Shearing Mechanics and the Influence of a Flexible Symphysis During Oral Food Processing in Sphenodon (Lepidosauria: Rhynchocephalia)

THE ANATOMICAL RECORD 295:1075 1091 (2012) Shearing Mechanics and the Influence of a Flexible Symphysis During Oral Food Processing in Sphenodon (Lepidosauria: Rhynchocephalia) MARC E.H. JONES, 1 * PAUL O HIGGINS, 2 MICHAEL J. FAGAN, 3 SUSAN E. EVANS, 1 AND NEIL CURTIS 3 1 Research Department of Cell and Developmental Biology, UCL, University College London, London, UK 2 Centre for Anatomical and Human Sciences, The Hull York Medical School, University of York, York, UK 3 Department of Engineering, Medical and Biological Engineering Research Group, University of Hull, Hull, UK ABSTRACT The New Zealand tuatara, Sphenodon, has a specialized feeding system in which the teeth of the lower jaw close between two upper tooth rows before sliding forward to slice food apart like a draw cut saw. This shearing action is unique amongst living amniotes but has been compared with the chewing power stroke of mammals. We investigated details of the jaw movement using multibody dynamics analysis of an anatomically accurate three-dimensional computer model constructed from computed tomography scans. The model predicts that a flexible symphysis is necessary for changes in the intermandibular angle that permits prooral movement. Models with the greatest symphysial flexibility allow the articulation surface of the articular to follow the quadrate cotyle with the least restriction, and suggest that shearing is accompanied by a long axis rotation of the lower jaws. This promotes precise point loading between the cutting edges of particular teeth, enhancing the effectiveness of the shearing action. Given that Sphenodon is a relatively inactive reptile, we suggest that the link between oral food processing and endothermy has been overstated. Food processing improves feeding efficiency, a consideration of particular importance when food availability is unpredictable. Although this feeding mechanism is today limited to Sphenodon, a survey of fossil rhynchocephalians suggests that it was once more widespread. Anat Rec, 295:1075 1091, 2012. VC 2012 Wiley-Periodicals, Inc. Key words: chewing; shearing; Sphenodon; symphysis; teeth; tuatara; mandible; mastication Additional Supporting Information may be found in the online version of this article. Grant sponsor: BBSRC; Grant numbers: BB/E007465/1, BB/ E009204/1, BB/E007813/1 [to S.E.E. (UCL), M.J.F. (University of Hull), P.O. (Hull York Medical School)]. *Correspondence to: Marc E.H. Jones, Research Department of Cell and Developmental Biology, Anatomy Building, Gower Street, UCL, University College London, London, WCIE 6BT, United Kingdom. E-mail: marc.jones@ucl.ac.uk Received 28 January 2012; Accepted 5 April 2012. DOI 10.1002/ar.22487 Published online 30 May 2012 in Wiley Online Library (wileyonlinelibrary.com). VC 2012 WILEY PERIODICALS, INC.

1076 JONES ET AL. INTRODUCTION Oral food processing can be defined as: Any behavior during which the size, shape, and (or) structural integrity of the [food or] prey item was changed via contact with the tongue, palate, jaws, and (or) teeth (McBrayer and Reilly, 2002; p 884). Under this definition virtually all lepidosaurs (tuatara, lizards, and snakes) process food orally to some extent (e.g. Günther, 1867; Throckmorton, 1976; Schwenk, 2000; Reilly et al., 2001; McBrayer and Reilly, 2002; Ross et al., 2007a,b, 2010) with specific behaviors including puncture crushing, palatal crushing, and side-to-side movements (McBrayer and Reilly, 2002). Sphenodon (the tuatara of New Zealand; Parkinson, 2002; Hay, 2003, 2010) in particular has a highly specialized feeding action whereby the lower jaw closes between two upper rows of teeth (marginal and palatal) before sliding forward to tear food apart with a shearing action analogous to that of a draw cut saw (Günther, 1867; Farlow, 1975; Robinson, 1976; Gorniak et al., 1982; Whiteside, 1986; Schwenk, 2000). Nothing comparable is known amongst other living amniotes, including within Squamata: lizards and snakes (Schwenk, 2000; Reilly et al., 2001). The feeding apparatus of Sphenodon includes a number of components: a large fleshy tongue; teeth that are fused to the crest of the jaw bone and are generally not replaced; marginal tooth rows that are relatively close to the midline; an enlarged chisel-like premaxillary tooth; anterior caniniform teeth on the maxilla, dentary and palatine; an enlarged palatine tooth row parallel to the maxillary row; prominent posterior flanges on maxillary and palatine teeth; small anterolabial and anterolingual flanges on dentary teeth; relatively thin enamel; and an articulation surface of the articular that is elongate by comparison with the quadrate-articular surface (Günther, 1867; Howes and Swinnerton, 1901; Gray, 1931; Robinson, 1976; Schwenk, 1986, 2000; Reynoso, 1996; Jones, 2008, 2009; Kieser et al., 2009; Jones et al., 2009a,b, 2011; Curtis et al., 2010a,b). An elongate wear facet on the labial surface of the dentary below the tooth row is sometimes present in adult animals reflecting tooth on bone contact coupled with proal jaw movement (Robinson, 1976). In the wild, Sphenodon takes a variety of prey items including ants, moths, caterpillars, spiders, beetles, snails, frogs, and lizards (Walls, 1981; Ussher, 1999; Parkinson, 2002), and large individuals are known to bite or saw the heads off sea birds (Walls, 1978, 1981; Dawbin, 1982; Gans, 1983; Cartland-Shaw, 1998; Cree et al., 1999). Direct observations of feeding have been described by several authors (e.g. Farlow, 1975; Walls 1981; Gorniak et al., 1982; Schwenk, 2000), with Gorniak et al. (1982) using electromyography to record jaw muscle activity. Feeding involves five stages: acquisition (prey capture), immobilization, processing, transport, and swallowing (De Vree and Gans, 1989). If the food item is small the tongue may be used to bring it into the mouth, but with larger prey the caniniforms and large chisel-like premaxillary teeth may be important for capture and immobilization (Gorniak et al., 1982). As the adductor muscles contract and the jaws close (during prey processing), the three tooth rows (two upper, one lower) allow three point bending to be applied to food items regardless of how worn the