0000010010- -??0100000000000-010?11?20010101??10?1??201?0???0??01101111100???????0- ?1????1?0??3??01?????2??????11?0??12??2?1?201200?????????0100?????1?00?1?1?1?2???0???0

Platyrhinops 11010000?011001?10000010?101002020000020000000001000000010010-- 11110000000000100?0?01?000010000???0?????010?1????01102110100000201?0-?0????010000??02- 0000100??0?110000102?2?1?200200?00-0????0100???0?1??0111010?2?1000??0

Amphibamus 31110000?011001010000000?10000202000002000000001100000001001???1111000000000010010?0??00 0110010?0010000?1100100?01110211010000?????0- 0??????10001??00?0000200??00110000?00?2?1?200200?00-00000010000?0??1?0100?1102?1?00??0

227 Micropholis 20(0 1)100000(0 1)110012100000(0 1)011?0{0 1}000000000000000000(0 1)1000000001000--1- 00000000{0 1}{0 1}000-000001?000010001100000000(0 1)10000?000110(1 2)110100000011110010??1000003?000?00001?????011010?0002?0?1200200??0- 00000010001?1120001100000201000100

Eoscopus 2?0100001{0 1}11001010000110111?000000000020(0 1)21000011000000010010--110100000000000- 0000???00001000000000000?0000100??111001101000000?{0 1}?0- ?0?0200?0001000000000100001011000???0200?(0 1)200200100-00000010????01110?21001(0 1)0201000100

Tambachia 2?01000001110010000001001100201000000000111000000000001000000- -110100000011010-0000???000010000?0001110??000?00???1100010000??1{0 1}1??0- 01????01000??0???00?00??????01?0???????????00?0?????????????????011001??????0?????0???

Triadobatrachus 31111---1--??????0--0100????????21???020-0101---0-0?20--?-000-- 12?1??????????????0011?0????1????10?????01?01?0001011-??{0 1}1?0????????0????????00103?012- 00?021----011010?1???20{0 1}1?002001{0 3}------?-10???0110001100?(1 2)12-1001?0?

Caudata ?2(0 1)11-----000012-0110100-11000002{1 2}0001--00001---0-1-21----1-0--120100-000(0 1)00010010011?(2 3)00(1 2)020{1 2}011011-010110(0 1)-{0 1}00?1110-1(0 1)2-0002-2-1-(0 1)-10{2 3}{0 1}1010(1 2)00(1 3)(0 1)0120000020100100111011222201120020013------1--1111(1 2)20(0 1)(1 2)(0 1)01120201000{0 1}00

Anura ?1111-----0-1----1--{0 1}100-110-0002100002000101---0-0-20----1-0--120100-000(0 1)000100100111000(1 2)0101011011-0101(0 1)011{0 1}00?011011(0 1)2-0002-{1 2}12-0-113--01001030012-000021----- 1101011122201120020013------1011001(1 2)3001000101211001100

Albanerpeontidae ??011-----01001211--1100?00?20001100012-00001---1-0-21---01-0--110100-000??1011011011- 2???0???????0????????????????????????002-2-1110213--01?000(0 1)?012?00002??????011001122220122002001????????--1????01???0???0??0??10?0?0?

Eocaecilia 330100-0021-1----0000000?1012100121001--00011---0-10010100?00-- 1201000000000010010011-100100010000111011?000000??011021{0 1}2000---2-??10103-- 1110002?002?00-02?????0111010122-201121020013??????????????01220121002?0?011001?1

228

APPENDIX 7 Humeral measurements.

Species Specimen Humeral Distal humeral Capitulum width number/Reference length (mm) width (mm) (mm) = L = W = C Ascaphus truei RM 4430 11.61 2.70 1.40 Leiopelma hochstetteri RM 2215 9.29 2.58 1.40 Prosalirus bitis MNA V 8725 11.30 3.00 1.60 Xenopus laevis RM 2230 16.00 3.41 1.85 Scaphiopus holbrookii RM 2425 17.80 4.00 2.50 Bufo americanus RM 4999 37.50 9.90 5.50 Acris crepitans Maglia et al., 2007 10.52 2.33 1.47 Dendropsophus koechlini RM 1226 5.37 1.09 0.57 Conraua goliath FMNH 248829 57.00 15.50 11.00 Rana catesbeiana RM 2785 41.50 11.70 6.80 Rana pipiens RM2812 16.00 3.50 2.00 Hynobius nigrescens MCZ 22513 10.10 2.77 1.31 Andrias davidianus FMNH 166872 39.00 11.00 5.50 Salamandra salamandra Francis, 1934 16.20 4.76 2.62 Necturus maculosus RM 5002 25.70 6.89 2.85 Ambystoma tigrinum RM 2161 13.50 4.30 2.30 Desmognathus sp. RM 1601 3.10 0.70 0.39 Dicamptodon tenebrosus CAS 210347 5.56 1.24 0.58 Eocaecilia micropodia MCZ 9163 4.25 1.18 0.55 Edops sp. MCZ 1781 187.50 115.40 54.20 Eryops megacephalus MCZ 1220 138.70 93.70 33.80 Acheloma sp. MCZ 2524 74.70 38.40 14.30 Cacops aspideophorus Williston, 1910 59.70 33.50 11.10 Dissorophus multicinctus MCZ 4176 77.20 36.50 15.80 Doleserpeton annectens FMNH UR 1321 8.50 2.70 1.40 Pantylus sp. MCZ 3692 26.90 13.90 4.00 Trihecaton howardinus Carroll & Gaskill, 1978 13.70 5.00 1.90 Cardiocephalus peabodyi MCZ 3692 8.25 2.75 1.10 ?Euryodus sp. FMNH PR 983 11.86 4.67 1.87 Scincosaurus sp. Carroll et al., 1998 5.80 2.50 1.09 Westlothiania lizziae1 Smithson et al., 1994 9.31 3.60 1.16 aguti Holmes, 1977 30.50 14.40 3.30 inc. sed.2 S12-7e 89.45 61.60 14.65 Protorothyridae inc. sed. Reisz, 1980 22.50 6.12 2.25 Chrysemys picta RM 5004 44.00 11.10 3.60 Petrolacosaurus sp. Reisz, 1981 53.33 15.46 5.40 Sphenodon punctatus FMNH 11113 36.90 14.30 4.50 Varanus albigularis RM 5003 83.50 32.60 6.50 Crotaphytus sp. RM 5000 18.60 4.70 1.10 Lacerta sp. RM 3556 6.59 1.74 0.52 Shinisaurus crocodilurus Conrad, 2006 18.93 5.50 1.80 Hesperosuchus agilis Colbert, 1952 94.00 13.87 4.37 Alligator mississippiensis (juv). RM 5005 41.70 9.70 3.50 Dimetrodon limbatus MCZ 1347 168.00 102.00 22.50 Eozostrodon sp. Jenkins & Parrington,1976 13.88 4.13 1.13 Didelphis virginiana RM 5020 62.80 18.50 5.30 Proterogyrinus scheelei Holmes, 1984 76.66 47.52 10.47 sp. Jarvik, 1996 123.88 62.50 15.63 Acanthostega gunnarii Coates, 1996 35.52 27.43 4.83

1Westlothiania was regarded as a stem amniote for the purposes of this study (Carroll and Holmes, 2007). 2Uncatalogued captorhinid specimen from Niger, field number is given, currently located in the Redpath Museum, Montreal.

229 APPENDIX 8

Comments and modifications to the characters listed by Vallin and Laurin (2004).

Character numbers and abbreviated descriptions (in bold) refer to character list by Laurin (1998), with additional characters listed in Laurin and Reisz (1999) and Vallin and Laurin (2004). The characters are followed by comments and changes made by the present author.

Character 2. Lateral lines present at least in adults (0); only in larvae (1); never present (2). The ontogeny of microsaurs was previously coded as {12} [either (1) or (2)]. This should be done for amphibamids as well. Dolseserpeton recoded {12}. Amphibamus seems to be best coded (2), based on Milner (1982).

Character 3. Dermal sculpturing: ‘honeycomb’ (0); cosmine (1); widely spaced pits (2); low bumps (3); smooth bone (4). Unlike the previous coding, Doleserpeton does not have honeycomb sculpturing on its dermal bones in any specimen. The bones are almost completely smooth. This is in stark contrast to the similar small-sized Amphibamus grandiceps. Doleserpeton: previous coding (0), new coding (4). Tersomius is recoded (3).

