q 1999 The Paleontological Society. All rights reserved. 0094-8373/99/2501-0004/$1.00

Paleobiology, 25(1), 1999, pp. 58–74

Mammal-like occlusion in conodonts

Philip C. J. Donoghue and Mark A. Purnell

Abstract.—Conodont element function and feeding mechanisms are of considerable paleobiologicalimportance, yet many details remain poorly understood and speculative. Analysis based on mor-phology, physical juxtaposition, and patterns of surface damage and microwear on pairs of Pa el-ements from individuals of Idiognathodus indicates that these elements crushed food by rotationalclosure, which brought the oral surfaces into complex interpenetrative occlusion. Other molariformconodont elements also functioned in this manner. Occlusion of this complexity is unique amongnonmammalian vertebrates, and is all the more surprising given that conodonts lacked jaws. Inaddition to enhanced understanding of food processing in conodonts, our analysis suggests thatmany details of conodont Pa element morphology, which underpin taxonomy and biostratigraphy,can now be interpreted in a paleobiological, functional context.

Philip C. J. Donoghue* and Mark A. Purnell. Department of Geology, University of Leicester, UniversityRoad, Leicester LE1 7RH, United Kingdom. E-mail: pcjd2@le.ac.uk and map2@le.ac.uk

*Previous address: School of Earth Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT,United Kingdom

Accepted: 20 September 1998

Introduction

Once described as little whatzit’s (Sweet1985), conodonts are now generally recog-nized as vertebrates (see Aldridge and Purnell1996 for a review), constituting one of the ear-liest, longest lived, and most diverse clades ofjawless fish. This reassignment has left cono-donts’ biostratigraphic and paleoecologicalutility undiminished but has dramatically in-creased their paleobiological significance: thecharacteristic phosphatic elements of cono-donts are among the first remains of verte-brate mineralized tissues in the fossil record(Sansom et al. 1992), and they provide the firstdirect evidence of macrophagy in early ver-tebrates (Purnell 1995). Feeding mechanismsare widely held to be of central importance inunderstanding the evolution and diversifica-tion of aquatic vertebrates (e.g., Pough et al.1996), so conodont element function, onceconsidered esoteric, now has an importantrole to play in understanding early vertebratediversity and in analyses of conodont evolu-tionary history. But many aspects of elementfunction remain poorly constrained and spec-ulative; if the study of conodonts is to realizeits paleobiological potential we need to un-derstand food acquisition and processing indetail.

In terms of species and individuals, theozarkodinid conodonts were the most diverseand abundant. We also know more aboutozarkodinids than any other group: almost allof the few known examples of fossilized co-nodont soft tissues are from taxa assigned tothe Ozarkodinida, and from modeling andanalysis of natural-assemblages preservingcomplete element arrays, the architecture ofthe apparatus is known in detail (Fig. 1) (Al-dridge et al. 1987; Purnell and Donoghue1997). In the ozarkodinid apparatus, the an-terior S and M elements (see Fig. 1) graspedfood (Aldridge et al. 1987; Purnell and Don-oghue 1997) and the P elements, particularlythe Pa elements at the posterior of the appa-ratus sheared and crushed (Purnell 1995). Un-like teeth in more familiar groups of verte-brates, conodont Pa elements were bilaterallyopposed across the sagittal plane, and naturalassemblages and clusters of diageneticallyfused elements indicate that in ozarkodinidsthe left (sinistral) element of the Pa pair lay be-hind the right (dextral) (Purnell 1995; Purnelland Donoghue 1997).

Whereas M and S elements are exclusivelycomblike in shape, many ozarkodinid cono-donts developed P elements with complexmolar-like morphology, and several authorshave suggested that they also developed in-

59MAMMAL-LIKE OCCLUSION IN CONODONTS

FIGURE 1. Position and orientation of the apparatus in ozarkodinid conodonts. The head is cut away to show thePa and Pb elements. The principal axes and the terms for orientation used in the text are also shown. Apparatusarchitecture after Purnell and Donoghue (1997).

terlocking occlusion (Jeppsson 1971; Nicoll1987; Weddige 1990; Purnell 1995). This hy-pothesis is supported by microwear patternsindicating that Pa elements were repeatedlybrought together with a degree of occlusalguidance that is difficult to reconcile with thefact that conodonts were jawless (Purnell1995). Occlusion is very uncommon amongvertebrates and has rarely been recorded out-side mammals. There are exceptions (e.g., twoCretaceous crocodyliforms [Clark et al. 1989;Wu et al. 1995], a Paleocene mammal-like rep-tile [Fox et al. 1992], two Triassic reptiles[DeMar and Bolt 1981; Carroll and Lindsay1985], a sauropod [Robinson 1956], a ptero-saur [Wild 1978], and lungfish, with complexdental plates that occlude [e.g., Kemp 1977]).In all these examples occlusion is much sim-pler than in mammal molars, and some au-thors have gone as far as suggesting that com-plex occlusal dentition is unique to mammals(Janis 1990; Smith 1993). Improvements in oc-

clusion during the early evolution of mam-mals were linked to changes in jaw structure,articulation, associated musculature, andbrain programs (Young 1978). The possessionof jaws at least is usually considered a prereq-uisite for occlusion, and the hypothesis ofcomplex interlocking occlusion in ozarkodi-nid conodonts is thus extremely surprisingand counterintuitive.

Were conodonts capable of mammal-likeocclusion? If so, how was element motion con-strained in the absence of jaws? Here, we ad-dress these questions through functional anal-ysis of morphology and patterns of surfacewear and damage on molariform Pa elementsdissected from conodont natural assemblages.