teeth are (Fig. 1, Fig. 1. Three-point bending on stiff or brittle materials as suggested by Evans (1980). (A) A planar food item is held between the upper and lower tooth rows. (B) The dentary teeth press food against the two upper rows of teeth. (C) Dentary continues to rise until the food item fractures because of tension on its dorsal surface. For abbreviations see Table 1. Table 1; Evans, 1980). This mechanism is particularly useful for dealing with stiffer but more brittle food items such as beetles with highly sclerotised shells (Lucas and Luke, 1984). Once the jaws are closed the large pterygoideus muscle contracts pulling the lower jaw forward (Fig. 2; Farlow, 1975; Robinson, 1976; Gorniak et al., 1982; Whiteside, 1986; Curtis et al., 2010a; not backwards as inferred by Günther, 1867; p 601). Material held or impaled by the teeth is sheared between the anterior flanges on the dentary teeth and the posterior flanges on the maxillary and palatine teeth (Robinson, 1976; Gorniak et al., 1982; Jones, 2009). The jaw is reopened and the tongue may be used to position the food before another shearing stroke (Walls, 1981; Gorniak et al., 1982, personal observation). This protraction of the lower jaw therefore involves draw cutting (Frazzetta, 1988; Abler, 1992) and although it is often referred to as propaliny (e.g. Jones, 2008) it is more specifically proal jaw movement because the power stroke is anteriorly directed (a posteriorly directed power stroke is termed palinal movement: Krause, 1982). Following a number of cycles that may vary according to prey type and size, the food is swallowed (Gorniak et al., 1982). Disagreement remains over some aspects of the shearing movement. After examination of anatomical material, both Günther (1867; p 600) and Robinson (1976; p 54) reported that the jaw symphysis was maintained by a fibrous ligament with no involvement of Meckel s cartilage. Both reasoned, as did Schwenk (2000; p 189), that the jaw joints and flexible symphysis accommodated a movement between the right and left mandibles that is necessary for their forward shearing movement (Günther, 1867; p 601; Robinson, 1976; p 53-54). However, based on observations of feeding in captive Sphenodon, Gorniak et al. (1982; p 337) disputed this suggestion and reported that they found no evidence that the lower jaws moved laterally or could rotate about their long axes during the jaw cycle. They argued that the lower jaw was constrained to move parallel to the upper tooth rows Gorniak et al. (1982; p 345) because of the narrow gap between those rows and the proximity of the mandibular coronoid process to the pterygoid flanges on the palate. Perhaps correspondingly, Schwenk (2000) also described the lower dentition as fitting precisely into the gap between the upper tooth rows.

ORAL FOOD PROCESSING IN THE REPTILE Sphenodon 1077 TABLE 1. Anatomical abbreviations AN ANGULAR ant.lab.fl Anterolabial flange ant.lin.fl Anterolingual flange apex Apex art.sur Articulating surface of the articular constr Constraint cpd Coronoid process of the dentary cvex.dep Convex depression D DENTARY CO CORONOID den.can.t Dentary caniniform tooth den.t Dentary additional tooth ECPT ECTOPTERYGOID fas Fascia lab.fl Labial flange lat.lob Lateral lobe lat.ri Lateral ridge lin.fl Lingual flange ltf Lower temporal fenestra M MAXILLA mamem m. Adductor mandibulae externus medialis mamep m. Adductor mandibulae externus profundus mames m. Adductor mandibulae externus superficialis mamp m. Adductor mandibulae posterior mdm m. Depressor mandibulae med.lob Medial lobe mk.gr Meckelian groove mpstp m. Pseudotemporalis profundus mpsts m. Pseudotemporalis superficialis mptat m. Pterygoideus atypicus mpttyml m. Pterygoideus typicus (middle lateral part) mpttymm m. Pterygoideus typicus (middle medial part) mpttyv m. Pterygoideus typicus (ventral part) mx.can.t Maxillary caniniform tooth mx.t Maxillary additional tooth ntch Notch orb Orbit P PALATINE pal.can.t Palatine caniform tooth pal.t Palatine tooth path Pathology po.lin.fl Posterolingual flange po.med.fl Posteromedial flange PRE PREARTICULAR pt.fl Pterygoid flange qu.cot Quadrate cotyle sbs Secondary bone skirt sym Medial surface of the symphysis t Tooth we.fct Wear facet Because of its limited distribution and iconic status in New Zealand, Sphenodon is listed in CITES Appendix 1 (CITES, 2011) and subject to conservation efforts (Parkinson, 2002; Ramstad et al., 2009; Besson and Cree, 2010). Potential for invasive in vivo work is therefore restricted and the work of Gorniak et al. (1982) would be difficult to replicate today. However, computer modeling approaches such as multibody dynamics analysis (MDA) provide an alternative means of investigating feeding behavior (Curtis et al., 2008, 2010a,b,c; Moazen et al., 2008a,b; Curtis, 2011). It allows the interaction of the three-dimensional geometries of the feeding apparatus, muscle forces, and contact forces to be recorded and visualized in detail during jaw movement. An MDA model of Sphenodon has previously been used to illustrate muscle arrangement (Curtis et al., 2009; Jones et al., 2009) as well as investigating muscle activation and function (Curtis et al., 2010a), bite force (Curtis et al., 2010c), potential for mechanoreception at the jaw joints (Curtis et al., 2010b), and the distribution of strain under feeding loads (Curtis et al., 2011). Here we use this model to investigate the movement of the lower jaws in Sphenodon during proal shearing to test the feasibility of three different restrictions on mobility at the mandibular symphysis. Museum specimens were also examined for jaw anatomy and wear patterns associated with this feeding behavior. MATERIAL AND METHODS Material Skeletal material of Sphenodon sp. was sampled from a number of collections, including the Grant Museum of Zoology, University College London, UK (LDUCZ); Natural History Museum, London, UK (BMNH); Southland Museum and Art Gallery, Invercargill, New Zealand (SMAG); and the personal collection of Dr. David Gower, Natural History Museum, London (DGPC): LDUCZ x036, x146, x343, x721, x723, x1176; BMNH 1972.1499, 1985.1212; SMAG Mackay collection 35, 54; DGPC 1, 2. A Sphenodon skull and lower jaws (specimen LDUCZ x036) were subjected to microcomputed tomography (micro-ct) at the University of Hull, UK, using a X-Tek