Baixe o app para aproveitar ainda mais
Prévia do material em texto
Origin and evolution of gnathostome dentitions: a question of teeth and pharyngeal denticles in placoderms Zerina Johanson1 and Moya M. Smith2 1 Palaeontology, Australian Museum, 6 College Street, Sydney, NSW 2010, Australia (E-mail : zerinaj@austmus.gov.au. ph: 61(2) 9320-6142; fax : 61(2) 9320-6050) 2 Craniofacial Development, Dental Institute, Kings College London, and Earth Sciences, University of Bristol E-mail : moya.smith@kcl.ac.uk: current address : MRC Centre for Developmental Neurobiology, Kings College London SE1 1UL, UK (Received 31 March 2004; revised 11 November 2004; accepted 16 November 2004) ABSTRACT The fossil group Placodermi is the most phylogenetically basal of the clade of jawed vertebrates but lacks a marginal dentition comparable to that of the dentate Chondrichthyes, Acanthodii and Osteichthyes (crown- group Gnathostomata). The teeth of crown-group gnathostomes are part of an ordered dentition replaced from, and patterned by, a dental lamina, exemplified by the elasmobranch model. A dentition recognised by these criteria has been previously judged absent in placoderms, based on structural evidence such as absence of tooth whorls and typical vertebrate dentine. However, evidence for regulated tooth addition in a precise spatiotemporal order can be observed in placo- derms, but significantly, only within the group Arthrodira. In these fossils, as in other jawed vertebrates with statodont, non-replacing dentitions, new teeth are added at the ends of rows below the bite, but in line with biting edges of the dentition. The pattern is different on each gnathal bone and probably arises from single odontogenic primordia on each, but tooth rows are arranged in a distinctive placoderm pattern. New teeth are made of regular dentine comparable to that of crown-gnathostomes, formed from a pulp cavity. This differs from semidentine previously described for placoderm gnathalia, a type present in the external dermal tubercles. The Arthrodira is a derived taxon within the Placodermi, hence origin of teeth in placoderms occurs late in the phylogeny and teeth are convergently derived, relative to those of other jawed vertebrates. More basal placoderm taxa adopted other strategies for providing biting surfaces and these vary substantially, but include addition of denticles to the growing gnathal plates, at the margins of pre-existing denticle patches. These alternative strategies and apparent absence of regular dentine have led to previous interpretations that teeth were entirely absent from the placoderm dentition. A consensus view emerged that a dentition, as developed within a dental lamina, is a synapomorphy characterising the clade of crown-group gnathostomes. Recent comparisons between sets of denticle whorls in the pharyngeal region of the jawless fish Loganellia scotica (Thelodonti) and those in sharks suggest homology of these denticle sets on gill arches. Although the placoderm pharyngeal region appears to lack denticles (placoderm gill arches are poorly known), the posterior wall of the pharyngeal cavity, formed by a bony flange termed the postbranchial lamina, is covered in rows of patterned denticle arrays. These arrays differ significantly, both in morphology and arrangement, from those of the den- ticles located externally on the head and trunkshield plates. Denticles in these arrays are homologous to denticles associated with the gill arches in other crown-gnathostomes, with pattern similarities for order and position of pharyngeal denticles. From their location in the pharynx these are inferred to be under the influence of a cell lineage from endoderm, rather than ectoderm. Tooth sets and tooth whorls in crown-group gnathostomes are suggested to derive from the pharyngeal denticle whorls, at least in sharks, with the patterning mechanisms co-opted to the oral cavity. A comparable co-option is suggested for the Placodermi. Key words : Gnathostomata, jawed vertebrates, crown-group gnathostomes, total-group gnathostomes, Placodermi, dentition, endoderm. Biol. Rev. (2005), 80, pp. 303–345. f 2005 Cambridge Philosophical Society 303 doi :10.1017/S1464793104006682 Printed in the United Kingdom CONTENTS I. Introduction ................................................................................................................................................. 304 II. Gnathostome dentitions and vertebrate phylogeny ................................................................................ 305 (1) Alternative theories on the origin of teeth in stem-group gnathostomes ...................................... 305 (2) Characters of the dentition in crown-group gnathostomes ............................................................. 308 (3) Tooth addition with replacement in jawed vertebrates ................................................................... 309 (4) Phylogenetic relationships of the Placodermi .................................................................................... 309 III. Placoderm dentitions : contribution of teeth and denticles .................................................................... 311 (1) Oral cavity : presence of teeth and denticles ..................................................................................... 311 (a) Arthrodira : arrangement of teeth ................................................................................................. 312 (i) Evidence in coccosteomorph arthrodires for the presence of teeth .................................. 312 (ii) Diversity and arrangement of structures in the coccosteomorph dentition .................... 315 (iii) Juvenile dentitions : toothed rows ......................................................................................... 315 (iv) Evidence in pachyosteomorph arthrodires for the presence of teeth ............................... 315 (b) Other arthrodires : tooth rows and denticulated ‘areas ’ ........................................................... 317 (i) ‘Primitive brachythoracids ’ : teeth and denticles ................................................................ 317 (ii) Phlyctaeniida : denticle fields .................................................................................................. 319 (iii) Actinolepida: tooth addition .................................................................................................. 321 (c) Phyllolepida: denticle fields ............................................................................................................ 321 (d) Acanthothoraci : supragnathal plates ............................................................................................ 321 (e) Antiarchi and Rhenanida : external dermal elements in the dentition .................................... 323 (f) Other denticles in the oral cavity : parasphenoid and paraotic plates ...................................... 325 (2) Pharyngeal cavity : presence of pharyngeal denticles ....................................................................... 325 (a) Gill arches and skeletal components ............................................................................................. 325 (b) Postbranchial lamina and denticle pattern .................................................................................. 327 IV. A developmental model from phylogenetic pattern ............................................................................... 328 (1) Ectodermal against endodermal patterning ...................................................................................... 328 (2) Tooth retention and tooth addition ................................................................................................... 329 (3) Tissue growth in teeth and denticles .................................................................................................. 330 (4) Adaptive growth and tissue remodelling ............................................................................................334 (5) The placoderm oral cavity ................................................................................................................... 336 (6) The placoderm pharyngeal cavity ...................................................................................................... 337 V. Conclusions .................................................................................................................................................. 339 VI. Acknowledgements ...................................................................................................................................... 340 VII. References .................................................................................................................................................... 