ANDFOSSILSSTRATANumber 61 • September 2016

An international monograph

series of palaeontology and

stratigraphy

Ulf J. Borgen and Hans A. Nakrem

Morphology, phylogeny and taxonomy of osteolepiform fish

Please consult the editor at an early stage regarding the suitability of thetopic, technical requirements, and financing. Final acceptance for publica-tion, however, will not be made until the manuscript has been refereed.

Detailed instructions for the preparations and organisation of manu-scripts are available on the Lethaia Foundation and John Wiley & Sons Inc.web pages. These should be followed prior to submission of manuscripts.

These standards are introduced with Fossils and Strata 59, 2013. They are different from those published earlier in the series and incorporate anumber of changes. They thus replace previous versions of the guide.

Instructions to authorsFossils and Strata publishes original research papers such as monographs,collections of thematic papers and/or a collection of papers that originatedfrom international conferences. Submitted manuscripts are considered forpublication on the understanding that they have not been submitted oraccepted for publication elsewhere.

Manuscripts. – Manuscripts should be written clearly and concisely and conform to the general style of the monograph series.

All contributions should be written in British (UK) English. Authors whoare unfamiliar with English are encouraged to seek help of a competent linguist prior to submission. A list of independent suppliers of editing servicescan be found at www.blackwellpublishing.com/bauthor/english_language.asp

Fossils and Strata follows the S1 (Systeme International d’Unities) unitswherever possible.

Fossils and Strata accepts monographs with systematic taxonomy and theaccuracy of synonymies together with high-quality illustrations are a highpriority for the journal. Manuscripts that are inadequately prepared and donot follow the rules of nomenclature will be returned to the author.

Copyright. – The Lethaia Foundation owns the copyright to Fossils and Strataand licenses it to John Wiley & Sons Inc. The authors must declare in writingthe transfer of all rights to copyright to the Foundation. The author(s) will beasked to complete a copyright transfer agreement at the time of acceptanceof the manuscript.

Submission of manuscriptsFossils and Strata accepts printed manuscripts. Thematic or multiple contri-butions should be submitted electronically including the manuscript, figuresand tables.

Paper submission. – Two paper copies, including all illustrations and tables,must be submitted, together with all computer-generated files prepared onreadable media. For safety, the authors should retain a further copy. Double-space all text.

Send the manuscript to: Fossils and Strata, Dr. Svend Stouge, GeologicalMuseum, Øster Voldgade 5–7, DK-1350 Copenhagen K, Denmark. E-mail:svends@snm.ku.dk.

Manuscript processingThe initial submitted manuscript is sent to referees. On the basis of the referees’ reports, the editor will decide whether to publish the manuscriptand the modifications that will be required.

The author receives galley proofs and page proofs. The corrected proofs should be returned to the Production Editor. The editor and publisher take no responsibility for errors in the proofs not discovered by theauthors.

Notes for authors

Morphology, phylogeny and taxonomy of osteolepiform fi sh

by

Ulf J. Borgen and Hans A. Nakrem

AcknowledgementsFinancial support for the publication of this issue of Fossils and Strata

was provided by the Lethaia Foundation and the Stensiö Foundation

Introduction ................................................................................................ 1General taxonomy ....................................................................................... 2

‘Crossoptgerygii’ and ‘Rhipidistia’......................................................... 2Osteolepiform subdivisions .................................................................. 3Terminology ............................................................................................ 4

Cranial roof ......................................................................................... 4Endocranium .................................................................................... 13Dermal bones in mouth roof ......................................................... 13Lower jaw .......................................................................................... 14Operculo-gular bones...................................................................... 15

Upper Palaeozoic fossil locality in the Oslo Region ............................ 16Stratigraphy ........................................................................................... 16

Descriptions ............................................................................................... 18Material, methods and repository ...................................................... 18

Genus Askerichthys n. gen. ............................................................ 20Askerichthys heintzi n. sp. .......................................................... 20

Genus Megalichthys Agassiz, 1835 ................................................ 62Megalichthys cf. M. hibberti Agassiz, 1835............................... 62Megalichthys pygmaeus Tranquair, 1879 ................................. 97Megalichthys syndentolaminaris n. sp. .................................... 99

Genus Latvius Jarvik, 1948 .......................................................... 100Latvius grewingki (Gross, 1933) .............................................. 101Latvius deckerti Jensen, 1966 .................................................. 114Latvius sp. Jensen, 1966 ............................................................ 121Latvius cf. L. niger Jensen, 1973 .............................................. 128

Genus Osteolepis Agassiz, 1835 .................................................... 130Osteolepis macrolepidotus Agassiz, 1835 ................................ 131

Genus Gyroptychius McCoy, 1848 .............................................. 131Gyroptychius milleri (Jarvik, 1948) ........................................ 131Gyroptychius groenlandicus Jarvik, 1950a ............................ 141

Morphologic comparisons and discussions ........................................ 160Fronto-ethmoidal shield ................................................................... 160

General shape ............................................................................ 160Interpremaxilla .............................................................................. 162Composition of naso-rostro-premaxillary complex ................. 162Shape of upper mouth margin ..................................................... 165Anterior palatal (premaxillary and rostral) teeth and

palatal processes ......................................................................... 166Defi nitions and occurrences of morphotypes of anterior palatal teeth ............................................................................... 166

Tooth replacement and evolutionary relationship between the anterior palatal dental morphotypes .............. 167

Number of premaxillary teeth ................................................. 169Systematic signifi cance of anterior palatal teeth ................... 170Anterior palatal processes ........................................................ 171

Postrostrals ...................................................................................... 172Variation and trends in postrostral pattern ........................... 172Taxonomic signifi cance of postrostral pattern ..................... 176

Nasals ............................................................................................... 178Number of nasal bones ............................................................. 178Positions of the bones in the nasal series .............................. 178

Circumnarial dermal bones and supraorbitals .......................... 180Variation in eusthenopterids ................................................... 180Reconstructing the circumnarial pattern in Osteolepis macrolepidotus ....................................................... 186

Reconstructing the pattern of supraorbitals in Osteolepis macrolepidotus .......................................................................... 188

Circumnarial and supraorbital pattern in other Devonian osteolepiforms ........................................................ 189

Homologizations of circumnarial bones in Megalichthys hibberti, Ectosteorhachis, Askerichthys and Cladarosymblema ..................................................................... 191

Supraorbito-tectal series of panderichthyids ........................ 196Frontals and dermosphenotics ..................................................... 197

Shape and composition of frontals ......................................... 197Shape, positions and extension of dermosphenotic bones ....199

Fenestrae .......................................................................................... 199Position of nasal fenestra ......................................................... 199Shape of nasal fenestra ............................................................. 200Processus dermintermedius ..................................................... 201Tubercles in nasal fenestra ....................................................... 202Presence or absence of pineal openings ................................. 202

Position of pineal foramen ....................................................... 203Shape of pineal openings .......................................................... 204Size of pineal fenestra and presence of pineal plates ........... 204

Sensory canals ................................................................................ 204Pitlines ............................................................................................. 205

Parietal shield ...................................................................................... 206Relative lengths of fronto-ethmoidal and parietal shields ...... 206Shape of parietal shield ................................................................. 207

Proportions ................................................................................. 207Protruding posterior margin ................................................... 209Position of spiracular incision ................................................. 212Relative length of the spiracular incision .............................. 213

Parietals............................................................................................ 214Shape of parietals ....................................................................... 214Parietal structures ...................................................................... 214

Intertemporals ................................................................................ 215Shape of interpemporal ............................................................ 215Anterior intertemporal margin and process in post-Devonian taxa ........................................................................... 216

Anterior intertemporal processes in Devonian forms ......... 222On the function of the intracranial joint ............................... 223Lateral shelf of intertemporal .................................................. 223Ventral surface of intertemporal ............................................. 224

Supratemporals ............................................................................... 228Shape............................................................................................ 228Ventral surface and lateral margin .......................................... 228

Spiracular structures ...................................................................... 231Extension of the spiracular opening ....................................... 231Types of spiracular openings ................................................... 232Possible muscle insertions around the spiraculum .............. 233

Extratemporal (= postspiraculars) .............................................. 236Sensory canals ................................................................................ 236

Course of sensory canal ........................................................... 236Position of canal openings ....................................................... 237

Pitlines ............................................................................................. 238Extrascapulars ................................................................................. 238Dermal palatal bones ..................................................................... 239Vomers ............................................................................................. 239

Review of taxa with known vomers ....................................... 239Transverse tusk positions on anterior vomers ...................... 243Confi guration of anterior vomers ........................................... 244Extensions and proportions of the posterior vomer ............ 245Dental lamina and teeth ........................................................... 245Intervomerine Canals ............................................................... 246Evolution of the vomers ........................................................... 247

Parasphenoid .................................................................................. 248Division of the parasphenoid .................................................. 248Paraspenoid dental plate .......................................................... 248Buccohypophysial foramen ...................................................... 250Pars anterior of parasphenoid ................................................. 250Parasphenoid shelves and processi ascendens ...................... 250Foramina on the shelf ............................................................... 253

Acrochordal depressions, additional dermal bones and basicranial muscles ............................................................ 254Occurrence of acrochordal depressions ................................. 254Function of acrochordal depressions ..................................... 255

Endocranium....................................................................................... 258Ethmosphenoid .............................................................................. 258

Proportions of ethmosphenoid ............................................... 258Subethmoidal fossae .................................................................. 258Nasobasal canal and palatonasal canal................................... 261Internal structures of nasal capsule ........................................ 261Postnasal wall ............................................................................. 263Orbitosphenoid wall of eusthenopterids ............................... 268Comparing orbitosphenoid region of Ectosteorhachis foordi with other cyclolepidoids than eusthenopterids .........271

Orbitosphenoid in Osteolepidoidei and other forms .......... 273Systematic signifi cance of orbitosphenoid structure ........... 277Posterior end of orbitosphenoid ............................................. 277

Otico-occipital ................................................................................ 278Cheek plate .......................................................................................... 280

Maxilla ............................................................................................. 280

Contents

Shape............................................................................................ 280Ventral margin and tooth distribution on maxilla ............... 282

Postorbital and squamosal ............................................................ 283Lachrymal ....................................................................................... 285Jugal .................................................................................................. 285

Shape............................................................................................ 285Contact with the orbit............................................................... 285

Quadratojugal ................................................................................. 287Lower jaw (= mandible) .................................................................... 287

Shape ................................................................................................ 287Proportions ................................................................................. 287Anterior notch ........................................................................... 288Antero-dorsal bulb .................................................................... 288Dorsal margin ............................................................................ 288Positions of dental margin and glenoid notch ...................... 290Retroarticular process ............................................................... 292

External structures ......................................................................... 292Dentary/Infradentary suture ................................................... 292Inter-infradentary sutures ........................................................ 294Dorsal cosmine notch, cosmine corners and postero-dorsal bone margin ................................................... 295

Horizontal pitline ...................................................................... 298Vertical pitline of infradentary 2 ............................................. 300

Sensory canal .................................................................................. 300Canal course ............................................................................... 301Confi guration of canals ............................................................ 301

Dentary ............................................................................................ 302Dentary teeth ............................................................................. 302Dentary shelf and dental fossa ................................................ 307

Coronoids ........................................................................................ 307Number of coronoid bones and tusks .................................... 307Replacement of coronoid tusks ............................................... 312Relative size of coronoid tusks ................................................ 313Relative positions of coronoid tusks ....................................... 315Coronoid dental ridge ............................................................... 317Intercoronoid fossae .................................................................. 321Anterior mandibular fossa ....................................................... 322

Prearticular...................................................................................... 322Presence of a pars anterior of the prearticular ..................... 322Anteriad extension and confi guration of pars anterior of the prearticular .................................................................... 323

Course of the pars anterior of the prearticular ..................... 329Anterior extension of prearticular dental plate .................... 330Shape of dental plate ................................................................. 331Prearticular dental plate confi guration and teeth................. 332Postero-ventral depression ....................................................... 332Posterior part of prearticular ................................................... 333

Parasymphyseal dental plate and adsymphysial fossa .............. 333Occurrence of Parasymphyseal dental plate ......................... 333Shape of Parasymphyseal dental plate .................................... 334Relations of Parasymphyseal dental plate to other bones ... 334Adsymphysial fossa ................................................................... 338

Infradentaries .................................................................................. 341Infradentary surfaces ................................................................ 341Posterior extension of infradentaries ..................................... 341

Meckelian bone .............................................................................. 342Pars mentomandibularis .......................................................... 342Ventral exposures ...................................................................... 342Pars articularis............................................................................ 342

Foramina ......................................................................................... 343Sensory canal foramina ............................................................ 343Foramina for nerves and vessels ............................................. 344

Sensory canal pore patterns ......................................................... 350Distribution patterns of sensory pores .................................. 350Pore distribution as an age indicator ...................................... 350Evolution of pore dispersion ................................................... 353Taxonomic signifi cance of pore number and distribution ....355Size and types of sensory pits .................................................. 357

Operculo-gular complex ................................................................... 357Median gular ................................................................................... 357Principal gulars ............................................................................... 360Submandibulars .............................................................................. 362Opercular and subopercular ........................................................ 365

Shoulder girdle .................................................................................... 367

Scapulocoracoid ............................................................................. 367Dermal shoulder girdle ................................................................. 368

Fins ...................................................................................................... 369 Caudal fi n ........................................................................................ 369Position of fi ns ................................................................................ 370

Scales, body proportions and vertebrae .......................................... 370Scales ................................................................................................ 370

Scale distribution ....................................................................... 370Scale morphology ...................................................................... 371

Body proportions ........................................................................... 371Vertebrae ......................................................................................... 371

Phylogeny and taxonomy of osteolepiforms ...................................... 372Taxonomic concepts and principles ................................................ 372

Characters ....................................................................................... 372Quantitative characters ............................................................. 372Primitive and derived characters, convergence, parallelism and homoplasy ..................................................... 372

Suffi cient, necessary, indicative,single and combined characters .................................................................................. 375

Diagnoses ........................................................................................ 376A phylogenetic model ............................................................................ 377A systematic model ................................................................................. 381

Order Osteolepiformes Berg, 1937 .................................................. 381Suborder Osteolepidoidei Moy-Th omas & Miles, 1971 ............... 382Family Th ursiidae n. fam. ................................................................. 383

Genus Thursius Traquair, 1888 ................................................ 384 Family Osteolepididae Cope, 1889 .................................................. 388Subfamily Glyptopominae Goodrich, 1909 .................................... 389