HMX 160 scanner and the following scan parameters: Beryllium target, 93 kv; ua 17; aperture 75%; 1,000 projections averaging out 32 frames per projection. To reduce beam hardening the X-rays were filtered through a 0.1 copper plate. Using image segmentation and analysis software (Amira 4.1, Mercury Computer Systems Inc, and Avizo 6.3, Visualization Sciences Group), the micro- CT dataset was used to generate a geometrically accurate three-dimensional model of the skull, lower jaws, and joint surfaces. Comparative Anatomy The jaws, teeth, and articular surfaces were examined using a Wild stereo binocular microscope, paying particular attention to tooth shape and tooth wear facets. A contour visualization of the articular and quadrate cotyle morphology was generated by segmenting the respective CT models into layers of equal thickness (0.48 mm, exact value restricted by voxel size). Each layer is parallel to the horizontal plane between the ventral tips of the quadrate long axis of the skull and was assigned a different color. Multibody Dynamics Analysis The three-dimensional geometries of the skull elements were imported into ADAMS multibody dynamics analysis software (MSC Software Corp), as previously described (Curtis et al., 2009, 2010a,b). The CT models of the lower jaws were imported as two separate geometries, attached at their anterior symphysis by an assigned constraint within the multibody software (Fig. 3A). Mass and inertial properties of all geometries were calculated automatically within the software, based on volumes and tissue density (of 1,050 kg/m 3 ). Muscles were represented by 110 spring elements positioned

1078 JONES ET AL. Fig. 2. Basic jaw movement in Sphenodon as observed by Gorniak et al. (1982) in left lateral view. Skull modified from Robinson (1976). according to observations of dissected material (Curtis et al., 2009, 2010c; Jones et al., 2009a). These included eight elements representing the m. depressor mandibulae, 48 representing the external adductors, 10 representing the posterior adductors, and 44 representing the internal adductors (Fig. 3B,C). Vertical and horizontal movement components of a biting cycle, comparable to that discussed in Gorniak et al. (1982), were assigned to motion markers within the multibody software. Following the dynamic geometric optimization (DGO) method (Curtis et al., 2010a,c), the muscles were activated so as to move the lower jaws along the motion path observed in captive living animals (Farlow, 1975; Gorniak et al., 1982; personal observation). Thus, DGO varies the force within each individual muscle spring, based on its instantaneous orientation to move the lower jaws in line with the desired motion markers (see Curtis et al., 2010a for a more comprehensive explanation of this method). This muscle activation method does not constrain the movements of the two Fig. 3. The multibody model. (A) The model of the lower jaw showing the position of the anterior constraint. (B and C) The muscle representation in (B) oblique dorsolateral view and (C) oblique ventrolateral view. lower jaws in relation to one another. These movements are instead influenced primarily by the constraint at the mandibular symphysis and the contact geometries

ORAL FOOD PROCESSING IN THE REPTILE Sphenodon 1079 within the skull, i.e. the quadrate-articular jaw joints and the teeth. The skull was also fixed in space at the neck during all simulations since these investigations were focused on lower jaw movements only. Contact was specified between the articulating surfaces of the quadrates on the skull and the articulars of the lower jaws, the pterygoid flanges and coronoid bones, the premaxillary teeth and lower jaw tips, as well as between corresponding teeth. To remove noise from subtle differences between the left and right jaw joints a mirrored version of the right quadrate cotyle and articular was used for the left side. Motion paths defined a movement that aimed to open the lower jaws to a gape of approximately 30 degrees, during which the lower jaws would translate backwards so that the quadrate cotyles contacted the anterior edges of the articular joint surfaces (Fig. 1A) (Supporting Information 1 and 2). The lower jaws would then close while maintaining this joint contact position until the teeth of the upper and lower dentitions came into contact. The motion path then aimed to move the lower jaws forward to replicate the proal shear observed in live animals. To allow the lower jaws to follow the defined motion paths the jaw muscles were activated in an appropriate sequence as defined by the DGO muscle model (Curtis et al., 2010c). To test the influence of symphysis flexibility on lower jaw kinematics and feeding mechanics in Sphenodon the model was run with three different setups with respect to the anterior mandibular symphysis. Setup 1 assigned a fixed constraint that rigidly attached the two lower jaws and prevented any relative movements between them. Setup 2 assigned a hinged constraint that permitted medial-lateral flaring movements (i.e. rotation about the z-axis only, see Fig. 3), while setup 3 assigned a spherical joint that permitted free rotation about all three axes (i.e. rotation about the x, y, and z axes, see Fig. 3). RESULTS Tooth Shape Tooth shape was consistent with previous descriptions (e.g. Robinson, 1976; Gorniak et al., 1982; Jones, 2008, 2009; Jones et al., 2009a,b). The dentary teeth vary in shape according to the degree of wear. Before wear, the main dentary teeth of Sphenodon (the posterior additional teeth) are usually described as resembling smooth anterodorsally directed cones without sharp edges (e.g. Robinson, 1976; Reynoso, 1996). This observation is usually correct but some Holocene jaws referred to Sphenodon possess dentary teeth that bear true flanges (anterolabially, anterolingually, posterolabially, and posterolingually) and show no sign of wear (Fig. 4B). Wear facets on the dentary teeth first appear on the labial and lingual surfaces and eventually also become apparent on the anterior surface (Reynoso, 1996; Jones et al., 2009a,b). This gives worn teeth the appearance of a pyramid with a rectangular base, four definite corners, and an anteriorly located apex (Fig. 4A,B). Further wear leads to the teeth being labiolingually narrower with an anteroposterior