Morphology,, Characters, and the Interrelationships of Basal Sarcopterygians

Transcription

1 CHAPTER 17 Morphology,, Characters, and the Interrelationships of Basal Sarcopterygians RICHARD CLOUTIER URA 1365 du CNRS Université des Sciences et Technologies de Lille Sciences de la Terre Villeneuve d'ascq, France PER ERIK AHLBERG Department of Palaeontology The Natural History Museum Cromwell Road, London SW7 5BD, United Kingdom I. Historical Background) A new chapter in sarcopterygian systematics opened in 1970 when Schultze (1970) published a brief overview of sarcopterygian tooth structure which explicitly used the distribution of derived characters to delineate monophyletic groups and to argue for tetrapod monophyly. Three years later, Andrews (1973) produced the first attempt at an overall cladistic analysis of the sarcopterygian fishes. Although it retained certain characteristics of previous analyses, such as a priori separation of lungfishes and tetrapods from "crossopterygians" and a tendency to search for "key characters," it centered on a wide-ranging and thorough review of character state distributions and character polarities. Andrews (1973) concluded that character incongruence and thus evolutionary parallelism was rife among the "crossopterygians," but that the skull roof pattern might provide a reasonable guide to their relationships. On this basis she presented fully resolved phylogeny (Fig. la) which divided the Crossopterygii into Binostia (actinistians and porolepiforms) and Quadrostia (osteolepiforms, rhizodonts, and onychodonts). However, she made no greater 1A11 generic taxa referred in the text are extinct with the exception of Polypterus, Latimeria, Protopterus, Lepidosiren, and Neoceratodus. daim for it than that, as it "seems to do least injustice to knowledge in its present state, it is proposed as a model for future discussion" (Andrews, 1973, p. 137). The rest of the decade saw graduai progress in sarcopterygian phylogenetics. Miles (1975) used parsimony arguments to place lungfishes and "crossopterygians" side by side as sister-groups in the clade Sarcopterygii. He included the Tetrapoda within Crossopterygii but did not consider the possibility that the latter group might be paraphyletic relative to the Dipnoi. However, in a subsequent paper (Miles, 1977) he placed the Actinistia as sister-group to the Dipnoi + Choanata. Schultze (1977) applied cladistic principles to the distribution of fin and limb characters among the Sarcopterygii. At the 26th Symposium of Vertebrate Palaeontology and Comparative Anatomy held in 1978 at Reading, England, Gardiner presented a paper on sarcopterygian cladistic phylogeny which started a series of methodological debates. As a response to a comment by Parrington, Patterson confirmed in support of Gardiner's arguments for a cladistic view of relationships that a lungfish shares more characters with a cow than with a salmon. "The salmon, the lungfish, and the cow" soon became familiar as a classic example of a three-taxon statement. The debate was continued in the pages of Nature (Halstead, 1978; Halstead et al., 1979; Gardiner et al., 1979). INTERRELATIONSHIPS OF FISHES 445 Copyright 1996 by Academic Press, Inc. Ail rights of reproduction in any form reserved.

2 446 RICHARD CLOUTIER AND PER ERIK AHLBERG A Osteolepif ormes t.s Eus thenop ter o Actinistia o rcs Dipnoi Rhizodontif ormes Actinistia Porolepiformes C3) o,r) Onychodontiformes Coelacanthiformes Porolepiformes LPowichthys Youngolepis Dipnoi Osteolepiformes Porolepiformes Tetrapoda Tetrapoda Dipnoi Actinistia On yc ho don ti or mes Povvichthys Youngolepis Porolepif ormes Osteolepiformes Onychodontia Actinistia _r_tetrapoda Panderichthyidae Osteolepif ormes Rhizodontida Porolepiformes Powichthys Youngolepis Panderichthyidae _J Diabolepis Tetrapoda Dipnoi FIGURE 1 Previously published sarcopterygian phylogenies. (A) Andrews (1973); note that this phylogenetic hypothesis does not consider either the Dipnoi or the Tetrapoda. (B) Rosen et al. (1981); a pattern-cladistic approach with fossil taxa (dashed unes) inserted after the analysis of the Recent forms. Eusthenopteren occupies a basal position, far removed from the Tetrapoda. (C) Panchen and Smithson (1987); (D) Schultze (1987); (E) Ahlberg (1991b). Phylogenies (C) (E) are based on fossil and Recent data. Osteolepiforms and "panderichthyids" (=elpistostegids; see text) consistently group with tetrapods, but otherwise there is little agreement between the three. "The Terrestrial Environment and the Origin of Land Vertebrates" (edited by A. L. Panchen), published in 1980, gave a snapshot of the growing influence of cladistic methodology on sarcopterygian phylogenetics. With hindsight the two most significant papers are those of Patterson (1980) and Gardiner (1980). The former was a powerful critique of noncladistic approaches to the problem of tetrapod origins, while the latter presented a cladogram where lungfishes were the living and fossil sister-group of tetrapods, and "crossopterygians" were paraphyletic. Forey (1980) used a pattern-cladistic approach to argue that Latirneria is a sarcopterygian, the sister-group of Dipnoi + Tetrapoda, rather than a chondrichthyan. The points made by Patterson (1980), Gardiner (1980), and Forey (1980) were developed further by Rosen et al. (1981) in a provocative paper (Fig. lb) which succeeded in starting a vigorous debate. The

3 17. Interrelationships of Basal Sarcopterygians 447 following years saw a surge of publications on this topic, many of them couched as rebuttals of Rosen et al. but nevertheless characterized by an explicitly cladistic approach (Holmes, 1985; Jarvik, 1980; Long, 1985, 1989; Maisey, 1986a; Panchen and Smithson, 1987; Schultze, 1981, 1987, 1991, 1994; Ahlberg, 1989, 1991b; Cloutier, 1990, 1991a,b; Chang, 1991a,b; Chang and Smith, 1992; Young et al., 1992; Ahlberg and Milner, 1994). Much of the interest focused on the threetaxon problem of the living groups and on the position of the Osteolepiformes (Figs. lc 1 e). Forey (1987) and Forey et al. (1991) continued to build on the work of Rosen et al. Alongside the systematic reviews came a sequence of descriptive works (Vorobyeva, 1977; Jarvik, 1980; Campbell and Barwick, 1982a,b, 1984, 1987, 1988; Chang, 1982, 1991b; Chang and Yu, 1984; Andrews, 1985; Schultze and Arsenault, 1985; Long, 1989; Ahlberg, 1989; Ahlberg et al., 1994; Clack, 1989, 1994a,b; Coates and Clack, 1990, 1991; Vorobyeva and Schultze, 1991) which significantly expanded the data set available for phylogenetic analysis. The development of computer programs for phylogenetic inference ushered in the present stage of the debate, characterized by exhaustive analyses of large data sets (Cloutier, 1990, 1991a,b; Ahlberg, 1991b; Forey et al., 1991; Lebedev and Coates, 1995; Schultze, 1994; Schultze and Marshall, 1993). On a parallel front, several workers have approached the three-taxon problem of lungfish, coelacanth, and tetrapod relationships from a molecular perspective (Meyer and Wilson, 1990; Gorr et al., 1991; Stock and Swofford, 1991; Hedges et al., 1993). At present the bulk of the molecular evidence favors a lungfish tetrapod sistergroup relationship (Meyer, 1995). The last 20-odd years of research into sarcopterygian phylogeny have produced an enormously enlarged data base as well as substantial improvements in phylogenetic methodology and practice. No complete phylogenetic consensus has developed, but several significant and probably permanent changes of opinion have taken place: There is almost universal agreement (but see Jarvik, 1980) about the status of the Sarcopterygii as a clade and about the characters which define the group (Rosen et al., 1981; Schultze, 1987; Panchen and Smithson, 1987; Ahlberg, 1991b). Placement by Rosen et al. (1981) of the Dipnoi as the living and fossil sister-group of the Tetrapoda has flot generally found favor. Most paleontologists favor the "traditional" osteolepiform tetrapod relationship, redefined in terms of shared derived characters (Schultze, 1987; Panchen and Smithson, 1987; Long, 1989; Cloutier, 1990, 1991a; Ahlberg, 1991b; Vorobyeva and Schultze, 1991; Young et al., 1992; Ahlberg and Milner, 1994). The majority of these authors remove the elpistostegids ("panderichthyids"; see below) from osteolepiforms and regard the osteolepiforms as the sister-group of the clade [elpistostegids + tetrapods]. Panchen and Smithson (1987), however, place the elpistostegids within the osteolepiforms. The main nonparticipants in this consensus are Chang (1991b), who regards tetrapods as the sister-group of ail other sarcopterygians, and Forey et al. (1991), who continue to place the lungfishes as the living and fossil sister-group of tetrapods. A number of authors have come to view the Porolepiformes and the Lower Devonian genera Youngolepis and Powichthys as immediate relatives of the Dipnoi (Maisey, 1986a; Ahlberg, 1989, 1991b; Cloutier, 1991b; Chang, 1991a,b; Chang and Smith, 1992). This is linked with the recognition of Diabolepis (Chang and Yu, 1984) as a primitive lungfish (Fig. le), although some authors (Forey et al., 1991) accept the latter proposition without agreeing with the former. Another group of workers (Schultze, 1987; Panchen and Smithson, 1987; Long, 1989) place the porolepiforms and Youngolepis as separate plesions on the stem to Eosteolepiforms + tetrapods] (Figs. lc and 1d). As regards the three-taxon problem of coelacanths, lungfishes, and tetrapods, most recent paleontological analyses favor a lungfish tetrapod sistergroup relationship (Maisey, 1986a; Panchen and Smithson, 1987; Cloutier, 1990, 1991a; Ahlberg, 1991b; Trueb and Cloutier, 1991; Forey et al., 1991). However, there is also support for a Recent sister-group relationship between coelacanths and tetrapods (Schultze, 1987; Long, 1989; Vorobyeva and Schultze, 1991) or between coelacanths and lungfishes (Chang, 1991b). While it is heartening to see emerging areas of agreement, the substantial remaining disputes cannot be ignored. These are largely due to differences in character interpretation and scope. Thus, the reassertion of the osteolepiform tetrapod relationship is directly linked to the rejection of the interpretation of Rosen et al. (1981) of osteolepiform snout anatomy in favor of Jarvik's (1942, 1980) model. Many of the characters used by Maisey (1986a), Ahlberg (1989, 1991b), Chang (1991a,b), and others to link porolepiforms and lungfishes are ignmed by Schultze (1987) and Panchen and Smithson (1987). Our purpose in this chapter is to review the state of knowledge for ail the major sarcopterygian groups and to present a phylogenetic analysis based on a large data set. Our respective contributions in part reflect our slightly different fields of research. Cloutier

4 448 RICHARD CLOUTIER AND PER ERIK AHLBERG provided the information about actinistians, onychodonts, and dipnoans, while Ahlberg was the principal contributor on tetrapods, elpistostegids, porolepiforms, and Y oungolepis + Powichthys. The character matrix represents our shared knowledge and derives from several sources (e.g., Ahlberg, 1991b); however, the largest component was the previously unpublished character matrix from Cloutier's Ph.D. thesis (1990). For this reason, and because he performed the phylogenetic analysis and interpretation, Cloutier takes the formai position as principal author. It was a salutary experience to discover, during the preparatory stages of this collaboration, that there were considerable and sometimes irreconcilable differences between our character tables. Some of these were due to difficulties with poorly preserved specimens, but in other cases the disagreement was purely one of interpretation. This may be symptomatic of the debate as a whole. The significance of these problems is considered in more detail in the Discussion. The Actinistia (=Coelacanthi, Coelacanthia, Coelacanthiformes, Coelacanthii, Coelacanthina) include approximately 125 species belonging to 50 genera (Cloutier and Forey, 1991). They range in time from the Middle Devonian to Recent and reached their maximum of diversity during the Lower Triassic (Cloutier and Forey, 1991). However, they are known from the fossil record only from the Middle Devonian to the Upper Cretaceous. The earliest actinistian known is Euporosteus eifelianus from the Givetian of Germany (Stensiô, 1937), whereas the youngest fossil species is Megalocoelacanthus dobiei from the latest Campanian to middle Maastrichtian of New Jersey (Schwimmer et al., 1994). Orvig (1986) identified a fragment of bone possibly referrable to an actinistian, on histological characteristics, from the Paleocene of southern Sweor. dei "Ideil.114 II. The Principal Sarcopterygian Groups The following section attempts to summarize most of our knowledge of sarcopterygian morphology and diversity (Fig. 2). The major omission is the tetrapod crown group, which is not discussed in detail; its diversity is so great that any attempt at a systematic overview would swamp the rest of the paper, and we felt that a full account of the known stem tetrapod genera was more important and timely. Detailed trea t- ments of crown tetrapod systematics can be found Carroll (1988) and elsewhere. A. Actinistia FIGURE 2 A representative range of early osteichthyans. (A) The actinopterygian Mimia toombsi (after Gardiner, 1984). (B) (I), Sarcopterygii; (B) the actinistian Rhabdoderma elegans (after Forey, 1981); (C) the onychodont Strunius walteri (after Jessen, 1966); (D) the tetrapod Ichthyostega groenlandica (modified from Jarvik, 1952, 1980); (E) the elpistostegid Panderichthys rhombolepis (after Vorobyeva and Schultze, 1991); (F) the osteolepiform Osteolepis macrolepidotus (after Jarvik, 1948); (G) the rhizodont,?strepsodus anculonamensis (modilied from Andrews, 1985); (H) the porolepiform Glyptolepis paucidens (original); (I) the dipnoan Dipterus valenciennesi (after Ahlberg and Trewin, 1995). Not to scale. Except for Rhabdoderma and Strepsodus, which are of Carboniferous age, ail these genera date from the Middle-Upper Devonian (Eifelian Famennian). den. Latimeria chalumnae is the only living representai:ive of this group and it has been found only in the Comoros Archipelago, Mozambique Strait, and Chalumna River (Republic of South Africa). Although the monophyly of the Actinistia has never been questioned, the diagnosis has been addressed repeatedly (Andrews, 1973; Forey, 1981, 1984; Maisey, 1986a; Panchen and Smithson, 1987; Cloutier, 1991a,b, 1996a). The Actinistia is monophyletic based

5 17. Interrelationships of Basal Sarcopterygians 449 on the following synapomorphies (Cloutier, 1996a): (1) absence of maxilla, (2) absence of surangular, (3) absence of branchiostegal rays, (4) absence of submandibulars, (5) presence of rostral organ, (6) numerous supraorbitals, (7) presence of extracleithrum (absent in the Diplocercidae), (8) triangular entopterygoid, (9) short dentary, (10) dorsal margin of angular elevated nto a process, (11) coronoid IV oriented vertically, and (12) anterior position of anterior dorsal fin. Characters 2 and 10 are possibly dependent characters, characters 1 and 9 are homoplastic with respect to the Dipnoi, character 3 is homoplastic with respect to the Onychodontida, and character 4 is homoplastic with respect to the Tetrapoda. Three additional characters are considered as potential synapomorphies for the actinistians (Cloutier, 1991a,b), but their condition remains unknown or unclear in Miguashaia bureaui: (13) presence of ventral process of lateral rostral, (14) tandem jaw articulation, and (15) posteriorly expanded U-shaped urohyal. In addition to the characters listed above, Forey (1991) considered the following three characters as synapomorphies of the Actinistia: (1) second dorsal and anal fin lobes both contain a skeleton of several segments which resemble the endoskeletons of the paired fins, (2) head divided by a prominent intracranial joint in which the otico-occipital part of the neurocranium extends anteriorly to form a track-andgroove joint with the basisphenoid, and (3) small premaxilla. The first character is probably absent in Miguashaia bureaui; although the median fin endoskeletons are flot preserved, the lack of a posterior dorsal fin lobe (and the very slight lobation of the anal fin) suggests that the endoskeletons were of a generalized sarcopterygian pattern (Ahlberg, 1992a); this character might be a synapomorphy of the clade [Actinistia except Miguashaial (Cloutier, 1996a). The condition of the second character cannot be determined in M. bureaui; thus, it remains a potential synapomorphy of the group. The size of the premaxilla in Miguashaia is comparable to that of other sarcopterygians. Schultze (1973) proposed Miguashaia bureaui (middle Frasnian, Escuminac Formation, Québec, Canada) to be the most primitive actinistian. According to Cloutier (1991a,b, 1996a), Miguashaia is an actinistian because it shares the aforementioned derived characters with other actinistians, and within this clade it is the sister-taxon of the rest of the group. The clade [Actinistia except Miguashaia] is characterized by the following characters: absence of intertemporal, lacrimal fused with jugal, otic canal passing in postparietal, unbranched distal ends of lepidotrichia, tail notochord straight and horizontal, epichordal and hypochordal lepidotrichia symmetrical, and presence of a supplementary caudal fin. Interrelationships among actinistians have been investigated recently by Cloutier (1991a,b) and Forey (1991). Although there are minor disagreements between them (notably the position of the Coelacanthidae), both phylogenies are strongly asymmetrical and have good stratigraphic correlation. Most actinistian classifications fail to reflect the phylogeny of the group, but Schultze (1993) erected a classification which reflects Cloutier's (1991b) phylogeny. Miguashaiidae Schultze 1993 is a monospecific family containing the middle Frasnian Miguashaia bureaui from the Escuminac Formation of Québec. The anatomy of Miguashaia is fairly well known (Cloutier, 1996a) with the exception of the neurocranium, gill arches, and axial skeleton. An isolated scale from the Givetian of Latvia is referred to the genus Miguashaia (Cloutier et al., 1996). Miguashaia does flot belong to the Coelacanthidae as suggested by Carroll (1988). Diplocercidae Stensiô 1922 is a monogeneric family including fewer than 10 taxa of Deyonian and Lower Carboniferous age. Nesides Stensiô has been synonymized with Diplocercides by Cloutier and Forey (1991). Traditionally the genera Euporosteus and Chagrinia have been incorporated within the Diplocercidae; however, based on their phylogenetic position (Cloutier, 1991a,b) they are removed from this family. Diplocercides is the only Devonian actinistian in which the neurocranium has been studied in detail (Jarvik, 1954, 1964; Bjerring, 1993). Hadronectoridae Lund and Lund 1984 is a family originally created to include the three Namurian actinistians Allenypterus, Hadronector, and Polyosteorhynchus from the Heath Formation (Bear Gulch fauna), Montana. Lund and Lund (1985) provided detailed descriptions of the anatomy of these genera. Cloutier (1991a) excluded Allenypterus in order to keep the family monophyletic. The presence of a series of bifurcating supraorbital pores associated with the supraorbital sensory canal corroborates the monophyly of this clade (Cloutier, 1991a). "Rhabdodermatidae" Berg 1958 is a paraphyletic family including mainly Carboniferous genera (e.g., Rhabdoderma and Caridosuctor). The Carboniferous Rhabdoderma elegans is the best known representative of this group (Forey, 1981).