Character 5. Orbit shape: ovoid (0); confluent with a large antorbital fenestra (1); open posteriorly (2). It is hard to justify the previous coding, where lysorophians, anurans, and salamanders are all coded as having the same “orbit shape”. They may all be “open” posteriorly, but whereas the orbit of salamanders and frogs is bordered laterally by the maxilla, this is not the case in lysorophians. Furthermore, the posteriorly “open” condition of this character is accounted for by other characters (e.g. loss of bones), and the baphetid condition is not shared with any other group. This character has therefore been excluded.

Character 6. Jaw joint position: posterior to the occiput (0); near to the level of the occiput (1); in front of the occiput (2). Triadobatrachus (previous coding 1) has a jaw joint positioned posterior to the occiput as in some modern frogs (e.g. Rana catesbeiana). This can be seen in ventral view (Rocek and Rage, 2000). Triadobatrachus recoded (0).

Character 8. Frontal orbital contribution between prefrontal and postfrontal: absent (0); present (1). This character may be problematic. The character is partly present in salamanders (prefrontal present). Unchanged.

Character 10. Anterior tectal: present (0); absent (1). The anterior tectal is absent (1) in Doleserpeton, Amphibamus and Apateon.

230 Character 13. Parietal-squamosal contact: absent (0); present (1). Squamosal-parietal contact. In Eocaecilia, there is a clear contact between these bones at mid-length of the parietal. In Triadobatrachus, the posterior edge of the fused fronto-parietal extends laterally to contact the squamosal. The part contacting the squamosal may be a fused tabular. Unlike the previous coding, the contact is actually absent in hynobiid salamanders (Duellman and Trueb, 1994). In lysorophians, the column-like squamosal contacts the parietal ventrally anterior to the tabular. The contact cannot be seen in dorsal view. Some microsaurs also have this contact in front of the tabular. In Brachystelechidae the parietal is apparently fused to the postparietal, judging from the size and position of its posterior parts. The contact with the squamosal occurs posterior to the tabular, unlike that of lysorophians. Due to the numerous problems with this character, it has been removed.

Character 15. Pineal foramen: present (0); absent (1). The frontoparietals are unfused along the midline in Discoglossus (Pugener and Maglia, 1997), obscuring the presence or absence of a pineal foramen. Recoded (-).

Character 17. Supratemporal: present (0); absent (1). There is a bone located well within the skull table in Eocaecilia, which was tentatively identified as the tabular bone by Jenkins et al. (2007). However, comparisons with early tetrapods (e.g. Clack, 2002) indicate that this is more likely the supratemporal bone. Note that the element in question is located lateral to the parietal-postparietal suture, and is completely excluded from the postero-lateral corner of the skull table unlike the tabular of most forms, but similar to the supratemporal. A related problem is the identity of the supratemporal bone of microsaurs. Whereas the supratemporal is completely absent in salamanders and anurans, this bone may be fused to the tabular in microsaurs, as alluded to by Carroll et al. (1998). Stating that the supratemporal is absent in all of these forms is misleading. Microsaurs, Adelogyrinidae and Eocaecilia have been recoded (?).

Character 18. Tabular: present (0); absent (1). Given that the supratemporal bone was previously coded as being absent in Eocaecilia, then the tabular should have been coded as “present” but this was not the case. Eocaecilia recoded (?). The tabular bone of microsaurs may be fused to the supratemporal, but has not been recoded here.

Character 19. Tabular occipital flange: absent (0); single, extending ventromedially (I); double (2); curved, articulates with postparietal or supratemporal and otic capsule, dorsal flange absent (3). The temnospondyls in question have only one tabular flange (not two), and are not similar to seymouriamorphs in this respect. The tabular occipital flange seems to be more dependent on the shape of the opisthotic than the authors are aware of. The large paroccipital process of the opisthotic in Doleserpeton allows only a short tabular flange in this form. All temnospondyls recoded (1).

231 Character 21. Ventrally bent, sculptured tabular horn: absent (0); present (1). Ventrally bent tabular horn. Doleserpeton and Amphibamus recoded (0). Tersomius recoded (1). It appears that this and many other characters are intimately connected with degree of ossification of certain bones. The number of characters associated with the tabular therefore seems to be unnecessary, but has been kept unchanged.

Character 22. Postparietal number: two (paired) (0); single, median (I); none (2). Postparietals are present in Eocaecilia, this form has been recoded (0). The postparietals are correctly coded as absent in anurans and caudates, in which the posterior part of the skull roof is open. However, this condition is not comparable to Brachystelechidae. In the latter family, the postparietals are most likely fused to the parietals. This conclusion is based on the position of tabular, and comparison to related forms (e.g. Rhynchonkos). The condition of the postparietals is uncertain in Triadobatrachus as well. Brachystelechidae and Triadobatrachus recoded (?).

Character 31. Temporal emargination: absent, area covered by opercular bones (0); present, bordered by squamosal, tabular, and (sometimes) supratemporal (1); absent, area covered by squamosal, supratemporal, and tabular (2); present, bordered by quadrate (3); present, bordered by squamosal (4). This is probably one of the most misleading characters in this study. By coding the presence of the notch differently if it is bordered by the squamosal, tabular, and supratemporal (1) or only the squamosal (4), the resemblance between temnospondyls and salientians is erased. New coding: Absent, covered by opercular bones (0); present, bordered by squamosal, and also tabular and supratemporal if these are present (1); absent, covered by squamosal, supratemporal and/or tabular (2); present, bordered by the quadrate (3). All forms formerly coded (4) has been recoded (1).

Character 32. Maxilla-quadratojugal contact: present (0); absent (1). This feature is present (0) in Eocaecilia, contrary to the previous coding (Jenkins et al. 2007). Contrary to the description of this character (but not the coding) the feature is also present (0) in Doleserpeton. This feature is inapplicable (-) to salamanders and these have been recoded as well. It is absent (1) in Utaherpeton, Brachystelechidae, and Rhynchonkos (Carroll et al., 1998).

Character 37. Lateral palatal tooth row: present, complete (0); incomplete (1); absent (2). Note that the authors coded lysorophians as having an incomplete (1) row of palatal teeth, based on the dentition of the vomers. This should be the case for salamanders and frogs as well, as these forms possess a row of pedicellate teeth on the vomers. Apateon also appears to have an incomplete secondary row (1). Coding caecilians as having the primitive condition (0) may be misleading as the palatal tooth row extends on the labial side of the choana in these, not the lingual side as in primitive tetrapods. Caecilians have been recoded

232 (3). New character state: palatal tooth rows extending on the labial side of the choanae (3).

Character 38. Posteromedial vomerine tooth row: absent (0); present (1). The transverse vomerine tooth row mentioned (not always positioned posteriorly) actually has a broader distribution within lissamphibia. New coding: transverse row of more than five teeth on vomer: absent (0); present (1). Present in Discoglossidae, Hynobiidae, caecilians and Doleserpeton.

Character 43. Lateral exposure of the palatine: absent (0); present (1). Amphibamus has a lateral exposure of the palatine (LEP). Recoded (1).

Character 45. Palatjne shagreen of denticles: absent (0); present (I). Doleserpeton does not have a shagreen of denticles on the palatine, rather it has a short row of teeth similar to the marginal dentition. Recoded (0).

Character 49. Transverse flange of pterygoid: absent (0); present (1). This data set has an unusual interpretation of the transverse flange of the pterygoid (see Laurin, 1998 p. 27). However this is interpreted, the condition of Amphibamus is more similar to Doleserpeton than indicated by the previous coding. Amphibamus recoded (0).

Character 55. Palatine position: excluded from interpterygoid vacuity (0); borders part of the ectopterygoid vacuity (1). In discoglossids, the palatine is sometimes fused to the maxilla (Pugener, 1997) other times to the vomer (Maree, 1945), but it always borders the interpterygoid vacuity. Recoded (1).

Character 61. Post-temporal fenestra: large (0): small, represented by a large foramen (1). This opening is large (0) in Dolseserpeton. It has been misleadingly coded as ‘small’ (1) in most modern amphibians, but the character is actually inapplicable because of the lack of roofing bones. Most of these have been recoded (-). There is a trough dorsal to the otic capsule in many anurans. This may be a remnant of the post-temporal fenestra (Chapter 2). There is a striking resemblance between Doleserpeton and anurans in this area.