crest replacing the apex (Reynoso, 1996; Jones et al., 2009a,b). The degree of tooth wear tends to be greater anteriorly because these teeth are older; additional teeth are added to the rear of the jaw bone during growth (Harrison, 1901; Robinson, 1976). Anteriorly there is also a large conical anterior caniniform tooth within a row of small and usually worn hatchling teeth (Robinson, 1976). The posterior additional maxillary teeth are conical with a posterolingual flange that extends beyond the anterior edge of the following tooth (Fig. 4C,D). The palatine teeth are slightly smaller but are essentially mirror images of the maxillary teeth (Fig. 4C; Robinson, 1976): conical with posterolabial flanges. In both the maxillary and palatine teeth there is some alternation in size anteriorly. Wear on the maxillary teeth may be limited to the posterolingual edge or more commonly includes the entire lingual surface (Fig. 4D). Similarly, wear on the palatine teeth can be located at the posterolabial edge but usually includes most, or all, of the labial surface. Wear on opposing maxillary and palatine teeth is often broadly comparable but, as with the dentary teeth, the wear on the anterior teeth is often greater than that on the posteriormost additional teeth (Fig. 4C). The anterior-most dentition of the maxilla may include a large conical caniniform and a group of smaller conical successional teeth. A conical caniniform tooth is also present at the front of the palatine tooth row. The anterior alveolar margin of the dentary in some individuals is polished (Fig. 4D; e.g. LDUCZ x146; LDUCZ x343; Reynoso, 1996), probably from contact with the posterior surface of the chisel-like premaxillary teeth during shearing (Robinson, 1976; Jones et al., 2009a). A similar wear facet can be found on the posteroventral surfaces of the premaxillary teeth (e.g. LDUCZ x036). As previously described and figured, the palatine tooth row is near-parallel with the longer tooth row of the maxilla (e.g. Fig. 4C; Robinson, 1976; Gorniak et al., 1982; Jones et al., 2009). The gap between the upper tooth rows ( groove of Robinson, 1976; trough of Gorniak et al., 1982) is about 1.5 mm wide, wider than the labiolingual dimensions of the dentary tooth bases, which tend to be closer to 1 mm (LDUCZ x036; Jones 2009). The Symphysis of the Lower Jaw The osteological component of the lower jaw symphysis is located at the anteromedial end of the dentary associated with a small ventral chin-like projection. It is comma-shaped with a smooth medial surface (Fig. 5A) and lies ventral to a notch at the anterior tip of the jaw (Fig. 5B; Jones et al., 2009a). According to Robinson (1976) ligamentous soft tissues span the gap between the right and left notches, as well as the short gap beneath the symphysis. The Meckelian groove runs along the lingual surface of the dentary just above the ventral margin and terminates at the posteroventral boundary of the symphysis (Fig. 5A). Overall it appears as though that the lower part of the symphysis would be more rigid than the upper part. Articular and Quadrate Articulation Although the articulation surface at the base of the quadrate has been referred to as a condyle (e.g. Jones et al., 2009a) it is probably better referred to as a cotyle. It is bilobate in posterior view with the medial lobe

1080 JONES ET AL. Fig. 4. The teeth of Sphenodon. (A) Stereopair of the dentary tooth row in labial, apical and lingual views; (B) stereopair of the two posterior dentary teeth; (C) stereopair of the two upper tooth rows in an articulated specimen; (D) the maxillary tooth row in lingual and labial view; (E) a complete lower jaw in labial view; (F) close up of the anterior alveolar margin in lingual view. (A, B) Specimen SMAG Mackay collection 35; (C) specimen DGPC2; (D) specimen SMAG Mackay collection 54; (E, F) specimen LDUCZ 146. Scale bar ¼ 5mm. being slightly larger and extending further ventrally (Fig. 6A,B). The articulating surface of the articular is somewhat saddle-shaped, comprising a long rounded ridge laterally and a convex depression medially (Fig. 6D,E). These features are oriented along a similar axis to the dorsal process of the surangular, coronoid process, and tooth row, and are both approximately twice the length of the surface on the quadrate cotyle. The convex depression of the articular narrows anteriorly so that contact between the articular and quadrate is greatest at the end of the proal shear phase, just before the jaws open.

ORAL FOOD PROCESSING IN THE REPTILE Sphenodon 1081 Fig. 5. The osteological component of the symphysis in an adult Sphenodon. (A) Medial view of the right dentary of specimen LDUCZ x146 (reversed). (B) Anterior view of a CT model based on the lower jaws of LDUCZ x036. The ventral margins are asymmetric because of a local pathology on the right ramus (previously figured in Jones and Lappin, 2010). Scale bars ¼ 2 mm in (A) and 10 mm in (B). The Coronoid and Pterygoid-Flange The pterygoid flange is formed by the lateral process of the pterygoid and the ventral process of the ectopterygoid and is located at the posterolateral corner of the palate (Günther, 1867; Jones et al., 2009a, 2011). At least in large adult specimens (skull length ¼ >45 mm) it projects further ventrally than either the premaxillary tooth or quadrate cotyle (Fig. 2; e.g. DGPC 2, LDUCZ x036). The lateral surface is very subtly convex so that the ventral most tip is slightly medially inclined. The coronoid is a large essentially flat bone that contributes to the medial surface of the lower jaw posterior to the tooth row and anterior to the adductor fossa (Günther, 1867; Jones et al., 2009a). The lateral surface of the coronoid contacts the dentary whereas its ventral end meets the angular anteriorly and prearticular posteriorly. Posteriorly the coronoid bifurcates and envelops the lateral process of the surangular. In medial view the pterygoid flange and medial face are most closely aligned when the jaws are closed and the quadrate cotyles are positioned on the posterior ends of the articular surfaces. At this point the pterygoid flange extends ventral to the coronoid bone, adjacent to the prearticular and angular (e.g. DGPC2). Fig. 6. The jaw joint in an adult Sphenodon (LDUCZ x036). (A) A posterior view of the skull. (B,C) The quadrate cotyle