Previous Work on Occlusion in Conodonts.—Direct observation of the occlusal process inconodonts is obviously not possible, and ex-amination of occlusal surfaces on opposed el-ements is not a simple matter. Natural assem-blages preserve together the skeletal remains

60 PHILIP C. J. DONOGHUE AND MARK A. PURNELL

of individual conodonts, including articulatedelement pairs, but they are held together byindurated sedimentary matrix or by diagenet-ic minerals, and the occlusal surfaces of ele-ments are not visible. Thus almost all previouswork on occlusion in conodonts has relied onindirect methods, and all previous functionalreconstructions of element pairs are hypo-thetical. Jeppsson (1971) for example, recon-structed Idiognathodus Pa elements as an inter-locking occlusal pair but based this on linedrawings of elements in Lindstrom (1964: Fig.43f,g). The elements upon which these draw-ings were based are approximately equal insize and came from the same formation, butalmost certainly did not come from a singleindividual. Furthermore, although the spac-ing of the supposedly interlocking ridges andfurrows was measured, their depth and heightare entirely hypothetical (Jeppsson 1979). Ni-coll (1987, 1995) took Jeppsson’s approach fur-ther by physically reconstructing opposedpairs of Pa elements belonging to a variety oftaxa. Pairs of sinistral and dextral elements ofsimilar size were selected from collections ofisolated elements and reconstructed accord-ing to patterns of articulation in fused elementclusters (Nicoll 1985). He then photographedthe element pairs and evaluated the nature ofthe fit between the elements’ opposed articu-lating surfaces. Nicoll (1987) concluded thatbecause the pairs in his study did not fit close-ly together conodont elements must have beenpermanently covered by soft tissue in life.However, the poor degree of interlocking be-tween the elements observed by Nicoll mayhave resulted from his use of discrete elementcollections; like Jeppsson (1971), the elementshe put together almost certainly came fromdifferent individuals and had not worked to-gether in life.

Weddige (1990) drew on these earlier stud-ies in proposing antagonistic ‘‘see-saw’’ inter-action between ozarkodinid P elements, andPurnell and von Bitter (1992) later proposedsimilar interaction and element motion for Vo-gelgnathus. Purnell (1995) used an opposedpair of Pa elements of Idiognathodus as a dia-grammatic illustration of bilateral occlusionand left-behind-right pairing in ozarkodinidconodonts. This was based on the model of

skeletal architecture subsequently publishedby Purnell and Donoghue (1997, 1998) but thedetails of interlocking occlusion between theplatforms of the opposed elements were hy-pothetical.

Material and Methods

Our analysis is based on elements of Idiog-nathodus (sensu Baesemann 1973) from naturalassemblages on bedding planes of the un-named black shale unit in the Modesto For-mation (McLeansboro Group, Pennsylvanian)below the La Salle Limestone Member at Bai-ley Falls, Illinois (locality 1 of Rhodes 1952).Together with coeval deposits in the vicinity,this represents the most prolific source of nat-ural assemblages (e.g., Du Bois 1943) and pro-vided the material for the recent reconstruc-tion of the ozarkodinid feeding apparatus(Purnell and Donoghue 1997, 1998). One ofthe element pairs illustrated here (Birming-ham University, Lapworth Museum speci-mens BU 2683a and BU 2683b; hereafter re-ferred to as the BU pair) was recovered usingthe methods outlined in the appendix. Theother illustrated pair (University of Illinoisspecimens UI X-1509a and UI X-1509b; here-after referred to as the UI pair) is from thesame horizon and locality but was recoveredby Rhodes (1952) (deposited with the figuredmaterial). We are unsure of the techniquesused to extract these elements from the matrixof the bedding plane assemblage, but we sus-pect that the process was essentially mechan-ical.

The details of occlusion in these elementpairs were investigated directly by placing thePa elements in opposition. Gum tragacanthwas used as a temporary adhesive to hold theelements together for SEM examination.

Note that throughout this paper we use an-terior and posterior not in the conventional ar-bitrary sense, but to indicate the in vivo ori-entation of the elements. The terms dorsal,ventral, sinistral, and dextral also refer to invivo orientation and position (Fig. 1).

Morphology, Occlusion, Motion, andMicrowear

Morphology.—In life, the paired elementswere opposed across the animal’s axis of bi-

61MAMMAL-LIKE OCCLUSION IN CONODONTS

FIGURE 2. Stereo pairs of the Pa elements of Idiognathodus studied. The overlays are traced from the elements andindicate approximate contouring of the surface (highest dot densities correspond to lowest topography). These havebeen inverted and superimposed on the opposed element to show the inverse correspondence of morphologicalfeatures when occluded. A, Sinistral element BU 2683a, 348. B, Dextral element BU 2683b, 348. C, Sinistral elementUI X-1509a, 352. D, Dextral element UI X-1509b, 353.

lateral symmetry (see Purnell and Donoghue1997, 1998), yet the elements dissected fromthe bedding plane assemblages do not exhibitmirror-image symmetry and differ in signifi-cant morphological details of their oral sur-faces (Figs. 2, 3). Compared with the sinistralelement of the BU pair (Figs. 2A, 3A), the plat-form of the dextral element (Figs. 2B, 3B) isslightly wider: the maximum width of theraised area forming the anterior side of theplatform is approximately 0.2 mm, whereasthe maximum width of the same area on thesinistral element is approximately 0.14 mm.

On both elements, these raised anterior mar-gins are strongly convex anteriorly (Figs.2A,B, 3A,B), and considerably wider than theraised area on the posterior side of the plat-form (up to twice as wide). They bear trans-verse ridges that are asymmetrical, their ven-tral faces steeper than the dorsal faces. Theposterior margins of the two elements are lessconvex. Between the raised margins, both el-ements have a slightly depressed medialtrough that extends from the end of the ven-tral blade to the dorsal tip of the element. Thecentral area of the platform (i.e., above the

62 PHILIP C. J. DONOGHUE AND MARK A. PURNELL

FIGURE 3. Wear and surface damage on the Pa elements of Idiognathodus, oblique ventral views. Areas of damageand wear are outlined in black, the dashed lines indicate equivocal areas of damage. See text for details. Imagesare photomontages of several photomicrographs. A, Sinistral element BU 2683a, 3140. B, Dextral element BU 2683b,3140. C, Sinistral element UI X-1509a, 3180. D, Dextral element UI X-1509b, 3180.

apex of the basal cavity) is covered with smallnodes, but it grades dorsally into the smoothmedial trough. The ventral part of the oralsurface, in the area around the junction of theplatform and blade, is dominated by a com-plex arrangement of topographic highs andlows that together form a series of alternating

dorso-ventrally oriented ridges and furrowsthat lie in subparallel alignment on either sideof the blade. In the sinistral element the ven-tral blade is aligned with the medial axis ofthe dorsal part of the platform and the ridgeflanking the blade on the posterior side isaligned with the posterior margin. In the dex-

63MAMMAL-LIKE OCCLUSION IN CONODONTS

tral element, however, the blade and its lateralridges and furrows are offset posteriorly withrespect to the dorsal portion of the platform.The platforms of the two elements are similarin length and differ only slightly in area (si-nistral platform area 0.2 mm2, dextral 0.23mm2).