Character 62. Paroccipital process: absent (0); present (1). This had been coded absent (0) for all lissamphibians, but Sigurdsen (2008 [Chapter 2]) has shown that the crista parotica of salientians probably is homologous to this structure. This feature is also well-developed in Dendrerpeton and Doleserpeton.

Character 63. Otic tube: absent (0); present ( 1). The meaning of this character is uncertain, and appears to constitute a simple postero-lateral expansion of the otic capsule. Whether the description of it as a “tube” in seymouriamorphs is warranted, this is certainly not the case in anurans, which had been coded similarly in the matrix. There is a lateral expansion of the otic capsule in anurans, but it appears different from that of seymouriamorphs in that the expansion

233 includes the area of the basicranial articulation, whereas in seymouriamorphs the expansion is behind this level. Until further studies confirm the structure of this region, this character is best left out. Deleted.

Character 64. Exoccipital-dermatocranium contact: absent (0); with postparietal (1); with parietal (2). Doleserpeton has an area of contact between the exoccipital and the postparietal (1).

Character 67. Basioccipital and exoccipital: indistinguishably fused in adults (0); suturally istinct throughout ontogeny (I); basioccipitat never distinct (2). The basioccipital of Doleserpeton is indistinguishable from the exoccipital in all specimens. This may mean that it is either fused in the adults (0), since all specimens are at least subadult, or that there is only one center of ossification (2). Recoded {02}.

Character 68. Basioccipital position: reaches edge of foramen magnum (0); excluded from edge of foramen magnum by median exoccipital contact (1). The basioccipital is indistinguishable from the exoccipital in Doleserpeton. Recoded (-) for Doleserpeton.

Character 72. Epipterygoid ossification: present (0); absent (1). The epipterygoid is present (0) in Doleserpeton.

Character 73. Epipterygoid and pterygoid: fused in adults (0); suturally distinct in adults (1). There is a suture (1) between the epipterygoid and pterygoid elements in Doleserpeton.

Character 74. Basicranial articulation: not fused, potentially mobile (0); sutured. immobile ( I ). The basicranial articulation is no more immobile in Doleserpeton and Tersomius than in Dendrerpeton. Recoded (0) for the amphibamids. Too much may be read into this character.

Character 75. Distal end of stapes: robust, links braincase to quadrate (0); slender, without contact with the quadrate (1). Amphibamus has the same type of stapes (1) as other temnospondyls (pers. obs.). The stapes of salamanders is not particularly close to the primitive condition, contrary to the previous coding. Its distal parts are not robust and it only rarely, if ever, contacts the quadrate, although it may sometimes contact the squamosal. In specimens investigated by the present author, the stapes did not contact the quadrate in either Hynobius or Necturus. New character state: stapes with slender stylus not connected to a tympanum, but sometimes contacting the squamosal (2). Salamanders recoded (2).

Character 76. Dorsal process of stapes: absent (0); confluent with the footplate (1); separated from the footplate by a deep notch (2). Amphibamus has no dorsal process of the stapes (0) (pers. obs.).

234

Character 79. Mandibular fenestrae: absent (0); small fenestrae present in splenial, postsplenial, and angular ( I ); large fenestra(e) present between angular, postsplenial, splenial, and prearticular (2). There is a small intramandibular fenestra between the pre-articular (called the cornoid by Reese, 1906) and the dentary of many salamanders. In cryptobranchids (and apparently hynobiids), this opening is located between the pre-articular and the angular. It appears to pierce the dentary only in Salamandra (Francis, 1934), but this is not the case in any of the specimens investigated by the present author (Cryptobranchus, Necturus, Ambystoma, see also digimorph.org). The situation seen in salamanders (and particularly in cryptobranchids and hynobiids) is very similar to the condition of Doleserpeton, in which a small fenestra is present between the pre-articular and the angular. New character state: small fenestra present between pre-articular and angular (if present) or dentary (3). Salamanders and Doleserpeton recoded (3).

Characters 80, 81,and 82. [80]Anterior coronoid: present (0); fused or absent (1). [81] Middle coronoid: present (0); fused or absent (1). [82] Posterior coronoid: present (0); fused or absent (1). There are three characters coding the presence or absence of the coronoids, and this appears to be superfluous. It is notable that the authors managed to code so many character similarities of the lower jaw shared by lysorophians and lissamphibians, and yet the lower jaws of these forms are dissimilar in general morphology. New character 80: number of coronoids: three (0); two (1); one (2); none (3). Characters 81 and 82 removed. However, a separate bootstrap analysis was run in PAUP of the modified supermatrix, but with these characters unchanged. The result was almost identical to the run in which the characters were excluded.

Characters 83 and 84. [83] Anterior splenial: present (0); fused or absent (1). [84] Postsplenial: present (0); fused or absent (1). As with the coronoids it is probably better to have one character representing the number of splenials. New character 83: number of splenials: two (0); one (1); none (2). Character 84 removed.

Character 87. Parasymphysial fangs: present (0); absent (1). The parasymphyseal “fangs” of Doleserpeton are actually a short row of pedicellate teeth similar to the marginal dentition. Such dentition is also present in the Cretaceous caecilian Rubricacaecilia (Evans and Sigogneau-Russell, 2001), showing a probable origin of the double tooth rows seen in the lower jaw of Eocaecilia. New character state: tooth row similar to marginal dentition at symphysis (2). Eocaecilia and Doleserpeton recoded (2).

Character 88. Coronoid fangs: present (0); absent (1). Coronoid fangs are absent (1) in Doleserpeton.

235 Character 89. Coronoid denticles: absent (0); present (1). Dentary: dentigerous (0); edentulous (1). Coronoid denticles are present (1) in Doleserpeton.

Character 91. Retroarticular process: small, directed posteriorly (0); large (1); mid-sized, extends posteroventrally (2). The retroarticular process is small (0) in Doleserpeton.

Character 92. Medial mandibular tooth row: on coronoids (0); absent (1); on pseudodentary (2). The medial pseudodentary tooth row in caecilians is probably an extension of the symphyseal tooth row discussed in character 87 (above). Unchanged.

Character 94. Marginal teeth: non-pedicellate (0); pedicellate (1). The marginal teeth of Tersomius are non-pedicellate (0), those of Amphibamus and Apateon may be pedicellate, or they are polymorphic (recoded {01}).

Character 95. Maxillary teeth: at least as large as dentary teeth (0); much smaller than dentary teeth (1). The maxillary teeth of Doleserpeton are similar in size to those on the dentary (0).

Character 96. Labyrinthine infolding: present (0); absent (1). Amphibamus does not have labyrinthine folding of the dentine (1).

Character 100. Neural arches and centra: discrete elements throughout ontogeny (0); fuse late in ontogeny (1); indistinguishably fused early in ontogeny (2). Coding lissamphibian vertebral elements as being fully fused early in ontogeny (2) appears to be incorrect (Duellman and Trueb, 1994; Pugener and Maglia, 1997; Carroll et al., 2004). Lissamphibians recoded (1).

Character 106. Anterior articular surface of atlantal centrum: no broader than posterior surface (0); broader than posterior surface (1). The anterior atlantal articulating surface is wider than the posterior one in Doleserpeton, recoded (1).

Character 109. Atlantal intercentrum: present (0); absent (1). The atlantal intercentrum is absent in Doleserpeton.

Character 111. Transverse process in mid-presacral vertebrae: absent (0); short (I); long (2). The length of the transverse process in Doleserpeton and Amphibamus is best described as short (1).

Character 112. Parapophysis: short, small facets (0); almost as long as the diapophysis (transverse process) (I ); absent or not distinct from diapophysis (2). The paraphysis (transverse process of centrum) is small in Doleserpeton (0).

236 Character 113. Capitulum articulation in mid-presacrals: on intercentrum of its segment (0); on intercentrum and pleurocentrum of its segment (1); on pleurocentrum of its segment (2); on pleurocentrum of segment in front (3); on the transverse process (4). The capitulum of the rib articulates with the paraphysis on the intercentrum in Doleserpeton (0). This is also the case with the first rib of Eocaecilia, but perhaps not the other ribs (Jenkins et al. 2007). Eocaecilia coded {01}.