in (B) posterior and (C) ventral view. (C, D) The posterior part of the lower jaw in (D) posterior and (E) dorsal view. Each layer of the contour models is 0.48 mm thick. Scale bars in (A) ¼ 10 mm and (B E) ¼ 5 mm. Multibody Dynamics Model Jaw movement, as predicted by MDA, was notably different for each of the three symphysial setups. The fixed constraint (setup 1) restricted jaw movement considerably. During proal shearing only 1 mm of anterior movement is possible and gape is constrained to about 10 degrees (Supporting Information 3). This is due to the geometric constraints between the two conforming surfaces of the jaw joint. During jaw opening the medial lobe of the quadrate cotyle prevents posterior movement of the lateral ridge of the articulation surface of the articular. The hinged constraint (setup 2) restricted rotation of the lower jaws about their long axes, but did permit relative medial-lateral movements between the two lower jaws. During jaw opening the lower jaws slide backwards by approximately 4 mm so that the front of the articulation surface of the articular is located beneath the quadrate cotyle (Supporting Information 4). This

1082 JONES ET AL. position is maintained until the onset of proal shearing. Also, as the jaws open the anterior angle between the lower jaws increases by approximately 4 degrees because the articulars move laterally away from each other as their surfaces pass along the quadrate cotyle. This movement is permitted by the hinged symphysis. During jaw closing the wider angle between the lower jaws is maintained. On closure the lower jaws slide forwards approximately 4 mm and the angle between the lower jaws reduces as they return to their original position. The spherical constraint (setup 3) behaves somewhat similarly to the hinged constraint (setup 2). During jaw opening there is a comparable 4 mm posterior translation of the lower jaw (Fig. 7A,B) (Supporting Information 5). There is also an increase in the symphysial angle of approximately 4 degrees (Fig. 7C,D). However, in contrast to setup 2, relative movement between the lower jaws is not limited to just one plane. There is also a 10-degree long axis rotation of each of the lower jaws as the lateral ridge of the articular rotates within the groove of the quadrate cotyle (Fig. 7C H) (Supporting Information 6). As the jaws translate anteriorly during proal shearing the dorsal edge of each jaw rotates medially and the dorsoventral jaw axis becomes more vertical. This in turn brings the anterodorsal tips of the lower jaws closer together (Fig. 8A,B) and also alters the mode of interaction between the lower and upper dentition. At the beginning of the shearing phase, the most intimate shearing contact occurs between the anterolabial flanges on the dentary teeth and posterolingual flanges of the maxillary teeth (Figs. 7E,G and 8E,G), but toward the end of the shearing phase it is between the anterolingual flanges of the dentary teeth and the posterolabial flanges of the palatine teeth (Figs. 7F,H and 8F,H). Although the medial surface of the lower jaw and the lateral surface of the pterygoid flange are generally in close proximity the flanges do not appear to restrict forwards or backwards movement of the lower jaw in any of the models. Usually the point of closest proximity is between the ventral tip of the flange and the boundary between the coronoid bone and angular, but at the end of the shearing phase in setup 3 it is between the midpoint of the flange and the central part of the coronoid. DISCUSSION Jaw Symphysis Flexibility The analysis described above using the multibody model of Sphenodon demonstrates that flexibility at the mandibular symphysis is necessary to replicate the jaw movements observed in the living animal. For the lower jaw to move backwards at jaw opening and forwards after jaw closure the saddle-shaped joint surface of the articular has to pass along the labiolingually concave quadrate cotyle. For this to be possible, the symphysial angle between the lower jaws must change. This point can be illustrated by representing the lower jaws as an open triangle with the apex at the mandibular symphysis. The two long edges of this triangle are straddled by two fixed constraints distal to the apex (i.e. the large medial lobes of the quadrate cotyles). For the apex of this triangle to move backwards, the angle between the two long edges must increase and for the tips to move Fig. 7. Movement predicted by MDA in the model with a spherical constraint at the symphysis (setup 3). Relative positions of the lower jaws at the (A, C, E, G) beginning, and (B, D, F, H) end of proal jaw movement. (A, B) The skull in lateral view, (C, D) ventral view, (E H), and dorsal view with a transparent cranium. (G, H) A close up of the left posterior dentary teeth and left palatine tooth row. (A D) The cranium is transparent but the lower jaws are opaque. (E, F) The right lower jaw and portion of the left dentary are opaque, most of the cranium is transparent but one portion across the palate (including parts of the maxilla, palatine, jugal, and pterygoid) is relatively less transparent. All scale bars ¼ 10 mm. forwards the angle must decrease. Without the potential for symphysial hinging, backwards translation of the lower jaws is restricted and both normal gape and subsequent proal shearing are inhibited. This supports