The sinistral element of the UI pair is bro-ken toward the dorsal end (Fig. 2C), presum-ably as a result of extraction from the shale.The platform of the dextral element (Figs. 2D,3D) is slightly wider than that of the sinistralelement (Figs. 2C, 3C). The raised area form-ing the anterior side of each element is orna-mented with transverse ridges. These ridgesare rather blunt on the sinistral element, butthose of the dextral element are sharp and de-scend toward the medial trough of the plat-form where they break up into nodes. Theraised areas forming the posterior margins ofthe elements are narrower, but approximatelyequal in height to the anterior margins. Theirornament is more nodose than ridgelike, thenodes rather flattened and forming a serratedposterior side to the medial trough. Both ele-ments have a deep, steep-sided medial trough,the floor of which is ornamented with smallnodes. As in the other element pair, the ven-tral portion of the oral surface, in the areaaround the junction of the platform and blade,takes the form of a series of alternating dorso-ventrally oriented furrows and nodose ridgeslying in subparallel alignment on either sideof the blade. In the sinistral element these areapproximately aligned with the axis and mar-gins of the dorsal part of the platform. In thedextral element, as in the other pair, they areoffset posteriorly. Both elements have amarked posterior protuberance at this point.

In both pairs, the ventral part of the plat-form of the Pa elements is morphologically themost complex part of the element; this is typ-ical of Idiognathodus sensu lato.

Occlusion of the Element Pairs.—Rather thaninfer the details of occlusion between opposedelements from their morphology, we physical-ly placed the elements together to observe oc-clusion directly. In each pair, the sinistral el-ement was positioned so that its blade lay be-hind that of the dextral element, as indicatedin natural assemblages (Purnell 1995; Purnell

and Donoghue 1997, 1998). Juxtaposed in thismanner, as the oral surfaces of the platformsare brought together they lock into position(Figs. 4D–F, 5D–F, 6).

Like the blades, the opposing platforms areoffset so that the sinistral platform sits slightlyto the posterior of its dextral counterpart(Figs. 4D,F, 5D,F). The principal point of artic-ulation is the ventral part of the platformaround the area where the blade joins the plat-form. The complex morphology of this arealeads to very precise articulation of the op-posing platforms; they interlock much moreclosely than the more dorsal parts of the oralsurfaces, largely because the depth and heightof the opposed articulating components allowconsiderable interpenetration (Figs. 4D–F, 5D–F, 6). The alternating dorso-ventrally orientedfurrows and ridges on either side of the blade(Figs. 2, 3, 6) also incorporate a number oftransverse protuberances that result in a verystable three-dimensional interlocking of theopposing platforms. Each morphologicalstructure is mirrored by an ‘‘inverted’’ or neg-ative structure in the opposing element, sothat each alternation of ridge and furrow ismatched by a furrow and ridge on the oppos-ing element (Figs. 2–6).

Rotating the elements about this point of ar-ticulation brings more dorsal areas of the oralsurfaces of the platforms together. They alsointerlock, but less precisely: in each pair theraised posterior margin of the dextral elementand the raised anterior margin of the sinistralelement occlude with the medial trough of theopposed element. Continued rotation, bring-ing the dorsal parts of the platforms together,causes the interlocking furrows and ridges ofthe complex ventral area of the platform todisengage (Figs. 4A–C, 5A–C).

The anterior edge of the dextral elementsand the posterior edge of the sinistral ele-ments are nonocclusal, protruding anteriorlyand posteriorly, respectively (Figs. 4, 5). Whenthe dorsal parts of the platforms on the BUpair are occluded, the transversely orientedridges of the opposing surfaces interdigitate.

Element Motion.—The complex interpenetra-tive interlocking of the furrows and nodoseridges in the ventral area of the platform ef-fectively restricts movement between the ele-

64 PHILIP C. J. DONOGHUE AND MARK A. PURNELL

65MAMMAL-LIKE OCCLUSION IN CONODONTS

←

FIGURE 4. Occlusal juxtaposition of the Idiognathodus Pa element pair BU 2683. A, B, C, Dorsal and middle partsof platforms occluded, dorsal, posterior, and ventral views respectively; D, E, F, Ventral area of platform occluded,dorsal, posterior and ventral views respectively. The sinistral element is on the left in all views. A, 377; B, 377; C,375; D, 377; E, 378; F, 377.

ments to simple opening and closure. Whenthe furrows and ridges are engaged, anterior-posterior motion between the elements is notpossible, as the elements can only move in thetransverse plane, articulating about the ante-rior-posterior axis. Viewed along this axis, thesurfaces of the platforms are convex and thewhole oral surface cannot be in simultaneouscontact (Figs. 4, 5). Instead the platforms comeinto contact by rocking from ventral to dorsal,the various matching morphological struc-tures interlocking as they meet. Concomitant-ly, as the platform surfaces occlude from ven-tral to dorsal, the surfaces of the blades movepast each other, then part; reversal of this mo-tion causes the platforms to occlude from dor-sal to ventral and the sides of the blades tooverlap once more (Figs. 4, 5, 7).

Surface Damage, Wear, and Microwear.—Be-cause the elements studied here are from ar-ticulated natural assemblages preserved inblack shales, they cannot have undergone postmortem transport, and surface damage can-not be the result of sedimentary abrasion.Thus the wear on the surfaces of these ele-ments must have been produced in vivo, pro-viding important and unequivocal corrobo-ration of surface wear observed on specimensfrom collections of disarticulated elements(Purnell 1995). Damage produced in vivo dur-ing the normal use of feeding structures pro-vides a type of evidence regarding functionfundamentally different from that derivedfrom interpretations of morphology; for fos-sils, it is the closest possible approximation todirect observation of function (Purnell inpress).