Genus Glyptopomus Agassiz, 1844 .......................................... 390Genus Megistolepis Obruchev, 1955 ....................................... 392Genus Shirolepis Vorobyeva, 1977a......................................... 393Genus Greiserolepis Vorobyeva, 1977a ................................... 394

Subfamily Osteolepidinae n. subfam. .............................................. 396Genus Osteolepis Agassiz, 1835 ............................................... 396Genus Latvius Jarvik, 1948 ....................................................... 397Genus Gogonasus Long, 1985b ................................................ 400Genus Geptolepis Vorobyeva & Lebedev, 1986a.................... 403Genus Peregrina Vorobyeva & Lebedev, 1986b .................... 404

Family Megalichthyidae Hay, 1902 .................................................. 406Subfamily Ectosteorhachinae n. subfam. ........................................ 412Genus Ectosteorhachis Cope, 1880 ................................................... 412Subfamily Megalichthyinae n. subfam. or Cope, 1940 .................. 415

Genus Cladarosymblema Fox, Cambell, Barwick & Long, 1995 ................................................................................. 415

Genus Megalichthys Agassiz, 1835 ......................................... 415Subfamily Askerichthyinae n. subfam. ........................................... 419

Genus Askerichthys n. gen. ...................................................... 419Suborder Cyclolepidoidei n. suborder ............................................. 420Superfamily Eopodoidea n. superfam ............................................. 420Family Gyroptychiidae Berg, 1958 Berg 1940 ................................ 422

Genus Gyroptychius McCoy, 1848 .......................................... 422Family Panderichthyidae Vorobyeva & Lyarskaya, 1968 .......... 429Subfamily Panderichthyinae n. subfam ...................................... 432

Genus Panderichthys Gross, 1941 ........................................... 432Genus Livoniana Ahlberg, Lukševičs & Lebedev, 2000 ....... 435

Subfamily Elpistosteginae n. subfam ........................................... 435Genus Elpistostege Westoll, 1938 ............................................. 436 Genus Tiktaalik Daeschler, Shubin & Jenkins, 2006 ............ 436

Family Chrysolepididae n. fam .................................................... 438Genus Chrysolepis Lebelev, 1983 ............................................. 438

Family Eusthenopteridae Berg, 1955........................................... 440Genus Tristichopterus Egerton, 1861 ...................................... 445Genus Eusthenopteron Whiteaves, 1881 ................................ 447Genus Eusthenodon Jarvik, 1952 ............................................. 448Genus Platycephalichthys Vorobyeva, 1959 ........................... 449Genus Jarvikina Vorobyeva, 1977a ......................................... 452Genus Marsdenichthys Long, 1985a ....................................... 453Genus Spodichthys Jarvik, 1985 ............................................... 455Genus Notorhizodon Young, Long, & Ritchie, 1992 ............. 456Genus Mandageria Johanson & Ahlberg, 1997 .................... 457Genus Cabonnichthys Ahlberg & Johanson, 1997................ 458 Genus Heddleichthys Snitting, 2008b ..................................... 460Genus Langlieria Clément, Snitting & Ahlberg, 2008 .......... 462

Superfamily Parapodoidea n. superfam. ........................................ 463

Family Medoevididae n. fam. ....................................................... 463Genus Medoevia Lebedev, 1995 .............................................. 463

Family Canowindridae Young, Long & Ritchie, 1992 .............. 464Genus Canowindra Th omson, 1973 ...................................... 465Genus Beelarongia Long, 1987 ................................................ 466Genus Koharalepis Young, Long & Ritchie, 1992 ................. 467

Superfamily Rhizodontoidea n. superfam. .................................... 468Osteolepiformes subord. indet. ......................................................... 468Family Lamprotolepididae Vorobyeva, 1977a ................................ 468

Genus Lamprotolepis Vorobyeva, 1977a ............................... 468 Osteolepiformes subord. & fam. indet. ....................................... 470

Genus Litoptychius Denison, 1951.......................................... 470Genus Lohsania Th omson & Vaughn, 1968 .......................... 471

Genus Sterropterygion Th omson, 1972 .................................. 472 Genus Megapomus Vorobyeva, 1977a .................................... 472Genus Thysanolepis Vorobyeva, 1977a ................................... 474Genus Viluichthys Vorobyeva, 1977a ...................................... 475 Genus Mahalalepis Young, Long & Ritchie, 1992 ................. 476Genus Platyethmoidea Young, Long & Ritchie, 1992 ........... 478Genus Vorobjevaia Young, Long & Ritchie, 1992 ................. 479Genus Sengoerichthys Janvier, Clément & Cloutier, 2007 ... 480Genus Criptolepis Worobjeva, 1975c ...................................... 481

Acknowledgements ................................................................................. 481References................................................................................................. 482Tables ........................................................................................................ 491Appendix. Abbreviations used in illustrations and tables................. 506

Morphology, phylogeny and taxonomy of osteolepiform fish

ULF J. BORGEN† AND HANS A. NAKREM*

Borgen, U.J. & Nakrem, H.A. 2016: Morphology, phylogeny and taxonomy ofosteolepiform fish. Fossils and Strata, No. 61. pp. 1–514. ISSN 024-1164.

Material of six osteolepiform genera is described, including Askerichthys n. gen., a newLate Carboniferous genus from Norway,Megalichthys Agassiz, 1835 from the Carbonif-erous of Great Britain, Latvius Jarvik, 1948 from the Late Devonian of Germany andLatvia, and Osteolepis Agassiz, 1835 and Gyroptychius McCoy, 1848 from the MiddleDevonian of Great Britain and northeast Greenland. New information on Eus-thenopteron foordi Whiteaves, 1881 from the Late Devonian of Canada is presented inthe morphologic discussions. On the basis of the descriptions and previous studies mor-phologic variation in osteolepiforms is recorded, and it is discussed whether these varia-tions are taxonomic or intraspecific. Morphologic clines are described and it isdiscussed whether they are trends. When possible, functional implications of the mor-phologic variations are suggested. In the phylogenetic and taxonomic section differenttypes of characters as well as use of these characters when reconstructing phylogeny andtaxonomy, is discussed. Consideration has been given as to whether diagnoses can beconstructed in a more informative way by stating whether characters are necessary, suf-ficient or indicative. A tentative phylogenetic model based on the morphologic infor-mation in this and other works is presented. This phylogeny leads to a taxonomic modelthat is expressed as a review of osteolepiform taxa with diagnoses formulated as statedabove. The order Osteolepiformes Woodward, 1932 is divided into two suborders,Osteolepidoidei Moy-Thomas & Miles, 1971 and Cyclolepidoidei n. suborder. Oste-olepidoidei includes the families Osteolepididae Cope, 1889, Thursiidae n. fam. andMegalichthyidae Hay, 1902. Osteolepididae is divided into Glyptopominae Goodrich,1909 and Osteolepidinae Cope, 1889. Megalichthyidae is divided into the subfamiliesEctosteorhachinae n. subfam., Megalichthyinae n. subfam. and Askerichthyinae n. sub-fam. Cyclolepidoidei includes Eopodoidea n. superfamily, Parapodoidea n. superfam-ily, and Rhizodontoidea. Eopodoidea includes the families Gyroptychiidae n. fam.,Panderichthyidae Vorobyeva, 1968, Chrysolepididae n. fam. and EusthenopteridaeBerg, 1955. Parapodoidea includes the families Canowindridae Young, Long & Ritchie,1992 and Medoevididae n. fam. Panderichthyidae is divided into Panderichthyinae n.subfam. and Elpistosteginae n. subfam. New species erected in this paper are the mega-lichthyids Askerichthys heintzi andMegalichthys syndentolaminaris.

Ulf J. Borgen Bergshamravn 330, SE-76010 Bergshamra Sweden; Hans A. Nakrem(h.a.nakrem@nhm.uio.no), Natural History Museum (Geology) University of Oslo Pb.1172 Blindern NO-0318 Oslo Norway; manuscript received on 14 April 2014; manuscriptaccepted on 28 November 2015.

†Deceased.*Corresponding author.

Introduction

The group of fishes called Osteolepiformes is of greatinterest as they are generally accepted ancestors ofprobably all tetrapods. Since also considered rela-tively primitive (Jarvik 1968a, p. 506) it is a centralgroup in the study of vertebrate evolution.

This work includes the following: (1) an introduc-tory part discussing the general taxonomy of oste-olepiforms, as used by other authors and as used inthis work; terminological problems; a review of thestratigraphic background as well as surroundingfauna and flora of a new probably Late Carboniferousgenus from Norway, and a review of material andmethods; (2) a descriptive part treating macrostruc-tures of mainly four groups: the new probably Late

Carboniferous Norwegian taxon, Carboniferousmaterial from Great Britain referred to MegalichthysAgassiz, 1835, Late Devonian material from Balticumand Bergisch Gladbach in Germany referred to differ-ent species of Latvius Jarvik, 1948, and Middle Devo-nian material from Great Britain and northeastGreenland referred, respectively, to Osteolepismacrolepidotus Agassiz, 1835, Gyroptychius milleriJarvik, 1948 and Gyroptychius groenlandicus Jarvik,1950a; (3) a discussion of variation and possibletrends in osteolepiform morphology as well as somefunctional interpretations of the morphology. Thispart also includes new descriptions of material of Eus-thenopteron foordiWhiteaves, 1881 and Panderichthysrhombolepis (Gross, 1930); and (4) a phylogeneticand taxonomic part that includes a discussion of

DOI 10.111/let.12188 © 2016 Lethaia Foundation. Published by John Wiley & Sons Ltd

concepts and methods in the study of phylogeny andtaxonomy, a suggested phylogenetic model for oste-olepiforms, and a resulting likewise tentative taxo-nomic model of osteolepiforms with some suggestedamended diagnoses.

General taxonomy

‘Crossopterygii’ and ‘Rhipidistia’

The taxon Order Osteolepiformes Berg, 1937 has in aclassical system been considered as belonging to theSuperorder Rhipidistia Cope, 1887 within the ClassCrossopterygii Cope, 1871 (Berg 1958; Romer 1966;Romer 1966; Vorob’eva & Obruchev 1967; Andrews& Westoll 1970b; Moy-Thomas & Miles 1971). Syno-nyms for ‘Osteolepiformes’ have been ‘Osteolepi-doidea’ (Romer 1966, p. 361), ‘Osteolepidiformes’(Romer 1966, p. 361) and ‘Osteolepidida’ (Andrews& Westoll 1970b, p. 479; Moy-Thomas & Miles 1971,p. 110; Andrews 1973, p. 174). Rhipidistia have beencharacterized by Moy-Thomas &Miles (1971, p. 113)by cranial dermal bone pattern, in having branchedlepidotrichia, in having many more lepidotrichiathan radials in the caudal fin, and in having internalnostrils (=choanae). As pointed out by Andrews(1973, p. 162) the choanae have been considered adistinctive character for Rhipidistia. Crossopterygiihas by most contemporary workers been included inthe taxon Sarcopterygii Romer, 1955 that includesalso lungfishes. In the system suggested by Andrews(1973) Rhipidistia includes three orders; Osteolepi-formes Berg, 1937, Porolepiformes Jarvik, 1942 andRhizodontiformes Andrew & Westoll, 1970b. Theformer two were considered as having choanae,whereas this was unclear in Rhizodontiformes and itsinclusion in Rhipidistia was tentative. Non-rhipidis-tian crossopterygian groups include Coelacanthi-formes (=Actinistia) and Onychodontiformes(=Struniiformes, Jessen 1966, p. 334).

The validity of Crossopterygii (Stensi€o 1963, p. 82;Jarvik 1968a, p. 515, 1968b, p. 226; Bjerring 1971, p.189) and Rhipidistia (Jarvik 1942, pp. 142, 284;Andrews 1973, p. 173) has been doubted. More mod-ern works that discuss or use these terms are Ahlberg(1991a), Cloutier & Ahlberg (1996, pp. 465, 468) andJanvier 1996 (pp. 198, 247). Ahlberg (1991a, p. 280)introduced a system where Rhipidistia has a newmeaning and where it includes the superdivisionsTetrapodomorpha Ahlberg, 1991 and DipnomorphaAhlberg, 1991. Tetrapodomorpha are forms withchoanae and includes Osteolepiformes, Rhizodontida(=Rhizodontiformes = Rhizodontoidea in the heresuggested system), Panderichthyida Vorobyeva, 1981(=Elpistostegidae = Panderichthyidae in the here

suggested system) and Tetrapoda. Dipnomorpha areforms without choanae and includes (Ahlberg 1991, p.280) porolepiforms, dipnoans and primitive generalike Powichthys Jessen, 1975 and Youngolepis Chang &Yu, 1981. Thus, Ahlberg considered that porolepi-forms were without choanae. There has been a longdispute (cf. Janvier 1996, p. 204) as to whetherporolepiforms show choanal passages or not. Bjerring(1991) apparently had solved this when he describedthe presence of a passage from the nasal sac to themouth roof (called fenestra exotremiscalis by Bjerring)in a specimen of Glyptolepis groenlandica Jarvik, 1972prepared by serial sectioning. However, Cl�ement(2001) claimed that Bjerring’s result was due to distor-tion of the Glyptolepis Agassiz, 1844 specimen anddescribed a specimen of Heimenia Ørvig, 1969 thatshowed no choanae. The authors do not in this worktake a stand in this dispute, but will mention the possi-bility that there – in this respect – may be variationwithin the porolepiform group. Thus, Rhipidistia inthe newmeaning proposed by Ahlberg (1991) includesalso forms without choanae. It is noteworthy howeverthat if Ahlberg’s (1991a, p. 280) system is modified inthe way that Panderichthyida and Rhizodontida areincluded in Osteolepiformes (as suggested in themodel below), Tetrapodomorpha includes Osteolepi-formes and Tetrapoda (see phylogenetic and taxo-nomic part). In this way ‘Osteolepiformes’ maycomprise only fishes the way that it was meant by Jar-vik (1942), and be used as a paraphyletic group.

The inclusion by Ahlberg (1991, p. 280) ofPorolepiformes together with Dipnoi in the superdi-vision called Dipnomorpha, and thus separated fromOsteolepiforms, is questionable. This is because ofapparent synapomorphies between osteolepiformsand porolepiforms in the presence of a series of sub-mandibulars between the gular plates and themandibular, and the presence of more or less deeppaired subethmoidal fossae. This is further discussedin the morphologic discussions.