ORAL FOOD PROCESSING IN THE REPTILE Sphenodon 1083 Fig. 8. Movement predicted by MDA by the model with a spherical constraint at the symphysis (setup 3). Relative positions of the lower jaws at the (A, C, E, G) beginning, and (B, D, F, H) end of proal jaw movement. (A, B) The skull in anterior view and (C F) anterolateral view. (G, H) A close up of the left posterior dentary teeth and left palatine tooth row in cross-section. (A, B) The cranium is transparent but the lower jaws are opaque. (E H) Most of the cranium is transparent whereas the entire right lower jaw and a portion of the left dentary and cranium (including parts of the maxilla, palatine, jugal, and pterygoid) are opaque. All scale bars ¼ 10 mm. previous inferences that the anterior symphysis is mobile in Sphenodon (Günther, 1867; Robinson, 1976; Schwenk, 2000). Gorniak et al. (1982; Fig. 6) reported the anterior translational movement of the lower jaw to be about 2.3 mm. As this was recorded from an individual approximately 20% smaller than the one modeled here (see Curtis et al., 2010c), we might expect translational movement in the MDA model to be at least 2.75 mm. This is much greater than the 1 mm permitted by the setup with the fixed symphysis (setup 1) but is easily accommodated by the hinged and spherical joints (setup 2 and setup 3) which both allowed 4 mm of movement. Note, the MDA model lacks ligaments around the jaw joints so that the movements are restricted only by the geometry of the articulating surfaces, as well as the pterygoid flanges and tooth rows. Therefore, the extent of movement exhibited by a model should be viewed as the maximum amount achievable without the presence of soft tissues. Relative movement between the jaws appears to be possible because the symphysis is held together mainly by fibrous connective tissue (Günther, 1867; Robinson, 1976). The osteological component of the symphysis in lepidosaurs has received relatively little direct attention although variation clearly exists (Evans et al., 2002; Evans, 2008; Zaher et al., 2009). The morphology observed in Sphenodon certainly differs from that of Varanus and many snakes where there is no clear bony symphysial surface (Rieppel and Zaher, 2000; Evans, 2008; Holliday et al., 2010). Sphenodon also differs from many squamates, including most agamids, pleurodont iguanians, lacertids, skinks, geckos, and Lanthanotus where the bony symphysis tends to be ovoid, dorsally located and without a dorsal notch (e.g. Pregill, 1984; Rieppel and Zaher, 2000; Evans et al., 2002; Evans, 2008; Hollenshead, et al., 2011). Amongst squamates, the greatest similarity to the symphysis of Sphenodon seems to be with certain chameleons (e.g. Chamaeleo) where both a dorsal notch and ventral chin may be present (Čerňanský, 2010). Understanding the functional significance of this variation in the bony symphysis requires a more detailed examination of the relationship between anatomy (bone and soft tissues) and mandibular movement. The soft tissue anatomy of the squamate symphysis has been surveyed by Young (1998; snakes) and Holliday et al. (2010; lizards). Holliday et al. (2010) found that the lizard symphysis was usually divided into dorsal and ventral components. The dorsal part contained a fibrocartilaginous pad and tightly packed parallel fibers whereas the jaws were joined ventrally by more loosely connected woven fibers and, in some taxa, fusion of the left and right Meckel s cartilage. They concluded that intermandibular movement was possible in most squamates with rotation (or hingeing) occurring between the anterodorsal tips of the dentaries. The degree of intermandibular flexibility may be linked to the degree of intracranial flexibility, the most flexible symphyses being found in snakes which also possess intramandibular hinges and highly mobile cranial joints (Young, 1998; Lee, 1999; Rieppel and Zaher, 2000). Sphenodon has little cranial flexibility (Jones et al., 2011) and yet setup 3 of our multibody analysis suggests that rotational movements (around the long axis of the lower jaw) also occur in Sphenodon. However, in contrast to lizards the centre of rotation is located anteroventrally rather than anterodorsally (Figs. 8 10). The notched upper part of the symphysis, supported by soft tissue, splays during jaw opening and retraction but closes slightly during jaw protraction (proal movement) (Figs. 8A,B and 9D,H,L). A detailed histological examination of the symphysial joint is required to assess the exact nature of the fibers in Sphenodon and to determine how they may be related to these movements. Although the Meckelian cartilages are fused anteriorly in the Sphenodon embryo (de Beer, 1937; Schwenk, 2000), this is apparently not the case in adults (Günther, 1867). Holliday et al. (2010) suggested that fused Meckel s cartilages stabilized the joint between the lower jaws

1084 JONES ET AL. Fig. 9. Schematic summary diagram of the lower jaw movements in Sphenodon as suggested by the MDA model with greatest freedom for the quadrate to follow the articulation surface of the articular (setup 3). (A) The skull in lateral view. Relative positions of the feeding apparatus at the (B E) beginning, (F I) middle, and (J M) end of proal shearing: (B, F, J) the articular relative to the quadrate cotyle, (C, G, K) dentary tooth rows relative to the upper tooth rows; (D, H, L) the right and left tips of the lower jaws; (E, I, M) the long axes of the right and left lower jaws (intermandiblar angle). in squamates such as iguanians and geckos that use their tongue for prey capture or drinking, respectively (although a more apomorphic behavior in geckos may be eye licking, Daza, personal communication, March 2012). As Sphenodon uses its tongue for prey capture (Gorniak et al., 1982; Schwenk, 2000) it does not fit this pattern. Flexibility of the symphysis in Sphenodon has implications for jaw mechanics during unilateral biting (Schwenk, 2000; Evans, 2008) which is known to occur in this genus (Gorniak et al., 1982; Curtis et al., 2010a). This is because compliance between the jaws would inhibit the transfer of forces between the working and balancing sides (Crompton and Hylander, 1986; Holliday et al., 2010). The frequency and importance of unilateral biting in other lepidosaurs has yet to be systematically surveyed. Shearing in Sphenodon The lower jaws of Sphenodon are analogous to a serrated blade characterized by a series of high and low points that fragment compliant material such as food by gripping and ripping (Abler, 1992). During jaw closure, penetration of food material is made easier by the blades running along the apicobasal axis of the teeth and nearparallel to the arc of jaw closure (Freeman and Leman, 2006; Jones, 2009). Some food is also cut because the blades on the lower and upper jaw meet obliquely with different approach angles (Evans and Sanson, 2003). The arrangement of the teeth (in series) provides a series of notches for trapping compliant food as found in mammalian carnassial teeth (Butler 1946; Lucas and Luke, 1984; Abler, 1992; Evans and Sanson, 2003;