On the dextral element of the BU pair (Fig.3B), surface damage is evident on the tips ofnodes and ridges toward the ventral part ofthe platform, particularly in the area wherethe blade joins the platform. Surface damageis most intense on the crest of the blade in thisarea. Damage is also present on the crests ofthe ridges in the posterior margin, but the

ridges on the anterior margin retain sharpcrests and primary polygonal micro-orna-ment. On the sinistral element (Fig. 3A), dam-age is also concentrated in the ventral part ofthe platform and is especially evident on thecrest of the blade. The transverse marginalridges toward the dorsal part of the platformare pristine, with primary polygonal micro-ornament. On the dextral element of the UIpair (Fig. 3D), damage is clearly evident alongthe crest of the blade and on the crests of theadjacent ridges. The posterior protuberancehas a marked central depression that is alsothe result of surface damage. Except for twosmall areas of damage, the ridges of the an-terior margin of the platform are pristine andhave particularly well-preserved polygonalmicro-ornament. Polygonal ornament is alsopreserved on the posterior margin, except foran area near the crest that is smooth, possiblyas a result of wear. The sinistral element (Fig.3C) also exhibits clear damage on the crest ofthe blade and tips of adjacent nodes and ridg-es. The anterior margin preserves some pri-mary polygonal micro-ornament, especiallybetween ridges, but has some small patches ofdamage on ridge crests. The posterior marginpreserves polygonal ornament on its medialface, but the crest is smooth, possibly as a re-sult of wear. There is a patch of surface dam-age toward the dorsal end of the medialtrough, but as this end of the platform is bro-ken, damage during preparation cannot be ex-cluded as the cause.

Surface damage on the elements providesan independent test of our reconstructed pat-tern of occlusion and hypothesis of elementmotion. The distribution of surface damageaccords well with the reconstructed pattern ofelement occlusion. The nonocclusal anteriormargin on the dextral elements and the pos-terior margin on the sinistral elements exhibitthe least damage and the best-preserved pri-mary micro-ornament. Possible wear on thecrest of the posterior margin of the UI sinistral

66 PHILIP C. J. DONOGHUE AND MARK A. PURNELL

67MAMMAL-LIKE OCCLUSION IN CONODONTS

←

FIGURE 5. Occlusal juxtaposition of the Idiognathodus Pa element pair UI X-1509. A, B, C, Middle parts of platformsoccluded, dorsal, posterior, and ventral views respectively; D, E, F, Ventral area of platform occluded, dorsal, pos-terior and ventral views respectively. The sinistral element is on the left in all views. A, 387; B, 376; C, 389; D,387; E, 376; F, 389.

element is slightly anomalous, but it may havebeen produced by contact with the posteriorside of the posterior margin of the dextral el-ement, which also exhibits possible wear.However, the damage in these two areas issomewhat equivocal as it is possible that thelack of polygonal ornament on these surfacesis original. On all the elements damage islightest toward the dorsal end of the platform,and heaviest on the blade and adjacent nodesand ridges (the crest of the blade would orig-inally have been denticulated but has beenworn almost flat). This is consistent with thestresses that would have been generated inthis area if it was the hinge point about whichthe element articulated, with the interlockingof opposed ridges and furrows constrainingelement motion to the transverse plane. Dam-age to the tips of nodes in the central area ofthe medial trough in the elements of the BUpair (Figs. 3A,B) is also consistent with our re-construction of occlusion, as this is an areawhere nodes opposed nodes.

In addition to the distribution of surfacedamage and wear on the element pairs, themicroscopic textures developed within facetscan also yield valuable information. Many ofthe areas of damage on these elements exhibitchipped or pitted textures, but in some areas,particularly where surface damage extendsdown the sides of the blade, for example, thepitted appearance is probably the result of anirregular fracture rather than pitted micro-wear. However, pitted microwear is well de-veloped on the crests of the blades of all ele-ments (the BU pair especially), on several ofthe damaged nodes on the platforms of the BUelements, and on the nodes and ridges flank-ing the blade on the UI elements (Figs. 3C,D).Such pitting or chipping in teeth is diagnosticof crushing or compression (Gordon 1982;Maas 1994). Smoothly polished surfaces, an-other characteristic microwear texture, are de-veloped on the blades of the elements: on theanterior side of the sinistral elements and on

the posterior side in dextral elements (i.e., onthe occlusal surfaces) (Fig. 8). Similar micro-wear has been observed in the same positionon the blades of Gnathodus bilineatus Pa ele-ments (Purnell 1995); it indicates element-on-element contact either in the absence of food(Teaford 1988) or with nonabrasive food(Rensberger 1978). The smooth areas on theposterior margins of the UI element platforms(Figs. 3C,D) may be the original surface tex-ture, or they could be exhibiting polished mi-crowear textures. Perhaps the most significantfeature of the microwear on the elements is theabsence of any finely striated or scratched tex-tures seen on some other conodont elements(Purnell 1995). Scratched textures are pro-duced by shearing (Gordon 1982; Teaford1988; Maas 1991), and although the lack ofscratching represents negative evidence, takenwith the development of pitted textures itsupports the hypothesis that element motionwas constrained to a single transverse plane.

Function

Analyses linked to apparatus architecture(Aldridge et al. 1987; Purnell and Donoghue1997), ontogeny of elements and apparatuses(Purnell 1993, 1994), and microwear patterns(Purnell 1995) all indicate that the Pa elementsof Idiognathodus, and other ozarkodinids, wereinvolved in the processing of food rather thanits acquisition. Our analysis of element mor-phology, the nature of occlusion and inter-locking of the elements, and the patterns ofsurface damage and microwear provide sev-eral lines of evidence from which we have in-terpreted in detail how the Pa elements ofIdiognathodus performed this function. All theevidence of our analysis of the natural pairsindicates that motion between the elementswas restricted to bilateral occlusion in thetransverse plane, and that this was accom-plished by rotation (Fig. 7). The complex in-terlocking morphology of the ventral part ofthe platform, the presence of pitted microwear

68 PHILIP C. J. DONOGHUE AND MARK A. PURNELL

←

FIGURE 6. Complex interlocking occlusion of the ven-tral area of the platform in the Pa element pairs of Idiog-nathodus. Dorsal views, sinistral elements to the left. Up-per image BU 2683, ; 3260; lower image UI X-1509, ;3280.

textures on occlusal areas of the platforms, thepresence of smooth-polished areas on theblades, and the absence of any evidence ofshearing all indicate that Idiognathodus Pa ele-ments processed food by crushing. The asym-metry of the transverse ridges on the dorsalparts of the platforms (particularly on the BUelements) indicates that the power-stroke inthe occlusal cycle involved closure of the plat-forms from ventral to dorsal, the steeper ven-tral faces of the ridges acting to prevent foodparticles from being squeezed dorsally duringclosure. The reverse action reset the elementsfor another powerstroke, analogous to theopening of jaws in preparation for biting.