Osteolepiform subdivisions

Osteolepiformes was traditionally (Jarvik 1942, p.241; Romer 1966, p. 361; 1980a, p. 202) divided intotwo families, Osteolepididae (Osteolepidae by Jarvikand others) and Eusthenopteridae (=Rhizodonti-dae = Tristichopteridae). The main character statethat has been used to distinguish these families isthat osteolepidids have scales with a rhombic exter-nally exposed surface, an oblique ridge on the innersurface, and with a groove along the border betweenthe exposed part of the scale and the part covered bythe neighbouring scales (Jarvik 1980a, fig. 138A3,A4). Eusthenopterids on the other hand have round

2 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA

scales with an inner central boss and without thegroove bordering the exposed part of the scale (Jar-vik 1980b, fig. 138B3, B4). Carroll (1988, p. 611)included a larger number of families in Osteolepi-formes (=Osteolepidoidea by Carroll), but stillincluded genera like Megalichthys Agassiz, 1835,Ectosteorhachis Cope, 1880, Glyptopomus Agassiz,1844 and Gyroptychius McCoy, 1848 in Osteolepidi-dae (Osteolepidae by Carroll 1988).

Moy-Thomas & Miles (1971, p. 110) divided theorder Osteolepiformes (=Osteolepidida by Moy-Thomas & Miles 1971) into two suborders Oste-olepidoidei, which included genera like OsteolepisAgassiz, 1835, Megalichthys Agassiz, 1835 and Gyrop-tychius McCoy, 1848 and Eusthenopteroidei, whichincluded the genera Eusthenodon Jarvik, 1952, Eus-thenopteron Whiteaves, 1881 and PlatycephalichthysVorobyeva, 1962. The suborder Osteolepidoideithen becomes roughly the same as in earlier workswas called the family Osteolepididae (=Osteolepi-dae). Vorobyeva (1977a) divided Osteolepididae inseveral subfamilies, like Osteolepidinae, Gyropty-chiinae, Glyptopominae, Megistolepidinae,Thysanolepidinae and Viluichthyinae. Young et al.(1992, pp. 9, 20) also included the new familyCanowindridae and used the family name‘Megalichthyidae’. ‘Megalichthyidae’ was also usedby Fox et al. (1995, p. 107). ‘Panderichthyidae’ wasused by Vorobyeva (1977a, p. 200), Schultze & Arse-nault (1985, p. 297) and Carroll (1988, p. 611) for afamily within Osteolepiformes. Panderichthyidaewas considered including Panderichthys Gross, 1941and Elpistostege Westoll, 1938. The latter genus wasby Romer (1947, p. 311) included in a labyrintho-dont and temnospondyl family called Elpistostegi-dae. Until it is known whether Elpistostege has finsor feet we do not know for certain whether it is afish or tetrapod, but the current view seems to bethat it is a fish (Vorobyeva & Lyarskaya 1968, p. 74;Schultze & Arsenault 1985, p. 297; Daeschler et al.2006, p. 759). Vorobyeva & Schultze (1991) elevatedPanderichthyidae to the order category as Pan-derichthyida or Elpistostegalia (Schultze 1996, p.316). In this work is used the family name Pan-derichthyidae instead of Elpistostegidae, because itappears to be in common use and because Pan-derichthys is a better known genus than Elpistostege.Thus, ‘Panderichthyidae’ is more informative aboutwhat is typical for the family. The order category forthis taxon is not used in this work because Pan-derichthyidae clearly belongs in the clade calledEopodoidea (cf. taxonomical discussion), which is asuperfamily within Osteolepiformes.

Coates & Friedman (2010, p. 402) suggested thenew name ‘Megalichthyiformes’ for ‘tetrapodomorph

sarcopterygians more closely related to Megalichthysthan to Eusthenopteron’. This is an indefinite defini-tion, indefinite because it does not include any diag-nostic characters. It also apparently suggests that‘Megalichthyiformes’ is synonymous with the earlierused ‘Osteolepidoidei’ (Moy-Thomas & Miles 1971).This makes ‘Megalichthyiformes’ redundant. Thesuggestion by Coates & Friedman (2010, p. 402) isalso unfortunate, because it suggests a taxon in thesame category as Osteolepiformes, which Mega-lichthys clearly is a part of. Besides, Megalichthys isamong the most specialized and atypical generawithin this taxon (cf. phylogenetic and taxonomicpart), and to use this genus name as the basis for thename of the larger group is therefore irrational.

Thomson (1969, table 1) presented a list of thegenera included in Osteolepididae. This was theosteolepiform family with the largest number of gen-era. Some of these genera have since been removedfrom the family, for instance Canningius (Jarvik1950a, p. 6), Bogdanovia Obrucheva, 1955 (Obru-cheva 1955; Vorob’eva & Obruchev 1967, p. 459;Cloutier & Forey 1991, p. 68) and Thaumatolepis(Obruchev 1941; Vorob’eva & Obruchev 1967, p.456). Besides, documented descriptions of theseforms apparently are lacking. They are not consid-ered in the discussions below.

Characters other than scale configuration thathave been suggested as typical for Osteolepididae intraditional meaning (Vorob’eva & Obruchev 1967,p. 449; Moy-Thomas & Miles 1971, p. 125; Voro-byeva 1977a, p. 122; Jarvik 1980a, pp. 205, 206) arethe presence of an extratemporal (cf. Jarvik 1980a, p.205), the presence of basal scutes and lobate pairedfins (Moy-Thomas & Miles 1971, p. 125), and asmooth cosmine cover (Vorobyeva 1977a, p. 123).With respect to the majority of the suggested addi-tional characters, their presence or absence isunknown in most osteolepidoids, and they are thusinefficient in a family diagnosis. This applies to mostreferences to endocranial configurations (Vorob’eva& Obruchev 1967, pp. 448–451; Vorobyeva 1977a, p.122; Jarvik 1980a, p. 205).

Eusthenopteridae is partly equivalent to whatsome authors (Romer 1966, p. 361; Jarvik 1985, p.10) called Rhizodontidae and others (Janvier 1996,p. 221; Ahlberg & Johanson 1997; Johanson & Ahl-berg 1997, 2001; Clement et al. 2008; Snitting2008a–d) called Tristichopteridae. ‘Rhizodontidae’was introduced by Traquair (1881) and ‘Tristi-chopteridae’ was introduced by Cope (1889, p. 855).Both are thus older than the name Eusthenopteridaewhich was introduced by Berg (1958). However, thetime of the introduction of a name does not bynecessity have priority over convenience when a

FOSSILS AND STRATA Morphology, phylogeny and taxonomy of osteolepiform fish 3

more convenient name has reached frequent use(International Code of Zoological Nomenclature1985, Chapter 6, Article 23:2). Some forms that pre-viously were included in Rhizodontidae have beenclaimed to differ significantly from both osteolepi-forms and porolepiforms (Andrews 1973, p. 144, fig.2; 1985, fig. 7), and have been included in Rhizodon-tiformes (=Rhizodontida). It is confusing and irra-tional to have the name Rhizodontidae connected tothe taxa that were included in Osteolepiformes andnot those that were included in Rhizodontiformes.Thus, another name should be used for these formsthan Rhizodontidae. Berg (1958) designed a diagno-sis for this family and included only some species ofEusthenopteron Whiteaves, 1881 in it. Because itseems impractical to retain the name Rhizodontidaefor a group not included in Rhizodontiformes ‘Eus-thenopteridae’ is here used for the remaining generafrom the Rhizodontidae. As mentioned is ‘Tristi-chopteridae’ used for this family in several relativelyrecent works. Firstly, Cope (1989) did not give anysort of definition of the group he called Tristi-chopteridae, and this family was the only groupwithin Rhipidistia. Thus, what Cope called Tristi-chopteridae is far from equivalent to the taxon Berg(1958, p. 94) called Eusthenopteridae. Besides, Eus-thenopteridae was in frequent use (e.g. Lebedev1995, p. 336) before the reintroduction of Tristi-chopteridae and is also more convenient becauseEusthenopteron Whiteaves, 1881, in contrast to Tris-tichopterus Egerton, 1861, is extremely well known.The name Eusthenopteridae thus associates directlyto what the family represents, whereas the reintro-duction of ‘Tristichopteridae’ was contrary to therules (International Code of Zoological Nomencla-ture 1985, Chapter 6, Article 23:2) because it isinconvenient. Thus, ‘Eusthenopteridae’ is moreinformative for this taxon than ‘Tristichopteridae’ or‘Rhizodontidae’, it is correct according to the rules,and it is therefore used in this work.

Terminology

A terminology for the morphology of osteolepiformshas emerged particularly through the works of Jarvik(1937, 1942, 1948, and subsequent works), and itseems reasonable to use this terminology as long asit is practical. Some of the parameters that weredefined and used by Jarvik (1948, fig. 12) cannot beused on specimens studied in this paper. This isbecause of differences in configuration and preserva-tion between specimens studied here and by Jarvik.Some new terms and parameters have therefore beendefined (Fig. 1, Appendix 1).

A phylogenetic reduction in number of bonesmay be due to fusion between bones, or the disap-pearance of one bone in combination with anexpanded growth of another bone that takes over thearea of the lost bone. Which of these processesoccurs is frequently difficult to say. Patterson (1977,p. 92) described the different interpretations of thisprocess. Jarvik (1980a, p. 250) described criteria thatstrongly suggested that bone reductions were fre-quently due to fusions. The term fusion will be usedhere even when it is unclear which of these two pro-cesses has occurred because no matter the process, afusion of areas has taken place.

When discussing taxonomic significance of diag-nostic characters basic logical terms are used. It is inthis work distinguished between sufficient charac-ters, necessary characters, characters that are bothsufficient and necessary, and indicative characters(cf. taxonomic part). This makes the significance ofthe characters more exact because they show theimplication of the used characters. A necessary char-acter is a character that is assumed to be present inall individuals of a taxon, but may also be present insome other taxa. A lack of a necessary charactermeans that the specimen under study is not a mem-ber of the taxon determined by the diagnosis. A suf-ficient character is sufficient to determine aspecimen to the taxon in question; it is not presentin any other taxon. In cladistic terms it is probablyequivalent to an autapomorphy. However, it is notby necessity present in all specimens of the taxon.An indicative character may not be sufficient or nec-essary but is seemingly more common in the taxonwith the diagnosis than in other taxa.

Cranial roof

General structureThe cranial roof consists of the fronto-ethmoidalshield (=ethmosphenoid shield), the parietal shield(=otico-occipital shield) and the extrascapular series.The fronto-ethmoidal shield includes premaxillae,rostrals, nasals, postrostrals (usually separated intoanterior and posterior bones), frontals, the supraor-bito-tectal series, and dermosphenotics. The pre-maxillae are paired bones along the upper jawmargin and usually carry the marginal teeth. Therostrals constitutes a series of bones posterior to thepremaxillae and they usually carry the ethmoid sen-sory canal. Yu (1998, figs 1, 2) reported that also ros-trals could be tooth-bearing. The nasals constitute aseries of bones between the frontals and the rostrals,and they carry the supraorbital sensory canal. Thepostrostrals are situated between the contra-lateral

4 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA

A

C

D

E

F

G

B

Figure 1. Sketches defining variables used in the tables. A, Fronto-ethmoidal shield in dorsal view. B, Fronto-ethmoidal shield in ventralview showing also the endocranium and parasphenoid. C, Parietal shield in dorsal view. D, Primary gular. E, Opercular and Subopercu-lar. F, Lower jaw in external view. G, Lower jaw in internal view. Abbreviations used in illustrations and tables are explained in Appen-dix 1.

FOSSILS AND STRATA Morphology, phylogeny and taxonomy of osteolepiform fish 5

nasal series, posterior to the rostrals and anterior tothe frontals.

The parietal shield normally includes paired pari-etals, intertemporals, supratemporals and extratem-porals. The parietals are situated on both sides of themedian line, the intertemporals are paired bones sit-uated lateral to the anterior part of the parietals andthe supratemporals are paired bones lateral to theposterior part of the parietals. The extratemporalsare situated lateral to, or somewhat postero-lateralto, the supratemporals. In osteolepiforms there areusually three extrascapulars situated posterior to thesupratemporals and parietals. These most posteriorbones of the cranial roof carry the posterior sensorycanal commissure. Skull roof parameters are defined(Fig. 1A, B; Appendix 1).

Premaxilla and rostralsJarvik (1942, pp. 346, 347) defined premaxilla androstral series by the respective presence on the pre-maxilla of a tooth row, and on the rostrals of theethmoid cross-commissure of the sensory canal (cf.Holmgren & Stensi€o 1936, p. 355). He also statedthat in Eusthenopteron Whiteaves, 1881 and Holopty-chius Agassiz, 1839 the premaxilla had fused withrostrals to constitute a rostro-premaxilla. In Eus-thenopteron even a nasal was included constituting anaso-rostro-premaxilla. Eusthenopteron and Holopty-chius differed in that the premaxilla of the formerhad fused with a median rostral leaving the more lat-eral rostral free, whereas in the latter it was theopposite. This pattern with separate premaxilla androstrals may be primitive for teleostomes (=Oste-ichthyes = Actinopterygii + Sarcopterygii) becauseit has been described also in both palaeoniscoids(Nielsen 1949, fig. 73; Gardiner 1963) and in coela-canthiforms (Millot & Anthony 1958, p. 38, fig. 8).Gardiner (1963, R.pmx, figs 1–5, 18) and Nielsen(1949) also indicated the tendency of fusions ofthese bones in palaeoniscoids. In later works, Gar-diner (1984), Ahlberg (1991a, p. 259), Vorobyeva &Schultze (1991, fig. 6) and Fox et al. (1995) use thename premaxilla for the bone that carries both sen-sory canal and tooth row, that is the bone that Jarvik(1942, p. 347, footnote) calls rostro-premaxilla.Johanson & Ahlberg (1997a, fig. 21b) reconstructedMandageria Johanson & Ahlberg, 1997 with a dis-tinct lateral rostral ventral to the fenestra exonasalis,and mesial to this lateral rostral they showed thedorsal part of a large bone they named premaxilla. Itseems logical to assume that the part of the bonenamed premaxilla that is situated mesial to the lat-eral rostral, is a more mesial rostral. This has eitherfused with the premaxilla, or had their suturetowards the premaxilla covered by cosmine. Lebedev

(1995) used the terminology used by Jarvik andcalled the bone that constitutes the upper mouthmargin naso-rostro-premaxilla. Jarvik’s terminologyis also provisionally followed in this work. The origi-nal premaxilla is that of tetrapods and we do not yetknow whether this is homologous with the fusedrostro-premaxilla we see in some osteolepiforms, oronly the tooth-bearing marginal bone. However, theapparent fusion of these bones in PanderichthysGross, 1941 (Vorobyeva & Schultze 1991, fig. 6),which is a member of the pretetrapod family Pan-derichthyidae, may indicate that the fused bone ishomologous to the tetrapod premaxilla.