ORAL FOOD PROCESSING IN THE REPTILE Sphenodon 1085 Anderson and LaBarbera, 2008; Anderson, 2009). Food that is not cut is gripped and held. During proal shearing, trapped food is cut (or ripped) between the anterior blades on the dentary teeth and posteroventral blades on the upper dentition. For such a draw-cutting system to be effective there is no need for the horizontal (forward) motion to be greater than the distance between the blades as once the blades have moved this distance there should (in theory) be nothing left to cut (Frazzetta, 1988; Abler, 1992). When the blades are subject to wear and the material to be cut is heterogeneous the situation can be more complex. Nevertheless, the distance that the lower jaws move forward during proal shearing (about 2 3 mm) corresponds with the length of most marginal teeth and hence the distance between the cutting edges on each tooth (e.g. Jones, 2009). There is obvious disagreement as to the sophistication of the shearing mechanism in Sphenodon. In a review of food processing, Reilly et al. (2001) described it as formidable and one of the best shearing jaws in amniotes. By contrast, Robinson (1976) characterized occlusion in Sphenodon as poor and Gorniak et al. (1982; p 345) stated that Sphenodon lacked occlusion comparable to mammals, speculating (1982; p 352) that tooth wear was as likely to reflect food-on-tooth wear as tooth-on-tooth wear. The results of our analyses question these latter assertions. Although the gap between the upper teeth is larger than the labiolingual dimensions of the dentary teeth, the orientation of the dentary teeth within the gap may change dynamically during jaw movement. The MDA model with the spherical joint (setup 3) permits the greatest flexibility, thus allowing the articulation surface of the articulars to follow the contours of the quadrate cotyle without restriction. During proal shearing, the lower jaws also undergo a long axis rotation that engenders a subtle dynamic form of occlusion between the dentary, palatine, and maxillary teeth (at least at the rear of the jaws). As the jaws close the anterolabial flanges on the dentary teeth cut against the posterolingual flanges of the maxillary teeth whereas toward the end of the shearing phase the anterolingual flanges of the dentary teeth cut against the posterolabial flanges of the palatine teeth. In life this would promote close contact between opposed blades and enhance the shearing action by focusing the cutting force between particular blades in turn (Figs. 8G,H, 9, and 10) (Lucas and Luke, 1984; Evans and Sanson, 2003). That Gorniak et al. (1982) did not note these movements may simply reflect the difficulty of perceiving subtle intracranial movements in a relatively small living animal, particularly when those movements may be masked by the skin and other soft tissues. As mentioned above, shearing mechanisms can be disrupted by heterogeneous material that offers uneven resistance, leading to irregular movements of the cutting tools. Such problems may be reduced by stabilizing or compensatory movements, in this case controlled by modifications to muscle activity (Curtis et al., 2010a), but also through physical constraints (in addition to those of the jaw articulation). The short distance between the pterygoid flange and medial surface of the lower jaw supports the hypothesis that the pterygoid flange acts as a brace to limit medial displacement of the lower jaw (Fig. 11; Iordansky, 1964; Taylor, 1992; Jones et al., 2011; Porro et al., 2011). Therefore, as well as being important for dealing with struggling prey this medial buttress may also help guide the jaws and ensure smoother shearing of heterogeneous food material. It is possible that the convex lateral surface of the pterygoid flange is shaped to maintain a consistent relationship with the flat surface of the coronoid during the 10- degree long axis rotation of the lower jaw (Figs. 9C,G,K and 11B,D) during shearing. The closer anteroposterior alignment of the coronoid and pterygoid flange just before jaw opening suggests that this is the time when guidance or support from the flange is most important (Fig. 10C,D). Nevertheless, the ventral extent of both the flange and coronoid bone ensure support is also provided at jaw closure and during shearing (Fig. 11A,B). The large chisel-like teeth may also serve as a constraint to jaw movement providing an anterior obstruction to the jaw tips as well as helping to retain food in the mouth. The wear facets on the dentaries and premaxillary chisel-like teeth demonstrate that contact certainly occurs between these elements. Oral Food Processing Oral food processing is particularly complex in mammals and involves an integration of elaborate dentitions, three-dimensional occlusion, and highly rhythmic chewing strokes (e.g. Hiiem ae and Crompton, 1985; Evans and Sanson, 2003; Ross et al., 2007a,b, 2010). Thus, although lepidosaurs are considered to chew it has been argued that the term mastication should be restricted to mammals (Ross et al., 2007a; p 133). Because the most sophisticated oral food processing is found in mammals it has been strongly linked to the evolution of endothermy or at least the maintenance of a high metabolic rate and activity level (e.g. Lumsden and Osborn, 1977; Lucas and Luke 1984; Crompton and Hylander, 1986; Reilly et al., 2001; Kemp, 2005; Fritz et al., 2010; see also footnotes 1 and 2). In a review of lepidosaur feeding, Herrel et al. (2001) wrote that a true mammalian-like power stroke is found only in Sphenodon. However, Sphenodon does not have a particularly high metabolic rate (Thompson and Daugherty, 1998) and it is generally considered to be a sit-and-wait predator (e.g. McBrayer and Reilly, 2002). Although Sphenodon does not comminute food to the same extent as many mammals, it certainly does process food to a greater degree than other lepidosaurs, including squamates such as Varanus that are considered to be particularly active 1 The pressence of endothermy in birds that lack extensive oral food processing is excused because of the internal food processing performed by the gizzard (Reilly et al., 2001). 2 Abler (1992) also speculated that the more complex jaw movements involved in sophisticated oral food processing might also be linked to the evolution of mammalian intelligence. Despite possessing jaw movement that is relatively more complex than many non-avian reptiles tuatara appear to be no more intelligent (Northcutt and Heath, 1973).