Character 114. Number of presacral vertebrae: 23 to 32 (0); fewer than 23 (1); 33 to 60 (2); more than 60 (3). The number of presacrals in Doleserpeton is uncertain, but it can be narrowed down to between 22 and 25 presacrals. Recoded {01}

Character 116. Rib length: less than four centra (segments) long (0); at least four centra long (1). This character fails to show the difference between the rib morphologies of tetrapods. There appears to be a tendency for relatively longer ribs in larger animals, so equally sized specimens should be compared. Amniotes have relatively long ribs with strong ventral curvature (1). This condition is also seen in most lepospondyls (except Adelogyrinidae). In modern amphibians (including Eocaecilia and Triadobatrachus) the ribs are very short and slightly posteriorly oriented. Modern amphibian ribs lack the strong ventral curvature seen in amniotes and most lepospondyls. This is also true for amphibamids and the early tetrapod Acanthostega, and probably represents the primitive condition (0). In Doleserpeton, the longest ribs are less than three centra long, and most are even shorter. New coding for rib morphology: Short ribs (shorter than three centra in small animals), curved slightly posteriorly (0). Long ribs with strong ventral curvature (1). Ichthyostega, Eryops and seymouriamorphs appear to have an intermediate condition {01}.

Character 119. Uncinate processes: absent (0); present (1). Uncinate processes are absent (0) in Doleserpeton and Dendrerpeton.

Character 120. Interclavicle: without a parasternal process (0); with a parasternal process (1); absent (2). The interclavicle apparently had no parasternal process in Doleserpeton (0).

Character 121. Clavicle: present (0); absent (1). The clavicle is present (0) in Doleserpeton.

Character 122. Clavicle shape: with a ventral plate (expansion) (0); without a ventral plate (1). Although the clavicle has a ventral plate (0) in Doleserpeton, it is relatively slender. According to Wellstead (1991), this feature is also present in lysorophians, and Asaphestera also appears to have the primitive condition (Carroll and Gaskill, 1978). The latter is also recoded (0) here. The clavicle of brachystelechids is unclear (?), and Pantylus appears to be somewhere in between {01}.

237

Character 123. Cleithrum: with a dorsal expansion (0); slender, without a discrete dorsal expansion (1); with a ventral expansion (2); absent (3). The cleithrum of Doleserpeton has a moderate dorsal expansion (0).

Character 124. Postbranchial lamina: present (0); absent (1). The postbranchial lamina of the cleithrum is absent (1) in Doleserpeton.

Character 129. Scapulocoracoid and cleithrum: coossified (at least in adults) (0); discrete elements (1). The scapulocoracoid and cleithrum are not coossified in Doleserpeton (1).

Character 131. Supinator process: low, indistinct (0): high, well defined (1). The supinator process is absent (0) in Doleserpeton, Amphibamus, Apateon, Eocaecilia and Triadobatrachus.

Character 132. Anterior humeral ridge: present (0); absent (1). The anterior humeral ridge(anterior keel of Coates, 1996) is absent (1) in amphibamids and Eocaecilia.

Character 133. Ectepicondyle: low, indistinct (0); high, well defined (1). Given the previous coding of the ectepicondyle as being “low and indistinct” (0) in salamanders and frogs, it should be coded as such in Eocaecilia and Triadobatrachus as well (pers. obs.).

Character 134. Pectoral tubercle: separated from proximal articular surface by finished (periosteal) bone (0); confluent with proximal articular surface (not separated from it by periosteal bone) (1). The homologies of this character are not clear. The pectoral (humeral) tubercle is a term that is most often used for mammals, and is rarely used for early amniotes and amphibians. The infraspinatus and supraspinatus muscles inserting on the tubercle are probably homologous with the supracoracoideus muscle of amphibians (Romer and Parsons, 1986). This muscle inserts on the deltopectoral crest in amphibians and reptiles. The authors are presumably referring to the presence or absence of a dorsal humeral process (crista dorsalis) which is the insertion point of the subscapularis muscle (Francis, 1934). This process is often, but not always, present in salamanders (absent in Necturus, Cryptobranchus and probably also in hynobiids), and never present in anurans. It is absent in Doleserpeton. Given the uncertainties of this character, and that much variation may stem from poor preservation, this character has been deleted.

Character 135. Entepicondyle: broad (0); narrow (1). The relative width of the entepicondyle (see description in Laurin [1998b]) is generally highly variable in tetrapods, because of the great variation in shaft length. Also, the entepicondyles of lissamphibians (including Eocaecilia) and most dissorophoids are directed distally. This has been introduced as a new character state (3). That of

238 Acanthostega appears to be intermediate {03}. The states of Amphibamus, Lysorophia and Rhynchonkos are uncertain (?).

Character 139. Olecranon process: absent or cartilaginous (0); ossified (1). A highly developed olecranon is found in Eocaecilia, Ecolsonia, Rhynchonkos and others. Most forms are actually intermediate. This character is confusing because of the highly variable degree of ossification of the olecranon. The olecranon of salamanders is variable, and that of Doleserpeton varies from completely unossified to almost completely ossified due to ontogenetic differences. Deleted.

Character 140. Ulna length: shorter than radius (0); equal or longer than radius (1). The ulnae of Doleserpeton and Amphibamus are about the same length as the radii (1).

Character 141. Number of digits in manus: none (0); eight (1): five (2); four (3). There are four digits in the manus of Doleserpeton (3).

Character 142. Number of pelvic ossifications: one (0); three (1); two (2). This character is incorrectly coded. Triadobatrachus is actually very similar to modern anurans in having two ossified elements in the pelvis, whereas the third element (the pubis) remains cartilaginous (2). Carroll et al. (1998) clearly show three ossifications (1) in Cardiocephalus, Microbrachis and Utaherpeton.

Character 143. Iliac blade: very short (0); with discrete dorsal and posterior flanges (1); with a single, posterodorsal flange (2); with a single, dorsal flange (3); with a long anterodorsal flange (4). The iliac blade of Doleserpeton has a single posterodorsal flange (2).

Character 144. Internal trochanter: absent or not represented by a discrete structure (0); present, separated from proximal articular surface by finished (periosteal) bone (1): not separated from proximal articular surface by finished bone (2). The internal trochanter of the femur of Westlothiania should not be coded as confluent with the articulating surface (2) as in anurans, but as separated (1) (Smithson et al., 1994). This is true for Hynobius (Carroll and Holmes, 2007), Eocaecilia (Jenkins et al. 2007), Doleserpeton and Amphibamus as well. The internal trochanter is not absent in Acanthostega (Coates, 1996). There are numerous problems with this character, and small changes in ossification can result in large differences in the way this character is coded. Removed.

Character 145. Intertrochanteric fossa: present, proximal femoral head concave ventrally (0); absent, proximal femoral head flat or convex ventrally (1). Salamanders and Eocaecilia do have an “intertrochanteric fossa” (0) unlike previous coding. The femora of modern salamanders are, in fact very similar to those of Doleserpeton. It is also present in Acanthostega (Coates, 1996). Only anurans should be coded as lacking this fossa (1).

239

Character 146. Femoral adductor blade: absent (0); present, long (1); present, reduced distally. The adductor blade is weakly ossified but present in Doleserpeton. This is also the case for most salamanders (e.g. Francis, 1934; Carroll and Holmes, 2007).

Character 151. Number of phalanges in second pedal digit: two (0); three (1). Doleserpeton has two (0) phalanges in the second pedial digit. The detail with which the pes is coded seems excessive compared to the manus, but has been left unchanged.

Character 159. Humeral length: equivalent to at least four dorsal centra (0); shorter than four dorsal centra (1). The humeri of Doleserpeton are longer than four centra (roughly five centra long). This is true for Triadobatrachus as well. Both recoded (0).

Character 160. Posterior margin of skull: concave (0); straight (1); convex (2); undulating (concave on each side, extends posteriorly in the midline) (3). This character is extremely problematic when applied to such diverse animals, as the elements constituting the posterior margin of the skull are not the same. It is unclear why Doleserpeton and Amphibamus have been coded differently when they are, in fact, nearly identical in this respect. Excluded.

Character 161. Number of premaxillary teeth: more than 12 (0); 6 to 11 (1); 5 or fewer (2). The character state (0) should presumably have been phrased “12 or more”, as species with 12 teeth would otherwise not be included. Unchanged.

240 APPENDIX 9

Comments and changes to the characters listed by Ruta and Coates (2004).

Character numbers and descriptions (in bold) are from Ruta and Coates (2004). These are followed by comments and changes made by the present author.