6 450 RICHARD CLOUTIER AND PER ERIK AHLBERG Laugiidae Berg 1940 includes the genera Laugia and Coccoderma and ranges in time from the Lower Triassic to the Lower Cretaceous. In contrast to its classification in Schultze (1993), Synaptotylus is excluded from this family in order to respect the monophyly of the Laugiidae. Whiteiidae Schultze 1993 is a family created to include the Triassic genus Whiteia from Madagascar and western Canada. A great deal of the anatomy of these forms remains to be redescribed, on the basis of available material. This might require a reidentification of the North American species (R. Cloutier, personal observation). Coelacanthidae Agassiz 1843 is a family in which, to keep its monophyly, Cloutier (1991b) included all actinistians sharing the common ancestor of Coelacanthus and Latimeria. However, Schultze (1993) restricted the definition of the family to include Coelacanthus, Axelia, Wimania, and Ticinepomis. The phylogenetic position as well as the definition of the Coelacanthidae differ between Cloutier's (1991b) and Forey's (1991) hypotheses. Mawsoniidae Schultze 1993 includes Triassic and Jurassic genera (Alcoveria, Diplurus, Chinlea, Mawsonia, and Axelrodichthys) forming a clade which is the sister-group to the Latimeriidae (Cloutier, 1991b; Forey, 1991). Diplurus newarki (Schaeffer, 1952) and Axelrodichthys araripiensis (Maisey, 1986b) are the best known representatives of this family. Latimeriidae Berg 1940 is a family including the only living representative (Latimeria chalumnae) as well as a few Jurassic and Cretaceous genera (Holophagus, Undina, Libys, Macropomoides, and Macropoma). Lambers (1992) provided a revision of the genus Libys. The gigantic, recently described Megacoelacanthus probably belongs to this family, based on the characteristics of the pterygoid and basisphenoid, and not to the Coelacanthidae as suggested by Schwimmer et al. (1994). Many authors have considered the actinistians as an evolutionarily conservative group (see Cloutier, 1991a,b) and Latimeria chalumnae as an example of a living fossil (Forey, 1984). As early as the Famennian, the actinistians had acquired their characteristic body shape. The anterior dorsal fin is located quite anteriorly compared to other sarcopterygians and is never lobated (Cloutier, 1996a). The anal and posterior dorsal fins are usually strongly lobed and contain endoskeletons and musculature which match those of the paired fins (Malot and Anthony, 1958) rather than the median fins of other sarcopterygians. Ahlberg (1992a) interpreted this pattern as evidence of a type of homeotic transformation, with paired fin structures being expressed at the anal and posterior dorsal fin sites. Actinistians have a diphycercal caudal fin, symmetrical dorsoventrally and possessing a supplementary caudal lobe; the only exceptions are Miguashaia, which has a heterocercal tail (plesiomorphic within the Actinistia), and Allenypterus which has a modified tapering diphycercal tail. Most of the synapomorphies diagnosing the clade concern the skull structure, primarily those parts related to the lower jaw and feeding mechanism. Lund and Lund (1985) and Lund et al. (1985) explained the conservatism of the jaw apparatus as a response to a specialized mechanism of suction feeding. Latimeria is the only living sarcopterygian to possess an intracranial joint (Lauder, 1980). Through geological time, the ethmosphenoid has become relatively longer than the otico-occipital part of the neurocranium (Forey, 1991). ln addition to the mechanoreceptive lateral line system, actinistians have a unique electroreceptive organ located in the anterior part of the ethmosphenoid, the rostral organ (Northcutt and Bemis, 1993). In Latimeria, this organ is probably used to localize prey; its use seems to be linked with a unique headstand behavior (Fricke et al., 1987). Among piscine sarcopterygians, the Actinistia is the only group in which the mode of reproduction has been documented in the fossil record. Oviparity has been documented by Schultze (1985) in the Carboniferous Rhabdoderma exiguum. This contrants with the Recent Latimeria, which is ovoviviparous. B. Dipnoi The lungfish clade, Dipnoi, is a universally recognized natural group whose record extends from the Lower Devonian (Pragian) to Recent (Schultze, 1992a). Approximately 280 species are divided into 64 genera, most represented only by tooth plates (ca. 125 species). The Dipnoi reached their maximum diversity during the Devonian (more than 85 species) and Triassic (more than 45 species). The 6 living species are classified into three genera: Protopterus (P. dolloi, P. annectens, P. aethiopicus, and P. amphibius) from tropical Africa (Greenwood, 1987), Lepidosiren (L. paradoxa) from South America, and Neoceratodus (N. forsteri) from Australia (Kemp, 1987). The oldest known members of the group include Uranolophus wyomingensis Denison 1968a (Beartooth Butte Formation, Wyoming), Diabolepis speratus (Chang and Yu, 1984) (Xitun Formation, Yunnan, China), and Speonesydrion iani Campbell and Barwick 1983 (Bloomfield Limestone, New South Wales, Australia) from the Pragian, and

7 17. Interrelationships of Basal Sarcopterygians 451 Sorbitorhynchus deleaskitus Wang et al (Dale Formation, Guangxi Province, China) and Dipnorhynchus suessmilchi (Etheridge, 1906) (New South Wales, Australia) from the Emsian. Interrelationships among dipnoans have only partially been assessed and remain highly debated (Miles, 1977; Marshall, 1987; Campbell and Barwick, 1990; Schultze et al., 1993; Schultze and Marshall, 1993; Long, 1993). The monophyly of the Dipnoi has never been chah lenged; however, since the discovery of Diabole pis speratus (Chang and Yu, 1984) the definition and diagnosis of the group have been debated (Maisey, 1986a; Campbell and Barwick, 1987; Panchen and Smithson, 1987; Schultze, 1987; Schultze and Campbell, 1987; Cloutier, 1990; Smith and Chang, 1990; Chang, 1991b). The Dipnoi are diagnosed by five uniquely shared derived characters (Cloutier, 1996b): the absence of marginal teeth on the lower jaw, the presence of tooth plates on the entopterygoids and prearticulars, a median B-bone located anteriorly or separating the parietals and postparietals, location of the anterior margin of the parietals well posterior to the orbits, and the presence of a labial cavity. Dipnoans are also characterized by the absence of vomerine fangs, maxilla, extratemporal, fossa autopalatina, and intracranial joint. Most of the characters diagnosing the Dipnoi reflect the peculiar nature of the dentition and cranial architecture. Miles (1977) considered Uranolophus wyomingensis to be the most primitive dipnoan. Campbell and Barwick (1984) regarded Speonesydrion iani and Dipnorhynchus suessmilchi as the most primitive genera but considered these as belonging to another lineage than Uranolophus. Their arguments were primarily stratigraphical. Schultze and Marshall (1993) used Dipnorhynchus suessmilchi as the functional outgroup for their phylogenetic analysis of lungfishes. The most controversial species in this respect is Diabolepis spera tus, which has been considered either as a primitive sarcopterygian related to Powichthys and Porolepiformes (Panchen and Smithson, 1987), a primitive dipnoan (Maisey, 1986a; Smith and Chang, 1990), the sistergroup of ail other dipnoans (Cloutier, 1990; Chang, 1991b), the sister-group of the Dipnoi (Chang and Yu, 1984; Ahlberg, 1991b), or a taxon of undetermined status (Campbell and Barwick, 1987; Schultze and Campbell, 1987). Campbell and Barwick (1990) ignored Diabolepis in their study on dipnoan phylogeny, whereas Schultze and Marshall (1993) discussed the significance of Diabolepis to basal dipnoan relationships without actually including it in the analysis. We regard Diabolepis as a dipnoan, the sister-group of ahl other Dipnoi. The clade [Dipnoi excepting Diabolepis] is supported by four synapomorphies: the palatal p0- sition of both anterior and posterior nares, the absence of premaxillae (though arguably homoplastic "premaxillae" have been identified in Ganorhynchus, Orlovichthys, and Scaumenacia; R. Cloutier (personal observation), the presence of C-bones, and the passage of the occipital sensory canal through both extrascapulars and postparietals. Dipnoan classification is unsettled because there is no consensus concerning the phylogeny of the group. Campbell and Barwick (1983, 1987, 1990) divide the Dipnoi into three lineages on the basis of dentition (tooth-plated, dentine-plated, and denticulated), but this phylogeny is unparsimonious (Schultze and Marshall, 1993; Cloutier, 1996b) and has not found general favor. Schultze's (1993) classification is based on the strict consensus tree presented by Marshall (1987, fig. 5). In order to provide a reasonably manageable taxonomic framework for the bewildering diversity of the lungfishes, we present here a list of families combining information from Miles (1977), Campbell and Barwick (1990), Long (1992), Schultze (1993), Schultze and Marshall (1993), and Cloutier (1996b). This should flot be regarded as a definitive statement of dipnoan taxonomy; several of the families are certainly or probably nonmonophyletic and require reevaluation. Diabolepididae Schultze 1993 has only one representative, the Early Devonian Diabolepis speratus, known exclusively from cranial material (Chang and Yu, 1984) and isolated dentitional elements (Smith and Chang, 1990). Its discovery had a crucial impact on our understanding of dipnoan porolepiform relationships (Maisey, 1986a; Cloutier, 1990; Ahlberg, 1991b). Uranolophidae Miles 1977 includes primitive Early Devonian dipnoans with a denticulated palate. As in Diabolepis and Dipnorhynchus, the B-bone separates the parietals but not the postparietals. Uranolophus wyomingensis is the only species belonging fo this family. It is known from a single complete specimen and numerous skulls and lower jaws (Denison, 1968a,b; Campbell and Barwick, 1988). Dipnorhynchidae Berg 1940 includes Early Devonian, dentine-plated dipnoans with a plesiomorphic skull roof pattern. The group comprises the genera Dipnorhynchus and Speonesydrion. Campbell and Barwick (1990) considered Speonesydrion as one of the basal members of their toothplated lineage. Ail species belonging to this family are known from partial skulls and lower jaws (Campbell and Barwick, 1982b, 1983, 1984). The

8 452 RICHARD CLOUTIER AND PER ER1K AHLBERG neurocranium of Dipnorhynchus suessmilchi is the best known of any Early Devonian lungfish. Chirodipteridae Campbell and Barwick 1990 is a Middle to Late Devonian family characterized by a peculiar type of dentition including dentine tuberosities arranged radially. The members are Chirodipterus, Pillararhynchus, Gogodipterus, and Palaedaphus. Chirodipterus australis (Gogo Formation, Lower Frasnian, Australia) is the best known representative (Miles, 1977). Stomiahykidae Bernacsek 1977 includes Middle to Late Devonian genera (Stomiahykus and Archaeonectes) with a large tusklike tuberosity at the anterior end of the mesial row of the entopterygoid tooth plate (Campbell and Barwick, 1990). Long (1992) considered the Stomiahykidae to be closely related to the Chirodipteridae. Dipteridae Owen 1846 is a paraphyletic group of Middle and Late Devonian forms with tooth plates, short posterior dorsal fin, and cosmine. Dipterus valenciennesi (Caithness Flagstone Group, Eifelian Givetian, Scotland) is the best understood representative, being known from whole bodies with well-preserved heads (Forster- Cooper, 1937; White, 1965) and parts of the porolepiform-like postcranial endoskeleton (Ahlberg and Trewin, 1995). More than 20 species referred to Dipterus are only known from isolated tooth plates (Schultze, 1992b). Rhynchodipteridae Berg 1940 includes Middle to Late Devonian, long-snouted dipnoans with denticulated palates. Rhynchodipterus, Griphognathus, and Soederberghia belong to this group. The toothplated, long-snouted genus Rhinodipterus is regarded by one of us (RC) as a rhynchodipterid, but some authors reject this view (Schultze, 1992a). The palatal condition of Iowadipterus remains unknown (Schultze, 1992a). The anatomy of Griphognathus has been described in detail (Miles, 1977; Campbell and Barwick, 1988). Fleurantiidae Berg 1940 includes the latest Givetian to Famennian dipnoans characterized by the following cranial features (Cloutier, 1996b): (1) rostral part of the skull elongated, (2) wide mouth gape, (3) single median E-bone, (4) long bone L 1 +L 2 extending medially to the M-bone, and (5) elongated entopterygoids bearing large conical teeth and numerous small denticles, both organized in radiating rows. Cloutier (1996b) discussed the interrelationships among fleurantiids (i.e., Fleurantia, Jarvikia, Andreyevichthys, and Barwickia). Fleurantia is the best known member of this family, although informative cranial material of Andreyevichthys is under study (R. Cloutier, personal observation). Phaneropleuridae Huxley 1861 includes Middle to Late Devonian lungfishes with enlarged bones B, C, and E, and a general absence of bone D (Scaumenacia, Phaneropleuron, and Pentlandia). The length of the posterior dorsal fin is more than one-quarter of the total length (Cloutier, 1996b). Schultze and Marshall (1993) considered this family to be paraphyletic. Phaneropleurids are closely related to fleurantiids. Uronemidae Traquair 1890 is a monogeneric Carboniferous group, represented by Ganopristodus (=Uronemus), characterized by highly modified tooth plates with one long lingual tooth ridge and reduced lateral rows (Smith et al., 1987) and a single bone replacing the intertemporal and supratemporal. Sagenodontidae Romer 1966 is composed of Carboniferous dipnoans with large bone B, reduced parietals and E-bones, and bone L, H-L2 in contact with bone B (Sagenodus). Tooth plates are formed by ridges rather than isolated teeth. The anal and posterior dorsal fins remain separated from the diphycercal caudal fin (Chorn and Schultze, 1989). Ctenodontidae Woodward 1891 is composed of Carboniferous dipnoans with a single large bone replacing the intertemporal and bones L, +L 2 (Ctenodus, Tranodis, and Straitonia). The skull roof pattern of Ctenodus and Tranodis is similar to that of the uronemids. The position of Straitonia is questionable because a single element is present between bones E and B as in Sagenodus. Conchopomatidae Berg 1940 is a monogeneric (Conchopoma), Carboniferous to Permian family characterized by a denticulated parasphenoid with a rounded anterior margin, concave anterior margin of bone B, and a single median fin fringe. Schultze (1975) revised the genus Conchopoma. Gnathorhizidae Miles 1977 is composed of Late Carboniferous to Early Triassic forms (Palaeophichthys and Gnathorhiza) in which the otic canal passes in the postparietal and with numerous cases of dermal bone reduction (large median bone E; single bone occupying space of bones 3, L,, L 2, and M; and single bone in place of intertemporal, supratemporal, and tabular). In the lower jaw, the oral canal passes in the infradentaries, whereas the mandibular canal is in an open groove (Schultze and Marshall, 1993). Gnathorhizids have been considered as the sister-group of the Lepidosirenidae by Lund (1970) but not by other authors (Ber-

9 17. Interrelationships of Basal Sarcopterygians 453 man, 1968; Miles, 1977; Schultze and Marshall, 1993). "Ceratodontidae" Gill 1872 is a paraphyletic group including Triassic to Tertiary species. The ceratodontid skull roof shows a reduced number of bones in the median and lateral series. More than 40 species of Ceratodus have been described, mostly from tooth plates. Ceratodontids may be paraphyletic with respect to the neoceratodontids. Schultze (1981) investigated the relationships among so-called ceratodontids (Ptychoceratodus, Microceratodus, Arganodus, Tellerodus, and Paraceratodus). Neoceratodontidae Miles 1977 includes Triassic (Epiceratodus) to Recent (Neoceratodus) species with a reduced postparietal and a single bone occupying the space of the parietal and bones L1, L2, and M. Schultze (1981) suggested that Asiatoceratodus is closely related to Ceratodus owing to the presence in both genera of a single bone replacing bones A, B, and C. Lepidosirenidae Bonaparte 1841 is an apomorphic family including two extant genera (Protopterus and Lepidosiren) and extinct representatives (e.g., the Cretaceous Protopterus regulatus). The skull is highly derived compared to most dipnoans (Miles, 1977; Schultze and Marshall, 1993). There are no cheek bones, nor skull roof bones lateral to the parietal and postparietal; the vomer is reduced to a patch of small, conical teeth; the nasal region is kinetic relative to the braincase; and the jaw adductor muscles attach above the skull roof. Devonian dipnoan morphology is fairly well documented owing to the material from the Gogo Formation of Australia (e.g., Chirodipterus australis and Griphognathus whitei), Caithness Flagstones of Scotland (Dipterus valenciennesi), and Escuminac Formation of Québec (Scaumenacia curta and Fleurantia denticulata). However, the anatomy of late Paleozoic, Mesozoic, and Cenozoic species remains poorly understood because most of the species are known only from isolated elements. The cranial anatomy of dipnoans is better understood than that of the postcranium but has generated much debate concerning the homology of the dermal bones. Forster-Cooper (1937) erected a neutral alphanumerical system of nomenclature for the skull roof (bones A F, H J, K Q, and X Z) and cheek (bones T, 1-11, and 13-14). Ahlberg (1991b) and Cloutier (1996b) agreed on the homologies of certain bones with that of other sarcopterygians: A= median extrascapular, X= intertemporal, Y1= supratemporal, Y2= tabular, J= parietal, I= postparietal, 1 = lacrimal, 4=postorbital, 8= squamosal, 9=preopercular, and 10= quadratojugal. In Recent forms the skull roof is greatly reduced compared to Paleozoic species, and the endocranium is cartilaginous rather than ossified. Dipnoans display a great deal of intraspecific variation, particularly in the skull roof pattern (Cloutier, 1996b) and probably the greatest amount of diversity in dermal skull roof pattern among sarcopterygians. The interpretation of the palatal and cheek bones differs greatly among authors; Rosen et al. (1981) argued for the presence of a maxilla, premaxilla, and choana in dipnoans based on their observation on Griphognathus whitei. Branchial and hyoid arches have been described by Miles (1977) for the Frasnian genera Chirodipterus and Griphognathus. Dipnoans are characterized by peculiar types of dentition. Most of them lack a marginal dentition; the maxillae are absent and the premaxillae, when present, are greatly reduced and bear only a few teeth (e.g., Scaumenacia, Andreyevichthys, and Ganorhynchus). Paired prearticular and entopterygoid tooth plates constitute the primary dentitional apparatus of most species; such tooth plates are known from the Emsian to the Recent. Two other distinct types of dentition have been reported: dentine-plated (e.g., Dipnorhynchus) and denticulated types (e.g., Uranolophus and Griphognathus). Campbell and Barwick (1983, 1987, 1990) asserted that dipnoans are divided into two lineages characterized by their dentition the tooth-plated and denticulated types. However, this hypothesis is unparsimonious (Schultze and Marshall, 1993; Cloutier, 1996b) based on the congruence of cranial and postcranial characters. In contrast to other gnathostomes, the teeth composing the plates are flot shed during growth but rather added anteriorly and laterally. This mode of growth allows ontogenetic studies because an adult carnes its own dental ontogenetic history (Cloutier et al., 1993). Because of the constant wear on the crushing and/or shearing surfaces, dipnoans have a hypermineralized tissue infilling the numerous pulp cavities in the tooth plate, the petrodentine (Smith, 1984). Several clearcut patterns, which might justify the term "trends," can be observed in the history of the Dipnoi. The earliest representatives of the group have, with the exception of Diabole pis (Chang and Yu, 1984), already acquired a wholly characteristic "lungfish head" featuring autostyly and a palatal bite. Their postcranial skeletons however seem hardly to be modified from the generalized sarcopterygian condition and bear a certain resemblance to those of porolepiforms (Denison, 1968a; Campbell and Barwick, 1988;

10 454 RICHARD CLOUTIER AND PER ERIK AHLBERG Ahlberg, 1989, 1991b, 1992b; Ahlberg and Trewin, 1995). During the Middle to Late Devonian, new dipnoan groups (Fleurantiidae, Phaneropleuridae) arise which have derived postcranial morphologies with long-based median fins (Ahlberg and Trewin, 1995; Cloutier, 1996b). All known Carboniferous and later lungfishes, other than Sagenodus (Chorn and Schultze, 1989), have very derived postcrania with diphycercal fin fringes rather than separate median fins (Ahlberg and Trewin, 1995). In parallel with this morphological change there is a presumed environmental shift from open marine environments such as Taemas or Gogo to "Old Red Sandstone" facies and finally to apparently nonmarine and oxygen-poor environments like coal swamps (Campbell and Barwick, 1988). Skeletal structures associated with air-breathing (cranial ribs and long parasphenoid stalk) are absent in the primitive marine lungfishes but appear during the Middle Devonian and seem to define a clade within the Dipnoi (Long, 1993). The dramatic changes in dipnoan median fin morphology during the Paleozoic were used by Dollo (1895) to infer the evolution of the group (as well as his principle of the irreversibility of evolution). Paedomorphosis has been suggested as a primary process in this group in relation to the fusion of the median fins, reduction of lepidotrichia, reduction of ossification (Bemis, 1984), and dentitional pattern (Cloutier et al., 1993; Long, 1993). However, in the absence of a fully resolved phylogeny (Schultze and Marshall, 1993) these hypotheses cannot be fully evaluated (Ahlberg and Trewin, 1995). Interestingly, the mode of growth of lungfish tooth plates (see previous discussion) allows the possibility of observing developmental heterochrony in phylogeny. Westoll (1949) and Schaeffer (1952) compared the evolutionary rates of dipnoans with the bradytelic evolution of actinistians. However, rates of evolution have never been calculated in a phylogenetic perspective for the Dipnoi. C. Tetrapoda Although Jarvik (1942, 1972, 1980) and Bjerring (1989, 1991) continue to argue for tetrapod diphyly and a urodele porolepiform relationship, the monophyletic status of the Tetrapoda is supported by a wealth of characters and is accepted by virtually all other workers (Schultze, 1970, 1981, 1987; Jurgens, 1973; Gaffney, 1979; Rosen et al., 1981; Shubin and Alberch, 1986; Panchen and Smithson, 1987; Vorobyeva and Schultze, 1991). It is also supported by molecular evidence (Hedges et al., 1993). The tetrapods are defined here as a clade characterized by the possession of limbs with digits rather than paired fins, a pelvis with a sacrum, and zygapophyses; early members can also be recognized by a suite of derived jaw characters (Ahlberg, 1991a, 1995; Ahlberg et al., 1994). This apomorphy-based definition encompasses the crown group and part of the stem group and would thus be seen as unsatisfactory according to the criteria of De Queiroz and Gauthier (1990). However, at present there is a sharp divide between early limbed vertebrates such as Ichthyostega and Acanthostega, whose membership in the tetrapod stem group can be taken as a well-founded starting assumption, and tetrapod-like "fishes" such as Panderichthys and Elpistostege, whose membership in the stem group needs to be tested. We therefore retain the traditional definition for the present. Note that a similar situation exists with respect to the Dipnoi. The tetrapods have a fossil record reaching back into the Frasnian (Warren and Wakefield, 1972; Ahlberg and Milner, 1994; Ahlberg, 1995). The "traditional" early tetrapod groups, Labyrinthodontia and Lepospondyli, were exposed as nonnatural during the past decade (Smithson, 1985; Milner et al., 1986; Panchen and Smithson, 1988). They have not been replaced by a new consensus. However, it is clear that all the Devonian genera, except perhaps Tulerpeton (Lebedev and Coates, 1995), fall outside the crown group. Within the crown group the Amniota and Lissamphibia are Recent sister-groups. The temnospondyls are members of the lissamphibian clade, while the anthracosaurs probably belong with the amniotes, but opinions differ as to the position of the loxommatids and the old "lepospondyl" groups (Milner et al., 1986; Panchen and Smithson, 1988; Trueb and Cloutier, 1991; Carroll, 1992; Ahlberg and Milner, 1994; Lebedev and Coates, 1995). The origin of the tetrapod crown group clearly antedates the first appearance of anthracosaurs and temnospondyls in the late Viséan (Ahlberg and Milner, 1994). The tentative assignment of the Russian Famennian tetrapod Tulerpeton to the amniote anthracosaur clade (Lebedev and Coates, 1995) suggests an even earlier date for the split. In the context of sarcopterygian interrelationships, the most interesting tetrapods are the Devonian genera. They are not generally placed in higher taxonomie categories, as their interrelationships are poorly resolved. We recognize eight genera: Ichthyostega Sâve-Sôderbergh 1932 is represented by numerous specimens from the Upper Famennian of eastern Greenland. Most of the skeleton except the manus is known, but the braincase is pe-