340 I. INTRODUCTION Classic theories of dental evolution are based on the acqui- sition of teeth at the jaw margins by the assumed evol- utionary migration of external skin denticles surrounding the mouth into the oral cavity (see Jollie, 1968). A develop- mental mechanism was provided to explain this evolution- ary pattern of change, from jawless to jawed fishes, in location of the denticles from skin to oro-pharynx and then to an enlarged, specialised set at the jaw margins (Reif, 1982). It was assumed that all denticles are homologous (those of the skin and oro-pharynx, as odontodes, Ørvig, 1977) and that the agreed phylogenetic order of their ap- pearance was skin denticles first, then oral denticles prior to teeth at the jaw margins, and lastly, pharyngeal denticles (Jollie, 1968). However, new fossil evidence on denticle dis- tribution relative to the origin of jaws has provided an alternative to this classic consensus view (van der Brugghen & Janvier, 1993; Donoghue & Smith, 2001; Purnell, 2001; Smith & Coates, 2001). Opinion is currently divided and there is active disagreement and debate as to the validity of these two theories (Smith and Johanson, 2003a, b ; Burrow, 2003; Miller, Cloutier & Turner, 2003; Young, 2003). Sire (2001) and Sire & Huysseune (2003) have also advocated a different definition of teeth, denticles and odontodes as dermal skeletal elements from studies of development in extant osteichthyans, including ‘extra oral teeth ’ (Sire, 2001). The main objective of this review is to present new data on the phylogenetic origins of teeth and pharyngeal den- ticles, focusing on the jawed vertebrate group Placodermi, phylogenetically basal jawed vertebrates and sister-taxon to the crown-group Gnathostomata (Fig. 1). We attempt to establish the phylogenetic sequence in which characters of the dentition are expressed within the Placodermi including pharyngeal denticle patterns comparable to those observed in the thelodont Loganellia scotica Traquair, 1898 (Johanson & Smith, 2003). We interpret our morphological observations in the context of a developmental model based on pattern- ing of tooth addition as identified in extant jawed vertebrates 304 Zerina Johanson and Moya M. Smith (Smith, 2003), using these data in the placoderm dentition to consider homology with teeth of crown-group gnatho- stomes. Our observations are placed in a phylogenetic context, based on recent work by Goujet & Young (1995) and Goujet (2001; Fig. 2) for higher-level placoderm phylogeny and Carr (1995) and Lelie`vre (1995) for arthrodiran relationships (Figs. 2, 4). Our central objective is to establish in which placoderm taxa teeth are present and to determine whether they are restricted to derived taxa within the group, as re- cently proposed (Smith & Johanson, 2003b). Using information from the Placodermi, we attempt to address some fundamental questions concerning the evol- utionary origins of jawed vertebrate dentitions. (1) Do oral denticles derive from denticles on the external dermal armour, and do these subsequently evolve, with co- option of their ectodermally derived pattern, into teeth at the jaw margins? Or (2), does this pattern derive exclusively from that of internal pharyngeal denticles, as such being endodermally derived? (3) Are pharyngeal denticles them- selves derived from the posterior migration of skin denticles into the caudal end of this region, or do they have a distinct and separate evolutionary history? Or (4), alternatively, are they derived from dermal denticles formed at the junctions of ectoderm with endoderm in the post-cranial region, either laterally at each pharyngeal pouch, or solely at the sites of the postbranchial laminae? II. GNATHOSTOME DENTITIONS AND VERTEBRATE PHYLOGENY (1) Alternative theories on the origin of teeth in stem-group gnathostomes Classic ideas on the evolutionary origins of teeth suggest that the denticles became enlarged to form teeth subsequent to their migration from external skin to oral margins, as the ectoderm invaginated to form the stomatodeum during development of the mouth (Fig. 3; Jollie, 1968). Develop- mental mechanisms proposed for this are that teeth devel- oped from an innovative tooth-making structure, an ectodermally derived tissue known as the epithelial dental lamina, such that teeth formed out of the bite and deep to the surface (Reif, 1982; Smith & Hall, 1993, see their Fig. 2). This tissue structure alone patterned the entire dentition, while denticles found posteriorly in the pharyngeal region originated later in the evolutionary process (Broili, 1933; Watson, 1937; Halstead Tarlo & Halstead Tarlo, 1965; Vertebrata Total Gnathostomata Jawed Vertebrates Crown Gnathostomata ?5 ?5 ?5 5 5 2, 4, 6 2, 6 3 1 2, 6 S A R C O P T E R Y G II A C T IN O P T E R Y G II A C A N T H O D II C H O N D R IC H T H Y E S O S T E O S T R A C I T H E LO D O N T I A N A S P ID A H E T E R O S T R A C I C O N O D O N TA LA M P R E Y S H A G FIS H C E P H A LO C H O R D A TA P LA C O D E R M I Osteichthyes Stem Gnathostomata Fig. 1. Cladogram showing relationships of jawed vertebrates, including the Placodermi and crown-group Gnathostomata (adapted from Donoghue et al., 2000, Purnell, 2001). Character states : 1, mineralised skeleton, odontode histogenesis ; 2, internal mineralised skeleton; 3, external mineralised skeleton ; 4, jaws ; 5, teeth ; 6, genes responsible for organising and patterning internal skeleton. ARTHRODIRA Brachythoraci Eubrachythoraci Pachyosteomorpha A can th o th o raci R h en an id a A n tiarch i P tycto d o n tid a P etalich th yid a A ctin o lep id a P h lyctaen iid a co cco steo m o rp h s 'primitive brachythoracids' M aideria B uchanosteus A ntineosteus H olonem a Fig. 2. Cladogram showing relationships of taxa assigned to the Placodermi (adapted from Goujet & Young, 1995 and Goujet, 2001) and relationships within the Arthrodira (see also Fig. 4). Note that in Goujet, 2001, the Rhenanida and Acanthothoraci are resolved as sister taxa. Origin and evolution of gnathostome dentitions 305 Jollie, 1968; Schaeffer, 1975; Burrow, 2003; Young, 2003). As such, these theories assume that mechanisms for the de- velopment and pattern regulation of skin denticles became co-opted to make teeth, involving an inductive ectodermal epithelium. Tooth-making ability has been linked to the acquisition of jaws, when the mandibular arch moved cau- dally together with the invagination of the ectoderm (Fig. 3; see Fig. 4 in Jollie, 1968). It was assumed that the inner denticulate surface of the first branchial arch was then used for food processing at an early phylogenetic stage, with these changeslinked to increasing predatory activity (Jollie, 1968). In recent years, several aspects of the classic theories (Jollie, 1968; Reif, 1982) have been challenged. For ex- ample, rather than accepting an evolutionary link between the development of a dentition and articulated jaws during increasingly active predation, Mallatt (1998) proposed the ventillatory theory, in which jaws evolved first for suction feeding, while teeth developed at a later stage of gnatho- stome evolution. Also, both systems, jaws and teeth, were proposed to be independent units in their development, and thus also change independently during evolution (Smith & Hall, 1993). Subsequently, it was recognised that although odontodes shared a homologous developmental programme for histogenesis, internal oro-pharyngeal denticles and ex- ternal skin denticles possessed different and divergent evol- utionary histories (Smith & Coates, 1998). Because these originate at a phylogenetically basal node in vertebrate phylogeny (Fig. 1), one could not be derived from the other. One topographic basis for the difference between oro- pharyngeal denticles and dermal skin denticles is the pro- posed phylogenetic independence of the visceral skeleton (splanchnocranium, Kardong, 2002) from the dermal and endoskeletal systems (Donoghue & Sansom, 2002). Donoghue and Sansom (2002) suggest that this distinction occurred at a level below the origin of the Vertebrata (a clade including the last common ancestor of hagfish, lam- preys, crown-group gnathostomes and all descendants of this ancestor, including fossil jawless fishes as well as the Antiarchi Arthrodira B rachythoraci Phlyctaeniida Phyllolepida A ctinolepida Ptyctodontida Petalichthyida Euantiarchi Yunnanolepids R henanida A canthothoraci B rindabellaspis W uttagoonaspis Fig. 4. Cladogram showing relationships within the Arthrodira. Adapted from Lelie`vre (1995) and Carr (1995). stom. stom. phar. phar. ph.d sp A B D E C t.pl phar. pter ce.h hy.h ba.h sp.sk PBL ph.d or.t 2nd.a 1st. a hy.a ma.a Fig. 3. (A–E) Diagrams to show generalised agnathan (A, B) and gnathostome (C, D) in longitudinal horizontal section, of types with both an armoured dermal skeleton (A, D) and one with separate denticles (placoid scales, B, C). (E) A transverse section through the pterygoid and hyoid arch in a gnathostome with denticulated pharyngeal plates (modified from Figs. 3, 4, 5 in Jollie, 1968). These show the limits of the ectoderm (hatched) and endoderm (shaded), as predicted by Jollie (1968). Double arrows mark the equivalent position of the postbranchial lamina (PBL) in placoderms. Establishing the positions of the ectoderm-endoderm boundary is proposed as a key to pattern information in development of oro-pharyngeal denticles and oral teeth (or.t). (B) Adapted from Jollie (1968) to show the position of pharyngeal denticles (ph.d) based on new infor- mation in Loganellia scotica. Other abbreviations : 1st.a, 2nd.a, first and second gill arches ; ba.h, basihyal ; ce.h, ceratohyal ; hy.a, hyoid arch ; hy.h, hypohyal ; ma.a, mandibular arch; phar., pharyngeal cavity ; pter, pterygoid ; sp, spiracle ; sp.sk, splanchno-skeleton; stom., stomodeum; t.pl, tooth plates. 