Character 9. Septomaxilla a detached ossification inside nostril: no (0); yes (1). The septomaxilla is a separate ossification in Doleserpeton (1).

Character 28. Maxilla contributing to orbit margin: no (0); yes (1). It is uncertain (?) if the maxilla contributes to the orbit rim in Amphibamus.

Characters 44, 45, 60, 66, and 71. These characters are based on the presence or absence of “strongly interdigitating” sutures between bony elements. Such sutures are (as admitted by the authors) highly variable between individuals and taxa, and also prone to misinterpretation, particularly in small specimens. These characters have been excluded.

Character 50. Postparietal/exoccipital suture: absent(0); present (1). There is probably a slight overlap of the postparietals and exoccipitals in Doleserpeton, but no suture. Unchanged.

Character 63. Separately ossified supratemporal: present (0); absent (1). The “tabular” of Eocaecilia may in fact be a separately ossified supratemporal (Jenkins et al. 2007), and should be coded as uncertain (?). Interpreting this bone as a supratemporal is actually more in line with the descriptions of both primitive tetrapods (e.g. Clack, 2002) and temnospondyls.

Character 67. Separately ossified tabular: present (0); absent (1). The “tabular” of Eocaecilia may in fact be the supratemporal (see above), and should be coded as uncertain (?).

Character 113. Posttemporal fossa occurring at occiput dorsolateral corner, delimited dorsally by skull table, not bordered laterally and floored by dorsolateral extension of opisthotic (0); fossa present near occiput dorsolateral corner, delimited dorsally by occipital flanges of tabular and postparietal and bordered laterally as well as ventrally by dorsolateral extension of opisthotic meeting tabular ventromedial flange (1); small fossa present near occiput ventrolateral corner, bordered laterally by tabular ventromedial flange, delimited dorsally by dorsal portion of the lateralmargin of the supraoccipital–opisthotic complex and floored by lateral extension of opisthotic (2); absence of fossa (3). The posttemporal fossa of Doleserpeton is bordered (and closed laterally) by the opisthotic, tabular and postparietal elements (1).

241 Character 123. Vomer with (0) or without (1) fangs comparable in size to, or larger than, marginal teeth (premaxillary or maxillary). Doleserpeton has no fangs on the vomers, but it does have rows pedicellate teeth on this element, as do modern amphibians. It should be coded similar to lissamphibians (1).

Character 167. Separately ossified supraoccipital: absent (0); present (1). Doleserpeton lacks a supraoccipital bone (0).

Character 175. Opisthotic forming a thickened plate together with the supraoccipital, preventing the exoccipitals from contacting the skull table: absent (0); present (1). The exoccipital contacts the skull table (0) in Doleserpeton.

Characters 198-200. Although the splenial is now known in Doleserpeton (Chapter 1), the relationship of this bone to the features in question (adductor fossa, coronoids) is still uncertain. Unchanged.

Character 205. Angular without (0) or with (1) mesial lamina. The angular is not extensive medially in Doleserpeton, but it is visible in this view. Recoded (1).

Character 206. Angular/prearticular suture: present (0); absent (1). The angular-prearticular suture is present (0) in Doleserpeton.

Character 208. Separately ossified surangular: present (0); absent (1). The surangular is at least partially fused to the articular in Doleserpeton. Coded (01).

Character 211. Prearticular/splenial suture: present (0); absent (1). A presplenial-prearticular suture is absent (1) in Doleserpeton.

Character 217. Middle coronoid with (0) or without (1) fangs comparable in size to or larger than marginal dentary teeth. The middle coronoid of Doleserpeton lacks fangs (1).

Characters 218-219. [218] Middle coronoid with (0) or without (1)small teeth (denticles) forming ontinuous shagreen or discrete patches and the basal diameter and/or height of which is less than 30% of that of adjacent marginal dentary teeth. [219] Middle coronoid with (0) or without (1) anteroposterior tooth row orientated subparallel to marginal dentary teeth and the basal diameter and/or height of which is 30% or greater than that of marginal teeth and twice or more that of denticles, if present. The middle coronoid of Doleserpeton has either large denticles or small teeth. Coding (?) is appropriate.

Character 221. Posterior coronoid with (0) or without (1) fangs comparable in size to or larger than marginal dentary teeth. There are no fangs (1) on the posterior coronoid of Doleserpeton.

242

Character 222. Posterior coronoid with (0) or without (1) small teeth (denticles) forming continuous shagreen or discrete patches and the basal diameter and/or height of which is less than 30% of that of adjacent marginal dentary teeth. There are no denticles (1) on the posterior coronoid in Doleserpeton.

Character 223. Posterior coronoid with (0) or without (1) anteroposterior tooth row orientated subparallel to the marginal dentary teeth and the basal diameter and/or height of which is 30% or greater than that of marginal teeth and twice or more that of denticles, if present. The posterior coronoid of Doleserpeton lacks a tooth row (1).

Character 224. Posterior coronoid without (0) or with (1) posterodorsal process. The posterior coronoid of Doleserpeton appears to have a posterodorsal process (1).

Character 225. Posterior coronoid exposed in lateral view: no (0); yes (1). The posterior coronoid of Doleserpeton is exposed (1) in lateral view.

Character 226. Posterodorsal process of posterior coronoid contributing to tallest point of lateral margin of adductor fossa: no (0); yes (1). The posterodorsal process of the posterior coronoid contributes to the the tallest point of the lateral margin of the adductor fossa (1) in Doleserpeton.

Character 227. Adductor fossa facing dorsally (0) or mesially (1). The adductor fossa actually faces dorsomesially in Doleserpeton. As this is probably what was meant by “mesial exposure” it has been coded as for other temnospondyls (1).

Character 228. Marginal tooth pedicely: absent (0); present (1). The pedicelly of the teeth of Amphibamus is uncertain (?).

Character 236. T-shaped dorsal expansion of cleithrum: absent (0); present (1). The dorsal expansion of the cleithrum of Doleserpeton is not T-shaped (0).

Character 237. Cleithrum with (0) or without (1) postbranchial lamina. The cleithrum of Doleserpeton lacks a postbranchial lamina (1).

Characters 240-242. Although some remains of the interclavicle of Doleserpeton are known, these characters are best left as uncertain (?). Unchanged.

Character 243. Transversely elongate grooves and ridges on central part of interclavicle ventral surface: absent (0); present (1). There are no transverse ridges (0) on the medial part of the interclavicle of Doleserpeton.

243 Character 245. Glenoid subterminal: yes (0); no (1). The presence of a “subterminal glenoid” is highly dependent on the degree of ossification and ontogenetic stage of the animal in question, as evidenced by the numerous Doleserpeton scapulocorocoids available. Excluded.

Character 246. Enlarged glenoid foramen: absent (0); present (1). The glenoid foramen of Doleserpeton is not enlarged (0) relative to that of other temnospondyls.

Character 247. Ventromesially extended infraglenoid buttress: absent (0); present (1). Doleserpeton has an infraglenoid buttress similar to that of Eryops (1).

Character 250. Distinct supinator process projecting anteriorly: absent (0); present (1). The supinator process is absent (0) in Doleserpeton.

Character 251. Sharp-edged, ventral humeral ridge: present (0); absent (1). The crista ventralis of modern amphibians is actually homologous with the deltopectoral crest of other tetrapods (Chapter 3). This crest is present (0) in Doleserpeton and Eocaecilia.

Character 256. Ectepicondyle ridge reaching distal humeral end: no (0); yes (1). The ectepicondyle reaches distally (1) in Doleserpeton.

Character 259. Position of radial condyle: terminal (0); ventral (1). The radial condyle (capitulum) of lissamphibians and Doleserpeton is large, and could be said to be situated terminally and ventrally. However, based on how this character is coded, the radial condyle of Doleserpeton and Eocaecilia should be coded as ventral (1). Close studies of the humeri of Triadobatrachus shows that the radial condyle in this form would have been situated in similar position, and this is clearl seen in the related Czatkobatrachus. Triadobatrachus coded (1).

Character 266. Humerus length greater (0) or smaller (1) than combined length of two and a half mid-trunk vertebrae. The humeral length of Doleserpeton is greater (0) than two and a half vertebrae.

Character 268. Radius longer (0) or shorter (1) than humerus. The radius of Doleserpeton is shorter (1) than the humerus.