11 17. Interrelationships of Basal Sarcopterygians 455 culiar and poorly understood (Jarvik, 1980). The pes has seven digits (Coates and Clack, 1990), the shoulder lacks a scapular blade (Jarvik, 1980), and the taul lepidotrichia (Jarvik, 1952). Ichthyostega is uniquely characterized by the possession of an unpaired median postparietal. Acanthostega Jarvik 1952 occurs alongside Ichthyostega in the upper Famennian of eastern Greenland. Long known only from two incomplete skulls, it is described in full from new specimens collected in 1987 (Clack, 1988, 1989, 1994a,b; Coates, 1991; Coates and Clack, 1990, 1991). Acanthostega's manus has eight digits (Coates and Clack, 1990), and the same may be truc for the pes (M. I. Coates, personal communication). The proportions of the forelimb elements are markedly more fishlike than those of Ichthyostega, and the lepidotrichial tail fin is even larger (Coates, 1995). The scapulocoracoid is comparable to that of Ichthyostega (Coates and Clack, 1991; M. I. Coates, personal communication). Tulerpeton Lebedev 1984 is known from a single incomplete body and a number of isolated bones, all from the upper Famennian Andreyevka-1 locality near Tula, central Russia (Alexeev et al., 1994). Tulerpeton has a manus with six digits (Lebedev, 1984). In certain other respects it resembles post-devonian tetrapods; the limb bones are slender and a scapular blade is present in the shoulder girdle (Lebedev, 1984; Lebedev and Coates, 1995). Associated bones from the site, which have not been formally attributed to Tulerpeton, show derived characters like open lateral line sulci (Lebedev and Clack, 1993) which are not present in the other Devonian tetrapods. Ventastega Ahlberg et al is described from cranial material collected at the upper Famennian localities of Pavâri and Ketleri in Latvia. Tetrapod clavicles, interclavicles, and ilia from these localities may also belong to this genus (Ahlberg et al., 1994). Ventastega is the only upper Famennian tetrapod known to possess coronoid fangs. In other respects it broadly resembles Ichthyostega and Acanthostega. Hynerpeton Daeschler et al is a genus of middle or upper Famennian age, based on a scapulocoracoid+ cleithrum from the Duncannon Member of the Catskill Formation, Pennsylvania. The shoulder girdle clearly belongs to a stem tetrapod and resembles that of Ichthyostega as well as the girdle fragments from Scat Craig. Metaxygnathus Campbell and Bell 1977 is represented by a single lower jaw ramus from the Cloghnan Shale of New South Wales, Australia, probably of lower Famennian age (Campbell and Bell, 1977). The jaw carnes coronoid fangs. The assignation of Metaxygnathus to the Tetrapoda has been disputed (Schultze and Arsenault, 1985; Schultze, 1987). However, its tetrapod nature is confirmed by a suite of derived characters which are shared with Acanthostega, Ichthyostega, and Ventastega but flot with sarcopterygian fishes (Ahlberg et al., 1994). Elginer peton Ahlberg 1995 is strictly speaking, known only from cranial remains, but it has been associated with postcranial tetrapod bones which probably also belong to it. This genus cornes from the upper Frasnian of Scat Craig, Scotland and is thus together with Obruchevichthys (sec following discussion) the earliest tetrapod known from skeletal remains. It has several autapomorphies including large size (skull length in excess of 40 cm; Ahlberg, 1995), triangular head shape with an acutely pointed snout, and, on the inner face of the mandible, a broad field of exposed meckelian bone ventral to the very narrow prearticular. The postcranial tetrapod bones from the site include an Ichthyostega-like tibia (Ahlberg, 1991a) and robust scapulocoracoids and ilia (Ahlberg, 1995). Obruchevichthys Vorobyeva 1977 is only known from two incomplete lower jaws, one from the upper Frasnian of Latvia and one from an unknown locality in western Russia. It shares several derived characters with Elginerpeton and appears to be the sister-group of that genus (Ahlberg, 1995). It is likely that the Elginerpeton Obruchevichthys clade (plesion Elginerpetontidae; Ahlberg, 1995) is the sister-group of all other Tetrapoda. Space does not permit us to list the post-devonian tetrapod groups. Apart from the aforementioned genera, the Devonian tetrapod record includes some well-preserved upper Frasnian trackways from Genoa River, Victoria, Australia (Warren and Wakefield, 1972) and?famennian trackways (with more than 150 footprints) from Valentia Island, southwestern Ireland (Stôssel, 1995). A supposed Lower Devonian trackway from Australia (Warren et al., 1986) cannot be confidently identified as belonging to a tetrapod, while the isolated Devonian "tetrapod footprint" described from marine Brazilian deposits by Leonardi (1983) is probably a starfish trace fossil (Rocek and Rage, 1994). Two other genera were described originally as Devonian tetrapods. Elpistostege Westoll 1938 has proved

12 456 RICHARD CLOUTIER AND PER ERIK AHLBERG to be a tetrapod-like fish (see following discussion), while Ichthyostegopsis Sâve-Siiderbergh 1932 appears to be synonymous with Ichthyostega. Ichthyostega, Acanthostega, Ventastega, Metaxygnathus, Hynerpeton, Elginerpeton, and Obruchevichthys retain primitive characters not seen in any later tetrapods. Ichthyostega, Acanthostega, and Ventastega share certain features such as a spade-shaped head, dorsally placed orbits, external nostril close to the jaw margin, reduction or loss of the lateral rostral bone, a closed palate with a mobile basal articulation, and entopterygoids which meet anteriorly in a midiine point between the vomers (Jarvik, 1980; Ahlberg et al., 1994; Clack, 1994a). The braincases of Acanthostega and Ichthyostega both show a basicranial fissure and some development of a cranial notochord. Ichthyostega, Acanthostega, and Tulerpeton possess more than five digits (Coates and Clack, 1990; Lebedev, 1984). Their humeri, like those of later tetrapods, are structurally comparable to those of osteolepiforms and rhizodonts but have a distinctive L shape (Andrews and Westoll, 1970a,b; Rackoff, 1980; Panchen, 1985; Panchen and Smithson, 1987; Ahlberg, 1989). A late Frasnian humerus from Scat Craig, associated with Elginerpeton (Ahlberg, 1991a; Ahlberg and Milner, 1994), is morphologically intermediate between those of Famennian tetrapods and osteolepiforms. Ichthyostega and Acanthostega lack scapular blades but have well-developed cleithra. Their vertebral columns are broadly similar to that of Eusthenopteron but have weakly developed zygapophyses (Andrews and Westoll, 1970a; Jarvik, 1980; Coates, 1995). Both genera have ribcages; the ribs of Ichthyostega are extremely broad, overlapping structures (Jarvik, 1980). D. Onychodontida The Onychodontida (=Onychodontiformes, Struniiformes) is known only from three genera (Grossius, Onychodus, and Strunius) ranging from the Pragian (Zhu and Janvier, 1994) to the Famennian (Schultze, 1993). Zhu and Janvier (1994) described a lower jaw from the Posongchong Formation of China as the oldest known onychodontid. Grossius is known from a single three-dimensional skull from the Frasnian of Spain (Schultze, 1973). Onychodus is represented primarily by parasymphysial tooth spirals and isolated bones (Jessen, 1966) but also by well-preserved articulated material from the lower Frasnian Gogo Formation of Western Australia (Andrews, 1973; Long, 1991). Jessen (1966) described two species of Strunius (S. walteri and S. rolandi) from the Frasnian of Germany (Upper Plattenkalk, Bergisch Gladbach) which at the moment remain the best described members of the group. Aquesbi (1988) described an onychodontid from Morocco represented by a single poorly preserved specimen. Material of Onychodus sp. from the Gogo Formation is being studied by S. M. Andrews (National Museums of Scotland, Edinburgh). The interrelationships among onychodonts have never been examined because of the lack of comparative material. The monophyly of the Onychodontida has never been addressed in detail. However, the presence of spiral parasymphysial teeth located dorsal to the dentaries has been suggested as a synapomorphy of the group (Jessen, 1966; Schultze, 1969, 1973). Aquesbi (1988) mentioned that the Onychodontida is characterized by the following: (1) a double series of long sigmoid parasymphysial teeth with striated or crenulated enamel, (2) a large infradentary bordering the dentary ventrally, (3) the absence of an interdavicle, and (4) a reduced opercular series. However, the Gogo Onychodus material contradicts character 2 (P. E. Ahlberg, personal observation). Smith (1989) proposed the organization of the enamel crystallites of the teeth into fine ribs with a superficial chevron pattern as an onychodontid synapomorphy. The skull combines characters similar to actinopterygians (e.g., well-developed dorsal process on the maxilla and large preoperculum oriented horizontally) and typical sarcopterygian features such as an intracranial joint. It seems likely that the "actinopterygianlike" characters are actually plesiomorphic osteichthyan traits. The sole family, Onychodontidae Woodward 1891, is coextensive with the Onychodontida. E. Porolepiformes The Porolepiformes are an exclusively Devonian group. The earliest known representatives are several species of Porolepis from the Pragian (= Siegenian; Harland et al., 1990) of the Rhineland and Spitsbergen (Schultze, 1993); the latest is Holoptychius sp. from the latest Famennian of central Russia (Alexeev et al., 1994), eastern Greenland (Bendix-Almgreen, 1976), and elsewhere. Schultze (1993) daims a Tournaisian record for Holoptychius on the basis of its occurrence in the Groenlandaspis Series of eastern Greenland. However, the attribution of this Series to the Carboniferous is questionable. Porolepiforms are absent from the Tournaisian of central Russia (Alexeev et al., 1994), North America, and Britain. Jarvik (1942) was the first worker to recognize that Porolepis, then the only known member of the family Porolepididae, shares many characters with Holoptychiidae such as Holoptychius and Glyptolepis. He united the Porolepididae and Holoptychiidae in the

13 17. Interrelationships of Basal Sarcopterygians 457 order Porolepiformes Berg (1937). The monophyly of this group has been accepted by ail subsequent authors except Maisey (1986a), who interpreted the porolepiforms as a paraphyletic assemblage of stem lungfishes. However, this interpretation was largely based on the characteristics of Powichthys and Youngolepis, which fall outside the Porolepiformes sensu Jarvik (1942). We define the Porolepiformes as a clade characterized by the possession of dendrodont teeth (Schultze, 1969; Panchen and Smithson, 1987), subsquamosals (Cloutier, 1990; Ahlberg, 1991b; Cloutier and Schultze, 1996), and a unique skull roof pattern in which the intertemporal and supratemporal are absent and the postotic sensory canal passes through the growth center of the postparietal bone (Ahlberg, 1992c). The clade thus defined is equivalent to Porolepiformes of Jarvik (1942), although the diagnostic characters are different. The porolepiforms are not a diverse group. We recognize eight genera, two in the Porolepididae and six in the Holoptychiidae: "Porolepididae" Berg 1940 is a paraphyletic group defined by the possession of cosmine. The best known genus is Porolepis Woodward 1891, which ranges in age from Pragian to Givetian (Schultze, 1993). It is unclear whether Porolepis is a clade or simply a paraphyletic assemblage of primitive porolepiforms. Heimenia Orvig 1969, also has cosmine, but the scale morphology is intermediate between those of Porolepis and the Holoptychiidae (Orvig, 1969). Only scales and a single lower jaw of Heimenia (Jarvik, 1972, pl. 12-6) have been figured or described to date. Holoptychiidae Owen 1860 is a clade defined by the possession of round scales, lack of cosmine, lack of median gular plate, and a relatively short ethmosphenoid cranial division (Ahlberg, 1992c). Glyptolepis Miller ex Agassiz 1841, ranges from the Eifelian to the early Frasnian (Lyarskaya, 1981). As traditionally defined, this genus may be paraphyletic with respect to other holoptychiids (Ahlberg, 1992c). Quebecius Schultze 1973, from the middle Frasnian of Miguasha, Québec (Cloutier et al., 1996), resembles Glyptolepis but is distinguished by a unique cheekplate pattern (Cloutier and Schultze, 1996). The early Frasnian genus Laccognathus Cross 1941, is defined by an autapomorphic dermal ornament composed of large tubercles with thick enamel (Orvig, 1957) and by the possession of very large infradentary foramina (Cross, 1941; Ahlberg, 1992c). Holoptychius Agassiz in Murchison 1839, ranges in time from the middle Frasnian (Jarvik, 1972; Cloutier et al., 1996) fo the end of the Devonian (Alexeev et al., 1994; see previous discussion). It has dermal ornament composed of laminar bone rather than dentine (Orvig, 1957). Duffichthys Ahlberg 1992c, is represented by autapomorphic lower jaws from the upper Frasnian of Scat Craig, Scotland. The Middle Devonian genus Hamodus Obruchev 1933 is only known from isolated, very large dendrodont teeth with barbed tips. A further holoptychiid genus, Paraglyptolepis Vorobyeva 1987, has been described from the Givetian of Estonia. However, on basis of the available material it is flot clear that this genus can be distinguished from Glyptolepis. In the Early and Middle Devonian, porolepiforms tend to be the largest predators in the faunas where they occur (Ahlberg, 1992b). Maximum size for the group seems to be close to 2m. They seem to have been the first sarcopterygian group to evolve elaborate branched lateral fine systems; their cranial sensory canals have numerous first- and second-order side branches, which cover almost the whole skull surface except the operculogular series. The gross morphology of the holoptychiids is extremely stereotyped (Ahlberg, 1992b). On the whole, porolepiform cranial anatomy is fairly similar to that of osteolepiforms. This is particularly true for the intracranial joint, which runs through the profundus foramen in both groups. However, the ethmosphenoid braincase block and lower jaw are much doser to those of Powichthys and Youngolepis (Jessen, 1980; Chang, 1982, 1991a; Ahlberg, 1991b). The basibranchial skeleton lacks a sublingual rod, unlike that of osteolepiforms (Jarvik, 1972). The vertebral column is lungfish-like, as are the archipterygial pectoral fins (Ahlberg, 1989, 1991b). However, the pelvic fins have asymmetrical endoskeletons of a more generalized sarcopterygian pattern (Ahlberg, 1989). The spread of anatomical information among the Porolepiformes is patchy. The postcranial endoskeleton has only been described from Glyptole pis, although personal observation (by P.E. Ahlberg) of Laccognathus specimens in the care of Emilia Vorobyeva shows a very similar vertebral column and fin supports. F. Powichthys and Youngolepis These two genera from the Early Devonian (Lochkovian Pragian) of Arctic Canada (Powichthys) and South China and Vietnam (Youngolepis) show affinities with both porolepiforms and lungfishes. They are known mostly from cranial remains, although the shoulder girdle of Youngolepis was described by Chang

14 458 RICHARD CLOUTIER AND PER ERIK AHLBERG (1991a) and a cleithrum associated with Powichthys was figured by Jessen (1980). Powichthys Jessen 1975, from Prince of Wales Island in the Canadian Arctic, is the more porolepiform-like of the two and was referred to the Porolepiformes by its discoverer (Jessen, 1975, 1980). However, it lacks the derived porolepiform skull roof pattern and has polyplocodont rather than dendrodont tooth folding. Powichthys resembles the porolepiforms most closely in the structure of its ethmosphenoid. It has a pair of well-developed internasal pits between the vomers, and there is a large profundus foramen in the postnasal wall. However, the snout also contains rostral tubuli like those in lungfishes (Jessen, 1980). An opercular series associated with the genus (but not formally attributed to it; Jessen, 1980) appears to have contained a preoperculosubmandibular bone similar to that of porolepiforms. The lower jaw (again not formally attributed, but very probably belonging to Powichthys) has a porolepiform-like parasymphysial tooth plate attachment and three infradentary foramina similar to those of Holoptychius and Laccognathus. However, the immobilized intracranial joint lies at the level of the trigeminal and lateral ophthalmic nerves, as in coelacanths. Youngolepis Chang 1982, from Yunnan, China (Chang, 1982, 1991b) and Vietnam (Tong-Dzuy Thanh and Janvier, 1990, 1994), has an extraordinary braincase which combines porolepiform- and lungfish-like features with apparent actinopterygian characteristics. The latter are most obvious around the posterior part of the braincase floor. There is a basicranial fissure rather than a fenestra. A "basicranial process" from the lateral commissure reaches forward toward a "processus descendens" from the sphenoid (Chang, 1982) just as in Mitnia (Gardiner, 1984). The sphenoid is pierced by separate foramina for the carotid and efferent pseudobranchial arteries. Although these features are otherwise known only from actinopterygian s, they are probably to be interpreted as plesiomorphic osteichthyan characters (Ahlberg, 1994). In Youngolepis their co-occurrence with an unconstricted cranial notochord, and an apparent remnant of the intracranial joint in the side wall of the braincase (Chang, 1982), raises the possibility that they are reversals from a fully developed intracranial joint. Other parts of the anatomy of Youngolepis seem to show a mixture of porolepiform, lungfish, and general sarcopterygian characters. The lower jaw resembles that associated with Powichthys and has infradentary foramina, the snout contains rostral tubuli as in Powichthys and lungfishes, the cheekplate is broadly osteolepiform-iike with extensive squamosalmaxillary contact, and the shoulder girdle has a flattened but essentially tripodal scapulocoracoid. G. Osteolepiformes The Osteolepiformes (Berg, 1937; Jarvik, 1942) is by far the most diverse of the extinct sarcopterygian groups though much less diverse than the Actinistia, Dipnoi, or Tetrapoda. Approximately 60 species from some 25 genera have been described, ranging in age from Middle Devonian (Eifelian) to Lower Permian (Sakmarian). Among them is Eusthenopteron foordi, the most thoroughly studied fossil sarcopterygian and one of the best known of all fossil vertebrates (Whiteaves, 1883, 1889; Goodrich, 1902; Jarvik, 1937, 1942, 1944a,b, 1954, 1963, 1980; Andrews and Westoll, 1970a). Yet for all this our anatomical knowledge of the osteolepiforms remains patchy; Eusthenopteron stands out against a host of incomplete and poorly understood genera. The overall impression given by the osteolepiforms is of a rather homogenous group of similar-looking fishes. However, this homogeneity does not necessarily imply monophyly; it is possible that the group Osteolepiformes is paraphyletic relative to the Rhizodontida, Elpistostegalia + Tetrapoda, or both. The cranial anatomy is best illustrated by Eusthenopteron, although comparable information on the neurocranium is available from Megalichthys (Romer, 1937), Gogonasus (Long, 1988a), and Medoevia (Lebedev, 1995). The dermal skull bones are well known in the Scottish Middle Devonian genera Osteolepis, Thursius, and Gyroptychius (Jarvik, 1948). The braincase is divided by an intracranial joint running through the foramen for the profundus nerve. There is only one external nostril on each side of the head, buta large palatal opening surrounded by vomer, dermopalatine, maxilla, and premaxilla appears to have transmitted a choana (Jarvik, 1942; Panchen and Smithson, 1987). This interpretation was challenged by Rosen et al. (1981), who tried to show that the size of the opening had been exaggerated by Jarvik. However, new evidence from acid-prepared specimens (Long, 1988a; Lebedev, 1995; P. E. Ahlberg, personal observation) corroborates Jarvik's description and shows that the opening bears a very close resemblance to the choanae of Devonian tetrapods (Jarvik, 1980; Clack, 1994a). The otoccipital braincase block broadly resembles that of actinistians but retains lateral otic fissures which end in large vestibular fontanelles. Dermal bone characteristics of the group include a large squamosal which separates the rather narrow preopercular from the maxilla. Cosmine is primitively

15 17. Interrelationships of Basal Sarcopterygians 459 present in osteolepiforms but has been lost in many genera. Vertebrae are either rhachitomous or ringcentra (Andrews and Westoll, 1970a,b). Ribs, if present, are short. The paired fin skeletons are short, uniserial metapterygia, and ail fin radiais are unjointed and unbranched (Andrews and Westoll, 1970a,b). The humerus is structurally very similar to that of basal tetrapods, although the actual shape is rather different (Ahlberg, 1991a,b). In the majority of osteolepiforms the paired and median fin bases carry enlarged scales, the so-called basal scutes. The classification of the Osteolepiformes is in urgent need of revision. Schultze (1993) divides them into the following families: "Osteolepididae" Cope 1889 is a paraphyletic group of primitive osteolepiforms. A typical representative is Thursius Sedgwick and Murchison Within this group, the megalichthyids can be recognized as a clade on the basis of several cranial characters (Young et al., 1992). The osteolepidids range in time from Devonian (Eifelian) to Permian (Sakmarian). Canowindridae Young et al is a clade characterized by a posteriorly broad postparietal shield, lateral extrascapulars which almost meet in the midline anteriorly, and exclusion of the postorbital from the orbital margin. Three canowindrid genera are known: Canowindra Thomson 1973, Beelarongia Long 1987, and Koharale pis Young et al At present the group appears restricted to the Upper Devonian of Australia and Antarctica. Tristichopteridae Cope 1889 (=Eusthenopteridae Berg 1940) is a clade characterized by the presence of posttemporal bones between the operculars and lateral extrascapulars. Tristichopterids also lack cosmine, have thin round scales with a median ridge on the inner surface, and possess a characteristic triphycercal caudal fin. The earliest known tristichopterid is Tristichopterus (Andrews and Westoll, 1970b) from the upper Givetian John O'Groats Sandstone, Scotland, while the latest is Eus thenodon from the upper Famennian Remigolepis Series of eastern Greenland (Jarvik, 1952). Marsdenichthys from the Frasnian of Australia is held by Long (1985) to be the most primitive known member of the group. Eusthenopteron is the only tristichopterid to have been studied in great detail, and relationships within the group remain obscure. Rhizodopsidae Berg 1940 is a group ranging from the Carboniferous to Permian. The best known genus is Rhizodopsis (Andrews and Westoll, 1970b; Moy-Thomas and Miles, 1971). H. Rhizodontida The Rhizodontida (Andrews and Westoll, 1970b) are a group of Devonian and Carboniferous fishes chiefly remarkable for their great size. The Scottish Lower Carboniferous form Rhizodus hibberti seems to have reached a length of 7 m (Andrews, 1985). Most known rhizodont specimens consist of partly or wholly disarticulated material. Only one complete individual, a juvenile of?strepsodus anculonamensis, has been described; it has an elongate body with small median fins clustered near the diamond-shaped symmetrical tai (Andrews, 1985, fig. 2; this paper, Fig. 2g). Despite our incomplete knowledge of rhizodont anatomy, the Rhizodontida can unambiguously be recognized as a clade. Synapomorphies of the group include robust lepidotrichia with extremely long unjointed proximal portions, and the presence on the cleithrum of a depressed posterior flange and an elaborate double overlap area for the clavicle (Andrews and Westoll, 1970b; Andrews, 1985; Long, 1989; Young et al., 1992). The humerus has a bulbous head which forms a bail-and-socket joint with the round glenoid, and the radiais of the pectoral fin skeleton are both jointed and branched (Andrews and Westoll, 1970b). Cosmine is always absent and the scales are round and thin. The best understood part of rhizodont anatomy is the pectoral girdle. Parts of the dermal skull have been described from?strepsodus and from Screbinodus (Andrews, 1985) and Barameda (Long, 1989). The bone pattern is broadly similar to that of osteolepiforms, but Barameda (the most complete rhizodont) shows some unusual characters such as an extratemporal which contacts the supratemporal ("intertemporal" of Long, 1989) and a reduced postrostral mosaic. Long (1989) interpreted Barameda as having two external nostrils on each side. However, the reconstructed pattern conflicts with evidence from a detached Strepsodus premaxilla (BMNH P364(2); P. E. Ahlberg, personal observation) which shows a continuous overlap area for the lateral rostral in the region where Long placed the anterior nostril. Andrews (1985) reconstructed an osteolepiform-like arrangement of bones in the narial region but felt uncertain whether one or two nostrils were present. It is also unclear whether rhizodonts possess a choana. The condition of the nasal region is important, as the supposed possession of two external nostrils was one of the main features