306 Zerina Johanson and Moya M. Smith Placodermi ; Fig. 1). Oro-pharyngeal denticles and teeth are included as part of the splanchnocranium, some fused to splanchnic bones of the palate and dermal bones of the jaws, along with the supporting branchial cartilages and their perichondral and endochondral bones (Nelson, 1969, 1970a ; Donoghue & Sansom, 2002; Kardong, 2002; Smith, 2003). As discussed below (Section IV.1), splanchnic structures are hypothesised to be dependent on endoderm, rather than ectoderm, for pattern information. Another observation supporting the divergent evolution- ary histories of internal and external skeletal systems is that the Conodonta (total-group Gnathostomata, Fig. 1) are the first vertebrate group to show skeletal mineralisation. Importantly, this occurs exclusively within the oro-pharyn- geal cavity, possibly associated with the splanchnoskeleton (Smith & Coates, 1998, 2000, 2001; Donoghue, Forey & Aldridge, 2000; Donoghue & Aldridge, 2001; Donoghue & Sansom, 2002). Conodont elements show both bilateral symmetry and antero-posterior differences as evidence of organisation or patterning. Given their known position in the conodont animal, ventral and posterior to the eyes, they are topographically co-incident with other feeding struc- tures ; the anterior elements could be part of the oral cavity, with the more posterior part of the pharyngeal apparatus (Donoghue & Purnell, 1999, see their Fig. 1). Other early total-group gnathostomes (Heterostraci, Anaspida, Osteostraci, Fig. 1) are characterised by a heavily min- eralised dermal armour and the apparent absence of internal dermal structures (Smith & Coates, 1998, 2000, 2001; Donoghue & Sansom, 2002). Although Donoghue & Sansom (2002) suggest that these internal and external el- ements can shift topographically between these regions, in- cluding from internal to external, in fact the two regions are distinct at the early phylogenetic stage represented by the Conodonta. As a new theory distinct from the classic one, Smith & Coates (1998, 2000, 2001) suggested that the essential pat- terning mechanisms required for the tooth sets characteristic of crown-group Gnathostomata evolved from the co-option of regulatory mechanisms associated with organised internal pharyngeal denticles in the jawless vertebrate Loganellia sco- tica Thelodonti ; Section IV.1). Significantly, as with the Conodonta, this was prior to the appearance of jaws. This new hypothesis essentially reversed the inferred polarity of dental evolution. Rather than an external origin from der- mal denticles with embryonic ectoderm recruited to make teeth at the mouth margins, an internal origin from embry- onic endoderm of the pharyngeal cavity was proposed. For example, the initial iterative pattern in the zebrafish Danio rerio pharyngeal region is dependent on endodermal signal- ling, which is also required to make the separate cartilages in the branchial arches (Piotrowski & Nu¨sslein-Volhard, 2000). Donoghue (2002) noted that the external skeleton (scales) is patterned, in a manner comparable to other dermally de- rived elements such as feathers, however, this patterning involving the ectoderm differs significantly from patterning associated with the splanchnocranium and with structures such as the denticle/tooth whorls. Despite these attempts to revise the theories on the origin of teeth, their origin from developmentally distinct pharyngeal denticles has been questioned in recent pub- lications. The first, by Purnell (2001), demonstrated that amongst other jawless vertebrates (Heterostraci, Fig. 1), oral denticles at the mouth margin of heterostracans intergraded morphologically with dentine ridges on the external head- shield. Thus, Purnell (2001) concluded that external dermal structures gave rise to oral denticles in heterostracans, as predicted by the more traditional theories of dental evol- ution. The heterostracan oral denticles also show a degree of organisation, such that the denticle tips are oriented out- wards, leading Purnell (2001) to suggest that the develop- mental differences between oro-pharyngeal denticles and skin denticles in thelodonts described by Smith & Coates (1998) were not significant. However, both thelodonts and elasmobranchs have comparably oriented denticles in the prenasal sinus and inhalent nasal openings, both respect- ively, derived from ectoderm (van der Brugghen & Janvier, 1993; Purnell, 2001). Denticles associated with the bran- chial arches in the pharyngeal region of Loganellia scotica are significantly different from the more anteriororo-nasal examples (Smith & Coates, 2001), forming joined sets or whorls with sequential addition from one direction only. Regulation of the pharyngeal system would appear to be more comparable to the patterned tooth sets in crown-group gnathostomes than either those of skin, oral or nasal den- ticles, and hence more likely to have been co-opted as a developmental system to pattern dentitions. Miller et al. (2003, Section II.4) have recently described the earliest known articulated shark in which they noted that the branchial denticles differed morphologically from teeth in the jaw. Tooth families were visible and showed newer teeth sitting in a space representing the dental lamina. We can conclude from Miller et al. (2003) that the tooth system was well regulated from a dental lamina and that this con- tinued along the length of the jaw to the more posterior tooth families, which are also distinctly different from the pharyngeal denticles. However, Miller et al. (2003) state that these posterior tooth sets are more similar to modified der- mal scales, based on absence of ‘ the dental membrane’. They suggested this evidence, and the differences between pharyngeal denticles and teeth, contradicted an origin of teeth from pharyngeal denticles. However, the term ‘dental membrane’ is obscure and not properly defined by the authors and we suggest that these posterior sets are teeth, developing from a superficial dental lamina, rather than scales. Differences between teeth and branchial denticles are misunderstood by the authors, as the concept advanced by Smith & Coates (1998, 2000, 2001) suggests that it is the regulatory and patterning mechanisms for sets of denticles that are co-opted, not just denticle morphology. Furthermore, denticles in the anterior mucous membranes near the mouth margin are polygonal in shape, differing morphologically from the rounded external skin denticles of the head, suggesting a separation, rather than continuation of, or gradation between, these denticulated areas. Contrary to the authors’ views, this does not support classic theories of dental evolution. In a phylogenetically based critique, Donoghue & Smith (2001), using new analyses, argued that Loganellia scotica is derived within the Thelodonti, rather than resolved to a Origin and evolution of gnathostome dentitions 307 position at the base of the group. Branchial whorls are not known from the more basal thelodonts and based on this phylogenetic resolution the derivation of tooth whorls from denticle whorls in Loganellia scotica is problematic. Similarities in organisation within the denticle whorls (Loganellia scotica) and tooth whorls (crown-group gnathostomes) observed by Smith & Coates (1998, 2000, 2001) would be convergently acquired. Burrow (2003), commenting on Smith & Johanson’s (2003b) identification of teeth in derived placoderm groups, suggested that more basal placoderm groups possessed not teeth, but tubercles, on the dental plates, organised in the same manner as the ‘ teeth ’ identified by Smith & Johanson (2003b). Burrow (2003) suggested that the tubercles on the dental plates are not randomly distributed in these basal placoderm groups, such as the Acanthothoraci, Rhenanida and Antiarchi (Figs. 2, 4), a contention discussed, and re- futed in Section III.1d–f. Burrow (2003, see her Fig. 1A) identified a gnathal plate from the Acanthothoraci and suggested that the arrangement of tubercles on this plate was comparable to external skin tubercles on the head and trunkshield plates. This led her to conclude that oral den- titions were derived from external skin tubercles, as suggested by classic theories of dental evolution. However, as noted by Smith & Johanson (2003a), the particular microvertebrate specimens Burrow (2003) illustrates as being an acantho- thoracid placoderm are of uncertain affinity, in all prob- ability acanthodian. Young (2003) also supported the derivation of all placo- derm dentitions from external skin tubercles and queried the characters specified by Smith & Johanson (2003b) to identify teeth in placoderms. He noted, as we have, that external dermal tubercles are also patterned (as discussed above, Donoghue, 2002), justifying this by illustrating a row of pointed tubercles along the edges of the arthrodiran spinal plate and antiarch pectoral fin. However, Young’s (2003) suggestion that these denticles satisfy the criteria used for the recognition of teeth by Smith & Johanson (2003b), is not adequately supported by his own figures (Fig. 1E–G in Young, 2003, also Stensio¨, 1948). We maintain that teeth (Young’s ‘denticles ’) in brachythoracid arthrodires differ from external tubercles in their spatiotemporal regulation and see no evidence to refute this (Smith & Johanson, 2003a, b). Differences between oral denticles on the phlyc- taeniid Dicksonosteus arcticus Goujet, 1975 (Fig. 10H) and tubercles on the external surface of headshield plates are readily apparent (see plate 5, 2b, 5a, in Goujet, 1975; Figs. 39, 41, 42, and plate 8.5B, 8.6A in Goujet, 1984a). External dermal elements are incorporated into the den- tition of two phylogenetically basal placoderm groups (Section III.1e), but these are not teeth, nor do they form part of a patterned dentition, nor are the dentitions of more advanced placoderms derived from these. (2 ) Characters of the dentition in crown-group gnathostomes Although teeth and skin denticles or tubercles are hom- ologous at the tissue development level (Ørvig, 1977; Schaeffer, 1977; Smith & Hall, 1990; Donoghue, 2002), the mechanisms regulating their respective developmental pat- terning (Donoghue, 2002) and specifying the positions for addition and replacement of teeth are non-homologous. As noted above, their divergent evolutionary histories (Smith & Coates, 1998, 2001; Smith, 2003) (Section II.1) can be traced to the base of the total group Gnathostomata and the Conodonta (Fig. 1). In the oral cavity both teeth and oral denticles are used to build dentitions in crown-group gnathostome taxa but the use of these terms as defined by Ørvig (1977) and Reif (1982) is often not