Jarvik (1942, p. 497, fig. 68E, D; 1980a, fig. 117)divided the naso-rostro-premaxilla into three topo-graphic parts, a pars dentalis, a pars facialis and apars palatina (p.d, p.f, p.pl, Fig. 106A).

Several forms show posteriorly directed processeson the palatal lamina (cf. Jarvik 1966, p. 78; 1980a,p. 171, fig. 82C), a median process and a pair of con-tra-lateral processes. The median process is usuallytooth bearing and is therefore probably associatedwith the premaxilla, but as already mentioned ros-trals can also be tooth-bearing (Yu 1988, figs 1, 2).Thus, it is possible that a tusk bearing median pro-cess is not only a premaxillary process, but that itconsists also of a rostral, and even an endocranialpart. Thus, this process is denoted antero-medianpalatal process (am.pl.pr, Fig. 13). It is uncertainwhether the lateral processes are parts of the pre-maxilla, of the rostrals or of both, and they are there-fore denoted ‘antero-lateral palatal processes’(al.pl.pr, Figs 76, 77, 95).

Mesial skull roofTwo different interpretations of the homologiesbetween the cranial roof bones of tetrapods and theosteolepiform fishes are in use, the so-called ortho-dox interpretation and the interpretation suggestedby Westoll (1938, 1943) and Romer (1941). Thesetwo interpretations imply different terminologies ofthese bones in osteolepiform fishes. The orthodoxinterpretation and terminology was generally usedbefore Westoll’s suggestion. The Westoll/Romer ter-minology and terminology (called W/R terminol-ogy) was initially used by British and Americanpalaeoichthyologists but has spread and is todayused by most workers in this field. However, ananalysis of this dispute by Borgen (1983) favouredthe orthodox interpretation, and the orthodox ter-minology is used in this work. This is, partly for rea-sons different from Borgen’s (1983), also theterminology used by Jarvik (1937, 1996). Laterworks supporting the W/R terminology are Schultze& Arsenault (1985, p. 294), Panchen & Smithson

6 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA

(1987, p. 410), Ahlberg (1991a, p. 246) and Daesch-ler et al. (2006). Janvier (1996, p. 262) found the W/R terminology credible, but pointed also outremaining problems with this terminology. Klem-bara (1992, 1993, 1994) and Jarvik (1996, p. 21) usedthe orthodox terminology.

At the transition from osteolepiforms to tetrapodsthe W/R terminology demands the following majorchanges in the bone pattern: (1) the whole osteolepi-form extrascapular series disappears completely; (2)the sensory canal commissure crossing theextrascapular series of the osteolepiforms is trans-ferred anteriad from the extrascapulars to the pairedparietals (postparietals in the W/R terminology) andsupratemporals (tabulars in the W/R terminology)of the tetrapods; (3) with the W/R interpretationand naming there has been a change in parietal posi-tion from a partly interorbital position in the oste-olepiforms to a mainly postorbital position (Borgen1983, fig. 1D) in tetrapods; (4) if the W/R terminol-ogy is correct the supraorbital sensory canal, whichin tetrapods normally penetrates the frontals andavoids the parietals (Bystrow 1935, figs 6–16; Borgen1983, fig. 4; Carroll 1988, fig. 9:14), have suddenlychanged its course because in osteolepiforms, andalso in other sarcopterygians (Jessen 1966, fig. 6; Jar-vik 1980a, fig. 184; Andrews et al. 2006, fig. 4), thesensory canal according to the W/R terminologypenetrates the parietals and avoids the postparietals;(5) ‘Anterior’ postrostrals of osteolepiforms havefused with at least some of the adjacent osteolepi-form nasals constituting the nasals of tetrapods, and‘posterior’ postrostrals have fused with adjacentnasals and have become frontals.

With the orthodox terminology the followingchanges are necessary: (1) the osteolepiformpostrostrals (both ‘anterior’ and ‘posterior’) andnasals fuse constituting the tetrapod nasals. (2) Intetrapods the parietals surround the pineal opening;in most osteolepiforms the frontals surround thisopening. Thus, the brain with the parapineal andpineal organs has stayed in the postorbital region,while the cranial roof bones have moved anteriad.(3) At the transition from osteolepiforms to tetra-pods there has in many tetrapods been an anteriadchange in the position of the frontals relative to theorbit.

The clear conclusion from comparing these listsis that the necessary number of changes at the tran-sition between osteolepiforms and tetrapods by theW/R terminology are distinctly larger than thosedemanded by the orthodox terminology. Thus, thelatter interpretation is more likely to be correct(more parsimonious). As will be shown below, theorthodox terminology is also more consistent with

the observable changes in the proportions of thecranium.

Five important aspects of this dispute(1) The alleged disappearance of the extrascapularseries at the osteolepiform-tetrapod transition thatfollows from the W/R interpretation. (2) The chang-ing positions of mesial cranial roof bones, frontals,parietals and postparietals. (3) With the W/R termi-nology, the unexplained change of the course of thesupraorbital sensory canal from penetrating the pari-etals and avoiding the postparietals in osteolepi-forms, to penetrating the frontals and avoiding theparietals in tetrapods. (4) The fusion at the oste-olepiform-tetrapod transition of the bones that inosteolepiforms are called postrostrals and nasals. (5)The transfer at the osteolepiform-tetrapod transitionof the pineal opening from interfrontal to interpari-etal positions.

These five aspects are in the following discussedsuccessively.

Alleged disappearance in tetrapods of the osteolepiformextrascapular bones. – Even if it is possible that der-mal bones may disappear as separate units, the rela-tive abrupt disappearance of the whole extrascapularseries is a radical step. Because this disappearancemakes no sense (cf. Pearson 1982, p. 37), andbecause no transitional morphotypes have beendescribed, it is too radical to be credible.

The orthodox interpretation claims a homologybetween extrascapulars of osteolepiforms and theseries of postparietals and tabulars in tetrapods. Thisis supported by (1) both series are dermal bones sit-uated posteriorly in the cranial roof, between theoccipital bones and the large paired bones posteri-orly in the cranial roof; (2) the pattern of the cranialroof bones of tetrapods where the sensory canal pat-tern is retained is exactly similar to that in osteolepi-forms, and in both patterns the extrascapular seriesof the osteolepiforms and the series of postparietalsand tabulars of tetrapods carry the posterior com-missure of the sensory canal; (3) both series showexactly the same variation in bone patterns, and (4)the bones of the two series have approximately thesimilar proportions. These points are in the follow-ing explained somewhat more detailed.

Between the occipital bones and the large pairedposteriormost cranial roof bones, which in man andall tetrapods are called parietals, is situated a seriesof dermal bones. In tetrapods these bones are calledpostparietals (=interparietals) and tabulars. In oste-olepiforms the extrascapular series is situated in theexact same position. It has been used by supportersof the W/R terminology as an explanation for the

FOSSILS AND STRATA Morphology, phylogeny and taxonomy of osteolepiform fish 7

claimed disappearance of the extrascapulars thatthey disappeared at the transition from fishes to tet-rapods because tetrapods developed a movable neckjoint. However, numerous tetrapods, including man,with a movable neck joint show dermal bones in thisposition (Sobotta-Becher 1956, fig. 77; Starck 1979,fig. 180; Carroll 1988, figs 9:14, 10:3, 10:15, 17:8,17:9, 17:13, 17:20, 17:22). These dermal bones in tet-rapods are the interparietals (=postparietals) and thetabulars.

In tetrapods where the sensory canal system hasbeen retained, the combined pattern of bones andsensory canals is nearly exactly the same as the pat-tern of osteolepiforms. This is seen in several stego-cephalians (S€ave-S€oderbergh 1935, figs 1, 31; 1937,figs 1, 3–5, 7C; Panchen 1970, fig. 1; Borgen 1983,figs 1, 2C, 4B, C; Ivachnenko 1987, figs 1A, D, 3A,5A; Carroll 1988, figs 9–14C, G, H, I; Klembara1992, fig. 2B). The pair of large bones anterior to theextrascapulars in osteolepiforms and anterior to thepostparietals and tabulars in tetrapods both showsthe X-pattern (Andrews 1973). The pair of largebones mesially in the X-pattern is called by W/R ter-minology parietals in tetrapods and postparietals inosteolepiforms, whereas in the orthodox terminol-ogy they are called parietals in both groups. Poste-rior to the X-pattern the extrascapulars carries theposterior sensory canal commissure in fishes, andthe series of postparietals and tabulars in tetrapodsdoes the same. Sensory canals may change coursebut the known changes are small, one sensory canalmoves from one bone to another (Borgen 1983, fig.7). Also, the Early Permian tetrapod Discosauriscus(Klembara 1992, fig. 2) shows, in addition to thesensory canal on the postparietal-tabular series, alsoapparent pitlines on the supratemporals and frontalsof the cranial roof that are reminiscent of the pitlinesof the supratemporal and frontals of osteolepiforms.Some say that osteolepiforms and other stego-cephalians than the Devonian are too far apart forcomparisons. However, stegocephalians developedfrom osteolepiforms so they are not that far apart.Besides, it is irrational to assume differences in bonepatterns where there are no differences. The patternof the bones and the sensory canal of osteolepiformsclearly were sufficiently stable to persist in many tet-rapods.

The postparietal-tabular series in tetrapods showexactly the same variation as the extrascapular boneseries among fishes, two, three or four bones. Howthis variation comes about follows from the configu-ration of these bones in an early ontogenetic stage ofthe postparietal-tabular series in man (Starck 1975,fig. 542). This series (called interparietals by Starck)has a basic number of four bone precursors that may

fuse in different patterns depending on what inci-sions, incision lateralis and/or incisura cranialis (cf.Starck 1975, fig. 542), between the precursors havebeen retained, and thus which of the four bones havefused. Examples showing the variation in this seriesin man are shown by Augier (1931, figs 141–148).There may be one, two or three bones in this series.Variation in primitive tetrapods is shown by Carrollet al. (2004, figs 6A, 11A). Exactly the same type ofvariation is shown in the extrascapulars of mostosteichthyans. Osteolepiforms, and other sarcoptery-gians (Jarvik 1980a, fig. 184; Jessen 1966, fig. 6;Andrews et al. 2006, fig. 4), suggest a fusion of thetwo mesial contra-lateral of the four bones with theresult of three extrascapulars. The presence of twoextrascapulars on a specimen of Thursius moy-tho-masi Jarvik, 1948 (see Jarvik 1948, fig. 63C) and inseveral early actinopterygians (Nielsen 1949, figs 21,64; Arratia & Cloutier 1996, fig. 6B) indicates afusion of the ipsilateral of the four bones. Moytho-masia nitida Jessen, 1968 showed all four bones inthe extrascapular series (Jessen 1968, fig. 1B). Thus,we have the same basic number of bones and varia-tions in the extrascapular series of most primitiveosteichthyan fishes as in the postparietal-tabular ser-ies of tetrapods. Like the median extrascapularamong fishes are sometimes divided into two, like inMoythomasia Jessen, 1968 (Jessen 1968, fig. 1B), sohas also the postparietal in fishes close to the fish/te-trapod transition like Elpistostege Westoll, 1938(Schultze & Arsenault 1985, fig. 7) and TiktaalikDaeschler, Shubin & Jenkins, 2006 (Daeschler et al.2006, fig. 3).

Schultze & Arsenault (1985, fig. 7) reconstructedin Elpistostege a hypothetic series of threeextrascapulars posterior to the postparietals. Thesebones will probably not be found when more com-plete specimens of Elpistostege are discoveredbecause the postparietals and the tabulars of Elpis-tostege probably are homologous with theextrascapulars, just as in Tiktaalik. In these twoforms the postparietals have extended anteriad dueto the prolongation of the snout, just as in Ichthyos-tega S€ave-S€oderbergh, 1932 where the two mesialbones still are fused into a median extrascapular(S€ave-S€oderbergh, 1932, fig. 15; Jarvik 1996, pl. 8).

The mentioned similarities between the variationin the postparietal/tabular series of tetrapods and theextrascapular series of osteichthyan fish groups aretoo great to be due to coincidence. Exceptions to thebasic pattern of four bones are seen in coelacanthi-forms (=actinistians) that may have an even largernumber of bones in the extrascapular series (Jollie1962, figs 4:33, 4:35; Jarvik 1980a, fig. 223). Thismay be retention of a more primitive morphotype.

8 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA

The bones that are called postparietals and tabu-lars in osteolepiforms according to the W/R termi-nology (=parietals and supratemporals in orthodoxterminology), are very different in proportions fromthe postparietals and tabulars in tetrapods. In tetra-pods the tabulars and postparietals are frequentlyabout equally long (Borgen 1983, figs 1, 2D, 4C;Schultze & Arsenault 1985, fig. 8C; Carroll 1988, fig.9:14) and when the tabulars sometimes are shorterthan the postparietals the difference is not great. Inosteolepiforms the bone that in W/R terminologyare called postparietals are much longer than themore lateral tabulars. Thus, this is quite differentfrom the postparietals and tabulars of tetrapods thatare usually of about the same length. If the pairedbones that in W/R terminology are called postpari-etals in osteolepiforms are homologous to the post-parietals in tetrapods this would mean that thepostparietals would have been shortened consider-ably at the transition from osteolepiforms to tetra-pods. This is illogical considering that a main changein the cranium as a whole, at the transition fromosteolepiforms to tetrapods, is a lengthening of thepreorbital part of the cranium, a lengthening that islargest along the median line. The lengthening of thepreorbital part, which is seen in stegocephalians(Carroll 1988, fig. 9:14) and in the osteolepiformfishes closest to the transition, Tiktaalik and Elpis-tostege (the subfamily Elpistosteginae, cf. taxonomicpart), should logically be followed by an anteriadexpansion of the dermal bones along the median lineand not a posteriad withdrawal of these bones. Thus,the postparietals in tetrapods do not fit as beinghomologous with the long bone that in osteolepi-forms according to the W/R terminology are post-parietals, but fit well being homologous to the bonesthat are called mesial extrascapulars.