16 460 RICHARD CLOUTIER AND PER ERIK AHLBERG which led Long (1989), Ahlberg (1991b), and Young et al. (1992) to place rhizodonts below osteolepiforms in the tetrapod stem group. A number of braincase fragments have been described from the Antarctic genus Notorhizodon (Young et al., 1992). They include a well-developed intracranial joint and closely resemble the corresponding parts of Eusthenopteron; this also applies to the palatoquadrate of Notorhizodon (Young et al., 1992). The lower jaw, known from Notorhizodon and in part also from Barameda (Long, 1989) and Strepsodus (Andrews, 1985), resembles those of osteolepiforms and porolepiforms. It has a very strongly developed symphysial fang pair on the dentary. Probably the earliest known rhizodont is Notorhizodon from the "Middle Late Devonian" Aztec Siltstone of Antarctica (Young et al., 1992), while the last known representative is Strepsodus from the Westphalian Coal Measures of England (Schultze, 1993). The only analysis undertaken to date of rhizodont interrelationships was that of Young et al. (1992), which produced the following topology: [Notorhizodon + [Barameda + [Screbinodus + [Rhizodus + Strepsodus]]]]. This hypothesis is biogeographically interesting in that the two genera judged to be most primitive both derive from East Gondwana. I. Elpistostegalia This group is generally known as Panderichthyida Vorobyeva 1989, but Elpistostegalia Camp and Allison 1961 has priority (Schultze, 1996). The name of the single constituent family should likewise be Elpistostegidae Romer 1947, rather than Panderichthyidae Vorobyeva and Lyarskaya The group is of great phylogenetic interest because its members display a melange of tetrapod-like and osteolepiform-like characters. It is stratigraphically restricted and of low diversity; at present we recognize only two genera and three species. Elpistostege Westoll 1938 is represented by the single species Elpistostege watsoni from the middle Frasnian of Miguasha, Québec, Canada. Two incomplete skulls and a section of vertebral column are known (Schultze and Arsenault, 1985) of this youngest elpistostegid. Panderichthys rhombolepis Gross 1941 was originally described on the basis of incomplete lower jaws from the early Fra snian of Latvia. The discovery of several complete specimens at Lode quarry in Latvia (Lyarskaya and Mark-Kurik, 1972) resulted in a series of descriptive and interpretive papers (Vorobyeva, 1977, 1980, 1986, 1989; Vorobyeva and Tsessarskii, 1986; Vorobyeva and Schultze, 1991; Vorobyeva and Kuznetsov, 1992; Worobjewa, 1975) which have made this the best known elpistostegid. Panderichthys stolbovi Vorobyeva 1960 is a slightly younger, though still early Frasnian, species from Russia that is known from a snout fragment and some incomplete lower jaws (Vorobyeva, 1960, 1962, 1971). Panderichthys stolbovi appears very similar to P. rhombolepis, but the two can be distinguished by their slightly different dermal ornament (P. E. Ahlberg, personal observation). In addition to these bona fide elpistostegids, Panderichthys bystrovi Gross 1941 and Obruchevichthys gracilis Vorobyeva 1977 have also been attributed to the group. The material of "Panderichthys" bystrovi cornes from the late Famennian locality of Ketleri in Latvia. The holotype, a mandibular fragment, needs to be redescribed; it certainly cornes from a sarcopterygian fish but does not appear to belong to an elpistostegid (P. E. Ahlberg, personal observation). The maxilla and premaxilla attributed to P. bystrovi by Vorobyeva (1962) actually belong to a tetrapod, Ventastega curonica (Ahlberg et al., 1994). Obruchevichthys, which is known only from two late Frasnian mandibular fragments, also appears to be a primitive tetrapod (Ahlberg, 1991a, 1995; Ahlberg et al., 1994; see below). Genuine elpistostegids are thus at present restricted to the early and middle Frasnian. Elpistostegid cranial anatomy resembles that of osteolepiforms in many respects. In Panderichthys rhombolepis and P. stolbovi (the two species where this region is known) the anterior end of the palate compares with that in Eusthenopteron: the vomers have welldeveloped posterior processes which suture to the sides of the parasphenoid and prevent the entopterygoids from meeting in the midline. This is quite different from the tetrapod pattern. The lower jaw likewise lacks obvious tetrapod characteristics (Ahlberg, 1991a) but resembles those of tristichopterids such as Eusthenodon (P. E. Ahlberg, personal observation) and Platycephalichthys (Vorobyeva, 1962) as well as that of the rhizodont Notorhizodon (Young et al., 1992). The bone and sensory line pattern around the external nostril matches that of Eusthenopteron (Vorobyeva and Schultze, 1991; Jarvik, 1980). The most obvious tetrapod-like structure in the elpistostegid skull is a pair of frontals anterior to the parietals (Westoll, 1938; Vorobyeva, 1977; Schultze and Arsenault, 1985; Vorobyeva and Schultze, 1991). It is worth noting in passing that the elpistostegid skull roof pattern furnishes powerful support for Westoll's (1938) terminology of dermal skull bones in osteichthyan fishes (Schultze and Arsenault, 1985;

17 17. Interrelationships of Basal Sarcopterygians 461 Ahlberg, 1991b). As in tetrapods, but unlike osteolepiforms, the parietals and postparietals of elpistostegids are immovably sutured together. The intracranial joint must thus have been immobile and was possibly obliterated altogether. Just as striking as these anatomical structures is the tetrapod-like morphology of the elpistostegid head. The skull is flattened and spade-shaped, the orbits are dorsal and crowned by bony "eyebrows," the interorbital skull roof is narrow and concave, and the external nostrils are almost marginal. These features must be related to the mode of life of the animais, and may indicate a shallow-water or marginal lifestyle (Ahlberg and Milner, 1994; Schultze, 1996). A similar situation obtains with respect to the postcranial skeleton; the straight tau, lack of separate dorsal and anal fins, and probably dorsoventrally flattened body of Panderichthys rhombolepis can ail be matched in Ichthyostega and Acanthostega and suggest that the elpistostegids may have had a capacity for terrestrial locomotion (Vorobyeva and Kuznetsov, 1992). The humerus and scapulocoracoid of P. rhombole pis combine tetrapod-like features with apparent autapomorphies (Vorobyeva and Schultze, 1991; Vorobyeva and Kuznetsov, 1992). Unfortunately these elements are unknown in the other two species. The vertebral column of P. rhombolepis has peculiar bladelike ribs which are sutured to the intercentra and neural arches. Pleurocentra are absent (Vorobyeva and Tsessarskii, 1986). Similar bladelike elements in a vertebral column attributed to Elpistostege are identified as neural arches by Schultze and Arsenault (1985), but these too appear to be ribs (R. Cloutier, personal observation). The combination of characters seen in the Elpistostegalia raises important phylogenetic questions. Their synapomorphies with tetrapods are striking and suggest that the two are sister-groups (Vorobyeva and Schultze, 1991; Cloutier, 1990; Ahlberg, 1991b; Ahlberg and Milner, 1994; but see Panchen and Smithson, 1987). Many of their plesiomorphic characters are general osteichthyan traits and thus unproblematical, but others are shared specifically with tristichopterids and rhizodonts and could pose a challenge to osteolepiform monophyly (see preceeding sections). Most interesting of ail are the characters that bear on the question of elpistostegid monophyly. Vorobyeva and Schultze (1991) propose five elpistostegid synapomorphies: (1) median rostral separated from premaxilla, (2) paired posterior postrostrals, (3) large median gular, (4) lateral recess in nasal capsule, and (5) subterminal mouth (= prominent snout). Character lis erroneous, as a comparable median rostral is developed in both osteolepiforms and basal tetrapods (Jarvik, 1980; Ahlberg, 1995). Characters 2, 3, and 4 are ail indeterminable in tetrapods (the postrostral mosaic and gular series have been lost altogether, while the nasal capsules are unossified and thus unknown in ail fossil tetrapods) and thus not really testable. This seems to leave a prominent snout as the only elpistostegid synapomorphy. Set against this character are a couple of features (intertemporal present in Panderichthys rhombolepis but absent in Elpistostege and known Devonian tetrapods; more tetrapod-like ornament in Elpistostege than in P. rhombolepis, P. E. Ahlberg, personal observation) which suggest the Elpistostegalia might be paraphyletic with respect to the tetrapods. At present our understanding of Elpistostege and P. stolbovi is too incomplete to allow the question of elpistostegid monophyly or paraphyly to be settled. However, the implications of the question are profound: if the elpistostegids are paraphyletic with respect to tetrapods, the many morphological details which give a common elpistostegid appearance to Elpistostege and P. rhombolepis (snout outline, shape and position of "eyebrows," etc.) will actually be attributes of the tetrapod stem lineage. A paraphyletic group Elpistostegalia would, in other words, provide much more detailed information about the earliest stages of tetrapod evolution than would an elpistostegid clade. The investigation of this area should be a priority for future sarcopterygian research programs. III. The Character Set The character set which we present is essentially a consensus list based on our earlier works (Ahlberg, 1989, 1991b; Cloutier, 1990), with the addition of a few characters from other sources (e.g., Chang and Smith, 1992). A total of 140 characters were combined (Appendix 1), and these include only those characters that can be recognized in early fossil sarcopterygians. The reasons for this approach are worth examining. The debate over the relative merits of fossil and recent data goes back two decades (Lovtrup, 1977; Patterson and Rosen, 1977; Patterson, 1981, 1982a,b; Rosen et al., 1981; Schoch, 1986; Doyle and Donoghue, 1986, 1987; Donoghue et al., 1989; Forey, 1987; Panchen and Smithson, 1987; Schultze, 1987, 1994; Campbell and Barwick, 1987, 1988; Gauthier et al., 1988; Huelsenbeck, 1991). It has often been clouded by conflation with the separate issue of whether fossils can be taken to represent actual ancestors of Recent forms. However, the view that fossil and recent organisms should ail be treated as terminal taxa in a phylogenetic analysis has gradually gained near-universal acceptartce.

18 462 RICHARD CLOUTIER AND PER ERIK AHLBERG There are two main schools of thought about the treatment of fossil data in a cladistic analysis. One argues that the cladogram should be constructed on the basis of character distributions among the living taxa and that fossils should only then be mapped onto the topology; the fossils are thus not allowed to modify the topology. This approach was applied by Patterson and Rosen (1977) to teleosts, and by Forey (1987) with certain reservations to sarcopterygians. The other school rejects this division and uses both fossil and Recent data in the initial analysis. Most paleontologists appear to fall into the latter school (Schultze, 1987; Panchen and Smithson, 1987; Vorobyeva and Schultze, 1991; Chang, 1991a,b). We follow Gauthier et al. (1988) in rejecting a priori primacy for the characters of Recent taxa. In some cases, the character combinations displayed by sequences of plesions manifestly have the capacity to overturn phylogenetic judgements based on living taxa alone, and we can see no justification for artificially preventing this outcome. White it is generally possible to get better and more de tailed anatomical information from Recent taxa than from fossils, this is not equally true for all characters. Features of adult skeletal morphology are often just as well understood in well-preserved fossils as in Recent organisms. The real disadvantage of fossils lies in the complete absence of certain kinds of data such as physiology, development, and in most cases soft anatorny and gene sequences. We designate these as "neontological" characters and use the term "paleontological" for such characters as can be detected in both living and fossil organisms. In practice, all character sets are affected by at least one of three types of problems which limit their usefulness in phylogenetic analysis. These are the following: Incomplete distribution: the characters are not known in all of the relevant taxa owing to anatomical incompleteness of the organisms. Poor understanding or definition of characters: the characters have limited "information content," making it difficult to distinguish homology from homoplasy. (3) Low number of characters: the character set is too small to provide adequate support for all nodes. Purely neontological data sets are usually not much affected by problem 3 but will suffer from 1 in direct proportion to the number of fossil taxa involved in the phylogenetic analysis. Paleontological data sets are often prone to problems 2 and 3. They are also affected by 1, in so far as fossil taxa are often incomplete. However, whereas "neontological" characters are typically known in all living taxa and unknown in all fossil ones, the gaps in a paleontological data set are determined by the preservation of different fossils and are likely to be more randomly distributed across the range of taxa. The Sarcopterygii includes three crown groups, namely the Tetrapoda (amniotes + lissamphibians), the Dipnoi (Neoceratodus, Protopterus, and Lepidosiren), and the Actinistia (Latimeria). The tetrapod and dipnoan crown groups date back approximately to the basal Carboniferous and the Lower Triassic, respectively (Ahlberg and Milner, 1994; Schultze and Marshall, 1993), whereas the monospecific actinistian crown group has no fossil record at all. However, each of these crown groups is associated with a recognized stem group which reaches back to the Devonian; there is no disagreement that Acanthostega and Ichthyostega are stem tetrapods, Dipnorhynchus and Uranolophus are stem dipnoans, and Diplocercides and Euporosteus are stem actinistians. Alongside the long-lived clades we find exclusively Paleozoic groups such as onychodonts, osteolepiforms, elpistostegids, porolepiforms, and rhizodonts. Some of these appear to be clades, but others may be paraphyletic taxa (Rosen et al., 1981; Young et al., 1992). We also have the more isolated Early Devonian genera Powichthys, Youngolepis, and Kenichthys (Jessen, 1975, 1980; Chang, 1982, 1991a,b; Chang and Smith, 1992; Chang and Zhu, 1993). The main phylogenetic uncertainties revolve around the relationships between the long-lived clades and various extinct groups; all the most debated phylogenetic nodes lie in or below the Devonian. Furthermore, outgroup-based phylogenetic analyses of the long-lived clades (Cloutier, 1991a,b; Forey, 1991; Schultze et al., 1993; Ahlberg and Milner, 1994; Lebedev and Coates, 1995) indicate that the most plesiomorphic and phylogenetically basal members of each clade are Devonian fossil genera. Fossil taxa thus occupy crucial positions in the analysis, and it is clear that paleontological characters will be very important for sorting out their relationships. However, we have decided to go one step further in omitting neontological data from the analysis altogether. The overall neontological data set divides naturally into morphological information and molecular data. Molecular phylogenetics is a relatively new field, but a number of workers have already tackled the threetaxon problem of lungfishes, actinistians and tetrapods (Meyer and Wilson, 1990, 1991; Gorr et al., 1991; Stock et al., 1991; Stock and Swofford, 1991; Sharp et al., 1991; Normark et al., 1991; Hedges et al., 1993). No consensus view has yet emerged from this research, and there are significant methodological disagreements within the field, although the bulk of the

19 17. Interrelat onships of Basal Sarcopterygians 463 molecular evidence seems to support a lungfishtetrapod sister-group relationship (Meyer, 1995). As a detailed discussion of the role of molecular data is given by Marshall and Schultze (1992) and Schultze (1994), we will flot give any further consideration to molecular data in this paper. In recent years the most important investigations of neontological anatomy have been those of Fritzsch (1987, 1988, 1992), Trueb and Cloutier (1991), and Northcutt and Bemis (1993). Fritzsch described some possible coelacanth-tetrapod synapomorphies from the structure of the inner ear, as did Northcutt and Bemis (1993) who focused on neurological characters; Trueb and Cloutier favored the topology [Actinistia + [Dipnoi + Tetrapodaff Our main reason for flot using these data is a wish to focus attention on the flood of new paleontological information which has become available during the past two decades (see Historical Background). A subsidiary consideration is the distribution of the data. Because neontological characters are unknown in ail the fossil taxa and can only support a few of the many nodes in the phylogenetic reconstruction, the use of large numbers of such characters seems likely to affect the analysis in unpredictable ways. Under the circumstances, we prefer to focus our present analysis entirely on the paleontological data set (Appendix 2). In effect, we want to see whether the "basal radiation" of sarcopterygians i.e., the short-lived groups and the early representatives of the surviving clades contains any obvious phylogenetic pattern. The results can then be compared with neontological (both molecular and morphological) and "total evidence" phylogenies for the Sarcopterygii in order to map out areas of agreement and disagreement. IV. Discussion The data matrix includes 140 characters and a total of 158 apomorphic character-states (Appendix 3). Appendix 1 provides the complete list of characters and their respective character-states. All characters were entered unordered and unweighted. All but one taxon (the actinopterygian Polypterus) are extinct ranging in lime from the Lower Devonian to the Upper Carboniferous (Appendix 2). Most of the data matrix (Appendix 3) was coded based on our respective observation of original material with the exception of Howqualepis (Long, 1988b), Speonesydrion (Campbell and Barwick, 1983, 1984), Dipnorhynchus (Campbell and Barwick, 1982a,b), Beelarongia (Long, 1987; Young et al., 1992), and Barameda (Long, 1989). Fifty-four most parsimonious trees at 277 steps were found using the heuristic search (C.I. = 0.578; C.I. excluding uninformative characters = 0.572; R.I. = 0.818) using the 140 characters coded for 32 taxa (including five outgroup taxa). The tree was rooted on a monophyletic outgroup including Polypterus, Cheirolepis, Mhnia, Moythornasia, and Hourqualepis. The Adams and strict consensus trees show the same topology (Fig. 3). Four topological variants were found: (1) among dipnoans, (2) at the base of the Tetrapodomorpha, (3) among osteolepiforms, and (4) among tetrapods. As the characters selected for the analysis were chosen for their potential to resolve relationships between (rather than within) acknowledged clades, the topological variation at variants (1) and (4) can be disregarded as unimportant. The monophyly of the Actinistia, Onychodontida, Dipnoiformes, Porolepiformes, Rhizodontida, Elpistostegalia, and Tetrapoda is corroborated (Table 1). However, the Osteolepiformes and Youngolepidida do flot appear to be monophyletic groups. The monophyly of most clades is robust (72% for the Elpistostegalia and Rhizodontida to 98% for the Actinistia and Dipnoi, based on 100 bootstrap replicates); a monophyletic Osteolepiformes has been replicated only 34%. The Actinistia is the sister-group of the remaining sarcopterygians (i.e [Onychodontida + Rhipidisfiai). Two clades constitute the Rhipidistia: (1) the Dipnomorpha including the Dipnoiformes and Porolepiformes and (2) the Tetrapodomorpha. In contrast to the conclusions of Chang and Smith (1992), Powichthys and Youngolepis do flot form a monophyletic group but rather consecutive plesions in the stem group of the Dipnoi; this pattern had been postulated by Ahlberg (1991b). Thus we consider the Dipnoiformes to include Powichthys, Youngolepis, and Dipnoi. There is no evidence for the monophyly of the Youngolepididae (Gardiner, 1984), Youngolepiformes (Chang and Smith, 1992), and Youngolepidida (Young et al., 1992). The dipnoan Diabolepis is considered to be the sister-group of the remaining Dipnoi as suggested by Cloutier (1990) and Chang (1991b). The Rhizodontida is the sister-group of the Osteolepidida as suggested by numerous authors (Cloutier, 1990; Ahlberg, 1991b; Vorobyeva and Schultze, 1991; Young et al., 1992). The interrelationships within the Osteolepidida are as follow: rosteolepiformes" + [Elpistostegalia + Tetrapodall. As first suggested by Schultze and Arsenault (1985), the Elpistostegalia is the sister-group of the Tetrapoda. However, in terms of extant organisms, the Dipnoi is the Recent sister-group of the Tetrapoda as suggested by Rosen et al. (1981). The distribution of characters is given for the major nodes concerning interrelationships among sarco-

20 464 RICHARD CLOUTIER AND PER ERIK AHLBERG.:c CC 0 sk.ch C).\' Ç.'.\ CC e cr, o c.,,:,.... so \ e c,. cc )., &,,,, 4', e c, c.,, Q, c, :-.' Q' t '. '.. '0 0 't, e>.e rt» (t) e.... se CD n,. q:,>.(1.. e'd'('''...! ''.' 0 e, Q9.'bre'e \ n,', C') e et, o z 0 rt, CD 0 \ O.\ e: () ) c, Ack. o p. c. çs 0 oo, ` e?:,,,, \,, (>,... rb, b cc \ c <2 k',:k.'c'l"- Q c9,0 e c, g. c,.5, g g" z.. '..: 0 e n, ( q> 0 k c, 0 k 0 u q. re.. ''.7) cc J,t, cc c2..' 0 'C 2 c C,.n`)`( e r'' (C 0 Z n ', )'', P., c) <zz., c, n sc,, r.5 n,. c, qy c,.e,,,.., >.,,.,,, ('', 0) (:), a\.,,,e.;,,,,,.2) b 0 :b :.- q \.,J 'o..) Q. o ck,, c; «z,e D ç.î,,; cs,..ç. Q., 0,..,'..,\ c ).. e, ' ç, ç (,, d<,,,,. (z,'' '.çs,.g, <>, 0 c, '' c., \z. O 'ZY 0 \c.)(`), 0.z;.. e; o, ' c., o. c'), 0 'es* ô N 0, `.O 'z' b '"`\'\ e..,. )'`;s7' c) 0,,ċ <tr \ O, O 0 o '.,- 0, o c) c 05,0 o c 0 0..,, et,, 4, e, o <., 2...Ç'c, ce,,p.,s \ '\. \. N ''' 0. 0, -\ \,.,/,,t, 0. Q. 0 (t \ '` )\-C.) 'CZ(' è;" -\,o'z' ' \- j" ecm,` e`'.-1,0cc,,c c;" o o,a (<, <;(t c) c)r'' e e- 4, j L FIGURE 3 Interrelationships of 28 sarcopterygian taxa. Adams and strict consensus tree based on the 54 most parsimonious trees at 277 steps [C.I. = 0.578; R.I. = 0.818]. pterygian higher taxa (Fig. 4). Complete lists of character changes for these nodes are given in Table 1. Only the uniquely shared, derived characters common to the 54 most parsimonious trees are discussed in this section unless controversial anatomical structures are involved. A. Sarcopterygii Romer 1955 Although unquestioned, the monophyly of the Sarcopterygii was supported by more than 30 characters (Table 1). Some of the synapomorphies represent the presence of new structures [tectals (char. 42), the TABLE 1 Distribution of Characters Common to the 54 Most Parsimonious Trees for Major Sarcopterygian Clades Taxa Uniquely shared derived characters Reversais Homoplasies Sarcopterygii 4, 18, 42, 49, 52, 54, 63(2), 88, 93(1), , 110, 120, 128 [Onychodontida + Rhipidistia] 93(2) Rhipidistia 56 Dipnomorpha 2, 41, 43, 130 Dipnoiformes 10, 77, 100, 119(1) [Youngolepis + Dipnoi] 17, 45, 108 Tetrapodomorpha 44, 71, 119(2), 122 Osteolepidida [Elpistostegalia + Tetrapoda] 15, 38, 67, 127 Actinistia 76, 95, 97, 138 Onychodontida 41(2) Porolepiformes 14(2), 51, 55, 63, 75, 101 Dipnoi 30, 65, 80 Rhizodontida 114, 132(2) Elpistostegalia 33 Tetrapoda 24, 39, 60-61, 68, 125 3, 28, 29(2), 34, 37, 40(2), 74, 81-85, 89, 94, 96, , 121, 124, , 35, 48, 64, 79, 92, 103, , 66, , 21, 29, , 74, 121, 135, 137 1, 31(1), 59, 78, , 81, 86, 89-90, , 82-84, , 79, , , , 25, 47, 58, 135(2) 19, 53, 58, 87, 91, 117, , 36, 57(2) 29, 73 20, 36, 37(2), , 14, 35, 79, 85, 92, 111 9, 19, 31(2), 47, 58, , 66, 112(2) Note. The category "uniquely shared derived characters" lists ail characters with a C.I. equal to 1. The names of Taxa are those used in the text and in Fig. 4. See trees in Figs. 3 and 4. Appendix 1 provides the character descriptions.