stringently applied (Smith, 1988). Distinction between the two terms involves developmental concepts and may be difficult to apply to fossil taxa (Schaeffer, 1977; Reif, 1982). This can cause confusion and lack of precision in comparisons between dentitions (for a discussion on lungfish dentitions see Smith, 1988). Distinctively new teeth form within a dental lamina below the biting surface of the jaw and in advance of participation in the bite. By contrast, denticles develop superficially, are widely dispersed, and form when a space becomes available due to general or local growth, or loss of a previous denticle. From these topo- graphical criteria, both the arrangement and relative timing of addition can be deduced. In particular, the position of putative tooth primordia can be inferred from the presence of a regulated and organised sequence of tooth addition, exemplified by tooth whorls in fossil taxa, but said to be absent in placoderms (Reif, 1982). In extant chondrichthyans (sharks, rays, and holocephalans), tooth ‘ families ’ or ‘sets ’ have been recognised, with the position of the tooth primordia in a deep epithelial fold (dental lamina). The newest teeth added to sets in early growth stages are the largest and the locations of tooth pri- mordia are not random, but part of an organised, timed sequence, regulated proximo-distally at the initiation stage of the dentition (Reif, 1976, 1984; Smith & Coates, 2001). Theoretical models for patterning tooth development and replacement in crown-gnathostome dentitions have recently been reviewed (Smith, 2003). Several taxa possess a stato- dont dentition, where teeth are not replaced but are retained in the tooth set (Jaekel, 1901; Patterson, 1992) and joined at their bases, with the newest teethshowing lyodont addition (from the lingual side of the jaw). These terms have been discussed for holocephalans (Patterson, 1992), other chondri- chthyans (Smith & Coates, 1998), acanthodians and placo- derms (Ørvig, 1973), and in general by Smith & Coates (1998, 2000, 2001) where the condition is considered to be primitive for jawed vertebrates. In many osteichthyians (bony fish), teeth are individually replaced except for tooth whorls at the jaw symphysis, but there are also taxa with a completely statodont dentition, notably the dipnoans (Denison, 1974; Smith, 1988). Tooth whorls have been noted in sarcopterygians [Psarolepis romeri Yu, 1998 (Zhu, Yu & Janvier, 1999), porolepiforms (Jarvik, 1972), onychodonts (Jessen, 1966; Long, 2001)] and a very limited number of actinopterygians (e.g. Howqualepis rostridens Long, 1988). In the fossil group Acanthodii, these tooth whorls occur not only at the jaw symphysis, but also as individual whorls, or sets, along the jaw margin in certain taxa [Brochoadmones milesi Bernacsek & Dineley, 1977 (Gagnier &Wilson, 1996)]. Other acanthodians (Family Ischnacanthidae; Long, 308 Zerina Johanson and Moya M. Smith 1986) possess rows of evenly spaced teeth of graded size, in which the largest represents the newly added tooth (Smith, 2003). The statodont condition is also recognisable in placoderm dentitions, with the cutting edges and tusks of the gnathalia maintained as continuous growth regions. In statodont dentitions teeth are added at the growth margins but the older teeth are not shed as they are worn down and incor- porated at the cutting edges, so that strictly in situ replace- ment does not occur. The contribution of teeth to these surfaces was proposed by Ørvig (1980a), and recently re- evaluated by Smith & Johanson (2003b), most notably within the Arthrodira. In this taxon examples of new teeth and older teeth, as part of the regulated tooth-addition unit, can all be observed in a variety of well-preserved taxa (see Section III.1a, b). (3 ) Tooth addition with replacement in jawed vertebrates Tooth replacement patterns in fish, amphibians and reptiles have been reviewed by Berkovitz (2000) and the develop- mental models for these by Smith & Coates (2001) and Smith (2003). The developmental mechanisms of this re- placement relative to the dental lamina (Reif, 1982), either continuous or discontinuous, and the regulatory genes con- trolling the pattern of tooth positioning are now being studied in fish (Fraser, Graham & Smith, 2004; Laurenti et al., 2004). Laurenti et al. (2004), reporting on the zebrafish pharyngeal teeth, implicate one gene in particular (Eve1) with epithelial initiation of the first tooth primordium in each dentate region. Fraser et al. (2004) report on epithelially expressed genes (Shh, Pitx2) important in creating both first and replacement teeth in the trout Oncorhyncus mykiss and show that the same genes are deployed in trout, in oral and pharyngeal teeth, at exactly equivalent stages of tooth de- velopment as in the mouse. They conclude that the genes needed to build a dentition are conserved from fish to mouse, thus a molecular tool kit for making teeth emerged early in vertebrate evolution. Amongst pharyngognath fish (functional teeth only found on the pharyngeal arches), the cichlids (Teleostei : Cichlidae) are noted for their phenotypic plasticity, forming in Astatoreochromis alluaudi fine sharp teeth or large molariform teeth in response to changes in diet as the teeth are cyclically replaced (Greenwood, 1965; Huysseune, 1995, 2000). Initial teeth form directly from the pharyngeal epithelium but tooth replacement is from the dental lamina deep in the bone, whereas other replacement pharyngeal teeth form from the dental epithelium of the predecessor tooth (Huysseune, Sire & Meunier, 1994; Huysseune, 1995; Huysseune & Sire, 1998). Most recently, this epithelium is considered to be a stem cell population (Huysseune & Thesleff, 2004) where together with odontogenic (tooth- producing) mesenchyme it can form a tooth primordium for a replacement tooth. There can be little doubt that in both the developmental and functional senses, true teeth form in pharyngognaths despite their location on the fifth branchial arch. Importantly, they are truly a part of the splanchno- skeleton and under the putative inductive influence of endoderm lining the pharyngeal cavity (Huysseune et al., 2002; Section IV.1). Recent comparisons between chondrichthyans and bony fishes have suggested that the concept of a ‘dental primor- dium’, or ‘ tooth clone ’ sensu lato, responsible for forming new teeth in each set, explains observed morphologies, particularly in bony fishes (Smith, 2003). In tooth replace- ment on the jaw of the trout Oncorhynchus mykiss (Berkovitz, 1977) and Polypterus senegalus (Cuvier, 1829) (Clemen, Bartsch & Wacker, 1998), the new teeth form from the dental epithelium at the base of the old one (comparable to the cichlids above). In these examples tooth addition loci are close to functional teeth rather than deep within a continu- ous dental lamina along the jaw margin as in elasmobranchs (Reif, 1982). Positions of the first tooth loci along the jaw determine where each new replacement tooth-germ forms. In this way, odontogenic cells (a clonal group, see Smith, 2003) of the initiator tooth buds form the primary (original) pattern and autonomously regulate the secondary replacement pattern. We suggest that these observations on the variability of development in osteichthyan and chondri- chthyan dentitions provide criteria for the recognition of different dentition patterns in fossil forms, for example, re- lated to identifiable locations of tooth initiation and addition. In Section III, we apply these criteria to the Placodermi. (4 ) Phylogenetic relationships of the Placodermi The Placodermi was first erected by M’Coy (1848) to in- clude the groups Antiarchi and Arthrodira, but also what we now recognise as a jawless fish, Psammosteus Agassiz, 1845 (Heterostraci). Traquair (1888) restricted antiarchs and arthrodires (the genus Coccosteus Agassiz, 1844) to the group Placodermata Owen, 1860, but later, Cope (1889) assigned the Antiarchi to the Ostracodermi, along with pteraspid (Heterostraci) and cephalaspid (Osteostraci) jawless fishes. The Antiarchi were believed to lack jaws, and Cope’s (1889) classification persisted for many years (Woodward, 1891; Dean, 1895; Goodrich, 1909). As well, Woodward (1891; Dean, 1895) allied the Arthrodira with the Dipnoi (lungfish). Thus, the Placodermi have not always been considered a monophyletic group. However, in-depth study of anaspid and cephalaspid jawless fishes made it clear that these were very different from the Antiarchi (Kiaer, 1924; Stensio¨, 1927) and shortly after, upper and lower jaw elements were discovered for the Antiarchi (Stensio¨, 1931). Stensio¨ (1927, 1931) returned to the usage of the Placodermi including only the Antiarchi and Arthrodira (or Euarthrodira, at that time including all other taxa currently assigned to the Placodermi, also Stensio¨, 1969) and suggested they were most closely related to elasmobranchs (Chondrichthyes). Placoderm monophyly was also questioned by workers who noted similarities between the Holocephali (Chondri- chthyes) and the Arthrodira (Stensio¨, 1925, 1934) and between the chimaeroids (Holocephali) and the placoderm group Ptyctodontida (Holmgren, 1942). However, some of these similarities have been shown to be non-homologous (Patterson, 1965, 1992; Gardiner, 1984b ; Forey & Gardi- ner, 1986), and currently the Ptyctodontida are retained within the Placodermi (Figs. 2, 4). Origin and evolution of gnathostome dentitions 309 Attempts to evaluate the relationships of taxa assigned to the Placodermi in a phylogenetic framework have been made by a variety of researchers (Miles & Young, 1977; Denison, 1978b; Young, 1980; Gardiner, 1984b ; Goujet, 1984b ; Forey & Gardiner, 1986; Goujet & Young, 1995). In these analyses, the Arthrodira, in its modern usage (Figs. 