Changes in position of the mesial cranial roofbones. – These changes probably are crucial in thisdispute. In the W/R interpretation the frontals aswell as the parietals have been claimed to havemoved posteriad, whereas the orthodox interpreta-tion presumes an anteriad transfer as a following ofthe anteriad prolongation of the snout. Changes inproportions of the endocranium must be followedby changes in the pattern of dermal bones and theanteriad prolongation of the snout must accordinglyhave the effect that the cranial roof bones along themedian line have moved anteriad. The positionalconnection between the length of the snout and thepositions of the bones is proven by the variation inthe cranial roofs of some stegocephalians (Borgen1983, fig. 4; Carroll 1988, fig. 9:14). For instance,Eryops Cope, 1887 and Rhinesuchus Broom, 1908

both have a long snout that has been followed by aprolonged frontal (Carroll 1986, fig. 9:14d, f). Alsothe parietal bones, the median extrascapulars(=postparietals) and surrounding bones haveextended anteriad as a consequence of the prolongedsnout. Shorter snout means that frontal bones donot extend far anteriorly to the orbits. This is clearlyshown in the bone pattern of Metoposaurus Lydek-ker, 1890 (Carroll 1986, fig. 9:14g) where the snoutis not much longer than in osteolepiform fishes andthe parietals and the frontals have about the sameantero-posterior positions in the cranium as in oste-olepiforms. Whether this similarity between Meto-posaurus and osteolepiforms is due to a reversion ofprimitive proportions or that this belongs to a lin-eage where the primitive proportions have remained,is irrelevant. It shows the correlation between thelength of the snout and the proportions of the cra-nial roof bones. A limited anteriad prolongation ofthe snout started already in the osteolepiform sub-family we have called Panderichthyinae n. subfam.,but is even more distinct in the subfamily that herecalled Elpistosteginae n. subfam. that fishes includesElpistostege and Tiktaalik (cf. Elpistosteginae in taxo-nomical part) where, as an answer to the prolongedsnout, the parietal has expanded to a level anteriorlyor nearly anteriorly to the orbits. Acanthostega Jar-vik, 1952 (Ahlberg et al. 2008, fig. 4), where thesnout is shorter than for instance in Tiktaalik, andthe anteriad extension parietals and postparietals arelikewise shorter (Clack 1994, fig. 11A; Daeschleret al. 2006, fig. 4d), represents an intermediate mor-phologic stage in this development.

The lengthening of a bone may be due to anteriorgrowth or anterior and posterior growth of the med-ian bones, but the result of the combined anteriorand posterior growth of these bones must be ananteriad transfer of the sutures between these bonesas a following up of the prolonged snout. Thus, aposteriad transfer of the parietals, as suggested bythe W/R interpretation, is illogical. Bystrow (1935,fig. 13) described the ontogenetic development ofthe cranium of Benthosaurus sushkini Efremov, 1929and showed that the growth that produced the longsnout was largely at the anterior part of the frontalsand the posterior part of the nasals. This is seenbecause the growth zones are shown by the surfacesculpture of the bones (Bystrow 1935, fig. 12). Inearlier stages in the ontogeny there was growth alsoat the anterior margin of the parietals (Bystrow1935, fig. 15).

Correspondence in position between the frontalsin Acanthostega on the one hand, and the naso-(pos-terior) postrostrals of the panderichthyid Pan-derichthys on the other, is probably considered a

FOSSILS AND STRATA Morphology, phylogeny and taxonomy of osteolepiform fish 9

main argument in the support of the W/R terminol-ogy (Janvier 1996, p. 262, fig. 6:5B, C). The anteriorposition of the frontals in Acanthostega (Ahlberg1991a, fig. 3B; Clack 1994, fig. 11A), Ichthyostega(Jarvik 1952, fig. 35B), Ventastega Ahlberg, Lukse-vicks & Lebedev, 1994 (Ahlberg et al. 2008, fig. 4),Tiktaalik (Daeschler et al. 2006, fig. 3), and Elpis-tostege (Schultze & Arsenault 1985, fig. 7), makes iteasy to confuse these bones with the posteriorpostrostrals in classical osteolepiforms like Eus-thenopteron foordi Whiteaves, 1881 or Osteolepismacrolepidotus Agassiz, 1835 and also the pan-derichthyid Panderichthys rhombolepis (Gross, 1930)(Vorobyeva & Schultze 1991, figs 4, 5). However,again it is important to note that there is virtually noprolongation of the snout in Osteolepis macrolepido-tus and Panderichthys rhombolepis, whereas in Tik-taalik and Elpistostege, as well as in many tetrapods,the snout is distinctly prolonged. When this fact istaken into consideration it is seen that the assumedhomology between the posterior postrostrals of oste-olepiforms and the frontals of tetrapods is incorrect.If we adjust for the prolonged snout (and the follow-ing anteriad transfer of the sutures) and the enlargedeyes in, for instance, Acanthostega or a tetrapod likeLyrocephalus Wiman, 1914 (cf. Borgen 1983, fig. 5),the position of the bones that in tetrapods are calledparietals will in osteolepiforms end in the position ofthe bones that supporters of the W/R terminologycall postparietals but in the orthodox terminologyare called parietals.

It is relevant that the intertemporal in the Car-boniferous tetrapod Baphetes orientalis Owen, 1854(Milner et al. 2009, fig. 3) has a position that is rem-iniscent of a small protrusion of the postfrontal ofTiktaalik (Daeschler et al. 2006, fig. 3). Thus, theposteriorly protruding part of the postfrontal in Tik-taalik probably is homologous to the intertemporal.The intertemporal may have fused with the posteriorsupraorbital and together these bones constitute thepostfrontal bone. This means that two lateral bones,the supratemporal and the intertemporal, are situ-ated more or less lateral to the main body of theparietals. The tabular is situated posterior to thesupratemporal. This interpretation of Tiktaalikwould also fit with that the anteriad continuation ofthe notch lateral to the parietals of which the ante-rior part (called ‘slightly separated scarf joint’ byDaeschler et al. 2006, p. 760, fig. 3) corresponds tothe spiracular slit of most osteolepiforms. In oste-olepiforms the spiracular opening normally (Mega-lichthys may be an exception) reaches anteriorly toor close to the boundary between inter- andsupratemporals (in orthodox terminology). This slitin Tiktaalik reaches to the boundary between the

supratemporal and the part of the postfrontal thataccording to the pattern of Baphetes orientalis (Mil-ner et al. 2006, fig. 3) is homologous to theintertemporal. Thus, the configuration of Tiktaalikcorresponds well to that of osteolepiforms whenusing the orthodox terminology.

The narrow slit in Tiktaalik, presumed homolo-gous to the spiracular slit in osteolepiforms, is poste-riorly continuous with a wider gap that probablybecomes the otic notch of tetrapods. Supporters ofthe W/R interpretation suggest that it is only thewider posterior gap that is the spiracular slit. A widespiracular opening has been suggested above as pos-sibly present in some specimens of Megalichthys hib-berti Agassiz, 1835 and in Gogonasus Long, 1985b(Long et al. 2006). However, according to Starck(1979, p. 162), large spiracular openings amongsharks are associated with bottom living forms. Thisis probably not the environment of Tiktaalik.

As mentioned, in both Elpistostege and Tiktaalikthe postparietals bones (=median extrascapulars inorthodox terminology) have expanded anteriad tocompensate for the changed proportions betweenthe pre- and postorbital parts of the cranium, just asthe median extrascapular has done in Ichthyostega(Carroll 1988, fig. 9:3a). Concerning Ichthyostega,Borgen (1983, p. 748) expressed some uncertainty inhow to name the bones of the cranial roof butexpressed support for Jarvik’s (1967) interpretation(Borgen 1983, fig. 6A). Now, the authors considerthe interpretation suggested by S€ave-S€oderbergh(1932, fig. 15; cf. Borgen 1983, fig. 6B) is more credi-ble with the exception that the bone situated pos-tero-lateral to the parietal and antero-lateral to themedian extrascapular is the supratemporal and not afusion between supratemporal and intertemporal, assuggested by S€ave-S€oderbergh. This is due to thatIchthyostega shows in S€ave-S€oderberghs (1932, figs15, 16) reconstruction posterior protrusions on thebone called supraorbital 2. These protrusions arereminiscent of the protrusion from the postfrontalin Tiktaalik (Daeschler et al. 2006, fig. 3) that was,because of Baphetes orientalis (Milner et al. 2009, fig.3), interpreted as an intertemporal that had fusedwith a supraorbital. This indicates that the boneS€ave-S€oderbergh (1932, fig. 15) considered a fusionof the inter- and supratemporal bones, instead, isonly the supratemporal. In this respect the authorsfind that the terminology for Ichthyostega also pre-sented by Carroll (1988, fig. 9:3a) is the most credi-ble one.

Change in the course of the supraorbital sensory canalnecessitated by the W/R interpretation. – Thesupraorbital sensory canal passing through the

10 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA

frontals and avoiding the parietals is the dominatingpattern among tetrapods. Instead the canal passesthrough the bones lateral to the parietals (e.g. Car-roll 1988, fig. 9:14c, g). If using the orthodox termi-nology for osteolepiforms this sensory canal passesthrough the same bones as they mostly do in tetra-pods. However, if using the W/R-terminology forosteolepiforms the sensory canal passes through theparietals and avoids the frontals (=posterior postros-trals in W/R terminology). Thus, this is also anexample where the use of the W/R terminologyentails another radical change that is neither indi-cated by transitory stages nor functionally explicable(cf. Pearson 1982, p. 37).

The occasional extension by the sensory canal tothe parietals of some tetrapods (Carroll 1988, fig.9:14i, j) may be a parallel to the sensory canal inva-sion of the parietals in Megalichthys, or a result of afusion of the parietal and intertemporal that proba-bly is seen in most porolepiforms (Jarvik 1972, p.100, figs 38, 43A, 45A). The sometimes transfer ofthe junction between supraorbital and suborbitalsensory canals to the postorbital is illustrated byBorgen (1983, fig. 7).

In Tiktaalik the dermosphenotic (intertemporal inW/R terminology) has disappeared as a separatebone (Daeschler et al. 2006, p. 760). Normally inosteolepiforms the supra- and infraorbital sensorycanals meet in the dermosphenotic but in Tiktaalikthere has probably been a transfer of this junction tothe intertemporal, possibly in the way shown by Bor-gen (1983, fig. 7).

Fusion of preorbital and postrostral bones of osteolepi-forms. – Concerning the apparent fusion of thebones in the postrostral region, which in osteolepi-forms are called nasals and anterior and posteriorpostrostrals, these bones become in the orthodoxinterpretation the nasals of the tetrapods. In theW/R interpretation they become both frontals andnasals. A support for the orthodox interpretation isthat there is apparently a trend towards fusionsbetween all ipsilateral postrostrals and nasals. Thishas the result that the small bones anterior to thepaired naso-postrostrals in Panderichthys and Elpis-tostege also fuse with the naso-postrostrals, and allthese bones become the tetrapod nasals. An indica-tion of this trend is seen in the rhizodontiformBarameda decipiens (Woodward, 1906) (Long 1989,figs 1, 2, 5) where there is on the left side only onepaired bone between the premaxilla (naso-rostro-premaxilla?) and frontal (orthodox interpretation)and in the claimed primitive sarcopterygian Mee-mannia Zhu, Yu, Wang, Zhao & Jia, 2006 (Zhuet al. 2010, fig. 3A). Also in the early palaeoniscoids

Cheirolepis Agassiz, 1835 (Pearson 1982, fig. 1;Arratia & Cloutier 1996, figs 1A, 2, 6, 9, 10),Moythomasia Jessen, 1968 (Jessen 1968, fig. 1; Arra-tia & Cloutier 1996, fig. 1C) and in other primitiveactinopterygians (Pearson 1982, fig. 3) all the boneslateral to the large median postrostral and betweenthe rostro-premaxilla and the frontals have fused.Thus, this is a trend parallel to the fusion of all thenasals in the same position at the transition fromosteolepiforms to tetrapods. The difference is thatin the actinopterygians, there still remains the largemedian postrostral that is common also in severalosteolepiforms. This is also seen in PolypterusLac�ep�ede, 1803 (Jarvik 1980a, fig. 235B; Arratia &Cloutier 1996, fig. 1B). Arratia & Cloutier (1996)have, contrary to Pearson (1982, fig. 1), transferredthe presumed erroneous W/R terminology of themedian paired bones in osteolepiforms on to theactinopterygians.

Schultze & Arsenault (1985) and Vorobyeva &Schultze (1991) showed that in Panderichthys andElpistostege, which are relatively close to the fish-tetrapod transition, the dermal bones between thefrontals (in orthodox terminology) and the (?ros-tro-)premaxilla consist of two pairs of bones. Theposterior pair of these bones, which are the bonesthat in fish terminology are called posteriorpostrostrals, fits as the frontal bones in the termi-nology followed by these authors after having fusedwith the adjacent nasals. The bones situated ante-rior to this pair of bones then become the nasals ofthe tetrapods after having fused with the adjacentnasals. This may be considered a support for theW/R interpretation and it is possible to imagine atransition between a panderichthyid Panderichthysrhombolepis and an early tetrapod like Acanthostegagunnari Jarvik, 1952 (Janvier 1996, fig. 6:5B;Schultze 1996, figs 7C1, D1). However, the manyindications that this interpretation is wrong suggestthat all the postrostral bones of osteolepiformsbecome parts of the tetrapod nasals.

Transfer of the pineal opening from an interfrontalposition to an interparietal position). – The positionof the pineal opening is, as stated by Parrington(1967, p. 233) and Jarvik (1967, p. 186), irrelevantfor the homologizations of the bones. Jarvik (1967,p. 197) explained that the reason for the change inposition of the pineal opening is that the main partof the brain including the diencephalon whose tec-tum includes the pineal and parapineal organs andhas remained in the postorbital part of the head,whereas the frontals and to some extent even theparietals have moved anteriad as a result of the rela-tive prolongation of the snout. The dermal bones

FOSSILS AND STRATA Morphology, phylogeny and taxonomy of osteolepiform fish 11

have moved anteriad, whereas the endocranium withthe brain has stayed behind (cf. Borgen 1983, fig. 2).