Much of the most significant information forunderstanding the details of how the elementsinteracted and functioned comes from theventral part of the platform. Morphologically,this area seems to be an adaptation to con-strain element motion to the transverse plane,but the surface damage and microwear devel-oped in this area provide more detail. Thecrest of the blade in this area of the elementbears clear evidence of breakage and pitting,indicating element-on-element contact andcrushing. This pitting cannot simply be the re-sult of crushing brittle foodstuffs as thiswould have left clear evidence in the form ofscratches on the sides of the blades, but as not-ed above, they are smoothly polished. Thedamage on the crests must have been pro-duced by direct contact of one blade crest onanother as the elements were brought togeth-er. This hypothesis is also supported by thelack of wear in the furrows adjacent to theblades. This indicates that at some phase in theocclusal cycle the ventral platform areas onopposed elements were parted. When partedthese areas would have provided no con-straints on element motion and could not haveacted together as a hinge. Thus the function ofthe morphologically complex ventral platformprobably changed through the occlusal cycle.

69MAMMAL-LIKE OCCLUSION IN CONODONTS

FIGURE 7. Operation of Idiognathodus Pa elements. A, Ventral blades partially occluded. B, Ventral area of the plat-form occluded and interlocking. C, Dorsal areas of the platforms occluded. Reversal of this motion resets the ele-ments for another powerstroke.

FIGURE 8. Surface damage and microwear on Idiognathodus Pa element BU 2683a. A, Oblique view showing bluntingand damage to the crest of the blade, ;3390. B, Enlargement of part of A showing where the primary striatedornament has been worn away to form smooth polished facets on the anterior surfaces of the denticles, ;3880.

70 PHILIP C. J. DONOGHUE AND MARK A. PURNELL

In the early stages of closure, the area func-tioned in simple crushing as the ridges andnodes were brought together, with the long,ventrally extended, flat-sided blades acting asthe primary means of element alignment. Asclosure continued, however, the furrows andnodose ridges would have started to interlock,and through the middle part of the crushingcycle this area would have provided a highlevel of constraint and prevented translationalmotion between the elements, thus maximiz-ing the efficiency of crushing between theplatforms. As the dorsal parts of the platformscome into contact, however, the interlockingfurrows and ridges disengage and occlusion ispoorly constrained (Figs. 4A–C, 5A–C). In life,occlusion may not have proceeded to thispoint, and this may account for the lack ofwear toward the dorsal ends of the platforms.

The cycle of closure in Idiognathodus is anal-ogous to the way in which interlocking occlu-sal molars operate in mammals (e.g., Cromp-ton and Hiiemae 1970). The surface morphol-ogy of the element platforms is also similar tothat of mammalian crushing teeth (cf. Rens-berger 1995). The raised ridges of mammalteeth act to reduce the surface area of tooth-food-tooth contact, concentrating appliedstress into a much smaller area and increasingthe efficiency of food breakdown (Rensberger1995). However, there must be a trade-off be-tween the apical angle of the ridges and themechanical strength of the brittle tissue fromwhich the elements are composed; the moreacute the apical angle, the higher the concen-tration of applied stress, hence the greater thelikelihood of brittle failure. A serious con-straining factor is thus the rheology of enameland its microstructure. Although enamel isthe most hard wearing of all vertebrate hardtissues, it is also one of the most brittle be-cause of its low organic content in comparisonwith other dental hard tissues such as ena-meloid, dentine, or bone. In Idiognathodus theproblem is exacerbated because unlike mostvertebrate teeth, conodont elements were al-most entirely composed of enamel (Donoghue1998).

Growth and Function of Idiognathodus Pa Ele-ments.—The Pa elements of Idiognathodus un-derwent considerable morphological change

during ontogeny (von Bitter 1972: Pl. 1; Pur-nell 1994: Fig. 4) and this must have affectedelement function. Juvenile Pa elements have amuch smaller platform than those of adults;they are basically blade-shaped with second-ary ridges above the basal cavity aligned par-allel to the blade (e.g., Purnell 1994: Fig. 4f),more closely resembling ancestral taxa. Thismorphology may have facilitated overlappingocclusion similar to that displayed by Gnath-odus bilineatus (e.g., Nicoll 1987: Pl. 5.3: Fig. 2),but without complex platforms these elementscould not have performed a crushing functionas efficiently as larger specimens. This musthave influenced prey selection. The smallestrecognizable juvenile may not, however, rep-resent a functional stage. Some studies of theinternal structure of conodont elements indi-cate periodic growth (Muller and Nogami1971, 1972; Zhang et al. 1997), and use mayhave been restricted to the end of each growthcycle (Donoghue and Purnell in press). Ma-ture platform morphology is attained at anearly stage in ontogeny (see von Bitter 1972:Pl. 1; Purnell 1994: Fig. 4) and platformgrowth in elements of more than 0.4 mm inlength is isometric (Purnell 1994). The small-est forms may thus represent an animal thatdied during initial growth of the feeding ele-ments prior to eruption. However, this canonly be rigorously evaluated by analysis ofmicrowear in relation to ontogeny.

Cyclical growth of the feeding elementscauses enormous problems in modeling his-togenesis, because it requires that the ele-ments be returned to the epidermis for sub-sequent growth to occur (Donoghue 1998).But how does this tissue-cover hypothesisstand up to our current view of element func-tion? Bengtson (1976, 1983) reconciled theneed for soft tissue cover and toothlike func-tion by suggesting that elements were evertedduring function and subsequently retractedinto an epithelial pocket. However, recent ad-vances in the understanding of conodont ele-ment growth (Muller and Nogami 1971; San-som 1996; Donoghue 1998) render Bengtson’sparadigm untenable, and we must now viewelements as either permanently or only peri-odically covered by soft tissue in a manner

71MAMMAL-LIKE OCCLUSION IN CONODONTS

analogous to growing denticles and scales(Donoghue 1998).