21 17. Interrelationships of Basal Sarcopterygians 465 l>'r / 0 CD ' n''.'" * n'. '' Zr 0 n. -.', e e,,,-,.. e,,,,.,?,e e (Y e,r) e.9 & e,.. d e,z() «e e 4.-- e e Ô ' SARCOPTERYGII DI PNOIFORM ES D IPNOM OR PHA RHI PI DISTIA TETRAPODOMORPHA OSTEOLEPID IDA FIGURE 4 Interrelationships of sarcopterygian higher taxa. The topology corresponds to the phylogenetic tree illustrated in Fig. 3. See Table 1 and text for the distribution of characters. squamosal (char. 54), the splenial (char. 93(1)), the jugal canal (char. 106), the humerus (char. 120), and basal plates in the dorsal fin supports (char. 128)]. In addition, structural and topographical changes occurred in the skull in comparison to basal actinopterygians: the premaxilla do flot form part of the orbit (char. 18), more than four sclerotic plates compose the sclerotic ring (char. 49), the dermohyal is absent (char. 52), the hyomandibular has two proximal articular heads instead of one (char. 88), the preopercular canal does not end at the dorsal margin of the preopercular (char. 105), and the mandibular canal does not pass through the dentary (char. 110). Characters 18, 52, 105, 106, 110, and 128 are complementary to the synapomorphies already identified for the Sarcopterygii (see Forey, 1980; Rosen et al., 1981; Gardiner, 1984; Maisey, 1986a; Panchen and Smithson, 1987; Schultze, 1987; Cloutier, 1990; Ahlberg, 1991b). The presence of cosmine (char. 1) and submandibulars (char. 64; Gardiner, 1984; Schultze, 1987) are flot sarcopterygian synapomorphies because actinistians lack these structures. Numerous branchial and hyoid characters used by previous authors (Forey, 1980; Maisey, 1986a; Panchen and Smithson, 1987) were not included in our analysis owing to the lack of information in the taxa analyzed. B. [Onychodontida + Rhipidistia] Of the nine characters supporting this clade, only one is not subsequently transformed. In addition to the splenial and angular, the infradentary series includes the postsplenial and surangular [char. 93(2)1. Based on this topology, the absence of surangular in actinistians (Cloutier, 1991a,b, 1996a) cannot be considered a synapomorphy of the group. Of special interest at this node is the presence of character 48 palatal opening ("choana") surrounded by the premaxilla, maxilla, dermopalatine, and vomer. The condition is known in Eusthenopteron, Panderichthys, tetrapods, and holoptychiids (P. E. Ahlberg, personal observation); the palatal opening is absent in dipnoans. Because the condition is unknown in onychodontids, it is possible that this character supports only the Rhipidistia. Nevertheless, the distribution of this character is contradictorv to that hypothesized by various authors (Panchen and Smithson, 1987; Schultze, 1987, 1991). The interpretation of a palatal opening in porolepiforms agrees with the identification of a fenestra endochoanalis in Glyptole pis by Jarvik (1972) and Bjerring (1991). Thus, if one accepts this interpretation as suggested by the distribution of the characters, it follows that the posterior external nostril and the choana are nonhomologous since onychodonts and porolepiforms have two external nares. Schultze (1987, 1991) and Chang (1991b) argued that porolepiforms lack a true choana because the palatal part of the fenestra ventrolateralis is covered by the vomer (and possibly the dermopalatines). However, Section 62 of Jarvik's grinding series of Glyptolepis groenlandica (Jarvik, 1972: fig. 8C) shows on both sides of the snout a palatal opening which communicates unambiguously with the posterior external nostril and the nasal cavity. Furthermore, a specimen of Holoptychius from Dura Den (Scotland) prepared by one of us (P. E. Ahlberg) shows a small round opening at the junction of the maxilla, premaxilla, dermopalatine, and vomer, which seems to correspond precisely with the choana reconstructed by Jarvik (1972) and Bjerring (1991) from the grinding series of Glyptolepis. This opening is similar in size to the external nostrils. Although this topic could benefit from further study, the currently available evidence thus supports Jarvik's and Bjerring's interpretation. C. Rhipidistia Cope 1871 Nine transformations corroborate the monophyly of the Rhipidistia (Table 1), of which only one is uniquely derived. In rhipidistians, the preopercular does not contact the maxilla (char. 56) because of a suture between the squamosal (or subsquamosals in porolepiforms) and the quadratojugal; in dipnoans in which the maxilla is absent, the preopercular does not reach the ventral margin of the cheek. In his overview of sarcopterygian tooth structure, Schultze (1970) identified three types of dentine folding (plicidentine) in order (1) to characterize certain sarcopterygian clades, (2) to suggest a close relationship between Osteolepiformes and Tetrapoda, and (3) to demonstrate the monophyly of the Tetrapoda. Among the three types, polyplocodont plicidentine

22 466 RICHARD CLOUTIER AND PER ERIK AHLBERG was said to be plesiomorphic, but no transformation series were inferred. Based on our tree, one has to consider that the dendrodont (in Porolepiformes) and eusthenodont types of folding (Schultze, 1970) evolved from a polyplocodont pattern. Thus the polyplocodont folding is not a synapomorphy of the clade [Rhizodontida + Osteolepidida] as suggested by Long (1989) and Young et al. (1992). D. Dipnomorpha Ahlberg 1991b The Dipnoi is closely related to the Porolepiformes (Maisey, 1986a; Ahlberg, 1989, 1991b; Cloutier, 1990, 1991a; Chang, 1991a,b; Chang and Smith, 1992) and not the sister-group of remaining sarcopterygians (Fig. ld; Schultze 1987, 1994) nor that of the Tetrapoda (Fig. lb; Rosen et al., 1981; Gardiner, 1984). In our analysis, the clade [Dipnoiformes + Porolepiformes] is supported by the following four characters: mesh canais of the cosmine pore-canal system without horizontal partition (char. 2), median extrascapular overlapping the lateral extrascapulars [char. 41(1)1, three or more tectals (char. 43), and presence of posterior branched radial complex associated with the posterior dorsal fin (char. 130). Ahlberg (1991b) was the first to propose a large suite of characters (17) to support the monophyly of the clade [Porolepiformes + [Powichthys + [Youngolepis + [Diabolepis + Dipnoi]]]] (Fig. le). Of the characters used by Ahlberg (1991b), only 9 were included in our analysis (our characters 31, 34, 59, 71, 79, 100, 123, 130, and 137); some were combined (e.g., char. 137) or simply deleted owing to the lack of morphological information. Only character 130 is fully congruent with Ahlberg's hypothesis. Because our analysis was performed at a lower taxonomie level, some of the characters considered as uniquely shared derived are homoplastic with respect to various taxa (e.g., character 123 is also present in the rhizodontid Bararneda). The absence of a contact between the supraorbital and the parietal (char. 34) was considered by Ahlberg (1991b) as a dipnomorph uniquely shared derived character. Recent studies of the basal actinopterygian Cheirolepis canadensis by Arratia and Cloutier (1996) show that a supraorbital is present in this species and that there is no contact between it and the parietal. Thus the polarity of the character is different than that of Ahlberg (1991b) as well as its distribution. The presence of preoperculosubmandibulars (char. 59) is frequently considered as a porolepiform synapomorphy (Jarvik, 1972; Vorobyeva and Schultze, 1991; Cloutier, 1990). Based on this analysis and Ahlberg (1991b), it is reinterpreted as a dipnomorph synapomorphy. It is likely that some of the 9-bones of dipnoans are homologous to the preoperculosubmandibular found in porolepiforms and Powichthys. E. Dipnoiformes Cloutier 1990 The Dipnoiformes is defined as the clade [Powichthys + [Youngolepis + Dipnoi]]. Powichthys and Youngolepis are consecutive plesions in the stem group of the Dipnoi; this pattern had been postulated by Ahlberg (1991b). The distribution of characters could be subject to further changes because ail the basal taxa of this clade (Powichthys, Youngolepis, Diabolepis, Dipnorhynchus, and Speonesydrion) are only known from incomplete specimens (mainly partial skulls). Characters 10, 77, 100, and 119 are congruently distributed; there are an additional seven characters. Chang and Smith (1992) and Chang and Zhu (1993) considered a broad marginal "tooth field" on the coronoids (char. 10) as a character shared by Youngolepis and Powichthys (also present in Kenichthys). Coronoids are absent in dipnoans. The presence of rostral tubuli is shared by Powichthys, Youngolepis, and basal dipnoans (char. 77). The infraorbital canal follows the dorsal margin of the premaxilla (char. 100); this condition might be related to the condition of character 17 at the following node. The proximal articular surface of the humerus is flat [char. 119(1)] rather than concave (plesiomorphic condition) or convex (Tetrapodomorpha synapomorphy). However, the distribution of character 119 could be an artifact of the selection of the taxa and the availability of information. Among basal dipnoiforms, the condition is only inferred in Youngolepis based on the condition of the glenoid fossa (Chang, 1991a). However, in advanced dipnoans the articular surface is concave (Schultze, 1987; Ahlberg, 1989). Within the Dipnoiformes, the clade [Youngolepis + Dipnoi] is supported by characters 17, 45, and 108 and five reversais (Table 1). The position of the premaxilla (char. 17) in Youngolepis is interpreted as a transitional state between a plesiomorphic condition in which it forms the anterior part of the upper maxillary arcade and the derived dipnoan condition where the premaxilla is absent or reduced to a small dentigenous part. The posterior naris is still external but located very close to the jaw margin [char. 45(1)]; this condition precedes the dipnoan one in which the posterior naris occupies a palatal position. On the lower jaw, the middle pit line developed into an enclosed oral canal or a structure of intermediate morphology (char. 108). F. Tetrapodomorpha Ahlberg 1991b The Tetrapodomorpha includes the Rhizodontida, Elpistostegalia, Tetrapoda, and the so-called osteo-

23 17. Interrelationships of Basal Sarcopterygians 467 lepiforms. Significant modifications involve the anatomy of the nasal region and anterior palate as well as the pectoral appendage. The Tetrapodomorpha shares a single external flans which corresponds to the anterior naris (char 44); howeyer, the condition is unknown in the Rhizodontida. The vomers articulate with each other medially (char. 71), and paired intervomerine pits are absent [char. 79(0)1. The anatomy of the humerus is modified: its proximal articular surface is convex [char. 119(2)1 and the deltoid and supinator processes are present (char. 122). Vorobyeva and Schultze (1991) refer to this clade as the Choanata; the main differences between their interpretation of the clade and ours concern the position of the Osteolepiformes and the distribution of some characters. They listed 21 characters of which 10 were used in our analysis (our characters 22-23, 41, 44, 48, 63, 66, 70, 119, and 140). The distribution of characters 44 and 119 are congruent in both hypotheses; however, Young et al. (1992) considered that the Rhizodontida lacks character 44. The presence of a palatal opening surrounded by the maxilla, premaxilla, vomer, and dermopalatine (char. 48) is interpreted in our analysis as a synapomorphy of the Rhipidistia because of its presence in holoptychiids. The median extrascapular is overlapped by the lateral extrascapulars (char. 41) flot only in members of this clade but also in actinistians. This relationship among the extrascapulars is considered to be plesiomorphic for the Sarcopterygii and is flot diagnostic for the Tetrapodomorpha as suggested by Jarvik (1980), Vorobyeva and Schultze (1991), and Young et al. (1992). Vorobyeva and Schultze (1991) define a character as "median gular always present" which they consider to be a synapomorphy of this clade; in our analysis, the presence of a median gular (char. 66) characterizes the rhipidistian node (and a large actinopterygian clade) and changes in holoptychiids and tetrapods. The polarity of character 66 is ambiguous because most basal actinopterygians possess a median gular with the exception of Polypterus. Based on our topology, the presence of a median gular is homoplastic with respect to actinopterygians and rhipidistians. Vorobyeva and Schultze (1991) also mention the presence of the posterior process on the vomer as a synapomorphy of this clade, although they note that the character is absent in some osteolepiforms. This character (char. 70) is in fact absent in the rhizodont Barameda (Long, 1989) as well as in the osteolepiform Medoevia (Lebedev, 1995) and Gogonasus (Long, 1988a), and the condition is unknown in Strepsodus and many osteolepiforms. The process is known to be present in tristichopterids, elpistostegids, and Crassigyrinus. Other characters listed by Vorobyeva and Schultze (1991), flot used in our analysis, deserve some comments. A separate median rostral is present in osteolepiforms, elpistostegids, and basal tetrapods (Jarvik, 1980; Ahlberg, 1995); thus a median rostral fused with the premaxilla is not a synapomorphy of this clade. A "long parasphenoid extending below oticooccipital region" is in fact restricted to crown-group Tetrapoda. The presence of seyen submandibulars is also found in porolepiforms (Cloutier and Schultze, 1996). G. Osteolepidida Boulenger 1901 The Osteolepidida are defined herein as the clade rosteolepiformes" + [Elpistostegalia + Tetrapoda]]. This "traditional" osteolepiform tetrapod relationship is defined in terms of shared derived characters and congruence of characters (Schultze, 1987; Panchen and Smithson, 1987; Long, 1989; Cloutier, 1990, 1991b; Ahlberg, 1991b; Vorobyeva and Schultze, 1991; Young et al., 1992; Ahlberg and Milner, 1994). This node is corroborated by character 70 and two reversals (chars. 113 and 124). The vomer has a distinctive posterior process (char. 70) which is lost in some basal tetrapods. (As mentioned above, this character is arguably primitively absent in some osteolepiforms and may thus in fact define a somewhat less inclusive clade.) The anocleithrum is exposed externally [char. 113(0)1; this character is considered as a reversai, although the plesiomorphic condition in actinopterygians deals with the postcleithrum. The mesomeres of the pectoral fin lack postaxial radiais [char. 124(0)1 in contrast to porolepiforms, dipnoans, and rhizodonts. In contrast to the hypotheses of Janvier (1980), Long (1985), and Vorobyeva and Schultze (1991), the monophyly of the Osteolepiformes is flot corroborated by our analysis; however, the paraphyly of the group is not demonstrated either. The monophyly of the osteolepiforms is jeopardized by the relative position of Eusthenopteron. Three equally parsimonious topologies have been obtained: Eusthenopteron is either the sister-group of (1) [Osteolepididae + Canowindridae], (2) [Elpistostegalia + Tetrapoda], or (3) [remaining osteolepiforms + [Elpistostegalia + Tetrapodall In topologies 2 and 3, the Osteolepiformes is paraphyletic, whereas in topology 1 the monophyly is demonstrated. In topology 1, the presence of a large median postrostral (char. 23) and basal scutes on the fins (char. 131) would be considered as osteolepiform synapomorphies. A close relationship with the clade [Elpistostegalia + Tetrapoda] is optimized by character 140 (presence of well-ossified ribs) and the reversai of character 35 (absence of extratemporal). In terms of anatomy, the third topology is less robust because the clade [[Osteolepididae + Canowindridae] + [Elpisto-

24 468 RICHARD CLOUTIER AND PER ERIK AHLBERG stegalia + Tetrapoda]] is supported by the reversai of two highly homoplastic characters: the presence of rhombic scales (char. 3) and the bilateral halves of the neural arch are separated (char. 136). H. [Elpistostegalia + Tetrapoda] The relationship between the Elpistostegalia and the Tetrapoda is well-corroborated. As mentioned by Schultze and Arsenault (1985) and Vorobyeva and Schultze (1991), the shape of the skull (char. 15) and the composition of the median series of skull roofing bones (char. 25) are diagnostic of this clade; the orbits are located dorsally, the interorbital distance is narrow and concave, and the skull as a whole is flattened. Paired frontals are present anterior to the parietals (char. 25); the superficially similar condition observed in Polypterus spp. is homoplastic with respect to this clade and these elements are not homologous to the frontals. The spiracle is present as a large, posteriorly open notch (char. 38) in the posterior part of the skull table enclosed between the tabular and the cheek bones. The unpaired fins (dorsal and anal fins) are lost (char. 127). The shape of the caudal fin is modified, the epichordal lepidotrichia being more developed than the hypochordal ones [char. 135(2)]. V. Conclusions The Sarcopterygii represents a well-diagnosed clade composed of seven unambiguous subclades (i.e., Actinistia, Onychodontida, Dipnoiformes, Porolepiformes, Rhizodontida, Elpistostegalia, and Tetrapoda) and one questionable taxon (i.e., Osteolepiformes). A cladistic analysis of 140 osteological characters yielded the following phylogenetic conclusions (congruence with published analyses indicated by references): The Actinistia is the sister-group of the remaining sarcopterygians (i.e., [Onychodontida + Rhipidistia]) (Panchen and Smithson, 1987). The Onychodontida is the sister-group to the Rhipidistia. The Dipnomorpha (including the Dipnoiformes and Porolepiformes) is the sister-group of the Tetrapodomorpha (Maisey, 1986a; Cloutier, 1990; Ahlberg, 1991b). The Dipnoiformes (including Powichthys, Youngolepis, and Dipnoi) is the sister-group of the Porolepiformes (Maisey, 1986a; Cloutier, 1990, 1991a; Ahlberg, 1991b; Chang, 1991b; Chang and Smith, 1992). The Rhizodontida is the sister-group of the Osteolepidida (Long, 1989; Cloutier, 1990; Ahlberg, 1991b; Vorobyeva and Schultze, 1991; Young et al., 1992). The interrelationships within the Osteolepidida are as follow: ["Osteolepiformes" + [Elpistostegalia + Tetrapoda]] (Schultze, 1987, 1991, 1994; Cloutier, 1990, 1991a; Ahlberg, 1991b; Young et al., 1992). The Elpistostegalia is the sister-group of the Tetrapoda (Schultze and Arsenault, 1985; Schultze, 1987, 1991, 1994; Cloutier, 1990; Ahlberg, 1991b; Vorobyeva and Schultze, 1991). In terms of extant organisms, the Dipnoi is the Recent sister-group of the Tetrapoda (Forey, 1980, 1987; Gardiner, 1980, 1984; Rosen et al., 1981; Maisey, 1986a; Cloutier, 1990, 1991a; Ahlberg, 1991b; Forey et al., 1991; Trueb and Cloutier, 1991). The classification proposed in this paper requires a reinterpretation of the diagnosis (characters) and definition (taxa) of already existing taxonomic categories. Available taxonomic names have been used and reassigned to clades that agree the closest to their original definition. We are not assigning Linnean rank to the taxonomic categories. Instead indentation signifies relative hierarchical rank. The classification of the Sarcopterygii based on this cladistic analysis is summarized as follows: Sarcopterygii Romer 1955 Actinistia Cope 1871 [Onychodontida + Rhipidistia] clade Onychodontidat Andrews 1973 Rhipidistia Cope 1887 Dipnomorpha Ahlberg 1991b Porolepiformes' Jarvik 1942 Dipnoiformes Cloutier 1990 Tetrapodomorpha Ahlberg 1991b Rhizodontida t Andrews and Westoll 1970b Osteolepidida Boulenger 1901 "Osteolepiformes t" Berg 1937 [Elpistostegalia + Tetrapoda] clade Elpistostegaliat Camp and Allison 1961 Tetrapoda Haworth 1825 VI. Summary Sarcopterygians are classified into three extant groups (i.e., Actinistia, Dipnoiformes, and Tetrapoda) and five extinct Paleozoic taxa (i.e., Onychodontida, Porolepiformes, Rhizodontida, Osteolepiformes, and Elpistostegalia); sarcopterygian fishes (excluding tetrapods) account for approximately 500 species belonging to approximately 160 genera. The diagnosis, taxo-