Character 269. Radius longer than (0), as long as (1), or shorter than (2) ulna. The radius and ulna of Doleserpeton are approximately equal (1) in length.

Character 271. Olecranon process: absent (0); present (1). The olecranon is hardly ossified in Karaurus, intermediate in Doleserpeton, Tridobatrachus and most anurans and caudates, and well-developed in Eocaecilia. The coding of this

244 character is somewhat misleading, and it may be too dependent on small individual differences in relative ossification. Excluded.

Character 278. Ischium contributing to pelvic symphysis: no (0); yes (1). The ischium contributes to the pelvic symphysis (1) in Doleserpeton.

Character 279. Number of pubic obturator foramina: multiple (0), single (1), or absent (2). Doleserpeton and Eryops have one (1) obturator foramen.

Character 284. Femur shorter than (0), as long as (1), or longer than humerus (2). The femur is longer than the humerus (2) in Doleserpeton.

Characters 286-287. [286] Outline of tibia medial margin shaped like a distinct, subsemicircular embayment contributing to interepipodial space and the diameter of which is less than one-third of bone length: absent (0); present (1). [287] Tibia without (0) or with (1) flange along its posterior edge. The tibia of Doleserpeton is of the same type as other temnospondyls (0).

Character 288. Fibula waisted: no (0); yes (1). The fibula of Doleserpeton is waisted (1).

Character 289. Ridge near posterior edge of fibula flexor surface: absent (0); present (1). There is a low posterior ridge (1) on the fibula of Doleserpeton. This ridge is frequently quite pronounced in salamanders, and Valdotriton is probably best recoded (?).

Character 290. Rows of tubercles near posterior edge of fibula flexor surface: absent (0); present (1). There are no tubercles (0) on the fibula of Doleserpeton.

Character 296. Cervical ribs with (0) or without (1) flattened distal ends. The cervical (anterior) ribs of Dolseserpeton are flattened (0) distally.

Character 303. Axis arch not fused (0) or fused (1) to axis (pleuro)centrum. In subadult Doleserpeton specimens the axis arch is not fused to the centrum, although it may be fused in fully adult individuals. Unchanged.

Characters 338-339. [338] Ossified lepidotrichia in caudal fin: present (0); absent (1). [339] Ossified epidotrichia in caudal fin: present (0); absent (1). Lepidotrichia are absent (1) in Doleserpeton.

245 APPENDIX 10

Modified supermatrix

Characters 1-161 are from Vallin and Laurin (2004), characters 162-500 from Ruta and Coates (2007), characters 501-719 from Anderson et al. (2008). The original order of the characters has been kept. For instance, character 10 from Ruta and Coates (2007) is equivalent to character 171 (10+161) here. See appendices 5-7 for individual character changes.

The following characters were excluded before analysis (using the Exclude command in PAUP and MrBayes) to avoid character repetition:

7, 10, 15, 17, 20, 22, 24, 25, 29, 31, 35, 38, 40, 43, 47, 48, 54, 61, 80, 81, 82, 83, 84, 85, 86, 88, 89, 95, 98, 104, 105, 111, 120, 124, 133, 136, 137, 138, 140, 141, 143, 144, 146, 155, 156, 159, 161, 180, 189, 196, 200, 217, 220, 228, 231, 238, 240, 249, 251, 258, 266, 281, 285, 294, 295, 300, 301, 311, 327, 332, 336, 345, 348, 353, 357, 362, 365, 369, 373, 377, 381, 389, 390, 405, 414, 429, 432, 463, 465, 469, 472, 477, 496, 497, 504, 505, 512, 513, 521, 522, 526, 528, 545, 552, 553, 554, 555, 559, 560, 561, 578, 583, 590, 591, 594, 599, 600, 615, 621, 637, 638, 649, 665, 679, 684, 692, 698, 703, 711, 712

In a separate run, characters 80-84 of Vallin and Laurin (2004) were included (see text). This did not significantly change the outcome of the analysis.

The following characters were of questionable value, and were therefore also excluded (see Appendices 5-7 for justification): 5, 13, 63, 134, 139, 160, 205, 206, 221, 227, 232, 406, 534.

Dimensions: number of taxa = 25, number of characters = 719

Acanthostega 000000100110000100000000010010100000001000001010000010000(01)01?10??01?00?0000010??00?0?0 01?0000000?01???????????1???10???000000110000010{03}00010101001????????00011000??0000001000 0000000001000000000120000111100000000000000001???000101000000000000011000000000000100000 0001110000000000001100000000000000000001000000000000000000000000100000000000000010000001 101010000000010010010001100110011000?000000000000110000010000000001000011000111000100001 10111010000100000000000000000000100000000000111100000001000101000100000100000?0000000000 10000000000000000000000100000000010--000000000000000-0000—000000000000000000000000000 00000000000000000000-010000000000?0000000000000000(0 1)00000000000000000000000000 000000000000000000000001000000

Proterogyrinus 1(01)20001001000?0000110001010000110?001????0001001?00010001001- 0001110011010????20000100001001000013100000000000?01011110000010110100010010011211102000 21??110000020??01001?1000010000000?00000000011000102110000001000000000000000000101?00000 0001000010000011100?00001000011002203000000{12}41??0???0???0010000001?0000001000000000000 0???000001001001010200000????10000011?0?10100010?0?0?010101010??1001010000111100000011100 1010120101100111110110110110101120001102111001000000000000110000000000111111000001010111 01000000(0 1)00{0 1}001?00010000000?000000000000000000000001000000010--000000000010000-

246 0020000000000????000?1000000?0??000000000000000000?00-00{0 1 2}001000000?00000000200001?0100000000000000?000000000010000000200100?1010010111000100

Dendrerpeton 2{12}0000101110000000101000010000100000202010001001?00001001001?1?1?????11??0???02???0000 0110010000?010100????????000?0110000011111?0011??10011?10102000?????1000030100000011100000 0000000000000000010000002110110010?00000000010000010011?000000001000010100000100(0 1)0001100002000020?0000004?0011?1100000110000001010000100010000000110002?0100100?1000002 01000??????0000???0110100010?0???0???0??????1000000000111000000011?01?01012100110011011010 011011111?120001??20110000000???0000010000000000011????0000011001110000000000110010000000 00?110000000000000000000010000000000000--100000000000000-000000?0000000000000000000000 10000?110000000000001110-00???1010001?00000000000001??10000000?00?0?0010 0000000000000000100220011001002?00?0??0

Asaphestera 2{1 2}40011011100-01?010000- 010000210001{01}02?100?10?10000??00????0??13210?11??0??????????001??0010001?512100???????{1 2}0??{12}11?0?00??111??0010?01000???1????????????0100310??01001?1000000000000000010000011?0 01021100000001000000101???1???0??00010?000001001000000101000010?10000000020011000004{01}? ???????????100000???????00??0000000000000???100101000?????????????????10000??????0??1010?????? ???????????00000002???111000111??0??010121?0????110??0010?0011?????{1 2}??????????0?10 000??100??????010??0??111?????1?001???1?122011-110111001000010000?00100000300010- 0001000010010000000-0--{0 1}10000001010000-01210002001000???01000?0??00???? 001010000?000?0100????20????110000??12?00102???????10001122??011??000???1110- ?0????102002100?1???200?10?????

Pantylus 2{1 2}01011011100-11?010000-0100002100012021100111-- 00001000311101003210011010???12?0?00001110010001151210?1110-1120301111010{0 1}1111101001000100113101020002????11100120??01101011010000000000000100100110001021101100 001000000101???1???000010000000001101100000101101000010000000003011000004?11000110000001 00001??????000100000000000000021001010000002002000011???0000001??011010101001???1???01111? 11001000120110210000011001101012110110111012010010011?1001?000???{12}0110010000111000001 110001?00011111111100011001?110011-11011100100001000000000101011001--0000000000010110000- 0--101000000222010-010100011020002000000011100010011?00000001000100011(0 1)10202010000001?01101010200000011000012202011101000?01111- 1010002010022002100110211000110

Cardiocephalus 2{1 2}41021011100-01?010000- 0100002100012021100110110000100011011001?210011??000110?????001?10010001?311?00???0- 11?0??{12}11?0??????11??001001110?1?1?20?0????????0101220??01101010000100000000100110000100 001021100000001010000111???1???000000010000001001?1?000101000???0100000000020?1001014{0 1}1100011000?1000000010100000010000001000000??2000101000000200200002????1100001??01101?10 11????????????????100100012?????????????????????????????????????????????????????????????????????? ??????????????????????????????????32011-110111001100010000?0001101121001--0001000010010(0 1)10000-0--121000?10222111011000003???0???????0??????0??0???????0????0?0111?1??0- 20????110002??01?00102???????100001220?0110000010????????1000?0-1121012?0??202010?????