2, 4), is resolved as one of the most derived groups within the Placodermi. Discrepancies amongst the phylogenies have mainly centred on the positions of the groups Antiarchi and Ptyctodontida and monophyly of the Acanthothoraci. In recent analyses, certain poorly understood groups have been excluded, for example, the Stensio¨ellida and Pseudopeta- lichthyida, with some question as to whether these belong to the Placodermi (Janvier, 1996a). With regards to the Ptycto- dontida, Miles & Young (1977) retained the group within the Placodermi, but suggested that their phylogenetically basal position was based on the presence of the pelvic and pre-pelvic claspers, otherwise present in chondrichthyans (holocephalians having both structures, elasmobranchs having the pelvic claspers alone). Other placoderms were considered derived in the absence of these claspers (Miles & Young, 1977). However, the chondrichthyan pelvic and pre-pelvic claspers are supported by cartilage, but in the ptyctodont Ctenurella, claspers consist only of dermal spines or plates, with no cartilaginous support. This led Forey and Gardiner (1986; also Gardiner, 1984b) to suggest that ptyctodont and chondricthyan claspers were not homol- ogous. In Forey and Gardiner’s (1986) analysis, ptyctodonts are still the most basal placoderm taxon, lacking a pre- pectoral process on the scapulocoracoid, dorsal narial openings and sensory canals in open grooves. However, in other analyses (Goujet, 1984b), the absence of dorsal narial openings is shared between arthrodires and ptyctodonts (see below) and the presence of sensory canals in tubes, rather than open grooves, is shared with petalichthyids (these are shown as reversals in Forey & Gardiner, 1986). In current analyses, the Acanthothoraci is resolved to the basal position within the Placodermi (Goujet & Young, 1995). In most phylogenies, the Antiarchi was considered to be the sister group of the Arthrodira, based on the greater length of the trunkshield relative to other placoderm taxa, considered to be the derived condition (Miles & Young, 1977; Denison, 1978b ; Young, 1980; Gardiner, 1984b ; Forey & Gardiner, 1986; Goujet & Young, 1995). By con- trast, Goujet (1984b) focused on characters of the head- shield, including the dorsal position of the nasal openings and the presence of a premedian plate, which led to a hy- pothesis of relationships between the Antiarchi, Rhenanida and Acanthothoraci. In the most recent available clado- gram, these three groups are resolved to sequential nodes near the base of the cladogram, again with Arthrodira as one of the most derived taxa (Figs. 2, 4 ; Goujet & Young, 1995). In previous phylogenies, the Acanthothoraci (=Palaeacanthaspida) was a monophyletic group, based on the posterior projection of the paranuchal plates of the headshield (Miles & Young, 1977; Young, 1980; Gardiner, 1984b). However, in the cladogram presented by Goujet (1984b), the Acanthothoraci are paraphyletic, with some taxa more closely related to antiarchs and others to the Rhenanida. Janvier (1996a) also noted that the group lacked any synapomorphies. Nevertheless, most recent phylogenies (Goujet & Young, 1995; Goujet, 2001) have retained a monophyletic Acanthothoraci, suggesting previous analyses (Goujet, 1984b) were based on poor data from incomplete specimens. Phylogenies addressing relationships of taxa assigned to the Placodermi have been varied in their resolution of indi- vidual groups, as has the position of the Placodermi within the clade of jawed vertebrates. Discussions in the upcoming sections follow recent phylogenies suggesting that placo- derms are phylogenetically the most basal group within the clade (Schaeffer, 1975; Schaeffer & Williams, 1977; Young, 1986; Coates & Sequeira, 2001; Goujet, 2001; Janvier, 2001; Zhu & Schultze, 2001). Other researchers have suggested a closer relationship of the Placodermi to chondrichthyans within the group Elasmobranchiomorphi (Stensio¨, 1925, 1927, 1931). Still others have proposed that placoderms and osteichthyans (sarcopterygians+ actinopterygians) are sister taxa, based on putative homol- ogies between placoderm trunkshield plates and ostei- chthyan shoulder girdle bones (Jarvik, 1944; Stensio¨, 1959), including a denticulated postbranchial lamina, among other characters (Forey, 1980; Gardiner, 1984b). However, these homologies have been questioned (e.g. Young, 1986) and are not supported by recent phylogenetic analyses (Zhu & Schultze, 2001). It is important to note that currently the monophyly of the Placodermi is based on a small number of characters that have not been tested in a phylogenetic analysis of the entire clade of jawed vertebrates (Goujet, 1982; Young, 1986; Janvier, 1996a ; Coates & Sequeira, 2001). Outgroup com- parison, necessary for determining character polarity, is also problematic. The recent analysis of Goujet & Young (1995) used a ‘hypothetical ancestor ’ to code the plesiomorphic state for all characters, which is not entirely satisfactory. Other researchers determined the plesiomorphic and de- rived conditions for various characters by evaluating their distribution within the Placodermi. For example, with re- gards to the trunkshield, Miles & Young (1977, p. 131) de- scribed the plates present in the ‘primitive placoderm’ by noting which plates were found among the widest range of placoderm groups. When outgroup comparisons are made, taxa utilised were often other jawed vertebrates (Miles & Young, 1977; Denison, 1978b, Gardiner, 1984b ; Forey & Gardiner, 1986). However, in order to determine character polarity in a group resolved to the basal node of a major clade like the jawed vertebrates, it is more appropriate to use the most closely related taxon outside this group, i.e. a jawless fish taxon. However, problems arise because these jawless fishes differ from members of the jawed vertebrate clade in many respects (Osteostraci) and are often known poorly, particularly internally (Heterostraci, Anaspida, Thelodonti). The Osteostraci are currently resolved as the sister group to the jawed vertebrate clade (Janvier, 1996b ; Donoghue et al., 2000; Donoghue & Smith, 2001). These have been used as an outgroup to polarise character states within the jawed vertebrates and less frequently, the Placodermi (Coates & Sequiera, 2001; Janvier, 2001; Johanson, 2002). For example, the osteostracan Norselaspis possesses a 310 Zerina Johanson and Moya M. Smith scapulocoracoid with an articular surface for a single fin radial and surrounding depressions for fin muscles and foramina for nerves and blood vessels comparable to the scapulocoracoid of jawed vertebrates (Janvier, 1985, 1996a). Comparisons between the course of these vessels in osteo- stracans and the Antiarchi suggested that the antiarch condition is plesiomorphic relative to the rest of the Placodermi and other jawed vertebrates, such that these taxa share a derived character that the Antiarchi lack (Johanson, 2002). This challenges placoderm monophyly but again has not been tested within a rigorous phylogenetic analysis. Placoderm non-monophyly could result in those taxa (Arthrodira) with teeth being resolved as more closely re- lated to crown-group gnathostomes than other placoderms. In this instance, teeth would not be independently derived within the placoderms, but consideration would still need to be given to the absence of teeth early in chondrichthyan and possibly acanthodian history (Turner & Young, 1987; Long, 1995; Janvier, 1996a ; Wilson, Hanke & Sahney, 1999; Williams, 2001). Miller et al. (2003) described the earliest articulated chondrichthyan currently known, from the early Devonian (Emsian), a specimen with up to fifteen spaced and organisedsets of teeth at the jaw margin. However, even earlier articulated shark specimens from the Early Devonian (Lockhovian) of the Canadian Arctic have been described as lacking teeth, leading Wilson et al. (1999) to propose an independent origin of teeth within the chondri- chthyans. As well, in some of the earliest known chondri- chthyan faunas from the Silurian of Russia and Asia, dermal scales are present but teeth are not. This includes taxa which may represent one of the most plesiomorphic chondri- chthyan groups, such as those assigned to the Mongolepida (Karatjute-Talimaa, 1973; Karatjute-Talimaa et al., 1990). Even earlier ‘chondrichthyan-like ’ scales, but not teeth, have been described from the Ordovician of Colorado (Sansom, Smith & Smith, 1996). III. PLACODERM DENTITIONS: CONTRIBUTION OF TEETH AND DENTICLES In current phylogenies, the Placodermi are the most plesio- morphic group to possess jaws and should provide import- ant data on the distribution of teeth and denticles at this stage of evolution, relative to jawless vertebrates such as Loganellia scotica and the Heterostraci and jawed vertebrates included in the crown-group Gnathostomata. We provide below new data and reassess published information on the feeding apparatus in all placoderm groups, together with a review of denticle distribution in the pharyngeal region based on Johanson and Smith (2003). One current opinion is that teeth (those derived from a dental lamina) are a synapomorphy of crown-group gnathostomes present in all jawed vertebrates except placo- derms (Denison, 1978b ; Reif, 1982; Goujet, 2001; Young, Lelie`vre & Goujet, 2001; Zhu & Schultze, 2001; Donoghue & Sansom, 2002; Burrow, 2003; Young, 2003). Equally, pharyngeal denticles are considered to be absent in placo- derms (Donoghue & Smith, 2001), although they are present in basal chondrichthyans and osteichthyans. Pharyngeal el- ements are largely unknown in acanthodians except for gill arches and accompanying rakers in Acanthodes bronni Agassiz 1833 (Miles, 1973). The absence of teeth in placoderms has been