25 17. Interrelat onships of Basal Sarcopterygians 469 nomic content, stratigraphic range, evolutionary trends, and classification of the eight sarcopterygian higher clades are described. Fifty-four most parsimonious trees were found using 140 osteological characters (referred to as "the paleontological characters") coded for 27 sarcopterygian basal taxa. The character distribution is discussed for the most parsimonious sarcopterygian topology: [Actinistia + Forolepiformes + Dipnoiformes1 + [Rhizodontida + ["osteolepiforms" + [ Elpistostegalia + Tetrapoda The controversial Devonian genera Youngolepis and Pounchthys are included in the Dipnoiformes; the Dipnoiformes together with the Porolepiformes constitute the Dipnomorpha which is the sister-group of the Tetrapodomorpha. Although osteolepiforms are closely related to the clade [Elpistostegalia + Tetrapoda], their monophyly is flot corroborated. The Elpistostegalia is the sister-group of the Tetrapoda, whereas dipnoans are the living sister-group to the tetrapods. Acknowledgments It is a pleasure to have this opportunity to acknowledge Colin Patterson's great contributions to vertebrate paleontology and systematics. He has played a major part-through his own work and through his influence on others-in placing the study of sarcopterygian interrelationships on a rigorous cladistic basis and thus shaping the field in which we work. On a personal level, we have both had the pleasure of working in the rigorous intellectual climate which he has encouraged and maintained at the Natural History Museum, London, and we owe him more pints of beer than we can readily remember. Thank you, Colin. This paper draws on some 10 years' worth of accumulated work, and there are many people who have helped and influenced us in various ways during that time. We thank, in no particular order, Hans-Peter Schultze, Jenny Clack, Peter Forey, Brian Gardiner, Michael Coates, Erik Jarvik, Philippe Janvier, Mahala Andrews, Chang Mee-Mann, Gloria Arratia, Oleg Lebedev, John Long and Marius Arsenault, as well as many others whom we cannot list here. Referen ces Ahlberg, P. E. (1989). Paired fin skeletons and relationships of the fossil group Porolepiformes (Osteichthyes: Sarcopterygii). Zool. J. Linn. Soc. 96, Ahlberg, P. E. (1991a). Tetrapod or near-tetrapod remains for the Upper Devonian of Scotland. Nature (London) 354, Ahlberg, P. E. (1991b). A re-examination of sarcopterygian interrelationships, with special reference to the Porolepiformes. Zool. J. Lion. Soc. 103, Ahlberg, P. E. (1992a). Coelacanth fins and evolution. Nature (London) 358, 459. Ahlberg, P. E. (1992b). The palaeoecology and evolutionary history of the porolepiform sarcopterygians. In "Fossil Fishes as Living Animais" (E. Mark-Kurik, ed.), pp Academy of Sciences of Estonia, Tallinn. Ahlberg, P. E. (1992c). A new holoptychiid porolepiform fish from the Upper Frasnian of Elgin, Scotland. Palaeontology 35, Ahlberg, P. E. (1994). The intracranial joint in vertebrate phylogeny. J. Morphol. 220 (ICVM-4 Abstr.), 319. Ahlberg, P. E. (1995). Elginerpeton pancheni and the earliest tetrapod clade. Nature (London) 373, Ahlberg, P. E., and Milner, A. R. (1994). The origin and early diversification of tetrapods. Nature (London) 368, Ahlberg, P. E., and Trewin, N. H. (1995). The postcranial skeleton of the Middle Devonian lungfish Dipterus valenciennesi. Trans. R. Soc. Edinburgh: Earth Sci. 85, Ahlberg, P. E., Luksevics, E., and Lebedev, 0. (1994). The first tetrapod finds from the Devonian (Upper Famennian) of Latvia. Philos. Trans. R. Soc. London, Ser. B 343, Alexeev, A. A., Lebedev, O. A., Barskov, I. S., Barskova, M. I., Kononova, L. I., and Chizova, V. A. (1994). On the stratigraphic position of the Famennian and Tournaisian fossil vertebrate beds in Andreyevka, Tula Region, Central Russia. Proc. Geol. Assoc. 105, Andrews, S. M. (1973). Interrelationships of crossopterygians. In "Interrelationships of Fishes", (P. H. Greenwood, R. S. Miles, and C. Patterson, eds.), pp Academic Press, London. Andrews, S. M. (1985). Rhizodont crossopterygian fish from the Dinantian of Foulden, Berwickshire, Scotland, with a re-evaluation of this group. Trans. R. Soc. Edinburgh: Earth Sci. 76, Andrews, S. M., and Westoll, T. S. (1970a). The postcranial skeleton of Eusthenopteron foordi Whiteaves. Trans. R. Soc. Edinburgh 68(9), Andrews, S. M., and Westoll, T. S. (1970b). The postcranial skeleton of rhipidistian fishes excluding Eusthenopteron. Trans. R. Soc. Edinburgh 68(12), Aquesbi, N. (1988). Étude d'un Onychodontiforme (Osteichthyes, Sarcopterygii) du Dévonien moyen (Eifelien) du Maroc. Bull. Mus. Na!!. Hist. Na!. [4] 10, Arratia, G., and Cloutier, R., (1996). Reassessment of the morphology of Cheirolepis canadensis (Actinopterygii). In "Devonian Fishes and Plants from Miguasha, Quebec, Canada" (H.-P. Schultze and R. Cloutier, eds.), pp Verlag Dr. Fredrich Pfeil, Munich. Bemis, W. E. (1984). Paedomorphosis and the evolution of the Dipnoi. Paleobiology 10, Bendix-Almgreen, S. E. (1976). Palaeovertebrate faunas of Greenland. In "Geology of Greenland" (A. Escher and W. S. Watt, eds.), pp Geol. Surv. Greenland, Copenhagen. Berg, L. S. (1937). A classification of fish-like vertebrates. Bull. Acad. Sci. URSS, pp Berg, L. S. (1940). Classification of fishes, both recent and fossil. Tr. Zool, lus!., Akad. Nauk SSSR 5, Berg, L. S. (1955). Classification of fish-like vertebrates, living and fossil. 2., corrected and enlarg. Tr. Zool, lus!., Akad. Nauk SSSR 20, Berg, L. S. (1958). "System der rezenten und fossilen fischartigen und Fische." Deutcher Verlag der Wissenschaften. VEB, Berlin (translation by W. Gross of Berg, 1955). Berman, D. S (1968). Lungfish from the Lueders Formation (Lower Permian, Texas) and the Gnathorhiza-lepidosirenid ancestry questioned. J. Paleontol. 42, Bernacsek, G. M. (1977). A lungfish cranium from the Middle Devonian of the Yukon Territory, Canada. Palaeontographica, Abt. A 157, Bjerring, H. C. (1989). Apertures of craniate olfactory organs. Acta Zool. (Stockholm) 70, Bjerring, H. C. (1991). Some features of the olfactory organ in the Middle Devonian porolepiform Glyptolepis groenlandica. Palaeontographica, Abt. A 219, Bjerring, H. C. (1993). Yet another interpretation of the coelacanthi-

26 470 RICHARD CLOUTIER AND PER ERIK AHLBERG form basicranial muscle and its innervation. Acta Zool. (Stockholm) 74, Camp, C. L., and Allison, H. J. (1961). Bibliography of fossil vertebrates Mem. Geol. Soc. Am. 84, Campbell, K. S. W., and Barwick, R. E. (1982a). A new species of the lungfish Dipnorhynchus from New South Wales. Palaeontology 25, Campbell, K. S. W., and Barwick, R. E. (1982b). The neurocranium of the primitive dipnoan Dipnorhynchus sussmilchi (Etheridge). J. Vertebr. Paleontol. 2, Campbell, K. S. W., and Barwick, R. E. (1983). Early evolution of dipnoan dentitions and a new genus Speonesydrion. Mem. Assoc. Australas. Palaeontol. 1, Campbell, K. S. W., and Barwick, R. E. (1984). Speonesydrion, an Early Devonian dipnoan with primitive toothplates. Palaeo 'chthyologica 2, Campbell, K. S. W., and Barwick, R. E. (1987). Paleozoic lungfishes-a review. J. Morphol., Suppl. 1, Campbell, K. S. W., and Barwick, R. E. (1988). Uranolophus: A reappraisal of a primitive dipnoan. Mem. Assoc. Australas. Palaeontol. 7, Campbell, K. S. W., and Barwick, R. E. (1990). Paleozoic dipnoan phylogeny: Functional complexes and evolution without parsimony. Paleobiology 16, Campbell, K. S. W., and Bell, M. W. (1977). A primitive amphibian from the Late Devonian of New South Wales. Alcheringa 1, Carroll, R. L. (1988). "Vertebrate Paleontology and Evolution." Freeman, New York. Carroll, R. L. (1992). The primary radiation of terrestrial vertebrates. Annu. Rev. Earth Planet. Sei. 20, Chang, M. M. (1982). The braincase of Youngolepis, a Lower Devonian crossopterygian from Yunnan, South-Western China. Ph.D. Thesis, University of Stockholm. Chang, M. M. (1991a). Head exoskeleton and shoulder girdle of Youngolepis. In "Early Vertebrates and Related Problems of Evolutionary Biology" (M. M. Chang, Y. H. Liu, and G. R. Zhang, eds.), pp Science Press, Beijing. Chang, M. M. (1991b). "Rhipidistians," dipnoans and tetrapods. In "Origins of the Higher Groups of Tetrapods: Controversy and Consensus" (H.-P. Schultze and L. Trueb, eds.), pp Cornell University Press (Comstock), Ithaca, NY. Chang, M. M., and Smith, M. M. (1992). Is Youngolepis a porolepiform? J. Vertebr. Paleontol. 12, Chang, M. M., and Yu, X. (1984). Structure and phylogenetic significance of Diabolichthys speratus gen. et sp. nov., a new dipnoan-like form from the Lower Devonian of Eastern Yunnan, China. Proc. Linn. Soc. N. S. W. 107, Chang, M. M., and Zhu, M. (1993). A new Middle Devonian osteolepidid from Qujing, Yunnan. Mem. Assoc. Australas. Palaeontol. 15, Chorn, J., and Schultze, H.-P. (1989). A complete specimen of Sagenodus (Dipnoi) from the Upper Pensylvanian of the Hamilton quarry, Kansas. In "Regional Geology and Paleontology of Upper Paleozoic Hamilton Quarry Area in Southeastern Kansas" (G. Mapes and R. Mapes, eds.), Kans. Geol. Surv. Guidebook Ser. 6, pp Clack, J. A. (1988). New material of the early tetrapod Acanthostega from the Upper Devonian of East Greenland. Palaeontology 31, Clack, J. A. (1989). Discovery of the earliest-known tetrapod stapes. Nature (London) 342, Clack, J. A. (1994a). Acanthostega gunnari, a Devonian tetrapod from Greenland; the snout, palate and ventral parts of the braincase, with a discussion of their significance. Medd. Gronl. Geosci. 31, Clack, J. A. (1994b). Earliest known tetrapod braincase and the evolution of the fenestra ovalis. Nature (London) 369, Cloutier, R. (1990). Phylogenetic interrelationships of the actinistians (Osteichthyes: Sarcopterygii): Patterns, trends, and rates of evolution. Ph.D. Thesis, University of Kansas, Lawrence. Cloutier, R. (1991a). Interrelationships of Palaeozoic actinistians: Patterns and trends. In "Early Vertebrates and Related Problems of Evolutionary Biology" (M. M. Chang, Y. H. Liu, and G. R. Zhang, eds.), pp Science Press, Beijing. Cloutier, R. (1991b). Patterns, trends and rates of evolution within the Actinistia. Environ. Biol. Fishes 32, Cloutier, R., (1996a). The primitive actinistian Miguashaia bureaui Schultze (Sarcopterygii). In "Devonian Fishes and Plants from Miguasha, Quebec, Canada" (H.-P. Schultze and R. Cloutier, eds.), pp Verlag Dr. Freidrich Pfeil, Munich. Cloutier, R., (1996b). Dipnoi: Sarcopterygii). In "Devonian Fishes and Plants from Miguasha, Quebec, Canada" (H.-P. Schultze and R. Cloutier, eds.), pp Verlag Dr. Friedrich Pfeil, Munich. Cloutier, R., and Forey, P. L. (1991). Diversity of extinct and living actinistian fishes (Sarcopterygii). Environ. Biol. Fishes 32, Cloutier, R., and Schultze, H.-P. (1996). Porolepiforms fishes (Sarcopterygii). In "Devonian Fishes and Plants from Miguasha, Quebec, Canada" (H.-P. Schultze and R. Cloutier, eds.), pp Verlag Dr. Friedrich Pfeil, Munich. Cloutier, R., Smith, M. M., and Krupina, N. I. (1993). Growth of the dental system of the Famennian dipnoan, Andreyevichthys epitomus from Russia: Morphometrics and morphogenesis of the entopterygoid tooth plate. In "The Gross Symposium. Scientific Sessions: Abstracts" (S. Turner, ed.). Universitè des Sciences et Technologies de Lille, Villeneuve d'ascq, France. Cloutier, R., Loboziak, S., Candilier, A.-M., and Blieck, A., (1996). Biostratigraphy of the Upper Devonian Escuminac Formation, eastern Québec, Canada: A comparative study based on miospores and fishes. Rev. Palaeobot. Palynol (in press). Coates, M. I. (1991). New palaeontological contributions to limb ontogeny and phylogeny. In "Developmental Patterning of the Vertebrate Limb" (J. R. Hinchliffe, J. M. Hurle, and D. Summerbell, eds.), pp Plenum, New York. Coates, M. I. (1995). The origin of vertebrate limbs. In "The Evolution of Developmental Mechanisms" (M. Akam, P. W. H. Holland, P. W. Ingham, and G. A. Wray, eds.), pp The Company of Biologists Limited, Cambridge, UK. Coates, M. I., and Clack, J. A. (1990). Polydactyly in the earliest known tetrapod limbs. Nature (London) 347, Coates, M. I., and Clack, J. A. (1991). Fish-like gills and breathing in the earliest known tetrapod. Nature (London) 352, Cope, E. D. (1889). Synopsis of the families of Vertebrata. Am. Nat. 23, Daeschler, E. B., Shubin, N. H., Thomson, K. S., and Amaral, W. W. (1994). A Devonian tetrapod from North America. Science 265, Denison, R. H. (1968a). Early Devonian lungfishes from Wyoming, Utah and Idaho. Fieldiana, Geol. 17, Denison, R. H. (1968b). The evolutionary significance of the earliest known lungfish, Uranolophus. In "Current Problems of Lower Vertebrate Phylogeny" (T. (rvig, ed.), Nobel Symp. 4, pp Almqvist & Wiksell, Stockholm. De Queiroz, K., and Gauthier, J. (1990). Phylogeny as a central principle in taxonomy: Phylogenetic definitions of taxon names. Syst. Zool. 39,

27 17. Interrelationships of Basal Sarcopterygians 471 Dollo, L. (1895). Sur la phylogénie des Dipneustes. Bull. Soc. Belge Géol. Paléontol. Hydrol. 9, Donoghue, M. J., Doyle, J. A., Gauthier, J., Kluge, A. G., and Rowe, T. (1989). The importance of fossils in phylogeny reconstruction. Annu. Rev. Ecol. Syst. 20, Doyle, J. A., and Donoghue, M. J. (1986). Seed plant phylogeny and the origin of angiosperms: An experimental cladistic approach. Bot. Rev. 52, Doyle, J. A., and Donoghue, M. J. (1987). The importance of fossils in elucidating seed plant phylogeny and macroevolution. Rev. Palaeobot. Palynol. 50, Etheridge, R. (1906). The cranial buckler of a dipnoan fish, probably Ganorhynchus, from the Devonian beds of the Murrumbidgee River, New South Wales. Rec. Aust. Mus. 6, Forey, P. L. (1980). Latimeria: A paradoxical fish. Proc. R. Soc. London, Ser. B 208, Forey, P. L. (1981). The coelacanth Rhabdoderma in the Carboniferous of the British Isles. Palaeontology 24, Forey, P. L. (1984). The coelacanth as a living fossil. In "Living fossils" (N. Eldredge and S. M. Stanley, eds.), pp Springer-Verlag, New York. Forey, P. L. (1987). Relationships of lungfishes. J. Morphol., Suppl. 1, Forey, P. L. (1991). Latimeria chalumnae and its pedigree. Environ. Biol. Fishes 32, Forey, P. L., Gardiner, B. G., and Patterson, C. (1991). The lungfish, the coelacanth and the cow revisited. In "Origins of the Higher Groups of Tetrapods: Controversy and Consensus" (H.-P. Schultze and L. Trueb, eds.), pp Cornell University Press (Comstock), Ithaca, NY. Forster-Cooper, C. (1937). The Middle Devonian fish fauna of Achanarras. Trans. R. Soc. Edinburgh 59, Fricke, H., Reinicke, O., Hofer, H., and Nachtigall, W. (1987). Locomotion of the coelacanth Latimeria chalumnae in its natural environment. Nature (London) 329, Fritzsch, B. (1987). The inner ear of the coelacanth fish Latimeria has tetrapod affinities. Nature (London) 327, Fritzsch, B. (1988). Phylogenetic and ontogenetic origin of the dorsolateral auditory nucleus of anurans. In "The Evolution of the Amphibian Auditory System" (B. Fritzsch, M. Ryan, W. Wilsczynski, W. Hetherington, and T. Walkowiak, eds.), pp Wiley, New York. Fritzsch, B. (1992). The water-to-land transition: Evolution of the tetrapod basilar papilla, middle ear, and auditory nuclei. In "The Evolutionary Biology of Hearing" (D. B. Webster, R. R. Fray, and A. N. Popper, eds.), pp Springer-Verlag, New York. Gaffney, E. S. (1979). Tetrapod monophyly: A phylogenetic analysis. Bull. Carnegie Mus. Nat. Hist. 13, Gardiner, B. G. (1980). Tetrapod ancestry: A reappraisal. In "The Terrestrial Environment and the Origin of Land Vertebrates" (A. L. Panchen, ed.), pp Academic Press, London. Gardiner, B. G. (1984). The relationships of the palaeoniscid fishes, a review based on new specimens of Mitnia and Moythomasia from the Upper Devonian of Western Australia. Bull. Br. Mus. (Nat. Hist.), Geol. 37(4), Gardiner, B. G., Janvier, P., Patterson, C., Forey, P. L., Greenwood, H., Miles, R. S., and Jefferies, R. P. S. (1979). The salmon, the lungfish and the cow: A reply. Nature (London) 277, Gauthier, J., Kluge, A. G., and Rowe, T. (1988). Amniote phylogeny and the importance of fossils. Cladistics 4, Gill, T. (1872). Arrangement of the families of fishes. Stnithson. Misc. Collect. 11, I-XLVI, Goodrich, E. S. (1902). On the pelvic girdle and fin of Eus thenopteron. J. Microsc. Soc. 45, Gorr, T., Kleinschmidt, T., and Fricke, H. (1991). Close tetrapod relationships of the coelacanth Latimeria indicated by haemoglobin sequences. Nature (London) 351, Greenwood, P. H. (1987). The natural history of African lungfishes. Morphol., Suppl. 1, Gross, W. (1941). Über den Unterkiefer einiger devonischer Crossopterygier. Abh. Preuss. Akad. Wiss., Mat.-Naturwiss. Kl. pp Halstead, L. B. (1978). The cladistic revolution-can it make the grade? Nature (London) 276, 759. Halstead, L. B., White, E. I., and MacIntyre, G. T. (1979). L. B. Halstead and colleagues reply Nature (London) 277, 176. Harland, W. B., Armstrong, R. L., Cox, A. V., Craig, L. E., Smith, A. G., and Smith, D. G. (1990). A Geologic Time Scale Cambridge University Press, Cambridge, UK. Hedges, S. B., Hass, C. A., and Maxson, L. R. (1993). Relations of fish and tetrapods. Nature (London) 363, Holmes, E. B. (1985). Are lungfishes the sister group of tetrapods? Biol. J. Linn. Soc. 25, Huelsenbeck, J. P. (1991). When are fossils better than extant taxa in phylogenetic analyses? Syst. Zool. 40, Huxley, T. H. (1861). Preliminary essay upon the systematic arrangement of the fishes of the Devonian epoch. Mem. Geol. Surv. U.K., Figures and Descr. Br. Org. Remains 10, Janvier, P. (1980). Osteolepid remains from the Devonian of the Middle East, with particular reference to the endoskeletal shoulder girdle. In "The Terrestrial Environment and the Origin of Land Vertebrates" (A. L. Panchen, cd.) pp Academic Press, London. Jarvik, E. (1937). On the species of Eusthenopteron found in Russia and the Baltic states. Bull. Geol. Inst. Univ. Uppsala 27, Jarvik, E. (1942). On the structure of the snout of crossopterygians and lower gnathostomes in general. Zool. Bidr. Uppsala 21, Jarvik, E. (1944a). On the dermal bones, sensory canais and pitlines of the skull in Eusthenopteron foordi Whiteaves, with some remarks on E. sâve-siiderberghi Jarvik. K. Sven. Vetenskapsalcad. Handl. [3] 21(3), Jarvik, E. (1944b). On the exoskeletal shoulder-girdle of teleostomian fishes, with special reference to Eusthenopteron foordi Whiteaves. K. Sven. Vetenskapsakad. Handl. [3] 21(7), Jarvik, E. (1948). On the morphology and taxonomy of the Middle Devonian osteolepid fishes of Scotland. K. Sven. Vetenskapsakad. Handl. [3] 25, Jarvik, E. (1952). On the fish-like tail in the ichthyostegid stegocephalians. Medd. Groenl. 114, Jarvik, E. (1954). On the visceral skeleton in Eusthenopteron with a discussion of the parasphenoid and palatoquadrate in fishes. Sven. Vetenskapsakad. Handl. [4] 5, Jarvik, E. (1963). The composition of the intermandibular division of the head in fish and tetrapods and the diphyletic origin of the tetrapod tongue. K. Sven. Vetenskapsakad. Handl. [4] 9, Jarvik, E. (1964). Specializations in early vertebrates. Ann. Soc. R. Zool. Belg. 94, Jarvik, E. (1972). Middle and Upper Devonian Porolepiformes from East Greenland with special reference to Glyptolepis groenlandica n. sp. Medd. Groenl. 187, Jarvik, E. (1980). "Basic Structure and Evolution of Vertebrates," Vol. 1. Academic Press, London. Jessen, H. (1966). Die Crossopterygier des Oberen Plattenkalkes (Devon) der Bergisch-Gladbach-Paffrather Mulde (Rheinisches Schiefergebirge) unter Berücksichtigung von amerikanischem und europâischem Onychodus-material. Ark. Zool. [2] 18,