Previous Interpretations of Occlusion and Func-tion.—The results of this study highlight anumber of errors in previous hypotheses of Paelement occlusion and function. The diagram-matic reconstructions of Jeppsson (1971: Fig.3) and Purnell (1995: Fig. 1) incorrectly showPa elements of Idiognathodus in oppositionwithout the anterior-posterior offset betweenthe platforms of the sinistral and dextral ele-ment. In this position the medial troughs ofthe elements would have opposed each otherand the occlusal surfaces of the elements couldnot have been brought into interactive artic-ulation. Our demonstration of accurate occlu-sion between Pa elements also indicates thatthe lack of close articulation in many of Ni-coll’s reconstructed pairs (1987) is an artifactof combining elements from different individ-uals. His hypothesis that the main functionalsurfaces of the platforms did not come intoclose contact and were covered in soft tissueis undermined by the lack of space betweenthe elements, and is effectively falsified by thetraces of damage observed at points of contactbetween the interactive surfaces. Previous in-terpretations of rotational occlusion betweenelements (Weddige 1990; Purnell and von Bit-ter 1992; Purnell 1995) are supported by ouranalysis of Idiognathodus, but the limits of el-ement movement demonstrated by the naturalelement pairs in this study indicate that it isunlikely that the ventral surfaces of the ele-ments would have parted to the degree illus-trated by Weddige (1990: Text-fig. 15b). Al-though his hypothesis is based on anotherozarkodinid, Polygnathus, this genus also ex-hibits complex morphology in the area of theplatform-blade junction that would have con-trolled element alignment and articulation.Purnell and von Bitter’s (1992) interpretationof Vogelgnathus campbelli, with the occlusalsides of the ventral blades as primary func-tional surfaces, implies that the powerstrokein V. campbelli was during closure of the ele-ments from ventral to dorsal, opposite to thatin Idiognathodus and most other taxa (see be-low).

Asymmetry and Function.—Gross asymme-try between sinistral and dextral elements of

the same species has been widely recognizedin the past, but all previous pairings have beeninferred from criteria such as common range,co-occurrence, and statistical association. Thisis the first demonstration of asymmetricallypaired elements from an individual conodont.The recognition of asymmetry in elementpairs can be important taxonomically. Thereare several examples where elements origi-nally assigned to different taxa have been rec-ognized as sinistral and dextral elements of asingle species (e.g., Voges 1959; Lane 1968;Klapper 1971; Klapper and Lane 1985; Sand-berg and Ziegler 1979; Kuz’min 1990). Lane(1968) recognized several possible types ofpairing in conodonts and erected a scheme ofclassification. This has proven useful, but it isnow clear that when small but functionallysignificant details of morphology are consid-ered, all ozarkodinid Pa elements are asym-metrical (symmetry Class III).

From a functional perspective, the impor-tance of recognizing asymmetry in element-pairs cannot be overstated (cf. Purnell and vonBitter 1992). Asymmetry in element morphol-ogy does not imply asymmetrical feeding be-havior in conodonts (Purnell 1995; contra Bab-cock 1993) but is related to their complex bi-lateral interaction. Perfect mirror image pairsof elements could not perform an efficienttooth-like function requiring occlusal contactbecause a degree of asymmetry is necessary toallow the functional pairs to interlock (‘‘com-plementary symmetry’’ of Weddige [1990];Purnell 1995). Preliminary work indicates thatozarkodinid conodonts evolved a consistentpattern of pairing, with the sinistral elementbehind the dextral (Purnell and von Bitter1992; Purnell 1995; Purnell and Donoghue1997), and the apparent absence of intrapop-ulation variation in asymmetry indicates thatthis phenomenon is not a manifestation ofhandedness. Only one example of the oppo-site asymmetry has so far come to light (a spe-cies of Idiognathodus [Stamm 1996], not Icriodusas reconstructed by Weddige [1990]).

Occlusion and Function in Other Taxa.—Themost important constraint on element motionin Idiognathodus was undoubtedly provided bythe ridge and furrow system of the ventralarea of the platform. Similar structures can be

72 PHILIP C. J. DONOGHUE AND MARK A. PURNELL

found in numerous other conodont taxa in-cluding gnathodids and some polygnathids.Siphonodella, for example, developed as manyas three or four ridges and intervening fur-rows parallel to the blade, and once occludedthese would have provided unmatched con-straint over the relative motion. The lack oftransverse structures in the complex may haveled to axial slippage between the elements un-less this was prevented by the transverse ridg-es on the dorsal part of the platform. Manytaxa, however, exhibit no morphological fea-tures apart from the ventral blade that wouldhave enhanced alignment between opposedPa elements (e.g., Palmatolepis). But microwearon the blade-shaped Pa elements of Ozarkodinaconfluens (Purnell 1995) indicates that, in someat least, the relative motion between the ele-ments was highly constrained and that dorsalparts of the elements met in a manner verysimilar to the platforms of Idiognathodus. Thedenticles of the opposing elements must havemet in an offset intermeshed arrangement,shearing food as they came together. However,the physical constraints on relative motionand the means by which occlusion was con-trolled remain unknown.

Conclusions

Morphology, physical juxtapositioning, pat-terns of surface damage, and microwear pro-vide compelling evidence that the Pa elementsof Idiognathodus crushed food by rotationalclosure constrained by complex interpenetra-tive occlusion of the ridges and troughs in theventral part of the platform. This supports thehypothesis that, even without jaws, conodontsdeveloped dental occlusion of mammal-likecomplexity. Other conodonts with similar mo-lariform Pa elements probably crushed food ina manner similar to that of Idiognathodus.