In most osteolepidoids the pineal fenestra andforamen is situated in an interfrontal position nearthe level of the orbits. In the group here called Eopo-doidea (cf. phylogenetic and taxonomic part) bothsome eusthenopterids and panderichthyids alsoshow an interfrontal position of the pineal opening(orthodox interpretation) but distinctly more poste-rior than normal in other osteolepiforms. In Eus-thenodon (Jarvik 1952, fig. 23) the pineal fenestrareaches nearly to or to the posterior frontal margin,and in Panderichthys rhombolepis (Schultze & Arse-nault 1985, fig. 8B; Vorobyeva & Schultze 1991, figs4, 5) and Elpistostege (Schultze & Arsenault 1985, fig.8A) the pineal foramen is positioned distinctly pos-terior to the orbits. Thus, we see in eopods a poste-riad transfer of the pineal openings. Interesting inthis connection is also the position of the pinealopening in some branchiosaurs (Boy 1972, pl. 1, fig.1, pl. 2, figs 1–3, figs 56, 69). Here the pineal openingis situated anteriormost in the interparietal suture,in the previously mentioned anterior part of theparietals that is at a level anterior to the laterally sit-uated supratemporal and (the presumed sometimespresent) intertemporal. It is thus situated in theabove-mentioned anterior expansion of the parietalsthat originally is due to the anteriad expansion ofthe tetrapod snout and that has occupied the posi-tion that in most osteolepiforms is held by the fron-tals (orthodox terminology). Even more revealing isthe observations referred to that among recent rep-tiles (Trost 1956, p. 323, figs 1a, b, 2c–e) and amongPermian amphibians (Boy 1972, pl. 2, fig. 1, fig.29B), the pineal opening in several cases is situatedin the suture between frontals and parietals, and in afew taxa (Trost 1956, fig. 1c) also in interfrontalposition but close to the suture with the parietals.Thus, there are several recorded examples of transi-tional stages in the transfer of this opening from aninterfrontal to an interparietal position. Schultze &Arsenault (1985, p. 295) claims that the pineal fora-men (parietal foramen by Schultze and Arsenault) isinterparietal in all tetrapods. This allegation is, asalready mentioned, disproved by Trost (1956).

To understand homologies correctly it is necessaryto follow the principle concerning comparative anat-omy stated by Pearson (1982, p. 37) of viewing inter-pretations of homologies in a holistic (alsofunctional) context instead of only concentrating onthe position of a pair of bones relative to, forinstance, the orbit or pineal foramen. The authorswill add to this principle that claimed morphologicchanges should preferably also be demonstrated bythe observed presence of intermediate morphologic

stages. The morphologic changes that are associatedwith the orthodox interpretation satisfy both theseprinciples. The W/R terminology satisfies neither ofthese principles.

It should be pointed out that the discussed dis-pute concerns the bone homologies at the transitionfrom osteolepiforms to tetrapods and that both theorthodox and W/R interpretations are based on aconsensus about the homologies and terminologyfor tetrapods. However, Bjerring (1995) reinter-preted the homologies of the skull roof among tetra-pods in a way that, if correct, has implications forthe correct terminology of these bones also in fishes.This suggestion is not discussed in this paper.

The definition of the nasal bones presumably isbased on man. Because of this, and because (as indi-cated above) the osteolepiform nasals (defined bythe sensory canal) may be homologous only with thelateral part of the tetrapod nasals (consideredhomologous with the nasals in man), it is perhapsmore correct to call the nasals of osteolepiforms ‘lat-eral nasals’, and the posterior and anterior postros-trals ‘median nasals’. However, because there is stilla disagreement concerning which interpretation ofthe skull roofing bones is correct, the normal oste-olepiform terminology with anterior and posteriorpostrostrals, and ‘nasals’ defined by the sensorycanal, is used in this work.

Circumnarial bones and supraorbito-tectalsThe circumnarial bones in Eusthenopteron foordiWhiteaves, 1881 include the lateral rostral and tectalbones. The tectals and the supraorbital constituteone series of bones, the supraorbito-tectal series.Thus, these bones are all discussed under the sameheading.

The number of tectals has been subject of uncer-tainty. In Eusthenopteron foordi a posterior tectal wasconsidered fused with the anterior supraorbital (Jar-vik 1944a, figs 6, 7; 1980a, fig. 119). This hypothesiswas based on a single specimen, SMNH P 31 (Jarvik1944a, p. 11, fig. 4). Bjerring (1979, fig. 1) followedthis interpretation, but later Jarvik (1980a, p. 159,fig. 116; E. Jarvik pers. comm. 1991) expresseduncertainty about it. The posterior tectal part of thisalleged compound bone, which Jarvik called thesupraorbito-tectal, constituted the preorbital corner(Jarvik 1980a, fig. 119). The pattern in this regionhas been studied by one of us (UB) (cf. morphologicdiscussions) and we will also use the term supraor-bito-tectal.

Jarvik (1980a, fig. 119) illustrated one single longanterior tectal in Eusthenopteron foordi. However,Jarvik (1966, fig. 13C) showed a groove that reachesthe dorsal margin of the nasal fenestra. Jarvik (1966,

12 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA

fig. 14; 1980a, figs 116, 120; E. Jarvik, pers. comm.1991) considered this a fracture and did not show asuture in this position. Also Bjerring (1979, fig. 1A)omitted this in a reconstruction, but included it in alater reconstruction (Bjerring 1989, fig. 2E). Thus,the literature may report two anterior tectals in Eus-thenopteron foordi. This was what Jarvik (1948, fig.17) showed in an illustration that showed the maxi-mum number of skull roof bones in the osteolepi-dids examined by him. Thus, osteolepiforms have atleast two anterior tectals, and a problem is how toname them. In this work all tectals anterior to theposterior tectal are called ‘anterior tectals’. Whenthere is more than one bone they are numberedfrom the anterior end of the series. Thus, when oneform shows one anterior tectal and another formshows two or more, the homologies between thesebones are unknown. Thus, the anterior tectal 1 inone form is not necessarily homologous with theanterior tectal 1 in another form. Descriptions ofEusthenopteron foordi specimens and following dis-cussions that are relevant to this problem are pre-sented in the morphologic discussions.

In works treating particularly megalichthyids theterms ‘prenarial and postnarial bones’ have beenused (Thomson 1964a; Jarvik 1966; Schultze 1974).These names are not part of the regular series ofbones that constitute the dermal cranium but areused when the homologies of the circumnarial bonesin the regular system of bones are unknown. Obvi-ously prenariale surrounds the anterior and postnar-iale the posterior part of this fenestra.

The maximum number of supraorbitals was byJarvik (1948, fig. 17) given as three. In later works,his terminology has varied somewhat (Jarvik 1980a,figs 115, 116, 119). Here, the authors will use thename anterior supraorbital and number the poste-rior supraorbitals from the anterior end.

Anterior intertemporal processThe process situated antero-lateral on the parietalshield, referred to by Bjerring (1972, p. 80) as the‘frontodermosphenotic process’, has been called byBorgen (1983) ‘anterior intertemporal process’. Thisis because it usually is advantageous to include inthe term for a structure the name of the bone onwhich the structure is situated, and also its positionon the bone.

Endocranium

General structureThe endocranium of osteolepiforms (Jarvik 1980a,fig. 86) consists of an anterior moiety that is sepa-rated from a posterior moiety by the so-called

intracranial joint. The anterior moiety, which iscalled the ethmosphenoid, includes the ethmoid andsphenoid (also called orbitotemporal) parts of theendocranium. The border between these two parts isdefined by the posterior extension of the postnasalwall. Jarvik (1942) established a detailed and welldefined terminology for the different parts of theethmosphenoid of osteolepiforms.

The posterior moiety, which is called the otico-occipital part, consists of the otic and occipitalregions, respectively. The border between these tworegions is defined (Jarvik 1980a, p. 39) by the open-ing for the vagus nerve.

Dorsal fossa of otic region. – The terminology con-cerning some endocranial fossae of the otic regionwas recently clarified by Bjerring (1984) who showedthat the term ‘fossa bridgei’, as previously used onosteolepiforms (Thomson 1965, fig. 1; Bjerring 1972,figs 8, 9; Vorobyeva 1977a, figs 9, 11; Jarvik 1980a,figs 77, 88), should be changed. Bjerring suggested‘fossa supra-auditiva’ which will be used here.

Dermal bones in mouth roof

General structureThe dermal mouth roof includes vomers, parasphe-noid, entopterygoid, ectopterygoid, dermopalatine,paraotic plates and subotic plates. Also additionaldermal plates situated in the mouth roof, and whichmay be more or less firmly attached to endocraniumand gill arches are included here. Measured variableson the parasphenoid are defined (Fig. 1B).

Vomers. – Bjerring (1991) suggested that the pairedtooth and tusk bearing bones situated anteriorly onthe palate, which in both osteolepiforms andporolepiforms generally have been denoted vomers,are not homologous with the vomers as they origi-nally were defined in man. Bjerring (1991, p. 227)stated that the vomer in man is the result of a fusionof two contra-lateral bones and this bone was situ-ated beneath both the ethmoid and orbito-temporalparts of the endocranium. Because the nasal sac inman (and other mammals) has expanded posteriadand thus the posterior border of the ethmoid regionhas moved posteriad, the original position of thevomer in mammals may mainly have been subjacentto the interorbital region. He compared this with thevomers in urodeles and porolepiforms. In urodeles apalatal bone consisted of an anterior part situatedbeneath the ethmoid, and a posterior part that runsalong the parasphenoid. In the porolepiform Glyp-tolepis groenlandica Jarvik, 1972 the so-called vomeris situated beneath the ethmoid region. This bone in

FOSSILS AND STRATA Morphology, phylogeny and taxonomy of osteolepiform fish 13

Glyptolepis was homologized with the anterior partof the bone in urodeles. The posterior part of theurodele bone was homologized with fused dentalplates. The latter, which is situated in theorbitosphenoid region, is interpreted as the homo-logue of the vomer in man. The bone in urodeleswas considered a composite bone, and that inporolepiforms a single bone called ‘dacnil’. Fromthis he concluded (Bjerring 1991, fig. 14) that thebones usually called vomers in osteolepiforms werenot vomers, but also the result of fusion by a subeth-moidal dacnil and a suborbitosphenoid vomer.

As shown below, the vomer in osteolepiforms isdivided into two ipsilateral bones that are provision-ally called anterior and posterior vomers. For tworeasons this terminology is retained, in spite of Bjer-ring’s results: (1) the gap from man to porolepi-forms and osteolepiforms is a long gap for directhomologizations, particularly when one part is astructure as modified as the human cranium. Thiscauses uncertainties. An apparently unsolved prob-lem in Bjerring’s interpretation is where the bone hecalled dacnil is incorporated in the human skull.One credible alternative is that the human vomer isthe result of a fusion between the two dermal unitsin the osteolepiform palate (dacnil and vomer inBjerring’s terms), as it may be in urodeles. If so, theterms anterior and posterior vomers seem fitting; (2)because of the mentioned uncertainties, and becausethe term vomer is currently used for the dermalbones in this region, the authors have, also for thesake of simplicity, used the terms anterior and poste-rior vomer.

Lower jaw

Measured variables are defined in Figure 1. Mostlength measurements on the jaw are taken parallelto an axis defined by the antero-dorsal and pos-tero-dorsal corners. Heights are measured verticalto this axis. The two faces of the lower jaws may becalled lateral and mesial sides, dorsal and ventralsides, labial and lingual sides, or external and inter-nal sides. The latter alternative is the simplest andleast ambiguous and will be used here. The twomargins separating the two faces are called ventro-mesial and dorsal margins, respectively. The part ofthe dorsal margin that carries teeth is called dentalmargin.

In all osteolepiforms and also many other oste-ichthyans the lower jaws show a more or less devel-oped antero-dorsal mandibular expansion. Thisusually consists to a large extent of the anterior partof the Meckelian bone, also called the men-tomandibular, but also to a varying extent of other

bones like the anterior part of the prearticular andthe dentary.

The terminology for the infradentaries differsamong different authors and needs some comments.The earlier most used terminology called the fourinfradentaries, from the anterior end, ‘splenial’,‘postsplenial’, ‘angular’ and ‘surangular’. This termi-nology was used by Jarvik (1937), Gross (1941),Nilsson (1943, 1944), Westoll (1943), Romer (1966,p. 50), Vorob’eva and Obruchev (1967), Thomson(1964a) and Miles (1971), with the difference thatJarvik, Gross and Nilsson used ‘supra-angular’instead of ‘surangular’, and Nilsson used ‘presple-nial’ instead of ‘splenial’. To get a simple system freefrom earlier used terms Jarvik (1944) introduced ‘in-fradentary 1’, ‘infradentary 2’, ‘infradentary 3’ and‘infradentary 4’. This system has later been adoptedby Gross (1956), Jessen (1966, 1973), Vorobyeva(1977a) and Young et al. (1992). An advantage withthe number system is that it emphasizes that thebones are parts of a series of bones. One theoreticaldisadvantage is that there is a possibility that after areduction of bones in a series a single bone in a jawshould have the name, for instance, infradentary 3something which, when the jaw is studied in a non-evolutionary context, may seem awkward. Anotherdisadvantage with the number system is that it is notconsistent with the terminology used for tetrapods.In some relatively recent works (Fox et al. 1995;Long et al. 1997; Zhu & Schultze 1997) the previ-ously used terminology was reintroduced. Theauthors will use the number system in this work.

A problem similar to that concerning theinfradentaries exists with respect to the coronoids.These were named, from anterior to posterior, ‘pre-coronoid’, ‘intercoronoid’ and ‘coronoid’ (Gross1941; Nilsson 1943, 1944). A number system is usedby several authors like Jarvik (1944a), Vorobyeva(1962; 1977a, b), Jessen (1966, 1973), Young et al.(1992) and Chang & Zhu (1993). With respect tocoronoids the number system may be even moreproblematic than with the infradentaries, since it hasbeen claimed that there originally were more thanthree coronoids (Zhu & Yu 2004, fig. 8) of whichone may be a bone carrying the parasymphysealtooth plate (Jarvik 1972, p. 116). If correct we thushave a coronoid bone anterior to coronoid 1, some-thing that constitutes an obvious terminologicalproblem. However, since the homologies betweenthe three osteolepiform coronoids and possible otherbones in this series present in other groups isunclear, as is also the relationship to the parasym-physeal tooth plate, and also because this systemnow seems current, the number system is used herealso for the coronoids.