28 472 RICHARD CLOUTIER AND PER ERIK AHLBERG Jessen, H. (1975). A new choanate fish, Powichthys thorsteinssoni n.g. n.sp., from the early Lower Devonian of the Canadian Arctic Archipelago. Colloq. Int. C. N. R. S., 218, Jessen, H. (1980). Lower Devonian Porolepiformes from the Canadian Arctic with special reference to Powichthys thorsteinssoni Jessen. Palaeontographica, Abt. A 167, Jurgens, J. D. (1973). The morphology of the nasal region of Amphibia and its bearing on the phylogeny of the group. Ann. Univ. Stellenbosch, Ser. A 46, Kemp, A. (1987). The biology of the Australian lungfish, Neoceratodus forsteri (Krefft 1870). J. Morphol., Suppl. 1, Lambers, P. (1992). On the ichthyofauna of the Solnhofen Lithographie Limestone (Upper Jurrasic, Germany). Ph.D. Dissertation, Rijksuniversiteit Groningen, Groningen, The Netherlands. Lauder, G. V. (1980). On the evolution of jaw adductor musculature in primitive gnathostome fishes. Breviora 460, Lebedev, O. A. (1984). The first find of a Devonian tetrapod in the U.S.S.R. Dokl. Akad. Nauk SSSR 278, Lebedev, O. A. (1995). Morphology of a new osteolepidid fish from Russia. Bull. Mus. Natl. Hist. Nat., Sect. C, [4] 17, Lebedev, O. A., and Clack, J. A. (1993). Devonian tetrapod remains from Andreyevka, Tula, Russia. Palaeontology 36, Lebedev, O. A., and Coates, M. I. (1995). The postcranial skeleton of the Devonian tetrapod Tulerpeton curtum Lebedev. Zool. I. Linn. Soc. 114, Leonardi, G. (1983). Notopus petri nov. gen., nov. sp.: Une empreinte d'amphibien du Dévonien au Parana'. (Brésil). Geobios 16, Long, J. A. (1985). The structure and relationships of a new osteolepiform fish from the Late Devonian of Victoria, Australia. Alcheringa 9, Long, J. A. (1987). An unusual osteolepiform fish from the Late Devonian of Victoria, Australia. Palaeontology 30, Long, J. A. (1988a). Late Devonian fishes from Gogo, Western Australia. Natl. Geogr. Res. 4, Long, J. A. (1988b). New palaeoniscoid fishes from the Late Devonian and Early Carboniferous of Victoria. Mem. Assoc. Australas. Palaeontol. 7, Long, J. A. (1989). A new rhizodontiform fish from the Early Carboniferous of Victoria, Australia, with remarks on the phylogenetic position of the group. I. Vertebr. Paleontol. 9, Long, J. A. (1991). Arthrodire predation by Onychodus (Pisces, Crossopterygii) from the Late Devonian Gogo Formation, Western Australia. Rec. West. Aust. Mus. 15, Long, J. A. (1992). Cranial anatomy of two new Late Devonian lungfishes (Pisces: Dipnoi) from Mt. Howitt, Victoria. Rec. Aust. Mus. 44, Long, J. A. (1993). Cranial ribs and the origin of dipnoan air-breathing. Mem. Assoc. Australas. Palaeontol. 15, Lovtrup, S. (1977). "The Phylogeny of Vertebrata." Wiley, London. Lund, R. (1970). Fossil fishes from Southwestern Pennsylvania. Part I: Fishes from the Duquesne Limestones (Conemaugh, Pennsylvania). Ann. Carnegie Mus. 44, Lund, R., and Lund, W. L. (1984). New genera of coelacanths from the Bear Gulch Limestone (Lower Carboniferous) of Montana (U.S.A.). Geobios 17(2), Lund, R., and Lund, W. L. (1985). Coelacanths from the Bear Gulch Limestone (Namurian) of Montana and the evolution of the Coelacanthiformes. Bull. Carnegie Mus. Nat. Hist. 25, Lund, W. L., Lund, R., and Klein, G. A. (1985). Coelacanth feeding mechanisms and ecology of Bear Gulch coelacanths. C. R. Neuvi. Congr. Int. Stratigr. Géol. Carbonifère 5, Lyarskaya, L. A. (1981). Baltic Devonian Placodermi Asterolepididae. Zinatne, Riga. Lyarskaya, L. A., and Mark-Kurik, E. (1972). Eine neue Fundstelle oberdevonscher Fische im Baltikum. Neues Jahrb. Minerai., Geol. Palaeontol., Monatsh. 7, Maisey, J. G. (1986a). Heads and tails: A chordate phylogeny. Clad s- tics 2, Maisey, J. G. (1986b). Coelacanth from the Lower Cretaceous of Brazil. Am. Mus. Novit. 2866, Marshall, C. R. (1987). Lungfish: Phylogeny and parsimony. J. Morphol., Suppl. 1, Marshall, C. R., and Schultze, H.-P. (1992). Relative importance of molecular, neontological, and paleontological data in understanding the biology of the vertebrate invasion of land. J. Mol. Evol. 35, Meyer, A. (1995). Molecular evidence on the origin of tetrapods and the relationships of the coelacanth. Trends Ecol. Evol. 10(3), Meyer, A., and Wilson, A. C. (1990). Origin of tetrapods inferred from their mitochondrial DNA affiliation to lungfish. J. Mol. Evol. 31, Meyer, A., and Wilson, A. C. (1991). Coelacanth's relationships. Nature (London) 353, 19. Miles, R. S. (1975). The relationships of the Dipnoi. Colloq. Int. C. N. R. S. 218, Miles, R. S. (1977). Dipnoan (lungfish) skulls and the relationships of the group: A study based on new species from the Devonian of Australia. Zool. I. Linn. Soc. 61, Miller, H. (1841). "The Old Red Sandstone or New Walks in an Old Field." Johnstone, Edinburgh. Milner, A. R., Smithson, T. R., Milner, A. C., Coates, M. I., and Rolfe, W. D. I. (1986). The search for early tetrapods. Mod. Geol. 10, Millot, J., and Anthony, J. (1958). "Anatomie de Latimeria chalumnae," Tome I. Squelette et muscles. CNRS, Paris. Moy-Thomas, J. A., and Miles, R. S. (1971). "Palaeozoic Fishes," 2nd ed. Chapman & Hall, London. Murchison, R. I. (1839). "The Silurian System." John Murray, London. Normark, B. B., McCune, A. R., and Harrison, R. G. (1991). Phylogenetic relationships of neopterygian fishes, inferred from mitochondrial DNA sequences. Mol. Biol. Foui. 8, Northcutt, R. G., and Bemis, W. E. (1993). Cranial nerves of the coelacanth, Latimeria chalumnae (Osteichthyes: Sarcopterygii: Actinistia) and comparisons with other Craniata. Brain, Behav. Evol. 42, Suppl. 1, Obruchev, D. (1933). Description of four new fish species from the Devonian of Leningrad Province. Mater. Cent. Sci. Geol. Prospect. Inst., Palaeontol. Stratigr. Mag. 1, Orvig, T. (1957). Remarks on the vertebrate fauna of the Lower Upper Devonian of Escuminac Bay, P.Q., Canada, with special reference to the porolepiform crossopterygians. Ark. Zool. [2] 10, Orvig, T. (1969). Vertebrates from the Wood Bay group and the position of the Emsian-Eifelian boundary in the Devonian of Vestspitsbergen. Lethaia 2(4), Orvig, T. (1986). A vertebrate bone from the Swedish Palaeocene. Geol. Foeren. Stockholm Foerh. 108, Owen, R. (1846). "Lectures on the Comparative Anatomy and Physiology of the Vertebrate Animals," Part I. Longman, Brown, Green and Longman, London. Owen, R. (1860). "Palaeontology, or a Systematic Summary of Extinct Animals and their Geological Relations." Black, Edinburgh. Panchen, A. L. (1985). On the amphibian Crassigyrinus scoticus Watson from the Carboniferous of Scotland. Philos. Trans. R. Soc. London Ser. B 309,

29 17. Interrelationships of Basal Sarcopterygians 473 Panchen, A. L., and Smithson, T. R. (1987). Character diagnosis, fossils, and the origin of tetrapods. Biol. Rev. Cambridge Philos. Soc. 62, Panchen, A. L., and Smithson, T. R. (1988). The relationships of the earliest tetrapods. In "The Phylogeny and Classification of the Tetrapods" (M. J. Benton, ed.) Vol. 1, pp Oxford University Press (Clarendon), Oxford. Patterson, C. (1980). Origin of tetrapods: historical introduction to the problem. In "The Terrestrial Environment and the Origin of Land Vertebrates" (A. L. Panchen, ed.), pp Academic Press, London. Patterson, C. (1981). Significance of fossils in determining evolutionary relationships. Annu. Rev. Ecol. Syst. 12, Patterson, C. (1982a). Morphological characters and homology. Syst. Assoc. Spec. Vol. 21, Patterson, C. (1982b). Classes and cladists or individuals and evolution. Syst. Zool. 31, Patterson; C., and Rosen, D. E. (1977). Review of ichthyodectiform and other Mesozoic teleost fishes and the theory and practice of classifying fossils. Bull. Am. Mus. Nat. Hist. 158, Rackoff, J. S. (1980). The origin of the tetrapod limb and the ancestry of tetrapods. Syst. Assoc. Spec. Vol. 15, Rocek, Z., and Rage, J. C. (1994). The presumed amphibian footprint Noto pus petri from the Devonian: A probable starfish trace fossil. Lethaia 27, Romer, A. S. (1937). The braincase of the Carboniferous crossopterygian Megalichthys nitidus. Bull. Mus. Comp. Zool. 82, Romer, A. S. (1947). Review of the Labyrinthodontia. Bull. Mus. Comp. Zool. 99, Romer, A. S. (1966). "Vertebrate Paleontology," 3rd ed. University of Chicago Press, Chicago. Rosen, D. E., Forey, P. L., Gardiner, B. G., and Patterson, C. (1981). Lungfishes, tetrapods, paleontology, and plesiomorphy. Bull. Am. Mus. Nat. Hist. 167, Sâve-Süderbergh, G. (1932). Preliminary note on Devonian stegocephalians from East Greenland. Medd. Groenl. 94, Schaeffer, B. (1952). Rates of evolution in the coelacanth and dipnoan fishes. Evolution (Lawrence, Kans.) 6, Schoch, R. M. (1986). "Phylogeny Reconstruction in Paleontology." Van Nostrand-Reinhold, New York. Schultze, H.-P. (1969). Die Faltenzâhne der rhipidistiiden Crossopterygier, der Tetrapoden und der Actinopterygier-Gattung Lepisosteus. Palaeontogr. Ital. 65 [N. S. 35], Schultze, H.-P. (1970). Folded teeth and the monophyletic origin of tetrapods. Am. Mus. Novi. 2408, Schultze, H.-P. (1973). Crossopterygier mit heterozerker Schwanzflosse aus dem Oberdevon Kanadas, nebst einer Beschreibungvon Onychodontida-Resten aus dem Mitteldevon Spaniens und aus dem Karbon der USA. Palaeontographica, Aht. A 143, Schultze, H.-P. (1975). Die Lungenfisch-Gattung Conchopoma (Pisces, Dipnoi). Senckenbergiana Lethaea 56, Schultze, H.-P. (1977). The origin of the tetrapod limb within the rhipidistian fishes. In "Major Patterns in Vertebrate Evolution" (M. K. Hecht, P. C. Goody, and B. M. Hecht, eds.), pp Plenum, New York. Schultze, H.-P. (1981). Hennig und der Ursprung der Tetrapoda. Palaeontol. Zh. 55, Schultze, H.-P. (1985). Reproduction and spawning sites of Rhabdoderma (Pisces, Osteichthyes, Actinistia) in Pennsylvanian deposits of Illinois, USA. C. R. Neuv. Congr. Int. Stratigr. Géol. Carbonifère 5, Schultze, H.-P. (1987). Dipnoans as sarcopterygians. J. Morpho!. Suppl. 1, Schultze, 1-1.-P. (1991). A comparison of controversial hypotheses on the origins of tetrapods. In "Origins of the Major Groups of Tetrapods: Controversies and Consensus" (H.-P. Schultze and L. Trueb, eds.), pp Cornell University Press (Comstock), Ithaca, NY. Schultze, H.-P. (1992a). A new long-headed dipnoan (Osteichthyes) from the Middle Devonian of Iowa, USA. J. Vertebr. Paleontol. 12, Schultze, H.-P. (1992b). Dipnoi. In "Fossilium catalogus I:Animalia" (F. Westphal, ser. cd.), Pars 131. Kugler Publications, Amsterdam and New York. Schultze, H.-P. (1993). Osteichthyes: Sarcopterygii. In "The Fossil Record 2" (M. J. Benton, ed.), pp Chapman & Hall, London. Schultze, H.-P. (1994). Comparison of hypotheses on the relationships of sarcopterygians. Syst. Biol. 43, Schultze, H.-P., (1996). The elpistostegid fish Elpistostege, the closest the Miguasha fauna cornes to a tetrapod. In "Devonian Fishes and Plants from Miguasha, Quebec, Canada" (H.-P. Schultze and R. Cloutier, eds.), pp Verlag Dr. Pfeil, München. Schultze, H.-P., and Arsenault, M. (1985). The panderichthyid fish Elpistostege: A close relative of tetrapods? Palaeontology 28, Schultze, H.-P., and Campbell, K. S. W. (1987). Characterization of the Dipnoi, a monophyletic group. J. Morphol. Suppl. 1, Schultze, H.-P., and Marshall, C. R. (1993). Contrasting the use of functional complexes and isolated characters in lungfish evolution. Mem. Assoc. Australas. Palaeontol. 15, Schultze, H.-P., Cloutier, R., and Marshall, C. R. (1993). Contrasting the use of functional complexes and isolated characters in lungfish evolution. Assoc. Australas. Palaeontol., K. S. W. Campbell Syinp., Abstr. Vol., p. 21. Schwimmer, D. R., Stewart, J. D., and Williams, G. D. (1994). Giant fossil coelacanths of the late Cretaceous in the eastern United States. Geology 55, Sedgwick, A., and Murchison, R. I. (1828). On the structure and relations of the deposits contained between the Primary Rocks and the Oolitic Series in the North of Scotland. Trans. Geo!. Soc. London [2] 3, Sharp, P. M., Lloyd, A. T., and Higgins, D. G. (1991). Coelacanth's relationships. Nature (London) 353, Shubin, N. H., and Alberch, P. (1986). A morphogenetic approach to the origin and basic organization of the tetrapod limb. Evol. Biol. 20, Smith, M. M. (1984). Petrodentine in extant and fossil dipnoan dentitions: Microstructure, histogenesis and growth. Proc. Linn. Soc. N. S. W. 107, Smith, M. M. (1989). Distribution and variation in enamel structure in the oral teeth of sarcopterygians: its significance for the evolution of a protoprismatic enamel. Hist. Biol. 3, Smith, M. M., and Chang, M. M. (1990). The dentition of Diabolepis speratus Chang & Yu, with further consideration of its relationships and the primitive dipnoan dentition. J. Vertebr. Paleontol. 10, Smith, M. M., Smithson, T. R., and Campbell, K. S. W. (1987). The relationships of Uronemus: A Carboniferous dipnoan with highly modified tooth plates. Philos. Trans. R. Soc. London 317, Smithson, T. R. (1985). On the relationships and morphology of the Carboniferous amphibian Eoherpeton watsoni. Zool. J. Linn. Soc. 85, Stensiô, E. (1922). Über zwei Coelacanthiden aus dem Oberdevon von Wildungen. Palaeontol. Zh. 4, Stensiô, E. (1937). On the Devonian coelacanthids of Germany with

30 474 RICHARD CLOUTIER AND PER ERIK AHLBERG special reference to the dermal skeleton. K. Sven. Vetenskapsakad. Handl. 3(16), Stock, D. W., and Swofford, D. L. (1991). Coelacanth relationships. Nature (London) 353, Stock, D. W., Moberg, K. D., Maxson, L. R., and Whitt, G. S. (1991). A phylogenetic analysis of the 18S ribosomal RNA sequence of the coelacanth Latimeria chalumnae. Environ. Biol. Fishes 32, Stôssel, I. (1995). The discovery of a new Devonian tetrapod trackway in SW Ireland. J. Geol. Soc., London 152, Thomson, K. S. (1973). Observations on a new rhipidistian fish from the Upper Devonian of Australia. Palaeontographica, Abt. A 143, Tong-Dzuy Thanh, and Janvier, P. (1990). Les Vertébrés du Dévonien inférieur de Bac Bo oriental (Provinces de Bac Thaï et Lang Son, Viêt Nam). Bull. Mus. Natl. Hist. Nat., Sect. C [4] 12(2), Tong-Dzuy Thanh, and Janvier, P. (1994). Early Devonian fishes from Trang Xa (Bac Thai, Vietnam) with remarks on the distribution of the vertebrates in the Song Cau Group. J. Southeast Asian Earth Sci. 10, Traquair, R. H. (1890). List of the fossil Dipnoi and Ganoidei of Fife and the Lothians. Proc. R. Soc. Edinburgh 17, Trueb, L, and Cloutier, R. (1991). A phylogenetic investigation of the inter- and intrarelationships of the Lissamphibia (Amphibia: Temnospondyli). In "Origins of the Higher Groups of Tetrapods: Controversy and Consensus" (H.-P. Schultze and L. Trueb, eds), pp Cornell University Press (Comstock), Ithaca, NY. Vorobyeva, E. I. (1960). New data on the crossopterygian fish genus Panderichthys from the Devonian of the USSR. Paleontol. Zh. 1, Vorobyeva, E. I. (1962). Rhizodont crossopterygians of the Main Devonian Field. Tr. Paleontol. Inst. 94, Vorobyeva, E. I. (1971). The ethmoid region of Panderichthys and some problems of the cranial morphology of crossopterygians. Tr. Paleontol. Inst. 130, Vorobyeva, E. I. (1977). Morphology and nature of evolution of crossopterygian fishes. Tr. Paleontol. Instit. 163, Vorobyeva, E. I. (1980). Observations on two rhipidistian fishes from the Upper Devonian of Lode, Latvia. Zool. J. Linn. Soc. 70, Vorobyeva, E. I. (1986). The current state of the problem of amphibian origin. In "Studies in Herpetology" (Z. Roêek, ed.), pp Charles University, Prague. Vorobyeva, E. I. (1987). Porolepid crossopterygian from the Middle Devonian of Estonia. Paleontol. Zh. 1987(1), Vorobyeva, E. I. (1989). Panderichthyida-a new order of Paleozoic crossopterygian fishes (Rhipidistia). Dokl. Akad. Nauk SSSR 306, Vorobyeva, E. I., and Kuznetsov, A. (1992). The locomotor apparatus of Panderichthys rhombolepis (Gross), a supplement to the problem of fish-tetrapod transition. In 'Fossil Fishes as Living Animais" (E. Mark-Kurik, ed.), pp Academy of Sciences of Estonia, Tallinn. Vorobyeva, E. I., and Lyarskaya, L. A. (1968). Crossopterygian and dipnoan remains from the Amata beds of Latvia and their burial. ln "Ocherki po filogenii i sistematike iskopaemykh ryb i beschelyustnykh" (D. V. Obruchev, ed.), pp Nauka, Moscow. Vorobyeva, E. I., and Schultze, H.-P. (1991). Description and systematics of panderichthyid fishes with comments on their relationship to tetrapods. In "Origins of the Higher Groups of Tetrapods: Controversy and Consensus" (H.-P. Schultze and L. Trueb, eds.), pp Cornell University Press (Comstock), Ithaca, NY. Vorobyeva, E. I., and Tsessarskii, A. A. (1986). On origin of vertebrae in lower tetrapods. Zh. Obshch. Biol. 47, Wang, S., Drapala, V., Barwick, R. E., and Campbell, K. S. W. (1990). A new Early Devonian lungfish, Sorbitorhynchus deleaskitus n. gen. et sp., from Guangxi, China. Paleobiology 16, Warren, A., Jupp, R., and Bolton, B. R. (1986). Earliest tetrapod trackway. Alcheringa 10, Warren, J. W., and Wakefield, N. A. (1972). Trackways of tetrapod vertebrates from the Upper Devonian of Victoria, Australia, Nature (London) 238, Westoll, T. S. (1938). Ancestry of the tetrapods. Nature (London) 141, 127. Westoll, T. S. (1949). On the evolution of the Dipnoi. In "Genetics, Paleontology and Evolution" (G. L. Jepsen, G. G. Simpson, and E. Mayr, eds.), pp Princeton University Press, Princeton, INJ. White, E. I. (1965). The head of Dipterus valenc ennesi. Bull. Br. Mus. (Nat. Hist.) Geol. 11, Whiteaves, J. F. (1883). On some remarkable fossil fishes from the Devonian rocks of Scaumenac Bay, P. Q., with descriptions of a new genus and three new species. Can. Nat. Q. J. Sci. [N.S.J 10, Whiteaves, J. F. (1889). Illustrations of the fossil fishes of the Devonian rocks of Canada. Part II. Trans. R. Soc. Can. 6, Woodward, A. S. (1891). Catalogue of the Fossil Fishes in the British Museum (Natural History), Part II. Br. Mus. (Nat. Hist.), London. Worobjewa, E. (1975). Bemerkungen zu Panderichthys rhombolepis (Gross) aus Lode in Lettland (Gauja Schichten, Oberdevon). Neues Jahrb. Minerai., Geol. Palaeontol., Monatsh. 1975, Young, G. C., Long, J. A., and Ritchie, A. (1992). Crossopterygian fishes from the Devonian of Antarctica: Systematics, relationships and biogeographic significance. Rec. Austr. Mus., Suppl. 14, Zhu, M., and Janvier, P. (1994). Un Onychodontide (Vertebrata, Sarcopterygii) du Dévonien Inférieur de Chine. C. R. Séances Acad. Sci., Ser , Appendix 1: Character List Cosmine. 0 = absent; 1 present. Mesh canais. 0 = pore cavity with horizontal partition; 1 = pore cavity without horizontal partition. Condition of scales. 0 = rhombic scales; 1 = rounded scales. Peg on rhombic scale. 0 = narrow; 1 = broad. Boss on internai face of scale. 0 absent; 1 = present. Ganoine. 0 = absent; 1 = present. Acrodin. 0 = absence of acrodin caps on teeth; 1 presence of acrodin caps on teeth. Marginal teeth on dentary. 0 = present; 1 = absent. 9. Dental plate. 0 = denticles on entopterygoid, or naked bone; 1 = tooth plate on entopterygoid; 2 = dentine plate on entopterygoid.