Our hypothesis of element motion, func-tion, and food processing is better constrainedand more detailed than previous hypothesesof feeding proposed for conodonts or any oth-er groups of fossil agnathans, but it also hasimportant implications for more general un-derstanding and analysis of morphology andfunction in conodonts. Features of the oralsurface of conodont Pa elements that havelong been used, somewhat arbitrarily, as tax-

onomic characters, and that thereby underpinthe biostratigraphic utility of conodonts, canfinally be understood in a paleobiological con-text.

Acknowledgments

We would like to thank R. J. Aldridge forproviding some of the material on which thisstudy was based and for discussion during thecourse of this work. Our concepts of elementfunction have also benefited from discussionwith members of the Pander Society includingJ. Dzik, M. Lindstrom, K. J. Muller, R. S. Ni-coll, P. H. von Bitter, and W. Ziegler. J. M.Rensberger and M. F. Teaford have providedvaluable feedback regarding microwear on co-nodont elements. M. Fasnacht provided ref-erences on nonmammalian occlusion. S. D.Sroka, D. B. Blake, and R. D. Norby lent ma-terial from Bailey falls. We thank Anita Lof-gren and Ivan Sansom for their reviews of themanuscript. During this work P. C. J. D. hasbeen funded by a University of Leicester Stu-dentship and an 1851 Research Fellowshipfrom the Royal Commission for the Exhibitionof 1851; M. A. P. has been funded by a Nation-al Evnironment Research Council FellowshipGT5/F/92/GS/6, an Advanced FellowshipGT5/98/4/ES, and by J. A. Purnell.

Literature Cited

Aldridge, R. J., M. P. Smith, R. D. Norby, and D. E. G. Briggs.1987. The architecture and function of Carboniferous polyg-nathacean conodont apparatuses. Pp. 63–76 in R. J. Aldridge,ed. Palaeobiology of conodonts. Ellis Horwood, Chichester,England.

Aldridge, R. J., and M. A. Purnell. 1996. The conodont contro-versies. Trends in Ecology and Evolution 11:463–468.

Babcock, L. E. 1993. Trilobite malformations and the fossil re-cord of behavioral asymmetry. Journal of Paleontology 67:217–229.

Baesemann, J. F. 1973. Missourian (Upper Pennsylvanian) co-nodonts of northeastern Kansas. Journal of Paleontology 47:689–710.

Bengtson, S. 1976. The structure of some Middle Cambrian co-nodonts, and the early evolution of conodont structure andfunction. Lethaia 9:185–206.

. 1983. The early history of the conodonta. Fossils andStrata 15:5–19.

Carroll, R. L., and W. Lindsay. 1985. The cranial anatomy of theprimitive reptile Procolophon. Canadian Journal of Earth Sci-ences 22:1571–1587.

Clark, J. M., L. L. Jacobs, and W. R. Downs. 1989. Mammal-likedentition in a Mesozoic crocodylian. Science 244:1064–1066.

Crompton, A. W., and K. Hiiemae. 1970. Molar occlusion andmandibular movements during occlusion in the American

73MAMMAL-LIKE OCCLUSION IN CONODONTS

opossum, Didelphis marsupialis L. Zoological Journal of theLinnean Society 49:21–47.

DeMar, R., and J. R. Bolt. 1981. Dentitional organization andfunction in a Triassic reptile. Journal of Paleontology 55:967–984.

Donoghue, P. C. J. 1998. Growth and patterning in the conodontskeleton. Philosophical Transactions of the Royal Society ofLondon B 353:633–666.

Donoghue, P. J. C., and M. A. Purnell. In press. Growth, func-tion, and the conodont fossil record. Geology.

Du Bois, E. P. 1943. Evidence on the nature of conodonts. Journalof Paleontology 17:155–159.

Fox, R. C., G. P. Youzwyshyn, and D. W. Krause. 1992. Post-Ju-rassic mammal-like reptile from the Palaeocene. Nature 358:233–235.

Gordon, K. D. 1982. A study of microwear on chimpanzee mo-lars: implications for dental microwear analysis. AmericanJournal of Physical Anthropology 59:195–215.

Janis, C. M. 1990. The correlation between diet and dental wearin herbivorous mammals and its relationship to the deter-mination of diets of extinct species. Pp. 241–259 in A. J. Bou-cot, ed. Evolutionary paleobiology of behavior and coevolu-tion. Elsevier, Amsterdam.

Jeppsson, L. 1971. Element arrangement in conodont appara-tuses of Hindeodella type and in similar forms. Lethaia 4:101–123.

. 1979. Conodont element function. Lethaia 12:153–171.Kemp, A. 1977. The pattern of tooth plate formation in the Aus-

tralian lungfish, Neoceratodus forsteri Krefft. Zoological Jour-nal of the Linnean Society 60:223–258.

Klapper, G. 1971. Sequence within the conodont genus Polyg-nathus in the New York lower Middle Devonian. Geologica etPalaeontologica 5:59–79.

Klapper, G., and H. R. Lane. 1985. Upper Devonian (Frasnian)conodonts of the Polygnathus biofacies, N.W.T., Canada. Jour-nal of Paleontology 59:904–951.

Kuz’min, A. V. 1990. Asymmetrical pairs of platform elementsof Polygnathus (conodonts). Paleontological Journal 1990:62–70.

Lane, H. R. 1968. Symmetry in conodont element-pairs. Journalof Paleontology 42:1258–1263.

Lindstrom, M. 1964. Conodonts. Elsevier, Amsterdam.Maas, M. C. 1991. Enamel structure and microwear: an experi-

mental study of the response of enamel to shearing force.American Journal of Physical Anthropology 85:31–49.

. 1994. A scanning electron-microscope study of in vitroabrasion of mammalian tooth enamel under compressiveloads. Archives of Oral Biology 39:1–11.

Muller, K. J., and Y. Nogami. 1971. Uber die Feinbau der Con-odonten. Memoirs of the Faculty of Science, Kyoto University,Series of Geology and Mineralogy 38:1–87.

. 1972. Growth and function of conodonts. Pp. 20–27 in24th International Geological Congress, Montreal.

Nicoll, R. S. 1985. Multielement composition of the conodontspecies Polygnathus xylus xylus Stauffer, 1940 and Ozarkodinabrevis (Bischoff and Ziegler, 1957) from the Upper Devonianof the Canning Basin, Western Australia. Bureau of MineralResources Journal of Australian Geology and Geophysics 9:133–147.