justified for three reasons. First, tooth whorls, or spirals, are absent from placoderm jaws; these whorls providing the structural evi- dence that a tooth-producing dental lamina is present in the crown-group gnathostomes, including the fossil group Acanthodii (Reif, 1982). For example, Young et al. (2001; also Young, 2003) recently emphasised the lack of order, or patterning (i.e. into whorls or spirals) of the denticles on the gnathalia of certain arthrodires (the group Phlyctaeniida; Figs. 2, 4). The second reason is the apparent absence of regular dentine in the dental tissues ; Young et al. (2001) emphasised the histological similarity of the denticles to the dermal tubercles on the external surface of the dermal plates of the head and trunkshield (see Section III.1.a, b). This variant of dentine has been described in both the arthrodire dentition and shield tubercles as semidentine (Ørvig, 1980a). Linked with this is the absence of a pulp cavity normally found in teeth, but not in dermal tubercles. Third, as universally accepted, an enamel covering is absent (Gross, 1957; Ørvig, 1973; Denison, 1978b). Contrary to these views, Smith & Johanson (2003a, b) described morphological and histological evidence for the presence of teeth in Placodermi, in the derived group Arthrodira, although the phylogeny indicated non-hom- ology with teeth in other jawed vertebrate clades (Figs. 1, 2). Because more phylogenetically basal placoderms such as the Antiarchi and Rhenanida (Figs. 2, 4) lack these tooth rows, teeth evolved independently within the group. The mor- phological evidence demonstrated that new teeth, with a gradual size increase and similar shape polarity, were con- sistently added to the ends of pre-existing rows, below the functional bite. Hence, the arthrodiran tooth row is very similar to a crown-gnathostome tooth whorl or family, and we have compared it to a statodont tooth row, as in dipno- ans. However, the more phylogenetically basal placoderms lacking these criteria cannot be said to possess teeth and are relevant in discussions of the origins of placoderm teeth, whether from external skin tubercles, or internal phar- yngeal denticles, because their edentate condition suggests arthrodiran teeth are neomorphic. (1 ) Oral cavity: presence of teeth and denticles We follow a developmental model of tooth addition to the dentition (Smith, 2003) and apply this to the Placodermi, to allow us to identify the evolutionary origins of the tooth- additive type of dentition and further to assess homologies with teeth in crown gnathostomes. This model is based on the statodont dentition, which essentially preserves the en- tire ontogeny of the dentition (e.g. the lungfish toothplate), allowing developmental and growth processes to be inferred and positions of newly added teeth to be identified (see Ørvig, 1980a). The functional part of the dentition, worn teeth, also tusks and cutting edges, forms from non-replaced Origin and evolution of gnathostome dentitions 311 teeth. The developmental model of tooth addition as pro- posed is defined as follows: (1) a tooth regional field (with odontogenic restriction boundaries) is associated with each dentate (gnathal) bone (Donoghue, 2002; Fraser et al., 2004). Within each field, a single separate primordium (dental placode, or clonal group of cells, gene expression centre ; Smith, 2003; Fraser et al., 2004; Laurenti et al., 2004) initiates the placoderm specific pattern for sequential tooth addition. Based on new evi- dence for the sequential, reiterative pattern of tooth initiation and morphogenesis of mammalian teeth (Jernvall et al., 2000), each primordium (dental placode) will act as a tooth signalling centre. (2) Genetic mechanisms responsible for the tooth re- gional field in (1) and the pattern of sequential, reiterative tooth development could be part of a hierarchical develop- mental module for tooth sets. In the early manifestation of this, one phylogenetic stage is oral denticles added to the margins of gnathal plates, but not in regular well-spaced rows. (3) New teeth develop in relation to the initial tooth site and are regulated as proposed in the clonal model for a discontinuous dental lamina (Reif, 1982; Smith, 2003) and by expression of epithelial genes (Fraser et al., 2004). (4) Teeth are added to a recognisable, statodont, tooth row linking the original tooth to later, regularly spaced functional teeth where each is supported by a new bone of attachment. Functional, worn teeth and newly added teeth are often incorporated and transformed into cutting edges and tusks (Section IV.3, 4) by an adaptive mechanism of pleromic growth of dentine into the bone (see Smith & Sansom, 2000). (5) New teeth of tubular dentine grow from the pulp, as in the crown-gnathostome-type. Only after further growth related to wear does this transform into semidentine (Section IV.4), the type that occurs in the dermal tubercles. Most aspects of this model can be readily recognised in the statodont dentition of arthrodiran taxa, and from the observed pattern of growth the genetic patterning processes are inferred. Placoderm feeding structures are organised into a single pair of lower jaw elements (infragnathals) and upper jaw elements known as anterior and posterior supragnathals (Fig. 5, IG, ASG, PSG respectively). Each infragnathal comprises a bony structure supported on Meckel’s cartilage ; the latter is endochondrally ossified to varying degrees in different taxa, but the infragnathal is a membrane bone and, therefore, ossified throughout its ex- tent (Figs. 6B, F,G and Figs. 7–9). The anterior and pos- terior supragnathals (Figs. 6A,C,D and 7A–G, I–K) are supported on the endocranium, on the ethmoid and auto- palatine (as part of the palatoquadrate) respectively.Homology of the upper feeding structures in the Antiarchi (Fig. 10I, J) and the Rhenanida (e.g. Nefudina qalibahensis Lelie`vre et al., 1995) to the supragnathals is problematic (Section III.1,e). Based on the current developmental model, the best evidence for teeth amongst placoderms is provided by the derived group Arthrodira (Fig. 2; Goujet & Young, 1995; Goujet, 2001). The Arthrodira can be divided into the Actinolepida, Phlyctaeniida and Brachythoraci (Fig. 4; Denison, 1978b ; Carr, 1995; Goujet, 2001) although only the Actinolepida and Eubrachythoraci (Fig. 4) possess dental morphologies most readily compared to the statodont condition of various chondrichthyans, acanthodians and osteichthyans. Dentitions of the ‘primitive brachythoracids ’ possess tooth rows in addition to areas or patches of den- ticles. Among the Arthrodira, only the Phlyctaeniida appear to lack tooth rows; instead denticles are added to the mar- gins of the gnathal plates of the dentition, showing some degree of regulation. A comparable pattern is seen within the group Phyllolepida. (a ) Arthrodira : arrangement of teeth As indicated in Figs. 2 and 4, the Arthrodira includes the Actinolepida, Phlyctaeniida and the Brachythoraci, the lat- ter containing two large groups, the coccosteomorphs and the pachyosteomorphs, along with the ‘primitive brachy- thoracids ’. Our descriptions start with the most derived arthrodiran taxa (Figs. 6–9, and 11), and proceed to the base of the clade, concluding with the Actinolepida (Fig. 10A–D; Denison, 1958). ( i ) Evidence in coccosteomorph arthrodires for the presence of teeth. Coccosteomorph taxa examined for this description include Kimberleyichthys whybrowi Dennis-Bryan & Miles, 1983, Harrytoombsia elegansMiles & Dennis, 1979, Incisoscutum ritchiei Dennis & Miles, 1981, Gogopiscis gracilis Gardiner & Miles, 1994, Compagopiscis croucheri Gardiner & Miles, 1994 and Bullerichthys fascidens Dennis & Miles, 1980 (Fig. 7I, J ; also Dennis-Bryan & Miles, 1983; Gardiner & Miles, 1990). The supragnathals of these coccosteomorph taxa possess rows of teeth, frequently two on the anterior and three on the posterior (Fig. 7A–G). These rows meet at (or radiate from) a single point on the supragnathal, representing both the growth centre of the gnathal element (gr.c) and the location of the original tooth primordium. Two or three ASG PSG IG Fig. 5. Coccosteus cuspidatusMiller, 1841 ex Agassiz ms. Anterior view of headshield, anterior and posterior supragnathals and infragnathals (all shaded). Adapted from Miles & Westoll (1968). Abbreviations : ASG, anterior supragnathal ; IG, infra- gnathal ; PSG, posterior supragnathal. 