1086 JONES ET AL. Fig. 11. A schematic diagram illustrating the relationship between the pterygoid flange and coronoid process based on results from setup 3. (A, B) At jaw closure and at the start of the shearing phase. (C, D) At the end of the shearing phase. In medial view (A and C) and anterior view (B and D). Of the cranium (light grey) only parts of the palatine, pterygoid, ectopterygoid, and quadrate are visible. Asterisk on the pterygoid flange indicates point of closest proximity. Scale bar ¼ 10 mm. Fig. 10. A schematic diagram illustrating the changes in the position of the tooth blades during shearing in Sphenodon. (A) At jaw closure and the start of the shearing phase. (B) At the end of the shearing phase. Maxillary, palatine, and dentary teeth are shown in ventral view but the dentary teeth are shown as if semitransparent. (Schwenk, 2000; Clemente et al., 2009). This suggests the link between food processing and high metabolism (and certainly endothermy) has been overstated. Oral food processing is advantageous for at least three reasons besides facilitating endothermy. First, it increases the surface area available for chemical breakdown (Bjorndal et al., 1990; Reilly et al., 2001). This in turn allows more efficient digestion and reduces the length of intestines or gut passage time needed to digest the same amount of food (Lucas and Luke, 1984). Surface coatings and internal structures that may be resistant to chemical breakdown can be opened or fractured mechanically (Hiiem ae and Crompton, 1985; Lucas, 2004). This is particularly important for herbivores and, as a result, food processing is also associated with increased omnivory and herbivory (e.g. Throckmorton, 1976; Throckmorton et al., 1981; Hiiem ae and Crompton, 1985; Bjorndal et al., 1990; Modesto, 1995; Sues and Reisz, 1998; Barrett, 2000; Reilly et al., 2001; Cooper and Vitt, 2002; Barrett and Upchurch, 2007; Upchurch et al., 2007). Although plant material can represent 14% of the material in Sphenodon scats its intake is probably accidental during prey capture (Walls, 1981; Cooper and Vitt, 2002). Second, oral processing allows tighter food packing as spaces within the food item are compressed. In addition, a limited space can accommodate a greater volume of material if that material is heterogeneous rather than homogenous in shape and size (Beckett, 2000; p 72). This again reduces the length of intestine required to deal with the same food intake. Finally, oral processing allows small pieces to be removed from a larger food item so that they fit inside the mouth and can be swallowed. This allows access to larger food items (Robinson, 1973; Lucas, 2004; Jones, 2008). All three factors help an animal when food is limited, either by allowing it to make the most of the food that is available or by widening dietary range. Robinson (1973) discussed shearing in Sphenodon (and other rhynchocephalians) as a means of accessing large food items. She referred to cranial kinesis in squamates as an alternate strategy to the same ends. Whether kinesis in squamates is linked to feeding on large food items is contentious (Schwenk, 2000; Metzger, 2002; Evans, 2008) and Sphenodon mainly eats small or medium sized arthropods (Walls, 1978, 1981; Ussher, 1999). However, average, modal, or preferred diet is not necessarily useful for interpreting the relationship between form and function because diet may change at times of hardship when selection pressures are at their greatest (Grant and Grant, 1993; Robinson and Sloan Wilson, 1998). Large adult Sphenodon are certainly known to eat vertebrate material (notably seabird chicks), particularly during periods when other sources of nutrients and water are scarce (Walls, 1978, 1981). Sphenodon populations have probably only been restricted to their current habitus for the past few hundred years (Hay et al., 2003) and it is unlikely that Sphenodon or its ancestral lineage evolved alongside sea birds (Crook, 1975). However, the ability to take a range of prey would have been advantageous in any resource challenged environment (Evans and Jones, 2010; Apesteguía and Jones, 2012). Sphenodon provides an example in which specialization of the feeding mechanism appears to permit a broader diet (Pregill, 1984). Evolutionary Perspective Historically, Sphenodon was considered to be representative of the ancestral reptilian condition (e.g.

ORAL FOOD PROCESSING IN THE REPTILE Sphenodon 1087 Fig. 12. A simplified cladogram of Rhynchocephalia plotted against time. Black bars represent the known fossil record, grey bars represent ghost ranges, faded grey bars indicate uncertain ghost ranges, and white bars indicate ghost lineages. The status of three characters associated with proal shearing found in Sphenodon are plotted next to taxa. A filled circle indicates state present, a line indicates absent, and a question mark indicates unknown. Clade names: A ¼ Lepidosauria; B ¼ Rhynchocephalia; C ¼ derived rhynchocephalians; D ¼ Sphenodontinae. Information taken from Evans (1980, 1990); Rasmussen and Callison (1981); Throckmorton et al. (1981); Fraser (1982, 1988); Carroll (1985); Carroll and Wild (1994); Reynoso (1993, 1996, 2005); Evans et al. (2001, 2002); Apesteguía and Novas (2003); Jones (2008); Evans and Borsuk-Bialynicka, 2009; Jones et al. (2009a,b). Please note that Tikiguania is no longer considered to represent a Late Triassic squamate (Hutchinson et al., in press), thus making material from the Early Jurassic of India potentially the oldest record for this clade (Evans et al., 2002). Williston, 1917; Crook, 1975) and Robb (1977) referred to Sphenodon as the most primitive and unspecialized reptile in existence. However, more recent research contradicts this (e.g. Gorniak et al., 1982; Gans, 1983; Whiteside, 1986; Thompson and Daugherty, 1998; Evans, 2003; Jones, 2006, 2008, 2009; Jones et al., 2009a, 2011; O Meally et al., 2009; Evans and Jones, 2010). The level of sophistication observed in the feeding mechanism described here further demonstrates that Sphenodon is not a generalized amniote: it is the only living amniote to possess a feeding mechanism that involves a propalinal (or proal) power stroke and bladed teeth on both the maxilla and palatine (Schwenk, 2000; Reilly et al., 2002). However, Sphenodon is also the only living member of the Rhynchocephalia (the sister taxon to Squamata, Evans, 2003; Rest et al., 2003), with all other named genera known exclusively from the Mesozoic (Jones, 2008; Evans and Jones, 2010). Several anatomical features associated with the Sphenodon-mode of shearing are found in its Mesozoic fossil relatives (Fig. 12). The enlarged palatine tooth row is present in all Rhynchocephalia and represents a diagnostic character for the clade (Evans, 2003; Jones, 2008). The elongate articulation surface of the articular is found in the most plesiomorphic members of the clade such as Diphydontosaurus and Gephyrosaurus from the Late Triassic and Early Jurassic of England and Wales respectively (Evans, 1980; Fraser, 1982; Whiteside, 1986; Jones, 2008), as well as in herbivorous eilenodontines from the Late Jurassic and Early Cretaceous of North America (Throckmorton et al., 1981; Rasmussen and Callison, 1981; Reynoso, 1996; Foster, 2003; Apesteguía and Novas, 2003). However, some derived taxa such as Clevosaurus hudsoni (Triassic, UK, Fraser, 1988) and possibly the pleurosaur Palaeopleurosaurus (Late Jurassic, Europe, Carroll, 1985) have a short articulation surface on the articular indicative of a different feeding mode. In Sphenodon the complete lower temporal bar probably serves to brace the quadrate and it has thus been considered associated with the proal shearing in this taxon (e.g. Whiteside, 1986). However, Gephyrosaurus which lacks a complete temporal bar demonstrates that it is not essential for proal jaw movement (Evans, 1980; Jones, 2008). Similarly, Clevosaurus and the fossil