14 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA

The anterior mandibular fossa is normally situ-ated posterior or postero-ventral to the antero-dor-sal mandibular expansion, dorsal to the pars anteriorof the prearticular and ventral to the anterior part ofthe coronoid ridge (a.m.fs, Figs 28, 31A, 61A, 73B).As shown in the descriptions and discussions belowthis fossa may sometimes also constitute a fenestraexposing the Meckelian bone. It apparently is theanteriormost of a series of cavities including also theintercoronoid fossae. Several names have been usedfor this fossa, ‘anterior dental fenestra’ (Watson1926, p. 252), ‘erste Vorcoronoidgrube’ (Gross 1941,p. 8), ‘anterior dentary fossa’ (Thomson 1964a, p.330, fig. 5B; Worobjewa 1975, fig. 3), ‘pit for vomertusk’ (Jarvik 1972, figs 49, 50; Young et al. 1992, fig.33) and ‘precoronoid fossa’ (Zhu & Yu 2004, fig. 4).Fox et al. (1995, fig. 47) used ‘anterior mandibularfossa’. The latter name on this fossa is used herebecause it distinguishes this fossa from the other fos-sae associated with the coronoids. The intercoronoidfossae are, contrary to the anterior mandibular fossa,largely surrounded by coronoids, whereas this ante-rior fossa is bordered ventrally and antero-ventrallyby the prearticular, sometimes the Meckelian bone(mentomandibular) and a parasymphyseal bone,posteriorly and dorso-laterally (sometimes evenpartly anteriorly) by the anterior coronoid andsometimes the dentary bone. It should not be con-fused with the adsymphysial fossa, which is a fre-quently less distinct fossa, situated anterior orantero-ventral to the anterior mandibular fossa andventral to the antero-dorsal mandibular expansion.The term precoronoid fossa, which may seem logi-cal, may be confused with the dental fossa on coro-noid 1 which earlier was called precoronoid.

‘Prearticular’ seems current and will be used here,even if Nilsson’s (1943, p. 15) arguments for insteadusing ‘gonial’ seem reasonable. The prearticular isusually divided into a smooth surfaced ‘pars ante-rior’ and a ‘pars dentalis’. The division between thesetwo parts of this bone is defined by the anteriormostpoint of the prearticular dental plate. Sometimesthere is also a ‘pars posterior’ adjacent to and evenposterior to the glenoid fossa.

Some authors (Vorobyeva 1977a, fig. 15C; Younget al. 1992, fig. 33A; Fox et al. 1995, p. 170; Longet al. 1997, fig. 38) have interpreted the pars anteriorof the prearticular as part of the Meckelian bone.This is treated in the below morphologic discussionsof the prearticular.

The above-mentioned parasymphyseal dentalplate is a small tooth-covered bone situated poste-rior to and near the symphysis, and on the antero-dorsal mandibular expansion. Among osteolepi-forms such bones were described by Vorobyeva

(1962, figs 19, 30) in Eusthenopteron saeve-soederber-ghi Jarvik, 1937, Eusthenodon wenjukowi Rohon,1889a, Platycephalichthys bischoffi Vorobyeva, 1962and Panderichthys rhombolepis (Gross, 1930) underthe term ‘praedentale’. Thomson (1964a, fig. 4) used‘crista dentalis’ and Jessen (1966, fig. 5B, C, pl. 10,fig. 1) used ‘adsymphysial dental plate’. Jarvik (1972,p. 113) discussed these structures and suggested tocall all such plates ‘parasymphyseal dental plates’, aterm previously (Jarvik 1962) used by him in a studyon porolepiforms. Until interrelationships betweensuch plates in different groups is cleared up it seemsreasonable to use a common term, and since ‘para-symphyseal dental plate’ appear to be the oldest, andin common usage (Jessen 1980, p. 186; Jarvik1980a), this name is used in this study.

The external surface shows three grooves thatsometimes are separate and sometimes continuous.One groove runs along the postero-dorsal margin.Another groove, which usually is in continuity withthe former, runs obliquely antero-ventrally on thejaw. Sometimes continuous with the latter there isan approximately vertical groove on infradentary 2.As pointed out by S€ave-S€oderbergh (1933, pp. 14,93) these grooves were previously interpreted assutures. Gradually it was established that these werepitlines but the early confusion is not surprisingbecause the suture between infradentaries and den-tary frequently runs partly together with the antero-ventrally running so-called horizontal pitline. How-ever, the suture continues to the postero-dorsal cor-ner of the jaw, whereas the pitline turns postero-ventrally before it reaches this corner. The frequentlyseparate pitline on infradentary 2 is, due to its nor-mal course, called ‘vertical pitline’. Distinguishingbetween suture and pitline may still be a problem.

Operculo-gular bones

Gular bonesThe gular bones include paired principal gulars ofdifferent shapes and an anteriorly situated mediangular.

SubmandibularsThe series of bones situated between the lower jawand the principal gulars have been called ‘lateralgulars’ (Moy-Thomas 1935) and ‘branchiostegalrays’ (Jarvik 1948; Young et al. 1992), but ‘sub-mandibulars’ seems now to be the current term (Jar-vik 1963; Moy-Thomas & Miles 1971, 1980a, b,1985; Schultze & Arsenault 1985; Fox et al. 1995;Ahlberg & Johanson 1997; Long et al. 1997) and wewill use this name. ‘Branchiostegal rays’ probablyshould be reserved for the rod-shaped bones of most

FOSSILS AND STRATA Morphology, phylogeny and taxonomy of osteolepiform fish 15

actinopterygians where the principal gulars are miss-ing. Pearson (1982, fig. 7) showed the intermandibu-lar dermal bones of different osteichthyan groups.The submandibulars are numbered, and in earlyworks, Jarvik (1948) considered the posteriormostbone as number 1. Later Jarvik (1980a, fig. 121C)reversed the succession. The posterior bone of thisseries, which is situated ventral to the subopercular,was by Jarvik (1980a, fig. 121) called ‘sub-mandibulo-branchiostegal plate’. Fox et al. (1995,fig. 17) called it ‘subopercular 2’. Because the reduc-tion of bone in the submandibular series takes placeanteriorly the authors have followed the early termi-nology by Jarvik and called the posteriormost bone,the one also called submandibulo-branchiostegalplate, for ‘submandibular 1’.

Opercular bonesThe opercular bones include a dorsal opercular anda ventrally situated subopercular. Relative heights ofthese bones vary.

Upper Palaeozoic fossil locality inthe Oslo Region

Geological setting

The study area lies within the Oslo Graben, whichwas formed during the Late Palaeozoic tectonicactivity comprising both extensional faulting andvolcanism. The Oslo Graben represents the northernextension of the northwest European basin system.The formation of the Oslo rift is related to the Varis-cian orogeny and the main graben of the OsloRegion was formed in the Late Carboniferous (Lar-sen et al. 2008). The initial rifting event was followedby the extensive volcanism, extensive rifting and theregion was uplifted by major batholiths in the latestPalaeozoic. The tectonic and volcanic activitieswithin the Oslo Region concluded in the Early Trias-sic (Larsen et al. 2008).

Stratigraphy

The Upper Palaeozoic sedimentary rocks in the OsloRegion overlie unconformably the Lower Palaeozoic(Cambrian–Silurian) Caledonian folded sedimentaryrocks. The Late Palaeozoic succession is referred tothe Asker Group (Fig. 3; Dons & Gy€ory 1967; Larsenet al. 2008) and is composed of the three formationsnamed from bottom to top, the Kols�as Formation,the Tanum Formation, and the Skaugum Formation(Fig. 3; Elder & Kanes 1966; Henningsmoen 1978;Larsen et al. 2008).

The Kols�as Formation (up to 20 m thick; Fig. 3) ismainly composed of red mudstone, sandstone, sub-ordinate conglomerate, limestone and minor anhy-drite characterizing a floodplain with fluvial streamchannel fill and lake depositional environments thataccumulated under arid conditions (Dons & Gy€ory1967; Henningsmoen 1978; Olausen 1981; Olaussenet al. 1994; Olaussen & Dahlgren 2007).

The overlying Tanum Formation is ca. 20 thick(Fig. 3). It overlies disconformably the Kols�as For-mation and is composed of thick cross-beddedquartz rich sandstone, pebbly sandstone and con-glomerate. Minor mudstone and fine-grained sand-stone, some of them with plant remains, are present.The interpretation of the depositional environmentof the Tanum Formation comprises floodplains anddeltaic deposited under semi-arid conditions (Dons& Gy€ory 1967; Olaussen et al. 1994).

The Semsvik locality

The Upper Palaeozoic fossil locality at Semsvik inthe township of Asker, Norway (59°51020.280’N,10°2406.130’E), is situated about 15 km. west of Oslo(Figs 2, 3). It was discovered on a field excursion in1931 by Olaf Holtedahl (Holtedahl 1931, p. 325).The measured stratigraphic succession is given inTable 1 and shown on Figure 3.

Of the beds exposed at Semsvik a conglomerate atthe bottom is considered part of the Tanum Forma-tion (Fig. 3), and the sandstone containing tuffs isreferred to the overlying Skaugum Formation (Elder& Kanes 1966, p. 2; Dons & Gy€ory 1967, p. 63; Hen-ningsmoen 1978, p. 14).

The new osteolepidoid specimens were collectedat the locality; fossil plants, lamellibranchs and otherfish remains have also found at the locality (Fig. 2B).

Plant fossilsThe plant fossils from Semsvik have been describedby Høeg (1936a, 1936b, 1936c). The flora consists ofsome poorly preserved ferns, pteridosperms of thegenus Neuropteris and some specimens similar toCallipteris or Allethopteris, airstems and rhizosomesof species of the equisetophyt Calamites, presumedleaves and reproduction organs of Calamites namedAsterophyllites, Calamostachys and Palaeostachys,leaves from species of Cordaites, seeds denotedSamaropsis, but considered being from a Cordaitesspecies, and conifers of the taxa Walchia and Ernes-tiodendron filiciforme (Schlotheim) Florin. Therewere also remains of Dicranophyllum, which is con-sidered a seed plant of unknown affinity.

In beds from the Asker Group at a locality nearTanum church (59°530440’N, 10°280450’E), silicified

16 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA

A

B

Figure 2. A, View of the Semsvik location in approximately western direction. The quarry in the Upper Palaeozic sediments is marked bythe white circle. The transition between sediments and basalts is shown in the mountain side. B, Details of the exposure at Semsvik location.

FOSSILS AND STRATA Morphology, phylogeny and taxonomy of osteolepiform fish 17

wood has been found, which Høeg (1936c) referredto Dadyxolon saxonicum (G€oppert) or Dadyxolonscrollianum (G€oppert).

Fossil faunaAfter a preliminary examination of the fish fossilsfrom Semsvik by Anatol Heintz, it could be statedthat they were remains of osteolepidoids, probably ofthe genus Megalichthys (Holtedahl 1931, p. 329).Later, Heintz (1934) provided a fuller description ofthe fish fauna. The osteolepidoid remains, somescales, a head plate and one lower jaw, were identifiedas Megalichthys sp., mainly because of their generalappearance and their Late Palaeozoic age (Heintz1934, pp. 181, 185). The material also included onepresumed shark tooth of a type previously calledPleuracanthus (concerning the use of this name seeZangerl 1981, pp. 64–65), a smaller presumed sharktooth that was indeterminable, a scale similar to thoseof holoptychids, a possible Pleuracanthus coprolite, apossible head plate and some possible neural archesof dipnoans, and some presumed palaeoniscoidremains that tentatively were determined to Amblyp-terus and Elonichthys. The palaeoniscoid materialconsisted of scales, usual and fulcra type, a ‘completespecimen’ consisting of the larger part of the body ofa fish, a fin fragment and two lower jaw fragments.

The lamellibranchs from Semsvik were tentativelydetermined as possible species of Palaeanodonta byDix & Trueman (1935, p. 26). Eagar (1994) restud-ied and revised the lamellibranchs assemblage andreferred them to Anthraconaia protracta Eagar, 1975.

Age of bedsSome doubts on the age of the sediment have pre-vailed. The presence of Ernestiodendron filiciforme(Schlotheim) Florin could indicate that the depositsat Semsvik should be referred to the Lower Permian(Høeg 1936a, p. 34). Also Dadoxylon saxonicumrecorded from the Tanum Church locality is mainlyan Early Permian (Rotliegendes) form (Høeg 1936c,p. 281).

All the described fish taxa are known both fromthe Carboniferous and Permian, but Amblypteruswas considered ‘more Permian than Carboniferous’(Heintz 1934, p. 192). On the assumption that theycorrectly had determined the lamellibranchs, Dix &Trueman (1935, p. 30) were certain that the age ofthe fauna was Permian. Thus, Henningsmoen (1978,p. 21) stated that the fauna and flora indicated anEarly Permian age.

However, from his restudy of the lamellibranchsEagar (1994) demonstrated that the fossil assemblagefrom the Tanum Formation was Late Carboniferous.Olaussen et al. (1994, p. 178) reviewed the results of

biostratigraphic studies and conclusively referred theTanum Formation to the Upper Carboniferous,which is followed here.

Descriptions

The descriptions are presented from Carboniferousto Devonian forms. This is because the Carbonifer-ous material consists of a larger number of well-pre-served specimens than the Devonian material. Inthis way it is possible to use descriptions from themany well-preserved post-Devonian specimens asbasis for the understanding of less well-preservedDevonian specimens. The material includes speci-mens from a new fish genus from the Upper Car-boniferous beds in Norway, specimens from theCarboniferous of Great Britain referred to Mega-lichthys Agassiz, 1835 specimens from Upper Devo-nian beds in Germany and Balticum referred todifferent species of Latvius Jarvik, 1948, one lowerjaw from Great Britain of the Middle Devonian Oste-olepis macrolepidotus Agassiz, 1835, several also Mid-dle Devonian specimens from Great Britaindetermined as Gyroptychius milleri Jarvik, 1948 andfrom Greenland referred tentatively to Gyroptychiusgroenlandicus Jarvik, 1950a.

Descriptions in this paper include onlymacrostructures. Discussions of possible signifi-cances of the described structures are mainlyincluded in the next section.

Material, methods and repository

During field work at Semsvik a detailed study of theprofile was made and rock samples from the differ-ent beds were collected (Figs 2, 3; Table 1). The fossilspecimens have been prepared mechanically, withvibro-tools and needles. On some specimens withonly one visible side, the exposed side was coveredby melted wax. After this artificial matrix had hard-ened the non-exposed side could be prepared.Chemical dissolution with acetic acid or sodiumhydroxide had no effect on the samples.

Prior to photography several specimens were whi-tened with ammonium chloride to improve contrast.The photographs have not been retouched. Threespecimens, two of Askerichthys (PMO 93549 andPMO 93553) and one of Megalichthys hibberti (HMG 8-52) were X-rayed to expose the course of sen-sory canals. The values used for the X-raying were35–40kV and 300–750mA.