312 Zerina Johanson and Moya M. Smith rows of teeth are present on the coccosteomorph infra- gnathal, one located at the symphysis and a second pos- teriorly along the dorsal jaw margin. A third medial row may be present between these in some genera (Fig. 7H). We interpret the point where these rows meet as the growth centre and location of the first tooth primordium (see Smith, 2003) and from this the position of each tooth row on the supra- and infragnathals is regulated. For example, on the anterior and posterior supragnathals, the position of new teeth is exactly reciprocated medially on the posterior pro- cess (pr.post) on both the left and right supragnathals in taxa such as Harrytoombsia and Compagopiscis (Fig. 7A,B, F; Miles & Dennis, 1979; Gardiner & Miles, 1994). In the supragnathal/infragnathal rows, the newest tooth is added to a specific location at the end of the row and most importantly, is formed away from the biting surface, and in advance of use on this surface (Figs. 7, l.t, 11E,G–I, t3–t5). These new teeth are inferred to form from separate pri- mordia located only at these sites, developed from the orig- inal or initial tooth primordium. The teeth erupt by differential growth of bone, prior to their incorporation into the functional biting surface (Fig. 7, bt.s), located on the infragnathal between the symphyseal (symph.) and marginal (marg.) tooth rows. Teeth from the third, medial row (med.), if present, are also incorporated into this biting surface and are better developed in certain taxa relative to others (com- pare Fig. 7H,L). In addition, the shape and polarity (orientation) of the teeth is constant through the row (e.g. Figs. 7A–H, 11E,G–I), and the base of each tooth is closely apposed to the adjacent one. It is noteworthy that original Fig. 6. (A–F) Dunkleosteus terrelli (Newberry, 1874). (A, C). Right posterior supragnathal, CMNH 5848, (A) lateral, (C) medial views. (B, E) CMNH 5698, left infragnathal, (B) lateral view, (E) closeup of marginal row of teeth. (D) CMNH 7888, anterior supragnathal, anterolateral view. (F) CMNH 5230, right infragnathal, lateral view. (G) Gorgonichthys clarki (Claypole, 1892), CMNH 7129, left infragnathal, lateral view. Larger unlabelled white arrows indicate the anterior direction. Abbreviations : bt.s, biting surface on gnathal ; CMNH, Cleveland Museum of Natural History, Cleveland; l.t., latest tooth to be added to the tooth row; marg., marginal tooth row; med., medial tooth row or tusk ; symph., symphyseal tooth row or tusk. Scale bars=1.0 cm. Origin and evolution of gnathostome dentitions 313 Fig. 7. For legend see opposite page. 314 Zerina Johanson and Moya M. Smith descriptions of these coccosteomorph gnathal plates by Dennis and Miles (1980) consistently refer to ‘ teeth’ and ‘ tooth rows’. ( ii ) Diversity and arrangement of structures in the coccosteomorph dentition. The above observations apply to several types of coccosteomorph dentition that otherwise function in differ- ent feeding types. For example, the dentition on the supra- gnathals of coccosteomorphs such as Harrytoombsia elegans (Miles & Dennis, 1979) must have occluded with that on the infragnathals, but strong shearing surfaces are not devel- oped (Fig. 7A–H). In Kimberleyichthys whybrowi (Dennis-Bryan & Miles, 1983), a deep shearing surface is formed (Fig. 7K) ; by comparison, the main biting surface of Bullerichthys fasci- dens (Dennis & Miles, 1980) is flatter and represents a duro- phagous, or crushing, dentition (Fig. 7I, J). Discrete tooth rows are readily observed on Harrytoombsia elegans and Compagopiscis croucheri (Fig. 7C) while in Kimberleyichthys why- browi, most of the row has been worn away and incorporated into the extensive shearing surface. However, two teeth are visible dorsal to this surface (Fig. 7K, l.t), produced prior to incorporation into the bite. This was one of the criteria for teeth produced within a dental lamina (Reif, 1982), and it is clear that the teeth in Kimberleyichthys whybrowi are produced in a similar manner. Interestingly, its predecessor in the tooth row appears just functional at the biting surface (Fig. 7K, larger arrow), indicating an ongoing, dynamic process associated with growth of the gnathal element. The anterior supragnathal of Bullerichthys fascidens shows teeth being added to two separate rows (see Fig. 10 in Dennis & Miles, 1980), but its posterior supragnathal, used as a durophagous dentition (Dennis & Miles, 1980), is very informative. Several rows of teeth are visible on the lateral surface of the plate (Fig. 7I), continuing ventrally to the biting surface. These lateral teeth are again produced out of the bite, in advance of their use on this surface. More im- portantly, the teeth are not added randomly, but are pro- duced at a set location on the dorsal surface of the posterior supragnathal. This is well below the oral surface, as new teeth are embedded in the newest bone (Fig. 7J, l.t) and here attach to the palatoquadrate, where a small space for soft tissue must have occurred for the tooth primordia of three adjacent rows to develop. ( iii ) Juvenile dentitions : toothed rows. Juvenile or sub-adult supragnathalsand infragnathals have been described from Incisoscutum ritchiei (Dennis & Miles, 1981), Eastmanosteus cal- liaspis Dennis-Bryan, 1987 (see her Fig. 18) and Coccosteus cuspidatus (Miles & Westoll, 1968). In Incisoscutum ritchiei, the posterior supragnathal retains the three main tooth rows identifiable in the adult form, with shorter rows associated with the lateral and medial ones (see Fig. 13A in Dennis & Miles, 1981). These additional rows are lost during growth to the adult morphology (see Fig. 10 in Dennis & Miles, 1981). On the juvenile infragnathal, a distinct symphyseal tooth is visible (see Fig. 12 in Dennis & Miles, 1981), as is a row along the dorsal margin of the bone. New teeth are added to the posterior end of this row, although only one tooth is present in the symphyseal row. This symphyseal tooth is the largest on the infragnathal and we consider this to be a new tooth, representing the original primordium, and site of subsequent addition to the symphyseal row tooth, forming the adult dentition. Additionally, new denticles are added along the inner face of the infragnathal, each with associated bone of attachment (see Fig. 12D in Dennis & Miles, 1981). In Coccosteus cuspidatus, by comparison, juvenile dentitions appear to have the same pattern of rows as seen in the adult dentition but none of the denticles present in Incisoscutum ritchiei (Miles & Westoll, 1968). The morphology of the juvenile dentition is important because it illustrates the early establishment of the symphy- seal and marginal tooth rows on the infragnathal, and lateral, medial and ventral rows on the supragnathal that persist and generate new teeth into the adult stages (Fig. 7). Similarity between the Incisoscutum ritchiei juvenile infra- gnathal and the infragnathal of a brachythoracid arthrodire similar to Buchanosteus fascidens has already been noted (Young et al., 2001). Other ‘primitive brachythoracids ’ (Carr, 1995), as well as certain pachyosteomorph arthrodires, the Phlyctaeniida and the Phyllolepida (Fig. 4) also possess upper and lower adult gnathalia in which new oral denticles are added only at the margins of the existing denticulated patch or area (Section III.b, c). The presence of these denticles may represent a retained juvenile condition. ( iv ) Evidence in pachyosteomorph arthrodires for the presence of teeth. In the pachyosteomorph arthrodires (Figs. 6, 8), den- titions include both tooth rows and denticulated areas, the latter occurring in the posterior region where the newest tooth is added to the marginal row either medially or lat- erally (Fig. 8E,G,H, lat.dent, m.dent). The number of tooth rows varies. For example, in the pachyosteomorph family Dunkleosteidae, the supragnathals and infragnathals of Dunkleosteus terrelli are characterised by well-developed shearing surfaces (Fig. 6 ; Newberry, 1889). In both Dunkleosteus terrelli and Gorgonichthys clarki (Claypole, 1892) a large tusk dominates the symphyseal region of the infra- gnathal (also present on the supragnathals), and a row of separate symphyseal teeth is absent. However, a row of teeth Fig. 7. Dental (gnathal) plates of coccosteomorph arthrodires. (A, B)Harrytoombsia elegansMiles & Dennis, 1979, WAM 70.4.254, left and right anterior supragnathals, ventral (occlusal) view. (C) Compagopiscis croucheri Gardiner & Miles 1994, WAM 94.8.1, posterior supragnathal, ventral view. (D, E) Harrytoombsia elegans, WAM 70.4.254, left suborbital plate with attached posterior supragnathal, (D), medial (internal), (E) lateral views. (F) Harrytoombsia elegans, WAM 70.4.254, posterior supragnathal, medial view. (G) Gogopiscis gracilis, WAM 94.5.1, posterior supragnathal, lateral view. (H) Gogopiscis gracilis, WAM 94.5.1, right infragnathal, medial view. (I, J) Bullerichthys fascidens Dennis & Miles, 1980, WAM 70.4.259, right posterior supragnathal, (I) ventrolateral view, ( J) dorsal view. (K) Kimberleyichthys whybrowi Dennis-Bryan & Miles, 1983, WAM 89.9.676, right anterior supragnathal, posterior view. (L) Bullerichthys fascidens, WAM 70.4.259, right infragnathal, medial view. Abbreviations as in Fig. 6, also : gr.c, growth centre on gnathal plate ; pr.post, posterior process on anterior and posterior supragnathals with accompanying teeth ; SO, suborbital plate ; WAM, Western Australian Museum, Perth. Scale bars=1.0 cm (D and E, L),=0.5 cm (A and B, F–H, K),=0.25 cm (C, I, J). Origin and evolution of gnathostome dentitions 315 Fig. 8. For legend see opposite page. 316 Zerina Johanson and Moya M. Smith is present at the posterior edge of the shearing surface of both the posterior supragnathal and the infragnathal, with each tooth incorporated into this surface (Fig. 6A–C,E,G), comparable to taxa such as the coccosteomorph Kimberleyichthys whybrowi (Fig. 7K). The teeth in Dunkleosteus terrelli are approximately equal in size along the tooth row, and are evenly spaced within the row. The incorporation of the anterior teeth into the shearing surface establishes a polarity of tooth addition in this taxon and in others [e.g. Gorgonichthys clarki, Fig. 6G] with these representing the functional teeth and those at the end of the row as the newly added or successor teeth. A medial row is represented on the infragnathal by a second tusk positioned between the sym- physeal tusk and the medial row of teeth. Ørvig (1980a) described the development of this tusk in substantial detail, as ingrowing pleromic dentine, forming in response to wear against the supragnathals (Fig. 12, tk). Although the shearing surfaces and large tusks result in a morphologically dramatic dentition for the Dunkleosteidae, the length of the functional shearing surfaces is maintained posteriorly by the addition of new teeth, as described above for the coccosteomorph arthrodires. Other pachyosteo- morph gnathals show even greater similarity to the cocco- steomorph condition. For example, in Heintzichthys gouldii (Newberry, 1885) and Gymnotrachelus hydeiDunkle & Bungart, 1939 (Fig. 8A–H; Carr, 1991, 1994), the symphyseal and marginal rows of teeth can be readily recognised on the infragnathal (e.g. Fig. 8E, F). On the posterior supragnathal of Heintzichthys gouldii (Fig. 8B) a row of widely separated teeth crosses what appears to be the margin of the plate. The teeth in the marginal row on the infragnathal are also evenly spaced and nearly all the same height posteriorly, although the teeth do become notably smaller anteriorly. Polarity of tooth addition on the infragnathal is indicated by this anterior decrease in size. As well, the thickness of the bone of attachment (Fig. 8A, b.att) is noticeably wider posteriorly, associated with each new tooth, eventually becoming thinner as it is incorporated into the infragnathal anteriorly. Carr (1991) illustrated an ontogenetic sequence for Heintzichthys gouldii (including Fig. 8A,C,D) which shows that teeth are eventually incorporated into a shearing sur- face, as in the Dunkleosteidae. It appears that at some point during this period of growth, tooth addition ceases com- pletely, with the shearing surface maintained by ingrowing pleromic dentine (Fig. 8D). The dentition of other pachyosteomorph taxa includes denticulated areas. In taxa such as Gymnotrachelus hydei (Carr, 1994; Fig. 8E–H) and Selenosteus brevis Dean, 1901 (Ørvig, 1980a) this area occurs posteriorly, such that the marginal row of the infragnathal continues posteriorly into this area. Thus, it is more difficult to identify the newest tooth added and the site of future tooth primordia. A comparable con- dition is described below in the ‘primitive brachythoracid’ Antineosteus lehmani Lelie`vre, 1984. Denticulated areas can also occur along the medial, or lingual, side of the infra- gnathal in pachyosteomorph taxa such as Protitanichthys sp. (personal observations ZJ and MMS, 2002). The location of this denticulated area is comparableto that described above for the juvenile Incisoscutum ritchiei, and will also be described in greater detail in Section III.(b) for Antineosteus lehmani. Denticulated areas of this type occur in adult and juvenile arthrodires, although in the coccosteomorph arthrodires, they appear to be limited to the juvenile morphology. (b ) Other arthrodires : tooth rows and denticulated ‘areas ’ The morphology of the dentition of these other arthrodires is variable, and is discussed in reverse phylogenetic order (Figs. 2, 4 ; Carr, 1995), from the phylogenetically basal ‘primitive ’ brachythoracids (e.g. Buchanosteus sp., Maideria falipoui Lelie`vre, 1995, Antineosteus lehmani) to the Phlyctaeniida and the basal arthrodire group Actinolepida. The Phyllolepida are described at the end of the section. ( i ) ‘Primitive brachythoracids’ : teeth and denticles. The den- titions of other brachythoracid arthrodires possess tooth rows comparable to the coccosteomorph and pachyosteomorph patterns, such as the ‘primitive brachythoracids ’ (Fig. 4; Carr, 1995) Maideria falipoui (Fig. 8L,M; Lelie`vre, 1995), Antineosteus lehmani (Fig. 9 ; Lelie`vre, 1984) and a ‘buchano- steid ’ arthrodire described by Young et al. (2001). On the anterior supragnathals of the ‘buchanosteid’ and Maideria falipoui (Fig. 8L,M), newer, larger teeth are arranged in rows along the anterior margin (t.row, dent), surrounding a Fig. 8. Dental (gnathal) plates of pachyosteomorph (A–H) and ‘primitive brachythoracid ’ (I–O) arthrodires. (A–D) Heintzichthys gouldii (Newberry, 1885) (A, B) CMNH 8037; (A) left infragnathal, lateral view, (B) posterior supragnathal, lateral view. (C) Heintzichthys gouldii, CMNH 8056, right infragnathal, lateral view, showing wear of tooth row. (D) Heintzichthys gouldii, CMNH 5728, left infragnathal, lateral view, showing extreme wear of tooth row (bt.s) and formation of symphyseal tusk. (E–H), Gymnotrachelus hydei Dunkle & Bungart, 1939; (E–G) CMNH 8051, right infragnathal, (E, F) medial view, (F) closeup of symphyseal tooth row and anterior end of marginal tooth row, (G) lateral view. (H) Gymnotrachelus hydei, CMNH 8084, medial view of denticulated area on posterior part of infragnathal. (I–K) ‘Saudi Arabian buchanosteid ’ MNHN ARB 239, (I) occlusal view anterior supragnathal, ( J, K) medial and lateral views of infragnathal. (L–M) Maideria falipoui Lelie`vre, 1995, headshield and anterior supragnathal plate in (L) ventral view. (M) Natural break through tooth in tooth row (t.row), showing pulp cavity (arrow). (N–P) Holonema westolli Miles, 1971 WAM 95.6.09. (N) medial view of gnathal plate associated with the infragnathal. (O, P) postbranchial lamina, anterior view. (O) entire lamina. (P) closeup of denticle rows near ontogenetic origin of lamina, indicated by larger white arrows in (O, P). (Q) Incisoscutum ritchiei Dennis & Miles, 1981, AMF121767, closeup of denticles on postbranchial lamina. Abbreviations as in Figs. 5–7, also : AMF, Australian Museum, Sydney; b.att, bone of attachment ; dent, denticles on gnathal plate ; lam.marg, dorsomedial margin of the postbranchial lamina where new denticles, either separate or in short rows, are generated ; lat.dent, m.dent, lateral and marginal field of denticles ; MNHN, Museum National d’Histoire Naturelle, Paris ; t.pipe, ‘ tooth pipe’ comprising dentition of the basal arthrodire Holonema ; t.row, tooth row. Scale bars=1.0 cm in all ( J and K share one) except=0.2 cm (M),=0.5 cm (I, Q). Origin and evolution of gnathostome dentitions 317 Fig. 9. For legend see opposite page. 318 Zerina Johanson and Moya M. Smith posterior patch of denticles (see Fig. 6 in Lelie`vre, 1995; and Figs. 2, 3 in Young et al., 2001). A tooth row can be seen on the infragnathals of the ‘buchanosteid’ (see Fig. 2C in Young et al., 2001), with the largest tooth visible occurring at the ventral part of the symphyseal region. A small patch of denticles also occurs along the lingual face of the infra- gnathal (see Fig. 4A, B in Young et al., 2001; personal ob- servations ZJ and MMS, 2001). This lingual area can be seen in other buchanosteid taxa (e.g. an undescribed taxon from Saudi Arabia), although on associated supragnathals of these taxa, tooth rows are present alternating with edentu- late areas (Fig. 8I ; see Fig. 3D–G in Young et al., 2001). On the infragnathal a marginal tooth row is present, while denticles can be observed both on the medial and lateral faces of the infragnathal (Fig. 8J,K, m.dent, lat.dent). This also characterises the ‘primitive ’ brachythoracid Antineosteus lehmani, as described below. Young et al. (2001, p. 675–676) describe a transformation sequence or series between the ‘primitive tuberculate tooth plates ’ of the new ‘buchanosteid ’ and the more specialised gnathal elements of eubrachythoracids, such as the cocco- steomorphs described above. In this proposed evolutionary sequence, the ability to produce enlarged teeth along the margins of the gnathal plate (described as ‘denticles ’ by Young et al., 2001) is retained. Subsequent loss of the pos- terior ‘denticle field ’ occurs in the ‘buchanosteid ’ and Maideria falipoui (Young et al. 2001, their Figs. 2, 3), to attain the morphology seen in coccosteomorphs (Fig. 7) including Coccosteus cuspidatus (Fig. 5, Miles & Westoll, 1968) or Incisoscutum ritchiei (Fig. 11A,E,G–I). Thus, Young et al. (2001, see their Figs. 3, 4) suggest that tooth rows on the anterior supragnathal of coccosteomorphs are homologous to the anterior and lateral tooth rows of the brachythoracid ‘buchanosteid ’ arthrodire. We agree with this interpret- ation, although Young et al. (2001) referred to these as rows of denticles, but we would maintain that coccosteomorph dentitions possess ordered teeth showing the characteristics described for the tooth rows or sets of other jawed ver- tebrates. We do not currently know if they are also composed of the same type of dentine, but each is added to only one locus on the gnathal bone. Therefore, contrary to Young et al. (2001) we would suggest that the anterior and lateral row in their ‘buchanosteid ’ and Maideria falipoui must also represent teeth. As in the coccosteomorph tooth row, teeth in the anterior row of the ‘buchanosteid ’ supragnathals are added to the (medial) end of the row (Young et al., 2001), are larger in this position, and also have an associated bone of attachment (personal observations ZJ and MMS, 2001). We noted that this bone of attachment can be identified and associated with each tooth in the row. As noted above for the pachyosteomorph arthrodire Heintzichthys gouldii, the bone appears to be thickest in association with the newer teeth but thinner along the row towards the oldest tooth, as this bone is remodelled into the main part of the gnathal bone. The addition of new larger teeth to the end of the tooth row can also be seen on the infragnathals of the ‘buchanosteid ’ (Young et al., 2001, see their Fig. 2C). We consider that this may relate to the symphysial tooth row, but is rather longer than in most taxa. In Maideria falipoui, quite large new teeth are seen at the lateral margin and a natural break shows a large pulp cavity in the teeth of this row (Fig. 8M). The infragnathal of Antineosteus lehmani, another ‘primitive brachythoracid’ (Fig. 4), is dominated by areas of denticles, but one prominent feature is a marginal ridge along the dorsal edge of the infragnathal (Fig. 9A, I, marg.). Nothing is known of the upper gnathal elements for Antineosteus lehmani. This elongate ridge is aligned with separate teeth, anterior, or symphyseal in position (marg., marg./symph.). The pro- cess by which these teeth are incorporated into the marginal ridge during growth (forming the functional biting surface) would seem to involve tooth addition at the anteriormost part of the infragnathal (at the symphysis). In addition to these teeth, new denticles
Compartilhar