31 17. Interrelationships of Basal Sarcopterygians 475 Dentition on coronoids. 0 = narrow marginal tooth row; 1 = broad marginal "tooth field." Spiral parasymphysial teeth. 0 = absent; 1 = present. Fang pairs in inner tooth arcade. 0 = absent; 1 = present. Fang pair on anterior end of dentary. 0 = absent; 1 = present. Plicidentine. 0 = absent; 1 = polyplocodont plicidentine; 2 = dendrodont plicidentine. Skull shape. 0 = lateral orbits, interorbital skull roof wide and arched; 1 = dorsal orbits, interorbital skull roof narrow and flat or concave. Premaxilla. 0 = present; 1 = absent. Position of premaxilla. 0 = marginal; 1 = ventral part turned in. Position of premaxilla. 0 = premaxilla forming part of orbit; 1 = premaxilla flot forming part of orbit. Maxilla. 0 = present; 1 = absent. Posterodorsal process of maxilla. 0 = present; 1 = absent. Shape of posterodorsal process of maxilla. 0 = smooth, convex posterodorsal margin; 1 = distinct posterodorsal angle. Position of median rostral. 0 = rostral does flot contribute to jaw margin; 1 = rostral contributes to jaw margin. Postrostrals. 0 = postrostral mosaic of small variable bones; 1 = large median postrostral, with or without accessory bones. Paired nasals meeting in midline of skull. 0 absent; 1 = present. Paired frontals. 0 = absent; 1 = present. E-bone. 0 = absent; 1 = present. C-bone. 0 = absent; 1 = present. Supraorbitals. 0 = absent; 1 = present. Number of supraorbitals (including the "posterior tectal" of Jarvik). 0 = one; 1 = two; 2 = more than two. B-bone. 0 = absent; 1 = present. Position of anterior margin of parietal. 0 = between or in front of orbits; 1 = slightly posterior to orbits; 2 = much posterior to orbits. Pineal opening. 0 = open; 1 = closed. Median supraorbital ridges ("eyebrows"). 0 = absent; 1 = present. Parietal-supraorbital contact. 0 = absent; 1 = present. Extratemporal. 0 = absent; 1 = present. Intertemporal. 0 = present; 1 = absent. Supratemporal series. 0 = single bone which contacts the extrascapular posteriorly and the intertemporal or dermosphenotic anteriorly; 1 = two bones (supratemporal and tabular) between extrascapular and intertemporal or postorbital; 2 = single bone (probably the tabular) in posterior position, bounded anteriorly by lateral extension of postparietal. Spiracle. 0 = small hole on kinetic margin between skull roof and cheek; 1 = large, posteriorly open notch. Extrascapulars. 0 = present; 1 = absent. 40. Number of extrascapulars. 0 = four; 1 = two; 2 = three; 3 = five. 41. Median extrascapular overlap. 0 = median extrascapular overlapped by lateral extrascapulars; 1 = median extrascapular overlaps the lateral extrascapulars; 2 = median extrascapular abuts the lateral extrascapulars. Tectals. 0 = absent; 1 = present. Number of tectals (not counting the "posterior tectal" of Jarvik; see char. 29). 0 = one; 1 = three or more. Anterior and posterior nares. 0 = both present; 1 = only anterior naris present. Position of posterior naris. 0 = external, far from jaw margin; 1 = external, close to jaw margin; 2 = palatal (palatal posterior flans of lungfishes deemed nonhomologous with tetrapod choana). Position of posterior naris. 0 = associated with the orbit; 1 = flot associated with the orbit. Position of anterior flans. 0 = facial; 1 = marginal; 2 = palatal. Palatal opening ("choana") surrounded by premaxilla, maxilla, dermopalatine, and vomer. 0 = absent; 1 = present. Number of sclerotic plates. 0 = four or less; 1 = more than four. Condition of lacrimal and jugal. 0 = separate bones; 1 = fused together. Prespiracular. 0 = absent; 1 = present. Dermohyal. 0 = present; 1 = absent. Postspiracular. 0 = absent; 1 = present. Squamosal and preopercular. 0 = one bone ("preopercular"); 1 = two separate bones. Subsquamosals. 0 = absent; 1 = present. Preopercular-maxillary contact. 0 = preopercular contacts maxilla (if maxilla absent, preopercular reaches ventral margin of cheek); 1 = preopercular does not contact maxilla (if maxilla absent, preopercular does not reach ventral margin of cheek). Quadratojugal. 0 = present, small; 1 = present, large; 2 = absent.

32 476 RICHARD CLOUTIER AND PER ERIK AHLBERG Jugal-quadratojugal contact. 0 absent; 1 = present. Preoperculosubmandibular. 0 = absent; 1 = present. Opercular. 0 = present; 1 = absent. Subopercular. 0 = present; 1 = absent. Branchiostegal rays. 0 = present; 1 = absent. Number of branchiostegal rays per side. 0 = 10 or more; 1 = two to seven; 2 = one. Submandibulars. 0 = absent; 1 = present. Width of submandibulars. 0 = narrow; 1 = broad. Median gular. 0 = present; 1 absent. Relative size of median gular. 0 = small; 1 = large. Lateral gular. 0 present; 1 = absent. Size of lateral gular. 0 = lateral gular and branchiostegal rays of similar size; 1 lateral gular covering approximately half the intermandibular space. Posterior process of vomer. 0 = absent; 1 = present. Articulation of vomer. 0 vomers do not articulate with each other; 1 = vomers articulate with each other. Articulation of pterygoid. 0 = pterygoids do not articulate with each other; 1 = pterygoids articulate with each other. Articulation of parasphenoid. 0 = parasphenoid not sutured to vomer; 1 = parasphenoid sutured to vomer. Denticulated spiracular groove on parasphenoid. 0 present; 1 absent. Buccohypophysial foramen of parasphenoid. 0 = single; 1 = double. Rostral organ. 0 = absent; 1 = present. Rostral tubuli. 0 = absent; 1 = present. Fossa autopalatina. 0 = absent; 1 = present. Paired intervomerine pits. 0 = absent; 1 = present. Labial cavity. 0 = absent; 1 = present. Dermal joint between parietal and postparietal. 0 = absent; 1 = present. Dorsal endoskeletal articulation between otoccipital and ethmosphenoid blocks of braincase. 0 = absent; 1 = present. Ventral endoskeletal articulation between otoccipital and ethmosphenoid blocks of braincase. 0 = absent; 1 = present. Basicranial fenestra with arcual plates. 0 = absent; 1 = present. Unconstricted cranial notochord. 0 = absent; 1 = present. Otico-sphenoid bridge. 0 = present; 1 = absent. Position of intracranial joint relative to cranial nerves. 0 = joint passes through profundus foramen; 1 = joint passes through trigeminal foramen. Condition of hyomandibular. 0 hyomandibular with one proximal articular head; 1 = hyomandibular with two proximal articular heads. Posttemporal fossae. 0 = absent; 1 = present. Postorbital process on braincase (equivalent to character A3 of Chang and Smith, 1992). 0 = present; 1 = absent. Dentary. 0 = long; 1 = short. 92. Anterior end of dentary. 0 not modified; 1 = modified into support for parasymphysial tooth whorl. 93. Number of infradentaries. 0 = one; 1 = two; 2 = four. Number of coronoids. 0 = four or more; 1 = three; 2 = two. Condition of most posterior coronoid. 0 = not distinctly differentiated from other coronoids; 1 = well developed and oriented vertically. Prearticular position. 0 = at posterior end of coronoid series, contacts dentary dorsally; 1 = ventral to the coronoid series, does not contact the dentary dorsally. Articulation of symplectic with articular. 0 = absent; 1 = present. Trajectory of supraorbital canal. 0 = canal passing between anterior and posterior nares; 1 = canal passing anterior to both nares. Contact of supraorbital canal. 0 = supraorbital and infraorbital canals in contact rostrally; 1 = canals not in contact rostrally. Relationship of infraorbital canal to premaxilla. 0 = infraorbital canal enters premaxilla; 1 = infraorbital canal follows dorsal margin of premaxilla. Trajectory of otic canal. 0 = otic canal does not pass through growth center of postparietal; 1 = otic canal passes through growth center of postparietal. Contact of otic canal. 0 = otic canal not joining supraorbital canal; 1 = otic canal joining supraorbital canal. Position of anterior pit line. 0 = anterior pit line on postparietal; 1 = anterior pit line on parietal Position of posterior pit line. 0 = posterior pit line on posterior half of postparietal; 1 = posterior pit line on anterior half of postparietal.

33 17. Interrelationships of Basal Sarcopterygians 477 Preopercular canal. 0 = preopercular canal ends at dorsal margin of preopercular; 1 = canal does not end at dorsal margin of preopercular. Jugal canal. 0 = absent; 1 = present. Position of infraorbital canal. 0 = ventral to anterior flans; 1 = dorsal to anterior naris. Pit lines of lower jaw. 0 = middle pit line not developed into enclosed canal ("oral canal"); 1 = middle pit line developed into enclosed oral canal or intermediate morphology. Pit line of lower jaw. 0 = anterior pit line flot developed into enclosed canal; 1 = anterior pit line developed into enclosed canal linking oral and mandibular canals. Trajectory of mandibular canal. 0 = mandibular canal passing through dentary; 1 = mandibular canal flot passing through dentary. Trajectory of mandibular canal. 0 = mandibular canal passing through most posterior infradentary; 1 = mandibular canal not passing through most posterior infradentary. Anocleithrum. 0 = element developed as postcleithrum; 1 = element developed as anocleithrum sensu stricto; 2 = element absent. Condition of anocleithrum/postcleithrum. 0 - exposed on surface; 1 = subdermal. Depressed lamina of cleithrum. 0 = absent; 1 = present. Dorsal end of cleithrum. 0 = pointed; 1 = broad and rounded. Relationship of clavicle to cleithrum. 0 = ascending process of clavicle overlaps cleithrum laterally; 1 = ascending process of clavicle wraps round anterior edge of cleithrum, overlapping it both laterally and mesially. Extracleithrum. 0 = absent; 1 = present. Interclavicle. 0 = present; 1 = absent. Proximal articular surface of humerus. 0 = concave; 1 = flat; 2 = convex. Endoskeletal supports in pectoral fins. 0 = multiple elements articulating with girdle; 1 - single element ("humerus") articulating with girdle. Entepicondylar foramen. 0 = absent; 1 = present. Deltoid and supinator processes. 0 = absent; 1 = present. Number of mesomeres in pectoral fin. 0 = three to five; 1 = seven or more. Trifurcations in pectoral fin skeleton (i.e., mesomeres carrying both pre- and postaxial radials). 0 = absent; 1 = present. Digits. 0 = absent; 1 = present. Pelvis contacting vertebral column. 0 = no; 1 = yes. Dorsal and anal fins. 0 = present; 1 = absent. Basal plates in dorsal fin supports. 0 = absent; 1 = present. Anterior dorsal fin support. 0 = separate radials and basal plate; 1 = single element. Posterior branched radial complex in posterior dorsal fin. 0 = absent; 1 = present. Basal scutes on fins. 0 = absent; 1 = present. Relative length of proximal unsegmented part of lepidotrichium. 0 = much less than segmented part; 1 = similar to segmented part; 2 = much greater than segmented part. Distal end of lepidotrichium. 0 = branched; 1 = single. Epichordal lepidotrichia in tail. 0 = absent; 1 = present. Relative size of epichordal and hypochordal lepidotrichia. 0 = epichordals less developed than hypochordals; 1 = epichordals and hypochordals equally developed; 2 = epichordals more developed than hypochordals. Neural arches. 0 = bilateral halves of neural arch separated; 1 = halves fused. Supraneural spines. 0 = present on thoracic and abdominal vertebrae; 1 = restricted to a few vertebrae at anterior end of column, or absent. Condition of intercentra. 0 = ossified; 1 = not ossified. Condition of pleurocentra. 0 = flot ossified; 1 = ossified. Ribs. 0 = absence of well-ossified ribs; 1 = presence of well-ossified ribs. Appendix 2: List of Genera Used in the Phylogenetic Ana lysis The genera are entered into the data set in alphabetical order so as to preclude any bias toward preconceived groupings during the phylogenetic analysis. The tetrapods, actinistians, and dipnoans included in the analysis are all stem-group members except Crassigyrinus, which is a probable crown tetrapod (Lebedev and Coates, 1995). Acanthostega: Late Devonian (Famennian) tetrapod from eastern Greenland. Most of the skeleton is known. Allenypterus: Early Carboniferous (Namurian) actinishan from Montana. Represented by complete, laterally compressed specimens.

34 478 RICHARD CLOUTIER AND PER ERIK AHLBERG Barameda: Early Carboniferous rhizodont from Victoria, Australia. Skull roof, palate, and some postcranial elements preserved as natural molds. Beelarongia: Late Devonian (Frasnian) canowindrid osteolepiform from Victoria, Australia. Skull roof, cheek, shoulder girdle, and pectoral fin preserved as natural molds. Cheirolepis: Middle to Late Devonian (Eifelian- Frasnian) actinopterygian known from Scotland and Québec. Represented by numerous complete, laterally compressed specimens. Crassigyrinus: Early Carboniferous (Viséan- Serpukhovian) tetrapod from Scotland. Largely complete except for the tait. Diabolepis: Early Devonian (Lochkovian) dipnoan from Yunnan, China. Only the skull roof, palate, and lower jaw have been described. Diplocercides heiligenstockiensis: Late Devonian (Frasnian) actinistian from Bergisch-Gladbach, Germany. Skull roof, cheek, and part of the axial skeleton are known. Diplocercides kaeseri: Late Devonian (Frasnian) actinistian from Hessen, Germany. The neurocranium has been described extensively. Dipnorhynchus: a dipnoan. The best known species is D. suessmilchii from the Lower Devonian (Emsian) of New South Wales, Australia. Skull roof, braincase, palate, and lower jaw are known. Our coding also includes data from D. kiandrensis and D. kurikae. Dipterus: Middle Late Devonian (Eifelian Frasnian) dipnoan from Scotland and Germany, represented by numerous complete bodies. Elpistostege: Late Devonian (Frasnian) elpistostegid from Québec, Canada. Known from two incomplete skulls and a piece of vertebral column. Eusthenopteron: widespread Late Devonian (Frasnian- Famennian) tristichopterid osteolepiform. The coding is based on E. foordi from Québec, which is represented by numerous complete bodies. Glyptolepis: Middle Late Devonian (Eifelian- Frasnian) holoptychiid porolepiform from Europe and Greenland. The whole body is known. Gyroptychius: Middle Devonian (Eifelian Givetian) "osteolepid" osteolepiform from Europe and Greenland. The whole body is known but only the dermal bones have been fully described. Holoptychius: Late Devonian (Frasnian Famennian) holoptychiid porolepiform of apparently worldwide distribution. Complete bodies, but endoskeleton rarely preserved. Howqualepis: Late Devonian (Frasnian) actinopterygian from Victoria, Australia. Natural molds of complete bodies. Ichthyostega: Late Devonian (Famennian) tetrapod from eastern Greenland. The dermal skull and most of the postcranium are known. Miguashaia: Late Devonian (Frasnian) actinistian from Québec, Canada. Represented by complete but crushed bodies. Munia: Late Devonian (Frasnian) actinopterygian from Gogo, Western Australia. Whole body known in outstanding, three-dimensional detail. Moythomasia: widespread Late Devonian (Frasnian) actinopterygian. Scored on basis of Gogo material comparable to that of Mimia. Onychodus: widespread Late Devonian (Frasnian) onychodont. The best material comprises skull bones and some postcranial elements from Gogo. Osteolepis: Middle Devonian (Eifelian Givetian) "osteolepid" osteolepiform from Scotland. Many complete bodies, but Little information about the internat skeleton. Panderichthys: Late Devonian (Frasnian) elpistostegid from Latvia and Russia. P. rhombolepis is represented by complete bodies from Lode, Latvia. Polypterus: a primitive Recent actinopterygian from equatorial Africa. Porolepis: a "porolepid" porolepiform from the Lower Devonian (Pragian Emsian) of Europe and Spitsbergen. The dermal skull, shoulder girdle, scales, and ethmosphenoid have been described. Powichthys: porolepiform-like genus from the Lower Devonian (Lochkovian) of Arctic Canada. Braincase and skull roof known. Associated lower jaw, operculogular elements, and palatoquadrate probably also belong to genus. Speonesydrion: Early Devonian (Siegenian) dipnoan from New South Wales, Australia. Only part of the skull has been described. Strepsodus: Carboniferous (Dinantian Westphalian) rhizodont from Europe and North America. Known from one complete but poorly preserved body and many isolated elements. Strunius: a small onychodont from the Upper Devonian (Frasnian) of Germany and Latvia. Represented by complete but rather poorly preserved bodies. Uranolophus: Early Devonian (Pragian) dipnoan from Wyoming. The skull roof, palate, lower jaw, and postcranial dermal skeleton are known. Ventastega: Late Devonian (Famennian) tetrapod from Latvia. Lower jaw, palate, and cheekplate are known; associated clavicles, interclavicles, and ilia probably also belong to the genus. Y oungolepis: sarcopterygian genus from the Lower Devonian (Lochkovian) of Yunnan, China. Head and shoulder girdle are known.

35 17. Interrelationships of Basal Sarcopterygians 479 Appendix 3: Data Set of 140 Characters for 28 Sarcopterygian Taxa Acanthostega Allenypterus Barameda Beelarongia Cheirolepis Crassigyrinus Diabolepis Diplocercides kaeseri D. heiligenstockiensis Dipnorhynchus Dipterus Elpistostege Eusthenopteron Glyptolepis Gyroptychius Holoptychius Howqualepis Ichthyostega Miguashaia Mirnia Moythomasia Onychodus Osteolepis Panderichthys Polypterus Porolepis Powichthys Speonesydrion Strepsodus Strunius Uranoloplats Ventastega Y oungolepis LOL? L0L LL101LL L0L1L1L0 OL1L LL ? 1? L0L1L010 OL1L ??L ????0?? ?010000??????0???001???0001? ???10?0??001010??0? "7 OLOL P P0001L0L ?000000L0000? OLOL ? LLL LL101LL01? L0L1L1L1 11 7?70001L ???0?? ? " OL I L LL? ? ? 1 I L0L1L01? OL1L ? I ILL? ? 1? 0010? L0L1L0 I? 1? 0100L1 2L000701LL1LLL ? L00011L000001LL1LLL ? ??01? OL010000??0???100101L ????101LL179097??9????????1001??? OL1L LL OL 1 L L L010 1L LLO?? OL1L L L010 OL L00000L0000L000001L0L0000? L0000? OL1L? L0L LL101LL11? LOL1LILO OL1L LL ?0010? L0L1L01? OL L00000L0000L000001L0L L00000 OL L00000L0000L000000LOL L00000 OL1L ? L0001L1 OILO 1 L ? LL ? OL L ILLI I? OL LLLLO1000L0010LOOL000LOL001000LLOLOOLIL000ILOLI L L ? ??? ??? ?0?00012L000001LL???L00001?? ?999000?11000?? OL1L ????0011???0? 7?100??1?? 770? L10? 701? OL1L ? ??0010 1? 11002? ?100??0001? 101LO I? LIOL000001LLILLL00011?? ??102120?7" OL? L ????? 1??? 70?? 771??? I? IL? 11700?? " ? ?10100????0110??0???10?1070??? A canthostega Allenypterus Baranzeda Beelarongia Cheirolepis Crassigyrinus Diabolepis Diplocercides kaeseri D. heiligenstockiensis D ipnorhynchus Dipterus Elpistostege Eusthenopteron Glyptolepis Gyroptychius Holoptychius Howqualepis lchthyostega Miguashaia Minda Moythornasia Onychodus Osteolepis Panderichthys Polypterus Porolepis Powichthys Speonesydrion Strepsodus Strunius Uranolophus Ventastega Y oungolepis L0? ?LL ??LL LLL0? r> ?7001L ?11?1???? ?`1991?9299? ? ?71??91110? ?99029? L ?000?00000L ?0??L00? 0? L? ??? ?0??000?00OLL??? LL L01?002111??111LLLOLL?? I LLL?10? " ?C ?? " ?????? ?101LOILL10000OLL?LL102LLL011L0010?? ?101001LL10000OLL??L102LLL011L ???? ? (i)100? ????002?0? ? ? ?? ?1????0?0???00010"9"?0?000???00?0??0LO?? L000000??0??L00?0?L? ? ILL? ??LL110?? 112L ? 111LLLO ?11? ?00111? ??00010?? LO LO LO 0000 L L ?000L00000L LOL ? 110? ? ?? ??? ?1? ??002?0?? ? L000000?7?7??1?? LL ? ILLL L010L L ?00000L ?? " ? ? ?? LL?000?OLL???102LLL?11L ? ?? ? ? ?LLL?? ?????? ???? 0 L1OILO1LL10000?LL?7?102LLL??1L00101??111? ??????0?0???000?01? ?100? ??0?10? ? ?229" Note. 0 = ples omorphic state; 1, 2, 3 = apomorphic states;? = character not available; L = logical impossibility; P = polymorphic states (0 and 1).