. 1987. Form and function of the Pa element in the cono-dont animal. Pp. 77–90 in R. J. Aldridge, ed. Palaeobiology ofconodonts. Ellis Horwood, Chichester, England.

. 1995. Conodont element morphology, apparatus recon-structions and element function: a new interpretation of co-nodont biology with taxonomic implications. Courier For-schungsinstitut Senckenberg 182:247–262.

Norby, R. D. 1976. Conodont apparatuses from Chesterian (Mis-

sissippian) strata of Montana and Illinois. Ph.D. dissertation.University of Illinois at Urbana-Champaign.

Pough, F. H., J. B. Heiser, and W. N. McFarland. 1996. Vertebratelife, 4th ed. Prentice Hall, Upper Saddle River, N.J.

Purnell, M. A. 1993. Feeding mechanisms in conodonts and thefunction of the earliest vertebrate hard tissues. Geology 21:375–377.

. 1994. Skeletal ontogeny and feeding mechanisms in co-nodonts. Lethaia 27:129–138.

. 1995. Microwear on conodont elements and macropha-gy in the first vertebrates. Nature 374:798–800.

. In press. Conodonts: functional analysis of disarticulat-ed skeletal structures lacking extant homologues. Pp. 129–146in E. Savazzi, ed. Functional morphology of the invertebrateskeleton. Wiley, Chichester, England.

Purnell, M. A., and P. C. J. Donoghue. 1997. Skeletal architectureand functional morphology of ozarkodinid conodonts. Phil-osophical Transactions of the Royal Society of London B 352:1545–1564.

. 1998. Architecture and functional morphology of theskeletal apparatus of ozarkodinid conodonts. Palaeontology41:57–102.

Purnell, M. A., and P. H. von Bitter. 1992. Blade-shaped cono-dont elements functioned as cutting teeth. Nature 359:629–631.

Rensberger, J. M. 1978. Scanning electron microscopy of wearand occlusal events in some small herbivores. Pp. 523 in P. M.Butler and K. A. Joysey, eds. Development, function and evo-lution of teeth. Academic Press, New York.

. 1995. Determination of stresses in mammalian dentalenamel and their relevance to the interpretation of feeding be-haviors in extinct taxa. Pp. 151–172 in J. Thomason, ed. Func-tional morphology in vertebrate paleontology. CambridgeUniversity Press, Cambridge.

Rhodes, F. H. T. 1952. A classification of Pennsylvanian cono-dont assemblages. Journal of Paleontology 26:886–901.

Robinson, P. L. 1956. An unusual sauropod dentition. Zoologi-cal Journal of the Linnean Society 43:283–293.

Sandberg, C. A., and W. Ziegler. 1979. Taxonomy and biofaciesof important conodonts of Late Devonian styriacus Zone, Unit-ed States and Germany. Geologica et Palaeontologica 13:173–212.

Sansom, I. J. 1996. Pseudooneotodus: a histological study of an Or-dovician to Devonian vertebrate lineage. Zoological Journal ofthe Linnean Society 118:47–57.

Sansom, I. J., M. P. Smith, H. A. Armstrong, and M. M. Smith.1992. Presence of the earliest vertebrate hard tissues in co-nodonts. Science 256:1308–1311.

Smith, K. K. 1993. The form of the feeding apparatus in terres-trial vertebrates: studies of adaptation and constraint. Pp.150–196 in J. Hanken and B. F. Hall, eds. The skull. Universityof Chicago Press, Chicago.

Stamm, R. G. 1996. Reversals of misfortune: a new species? ofIdiognathodus (Conodonta) based on functional surface mor-phology. In J. E. Repetski, ed. Sixth North American Paleon-tological Convention Abstracts of Papers. Paleontological So-ciety Special Publication 8:369. Smithsonian Institution Press,Washington, D.C.

Sweet, W. C. 1985. Conodonts: those fascinating little whatzits.Journal of Paleontology 59:485–494.

Teaford, M. F. 1988. Scanning electron microscope diagnosis ofwear patterns versus artefacts on fossil teeth. Scanning Mi-croscopy 2:1167–1175.

Voges, A. 1959. Conodonten aus dem Untercarbon I und II (Gat-tendorfia- und Pericyclus-Stufe) des Sauerlandes. Palaonto-logische Zeitschrift 33:266–314.

von Bitter, P. H. 1972. Environmental control of conodont dis-tribution in the Shawnee Group (Upper Pennsylvanian) of

74 PHILIP C. J. DONOGHUE AND MARK A. PURNELL

eastern Kansas. University of Kansas Paleontological Contri-butions 59:1–105.

Weddige, K. 1990. Pathological conodonts. Courier Forschung-sinstitut Senckenberg 118:563–589.

Wild, R. 1978. Die Flugsaurier (Reptilia, Pterosauria) aus derOberen Trias von Cene bei Bergamo, Italien. Bollettino dellaSocieta Paleontographica Italiana 17:176–256.

Wu, X.-C., H.-D. Sues, and A. Sun. 1995. A plant-eating crocod-yliform reptile from the Cretaceous of China. Nature 376:678–680.

Young, J. Z. 1978. Programs of the brain. Oxford UniversityPress, Oxford.

Zhang, S., R. J. Aldridge, and P. C. J. Donoghue. 1997. An EarlyTriassic conodont with periodic growth? Journal of Micro-palaeontology 16:65–72.

Appendix

Elements were removed from their indurated black shale ma-trix using a modified version of a technique established by Nor-by (1976). Samples were immersed in a solution of approxi-mately 10% sodium hypochlorite with 10 grams of sodium hy-droxide added per 100 ml to promote the reaction; after 24hours or more, the shale surface had usually disaggregated andthe conodont elements were readily removed from the matrix;shale still adhering to the elements was removed by repeatedtreatment with sodium hypochlorite. Invariably, sediment load-ing and compaction of the shale resulted in fracturing of ele-ments in situ, and all attempts at reattachment of broken frag-ments using organic and dental bonding resins have failed. Bed-ding plane assemblages with the least evidence of fracture, par-ticularly of Pa elements at the point where the platform andblade join, were therefore preferentially selected for study.