The fossil material collected from Semsvik inAsker belongs to the Natural History Museum inOslo. The numbers of these specimens have the

18 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA

prefix PMO. Material of Megalichthys is borrowedfrom the Natural History Museum in London andHancock Museum in Newcastle. Their specimenshave respective prefixes BMNH PV and HM. The

material of Latvius, and also other specimens fromthe Baltic Upper Devonian, belongs partly to Swed-ish Museum of Natural History, specimens with pre-fix SMNH, and partly to the Museum f€ur

A

C

D E

B

Figure 3. Maps showing the position of the fossil locality at Semsvannet W of the NW part of the Oslofjord. A, Map showing location offossil locality relative to the lake Semsvannet and Asker church. The asterisk shows the fossil locality and the square the Semsvik farm.Asker church is also marked. B, The inner region of the Oslo fjord also showing the position of the Asker church. C, Southern Norwaywith area shown in map B indicated by shading. D, A general section of the sediments at Semsvik and of the whole Asker Group belowthe basalt (B1) (Snorre Olausson pers. com., 2010). E, Legend.

FOSSILS AND STRATA Morphology, phylogeny and taxonomy of osteolepiform fish 19

Naturkunde Berlin, which have the prefix MB. Thematerial from Greenland has the prefix MGUH. Thespecimens are kept at the Natural History Museumof Denmark, University of Copenhagen, Denmark.For comparisons with materials from Palaeontologi-cal Museum of Uppsala (PMU) have been included;the Manchester Museum (MM); The SedgwickMuseum (SM); the Royal Scottish Museum (RSM);Australian National university (ANU); AustralianMuseum, Sydney (AMF); Commonwealth Palaeon-tological Collection (CPC), which is housed in theBureau of Mineral Resources, Geology and Geo-physics, Canberra, Australia; Queensland MuseumQMF, Peabody Museum, Yale University (YPM);Institute of Vertebrate Palaeontology and Palaeoan-thropology, Beijing (V).

Genus Askerichthys n. gen.

Type species. – Askerichthys heintzi n. sp.

Species included. – The type species is the only knownspecies of the genus.

Diagnostic characters. – (1) Lachrymo-maxillarynotch angular; necessary character. (2) Nasal fenes-tra situated close to the dorsal margin of thelachrymo-maxillary notch, and mostly situated pos-terior to the corner of the notch. (3) Large triangularretroarticular process on the lower jaw. (4) Twopairs of coronoid tusks and thus only two coronoids.Characters 2–4 are all necessary and probably suffi-cient characters.

Derivation of name. – The genus is named for thetownship of Asker from which the fish is known.

Askerichthys heintzi n. sp.

Figures 4–39, 110B, 117, 119D, 121D, 125F,126A, 127B, 158A, 133B, 134D, 137B, 143A,147A, 148A, 149H, 158A, 159A, 160A, 162A

and 164B

1934 Megalichthys sp. Heintz, pp. 181, 185, pl. 1,figs 8–12; fig. 2B.

1995 ‘Borgen’s new Norwegian genus’; Fox, Camp-bell, Barwick & Long, pp. 109–111, 129.

Holotype. – PMO 93555 (Figs 4, 5, 6, 21, 22, 23A, B,24A, B, 26, 32, 36, 37, 38), Tanum Formation, Sems-vik, Asker, Norway.

Material. – The holotype of Askerichthys heintzi isPMO 93555. It was collected by J. F. Bockelie at the

type locality at Semsvik in 1965, and consists of amain specimen PMO 93555a (Figs 4, 5) and itscounterpart PMO 93555b (Fig. 6). A posterior partof a left lower jaw that is prepared free from PMO93555a constitutes PMO 93555c (Figs 23B, 24B, 26),a probable left maxilla, stuck on a latex mould of thecounterpart, constitute PMO 93555d (Fig. 7) andthe right preopercular that is prepared free fromPMO 93555b is numbered PMO 93555e (Fig. 7B).The main specimen of the holotype shows remainsof a body seen in ventral view. It shows some oper-culo-gular bones, a complete lower jaw (Figs 5, 23,24), the left maxilla, an incomplete pectoral girdle(Clt, Figs 4, 37), scales from the ventral side, the cau-dal fin and remains of some median and paired fins(Figs 4, 6, 37).

There are four more or less complete specimens ofthe anterior cranial division (Figs 8, 9). PMO 73855(Figs 8A, 9A) is an imprint of a fronto-ethmoidalshield whose reference to Askerichthys is uncertain.PMO 93549, PMO 93553 and PMO 93554 are moretypical osteolepiform anterior cranial divisions (Figs8B–D, 9B–D, 10–13). There is one incomplete poste-rior cranial division PMO 73876 (Figs 15–19) that,like the holotype, was collected by J. F. Bockelie. Inaddition to the bones on the holotype there is oneseparate opercular PMO 73879 (Fig. 35), and threeseparate lower jaws, PMO 51005 that is incomplete(Heintz 1934, pl. 10), PMO 93546, and PMO 93548(Figs 23A–C, 24A–C, 25, 27–31). The latter speci-men is complete. These are paratypes.

Stratigraphic and geographic distribution. – The spe-cies is known from the Upper Carboniferous beds atSemsvik in Asker west of Oslo (Fig. 3). The beds arereferred to the Tanum Formation in the AskerGroup (Table 1, Fig. 3).

Derivation of name. – The species is named after thelate Professor Anatol Heintz, who was the first todescribe remains of this fish. Together with hisdaughter Natascha Heintz he kept vertebratepalaeontology alive as a science in Norway for aboutsix decades.

Diagnostic characters. – (1) Lachrymo-maxillarynotch angular; necessary character. (2) Nasal fenes-tra situated close to the dorsal margin of thelachrymo-maxillary notch, and mostly situated pos-terior to the corner of the notch. (3) Large triangularretroarticular process on the lower jaw. (4) Twopairs of coronoid tusks and thus only two coro-noids.

Characters 2–4 are all necessary and probably suf-ficient characters.

20 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA

Description

Fronto-ethmoidal shieldShape. – The general shape of the fronto-ethmoidalshields is shown in the illustrations (Figs 8–10).PMO 93553 (Figs 8C, 9C) is shorter and broaderthan PMO 93549 (Figs 8D, 9D). In PMO 93549 thedorsal mouth margin is visible from above. In PMO93553 and PMO 93554 the margin is not visible indorsal view, but this is probably due to the marginbeing secondarily bent underneath the anterior partof the shield because a fracture apparently resultingfrom such a break is visible. The shape of the sub-narial corners (sn.c, Figs 8D, 10B, C–E, 12B, 13)appears approximately right-angled on PMO 93553and slightly acute on PMO 93549 and PMO 93554.Anteriorly on the dorsal side of the shields the der-mal bones are covered by a more or less continuouscosmine sheet, which also covers several sutures.More posteriorly the sutures are visible.

In PMO 73855 (Figs 8A, 9A) the interpretationsof different bones are less obvious than on the otherspecimens. It is an imprint and thus cannot be pre-pared further. A notch on the left side of PMO73855 looks somewhat like an orbital fenestra (?fe.orb, Fig. 8A), and if so a pair of bones mesial tothis notch may be supraorbitals. An argumentagainst this interpretation is that the posteriormostbone has a groove that looks like a pitline (?pl.Fr,Figs 8A, 9A) suggesting that this bone is the frontal

(?Fr, Figs 8A, 9A). With this latter interpretation theabove-mentioned notch (?fe.orb, Fig. 8A) must be anotch between the frontal and the posteriormostnasal (?Na. 7, Fig. 9A). Such a notch is present alsoon PMO 93549 and PMO 93554, but is on thesespecimens much shallower (Figs 8C, D, 9C, D). Thislatter interpretation fits well with the shape of thecentral bone that is reminiscent of the median poste-rior postrostral of PMO 93549 and PMO 93553(M.p.pr, Figs 9C, D, cf. Fig. 8), and may thus repre-sent this bone (?M.p.pr, Fig. 9A, B). The rightantero-lateral part shows a possible fenestra exonasa-lis (?fe.ex, Figs 8A, 9A). Tentatively, this specimen isassigned to ?Askerichthys heintzi.

Premaxilla. – Neither the postero-dorsal suture ofthe premaxilla nor the postero-dorsal suture of therostro-premaxilla is distinguishable on the externalsurface of PMO 93549, PMO 93553 or PMO 93554(Fig. 8B–D). Except for the left lateralmost part onPMO 93553, the mouth margin of both PMO 93549and PMO 93553 is complete between the subnarialcorners. These two specimens have both been pre-pared in ventral view.

The anterior margin of the lachrymo-maxillarynotch can best be studied on PMO 93549 (Fig. 10A,B, D), but is observable also on PMO 93553 (Figs8C, 9C, 10E) and PMO 93554 (Figs 8B, 9B). On thelatter two specimens the margin shows no specificstructures except for a foramen (fr, Fig. 10E) that

Figure 4. Askerichthys heintziorum n. gen. et n. sp. Tanum Formation, Semsvik, Asker, Norway, Late Carboniferous. PMO 93555a. Mainspecimen of the holotype. Scale bar = 1 cm. Abbreviations explained in Appendix 1.

FOSSILS AND STRATA Morphology, phylogeny and taxonomy of osteolepiform fish 21

covers a large part of the margin. It cannot be deter-mined whether this is the opening for the infraor-bital sensory canal, or the so-called premaxillarycanal. The lack of other structures is partly becauseof incomplete preparation due to the fragility of thespecimen. On PMO 93549 this region is slightlydamaged, but shows more structures than the otherspecimens. A lateral distinct foramen is interpretedas the opening for the premaxillary canal (fr.c.Pm,

Fig. 10D). Mesial to this foramen there are some pitsthat probably are secondary, and mesialmost on themargin there are two foramina that have tentativelybeen interpreted as, respectively, the opening for theinfraorbital sensory canal (fr.io.sc, Fig. 10D) and abranch of this canal. A groove running along theventral margin of the narrow bar ventral to the nasalfenestra, possibly led the infraorbital sensory canal(gr.io.sc, Fig. 10D).

Figure 5. Askerichthys heintziorum n. gen. et n. sp. Tanum Formation, Semsvik, Asker, Norway, Late Carboniferous. Head section ofholotype PMO 93555a. The half right mandible in the upper left corner of the specimen was later prepared free and is PMO 93555c. Scalebar = 1 cm. Abbreviations explained in Appendix 1.

Figure 6. Askerichthys heintziorum n. gen. et n. sp. Tanum Formation, Semsvik, Asker, Norway, Late Carboniferous. PMO 93555b. Holo-type. Counterpart of PMO 93555a showing remains of the caudal fin and submandibulars. On this specimen, the main part of the fish isshown as an imprint. Scale bar = 1 cm. Abbreviations explained in Appendix 1.

22 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA

In ventral view the subnarial corner on PMO93553 (sn.c, Fig. 13) is positioned approximately atthe level of the mid-point of the lateral margin ofthe fenestra endochoanalis, whereas on PMO 93549it is positioned nearer to the anterior end of thismargin (sn.c, Fig. 12B).

On both PMO 93549 and PMO 93553 the ventralside of the mouth margin shows pits for premaxil-lary marginal teeth that are situated on a premaxil-lary shelf (sh.Pm, Fig. 13). Mesially this shelf isexpanded constituting the ventral part of an antero-median palatal process (am.pl.pr, Fig. 13). This pro-cess functions as the base for premaxillary tusks(tu.Pm, Figs 12B, 13). On PMO 93549 part of theprocess seems to be in contact posteriorly with ante-rior remnants of the parasphenoid (Fig. 12B). OnPMO 93553 such a contact is not observable (Figs12A, 13). There are no marginal teeth anterior to theantero-median palatal tusks on any of the two speci-mens.

PMO 93553 had on the right side apparently15 premaxillary teeth (t.Pm, Fig. 13) not count-ing the tusk (tu.Pm, Figs 12, 13). On the incom-plete left side of PMO 93553 there are observableremains of 9 teeth, but the ninth tooth is situ-ated in the same position as the ninth tooth onthe right side, suggesting a similar number onboth sides. On PMO 93549 the number of teethis less clear. An estimated number of teeth andempty pits on the right side suggests, not count-ing the tusk, 13–14 teeth. On the left side anestimate gives 11–13 premaxillary teeth.

On PMO 93553 the palatal lamina on the left sideshows a shallow groove posterior to the premaxillaryshelf. On PMO 93549 there is such a groove on bothsides. These grooves are rather indistinct, and maybe secondary features, but for reasons outlined inthe morphologic discussions they are tentativelyinterpreted as sutures between premaxilla and therostral series (s.Pm/Ro, Figs 12B, 13).

Laterally, on both sides, there are posterior expan-sions from the palatal lamina constituting the lateralwalls of the anterior palatal fenestra. These antero-lateral palatal processes (al.pl.pr, Fig. 13) are directedpostero-mesially. They are seen on both sides ofboth specimens, but are most distinct on the rightside of PMO 93553. This process meets antero-later-ally directed parts that have been interpreted asbeing endocranial. Since the ventralmost part of theprocess on the right side of PMO 93553 is situatedalso at a level ventral to the endocranial part, andsince the process on the left side of PMO 93553meets structures that possibly are remains of theanterior vomer (?Vo, Fig. 13), also the right antero-lateral palatal process possibly also met the anteriorvomer. A possible anterior suture of the anteriorvomer runs across the bone mesial to the left nasalcavity.

Rostrals. – If the above-mentioned groove observ-able on the palatal lamina is the suture between pre-maxilla and the rostral series, the part of the anteriorwall of the apical fossa dorsal to this groove is partof the rostral bones (?Ro, Fig. 13). The suture

A B

Figure 7. Askerichthys heintziorum n. gen. et n. sp. Tanum Formation, Semsvik, Asker, Norway, Late Carboniferous. A, Detail of holo-type. PMO 93555d. Part of the latex mould of PMO 93555b on which is stuck a dermal bone, probably the right maxilla. B, The rightpreopercular is prepared loose from PMO 93555b and is numbered PMO 93555e. Scale bar = 1 cm. Abbreviations explained in Appen-dix 1.

FOSSILS AND STRATA Morphology, phylogeny and taxonomy of osteolepiform fish 23

A

C

D

B

Figure 8. Fronto-ethmoidal shields in dorsal view. Tanum Formation, Semsvik, Asker, Norway, Late Carboniferous. A, Askerichthys?PMO 73855. B–D, Askerichthys heintziorum n. gen. et n. sp. B, PMO 93554; C, PMO 93553; D, PMO 93549. Scale bars = 1 cm. Abbrevia-tions explained in Appendix 1.

24 U. J. Borgen & H. A. Nakrem FOSSILS AND STRATA