Brachystelechidae 2{1 2}4(1 2)021111111-01?0000?--0(1 2)0000210011?02?000?00100010100011(0 1)1100-3200(0 1)11010001(0 1)2311??0?1--001(0 1)001151210??110-11?0311111010?1111101001?0110011?1210201- 21(01)(01)(02)10(01)10(013)20??01101?1000000001000000110000?121001121110?1?????????0001???1?? ?00000010000000100?0100001?1000???0100002200120?1000011??????????????????????????00????00000 000??????100101000000?00200002????110001??????0???????????????????????000000020?11210001111? 01?01???11011111101201?0?001?11101(1 2)000100?01?0010000??1000(0 1)0111?(0 1)01?000111111111000110011132011-10111100100???010000012102130001--00000000011120---00-0-- 12101-?100120011011110-3???00????100110-110010000000?0??0??001??11??101????1110 003?0120001020010000100011220?011000000001110-11???0102002201210?10020110?110

247 Rhynchonkos 2{1 2}4{1 2}021011100-01?010000-0100002100??00210001001-0010?00010(0 1)110013210011010001(0 1)210000001100000001131210011100112??221110?00???11?100100?11011?1{0 2}102000{2 3}10??10101220??011010100000000000000001?0000120001021100000001010000111???1???0000000100 00001001100000101000???01000000000(12)0?10000044?1000110000100000001010000000000000100000 00021001010000001002000021???1100000?1011010101001???011001?????100000002??????????????01?1 11??110111111012010010011?100120001002?11011000011100???111010000??11111??01?001???111330 11-110111001000010000?0010102121001--0001000010010010?00-0--121000010012000- 0101101310100010010011011000000011001000{01}0000100001110101010110002??02?00102???????100 00122-2011210001?3??????????0????12201210011020100010?

Eocaecilia 2240021011111-11??---00-0200002001003121000101--000001102001100-?2{0 2}- 111??00010?31121112--01211?1131??10111001110{01}2?0110?????111??0010?3110?1???10?0???????12- -1300??01001?100000000?0100000?1001010000??2011000000?000000001????????0000000001???????10 00001?1000000?100000000030?100000440101111100110000001??????00100101100000110002?0101100? 0000002001121???1100000??1??1???1???1???1???1???????11000000????????0???1??1?111??1111111110 1{1 2}0?????????100111001??????01000011?1000001110?0000000111111?00101???111330100-0021-1--- -0000000?1012100121001--00011---0-10010100?00--1201000000000010010011- 100100010000111011?000000??011021{0 1}2000---2-??10103--1110002?002?00-02?????0111010122- 201121020013??????????????01220121002?0?011001?1

Microbrachis 10?0011011101-01?010000-0?000021000120211001101100001000110101013210011010?010200000 00111001000115101000110-1120321111000111111010010?01000?311{12}020?0211{01}210011110? ?01001?110100000000000011000001000011211?1000101000000001???1???00000010000000?0011000001 010000000100002000020?100000{23}31110011000011100000111100001?0000010000000002?001010001 001002000011???10000011001101010100010101010101101100000002011121001?????01?01???01011111 0011001010011110012000100????0010000011000001110001?000111111111000100111132011- 100111011000010000?0010000010001--0000000011110000100-0--010000000(0 1)11000-010?00?201-001- 00000000-10000000001010000000010100111000002100100210110001020000001101011212101100 0000020110-00010000-102001100?21011110??00

Nectridea (1 2)(0 1)000(0 1 2)(1 3)(0 1)11(0 1)0(0 1)001(0 1)0(0 1)00001020(0 1 2)002(0 1)000(0 1)0021(0 1)001(0 1)(0 1)1000(0 1)0(0 1)(0 1)(0 1)0111101011210011?1(0 1)????2(2 3)1(0 1)110011(0 1)00(0 1)00011{1 5}121001210-112(01)2(01)1111000011111010111?01(01)0?1310??201?2110210011010??00001?10010000 010000001100001100001021101?1?????????0101???1???00?000?0000000110010001010100100001000020 00020?1000004401???11100011100001??????00000000000000000002?1000100?000000200001?????????? ??????????????????????????????000?0002011000000?111?11?0????1101111100120100?00112100120001? ?????0?10000?10111101110101?0000111111001001???11133011-110111001000010000?00120000{1 3}0011--00000000000120---0000--010100000012000-01??11?301-001?00100101- 0001?????00010000100?0???????????01?000011?01210001201111001000101222?11210000?20- 000?000002??0022001000010201000?10

Lysorophia 2240221001101-111000001-0212022-0011102100-101--001000002101100132100111-00011031111001-- 00100011{1 5}201000110-1120331111(0 1)(0 1)00111110100101?110003122020??2102210-- 1220??00001?10000100010000101?10001100001??1001100001010001??1???1???000000?0001???????110 001??????1??010000??0?130100000144110?0110000101?0001??????00000000000001???00210010100000 00002000021???110001???1??010101??1???1???1???????00000002011101001???1?0??111??1?011111001 10000?0011?100120001??????01100100?100000???000??10011111??110001??011133011----1110011-1-- 000000012000-20001--00001---0-1101110-000--120100000012000-012100031110001001001111 100?10100100100001001100111110203--1100002?0110001020000101001010212001121000003110 00100000010011001(0 1)11?12021110??00

Adelospondyli 1000011011101-01?1---00- 0?01?01(01)00012??1100110100000?0001101??01?1???11???????????00001??0010001?5101001?00- 112033-011100001111-11------0------11-1-{03}10??000?1?110100000?100?00101000?1??01?1 ?10?110000?000010?01???1???1?????0???0010?000?001?0??1000000010?0?120000??000000?3??????????

248 ????0???????????????????????????????0000101????????????1?????10000???0010??1010????????????????? 00001101011?0010????1??????????????????????????????????????????????0?101100?00????111?001?1??? 11????10?01????1??2?011-10021-1----0000000?00110000300000-00000010?111000000{0 1}10-- 0101???000?0000-01000012?1-0?1???0?0001-0000??????00??0?0000000?00??100(0 1)???100000??010?00102????????0001020220112??00???0-00000???0???2????????-????????????

Aistopoda 2{1 2}?00(1 2)(1 2)(0 1)11100-0(0 1)(0 1)0?00(0 1 2)0?010(0 1)002(0 1)000?1?21?001????00?0?0001??1-00-{0 3}1{0 2}??110?0?011031121111--00100011{1 5}121101000- 112023?111121-3?111-?1------0------?----100--310??01001?11010100010000001010000100000?2 1001100001000010001???001000000000001???????00000000100000001000?2100000100100144???????0? ??1000000???????000?00000000000?0???000001010000000200001????1?0000???1??01010100?????????? ??????0000(01)002011?????????1??????????????????????????????????????????????1000000?101???????01 01?0?00111111001100???11?230100-0010100101011000000012101020001--00001---0-00001001000-- 120100100010000-01200000001--????100?101?0-0????000000000100000001110-003--1110012?012- 000021----00111010122200121-01--0------00------

Colosteidae 100000100110000(01)00100000020(01)0020000?0010000000000000010010010101110001100001102?000 000001001001010100000???0002002101110000011101000000100112102020002000?10011310??00011?1 101010101100000000000011001012100110000000001000(01)???0001001010000000000110000000000000 0000100011000000100000022000001100000000000000000000010000010000000001000001001100001200 00001011001001100010100010000100101010111?0000100000010001000011100101011000110010012010 01001111111200010020?10100110000000001000000000001111??0000110101?1020100000100011000000 00000012(0 1)000300010000000000010000(0 1)000000--000000000000000-000000?000(0 1)000000000110(0 1)00000000000000000000000000100-001010010002?00000000000000??10100 000?0000211100000-0100000001110011011110100201010--0

Seymouria 2{1 2}0001101110010000201001010000110000202010001011100012-- 3111111111100110001021000000001110010000131110100000002010(12){01}1101001111111011100100 012112020002????12000(01)20??01001110000000000000000100000110001021100000(01)0100000000011 000010001?01000000100001000001010010001100002000020110000044101001100000110000011110?101 100000000000000020000010000?12002000011???1100001110110101010001010101010111110010000201 102?001101111110101211011001101201111101110?10200011020110000000000000001111100000011111 110000010101110100000000010010000000000000000001000010000000000001000000000-- 000000000012010-000000000000000000000001010011??0102000000000000000(0 1){0 1}- 200020001001?000000(01)020000100100000000001100000100110001001001020022101000000111100?0 0

Limnoscelis 2{12}3101101110000100300111010000210000202110011010110010001001?000331?011010????2?00100 00110010000131{12}10100?110?20{12}021110?00{01}1111?1011100100112112020002111012010020?? 01101110000000000000000100000110011121101001?00100000001???00000001?01001000100001000001 ?1000000110000000002021000004411??011?0001??0000010??00201?000000000000000210000110100100 0200001????100000???1??01010100????0101010????100100002011?????0101?0111010121101100110120 1111101110?101000110????0010000010000001111100000011111110000010101110101010001010010000 100000000000000100100000000000101100010000--121000001022010-0001000111- 00120001000010000?0??0000100000000000101(0 1)0-001020110000?00000100200000001000 0000000100000010011000??0100?0?0022101010010111000110

Branchiosauridae ??{04}00111111000010010?000010000100000102000000000001001103101?????????11??????????0??00?? ?0??1{0 1}?01?????0??????????11011?0001?11101001?????0?13{1 2}2???0??2000211001{1 3}010100001?1011001000010000001001011100002110000010?000000001???000100?1?01000000100001 01000001000000?1101111001201100000??000101110100000110010101000000011100000111002???????? ?100000210111??????0000???0??0???010????????????????1?000000?011101000??11?01???1??1?1111111 0110000?0011?????2?001??????0000001???00000???0?000000?11111100000110011131110000101000121 0000000?100100001000020???0000(0 1)1000000010010--110100000010000-0000???(0 1)00(0 1)00001000{0 1}0001?100??????1110010010????????0- ???????020010000?00?0000?0?0000000?????0?1200200?00-000??000????210000101?120211010—0

249 Ecolsonia 2{12}0000111110000100101000020000100000202010101000100001003101110112?0011?0?1???2000000 0?1100100001010100???????200??011100001111010111031101??1?1010?0?????1200001110000?1110000 000010100000000000121000021101000101000000001???0001000010000000010000101000001000000010 100120011011000004?0011?11000001?10000011?000010001000000011100?00101100010000020?1101??? 1?000001001?01010101010101010101???100000002011101000101111111101211011??1?0??01001001111 0?1?0??100{1 2}0??0?00000???0000010000000000011111100000??????10?11000001110011100001001 100201000000001121000000000001000000--10000000001{0 1}010-000000?000010001?0000000?000?0?? 00?11000100000000?0(0 1)0-0?00200?000??00010000000001011000????????????100???????? ?0000?0101?11??0?0?100110?0???

Eryops 2{12}0000101110000100101000010010100000202010001001100001001(01)010101221001101110102000 000001100100001010100010000?20011{0 1}1110000111101011100110113121010002????12- 000011000001110000000000001000100000110000021100000101000000001???00010001?0000000010000 1010000000010000100001000000110001044001101100000110000001010000100010000000110002001001 0001000002000001???100000010011010001010101010101011111001000000111000000011111111012110 01001101201001101?11011200011020110001000(01)0000000100000000000111111000001100111010100 00?1?1000-000000?001??0000000100?0000000000000000000000--100000000000010- 000001?0000000?0??00???0?001?0????01100010000?00????0???????0?0?01?00000000000000001000???? ??????000 0??00-000?000000??0???????0?10?1???0?000

Doleserpeton 2{1 2}40001111100001001000000100001000001121101101--00010110300101?112{0 2}- 011010101030000000211001110113111000010011100{01}1011000001111010010031100131210200?20?? ?10000101100000111010001000010000000001011100002110000010?000000001???00000000?010000001 000010100000100000011001111001201100000440111011111110010101??????0010111010000011101200 10110001000002111101???110000???0??0101(01)1010???01??0111111111000000011??1?00?011?01?111 ??1111111100110000?0011111?12000110????00000010?00000011(01)0100000011111110000011001113? 01000010110010000001001111000021000020000100011000000010010—12000000000000100100111 000110010?0010101?10001?0?00110211210000000110-110020110001?00(1 2)?0000200?0??110000112220102002000???000010100?010122002000110201000??0

Caudata 2240221-11111-1111---2--0112022-0010112101---1--001001-12111-00-122-0111-02011331121011-- 00101011{1 5}111101110-1121211011021-3-1110-00101311001312102000(2 3)0(0 1)2210-- ?1011100001?1001001001111?????1101010100??011?0?1?????????1??1???1???1?????????1???????00000 1??????1???100101100130?1000004?01011111101??????11??????01001?01000110111??200101100?00000 02001121???110000???1??1???1???1???1???1???????1?00000???1?????01111?01?111??1?1111111011000 0??011?10?121001?0????0000001?1100010???011??0000111111000101100111?2(0 1)11-----000012- 0110100-11000002{1 2}0001--00001---0-1-21----1-0--120100-000(0 1)00010010011?(2 3)00(1 2)020{1 2}011011-010110(0 1)-{0 1}00?1110-1(0 1)2-0002-2-1-(0 1)-10{2 3}{0 1}1010(1 2)00(1 3)(0 1)0120000020100100111011222201120020013------1--1111(1 2)20(0 1)(1 2)(0 1)01120201000{0 1}00

Anura 22(0 4)0212-11100--111---2--1-1222100000112101---1--001111113111-11-122-0111-11011031121111-- 1011101-{1 5}111001{0 2}10-1122411011020121111110010?311111323010101200031--- 01110100001?1?1?1???????1????011010??11????0?0??1?????????1??1???1???1?????1???1????????01001? ?????1???100000100130?110000??01010111101??????11??????00101?1100000011101??0100????0000002 10112????110010???1??1???1???1???1???1????????10?0000011?????10111?01?1110?1?11111100??1?00? 011??0?0?2??????????000010111000??????0101?0?0111????010001100111?1111-----0-1----1--{0 1}100- 110-0002100002000101---0-0-20----1-0--120100-000(0 1)000100100111000(1 2)0101011011-0101(0 1)011{0 1}00?011011(0 1)2-0002-{1 2}12-0-113--01001030012-000021-----1101011122201120020013------1011001(1 2)3001000101211001100

Triadobatrachus 2{1 2}30202-?1101-0111---?--0?12?21?0??0?????00101-?0011?1113101?11-??2?0111-?101103112111?-- 1?-?????{1 5}111001010-11222110110?0?11111?10010?311011?23010000?????1?-- 01??????????1?????????????????1101???11????0?1??1?????????1??1???1???1?????1???1????????01001??? ???????1????11?01?0?100000?????????1???101??101??????00001011000000111????0101100?0000002101 10????????10???1??1???1???1???1???1???????????????011?????1???1?01?111??1111111100110100?0111? 0?0?2000100{1 2}0??00000011?000000???0101?0001111111?10001???11131111---1--??????0--

250 0100????????21???020-0101---0-0?20--?-000--12?1??????????????0011?0????1????10?????0 1?01?0001011-??{0 1}1?0????????0????????00103?012-00?021----011010?1???20{0 1}1?002001{0 3}------?-10???0110001100?(1 2)12-1001?0?

Amphibamus ?2000010?1100001001?00000100001?000010201010100?000?011?3001????????0110?010102??00000?1?0 011{01}?11(03)10?00???????100110110000111110?001???11011312?02000200001?00130110000?1?1010 00100001000000?001011000002110000010?000000001???00000????010?000010000101000001000000110 0101100120?100000??01110111011110?010000110000000010100000111012?0101100?100000210110???? ??0000???0??0???010?0???0???0??????1?1000000011100000??11?01?111??1?111111101?0?0010011????? 2?00100????0000001?000000010000000000011111100000110011131110000?011001010000000?1000020 2000002000000001100000001001???1111000000000010010?0??000110010?0010000?1100100?011102110 10000?????0-0??????10001??00?0000200??00110000?00?2?1?200200?00- 00000010000?0??1?0100?1102?1?00??0

251