Recordsofthe WesternAustralian Museum23: 91-113(2005). Structure and function of the tooth plates of the Devonian lungfish Dipterus valenciennesi from Caithness and the Orkney Islands JanL. denBlaauwen^, RichardE.Barwick^andKentonS.W.CampbelF 'SwammerdamInstituteforLifeSciences,UniversityofAmsterdam,Kruislaan406, 10985m,Amsterdam, TheNetherlands,[email protected] ^SchoolofEarthandMarineSciences,AustralianNationalUniversity,Canberra,ACT0200,Australia, [email protected],[email protected] Abstract-TheteethoftheMiddleDevonianDipnoanDipterusvalenciennesi are described from new material from Caithness and the Orkney Islands, Scotland.ThebiostratigraphyoftheOldRedSandstoneinthesetwoareas is described on thebasisofnew information.Thepallialdentineis madeupof groups of hard clusters of material. The core dentine in the tooth plates is nowunderstood intermsofthedevelopmentofindividualelementsmaking up the structure of the dentine. The first deposited material is interstitial dentine, and the second is transparent dentine which is deposited from the pulp canals against the interstitial dentine. All the core dentine is perforate. Denteons continue to the tip of the tooth, and dentine tubules run from the pulp canals through the transparent dentine to the pallial dentine. The structure is not thatof petrodentine. The difficulties ofusing living material for the understanding of dentine in remote structures in time are outlined. Therelationshipsoforganismsafterthedevelopmentofnewpalatalbitingin gnathostomesisdiscussed. INTRODUCTION Following White (1965) we consider Dipterus DipterusvalenciennesiSedgwickand Murchison, valenciennesiasavalidname. 1829 from the Middle Old Red Sandstone of Scotland was studied by White (1965). References STRATIGRAPHY to previous work can be obtained from his paper. Later work by Schultze (1975) and Ahlberg and A stratigraphic table showing the distribution of Trewin(1995)isavailable. the Middle Old Red Sandstone is attached (Figure The solid-snouted, cosmine-coated specimens 1)- from the Thurso Flagstones, described by Agassiz D. valenciennesi is well known from Eifelian or in 1844 as Polyphractusplatycephalus, was one of Givetian cyclic sequences in the Orcadian Basin. the reasons for Pander (1858) and Watson and Day Thespeciesiscommoninthefish-bearinglaminites (1916) to use the specific name Dipterus of Achanarras (Forster-Cooper 1937; Trewin 1986) platycephalus. More recently Westoll (1949) and the equivalent laminites on Orkney, the described theskull roofpatternsofspecimensfrom Sandwick fishbed (Trewin 1976). These laminites Banniskirk (Caithness) and found them sufficiently were deposited in deep water in an extended lake aberrant to separate them into a new species, in the Orcadian Basin. Small specimens lack Dipterus braebypygopterus, and he revived cosmine on the scales and dermal plates, but Agassiz's specific nameplatycephalus for all other specimens20cmoroverinlength,havecosmineon Scottish specimens of Dipterus. He advocated the partofthescalesontheventralsideandpartofthe abandonment of the name Dipterus valenciennesi. dermal bones. Fully grown ones have a complete White (1965) concluded that the braebypygopterus cosminecover. Dipterusis also found inthecalcitic pattern was a variation on other specimens that nodules from the Moray Firth area where the occurred at Banniskirk and elsewhere. The only sedimentsshow fluvial domination in a southward other genus comparable with Dipterus is a new extension of the Achanarras fishbed (Trewin and genus to be described by Newman and den Thirlwall2002). Blaauwen from the Middle Old Red Sandstone of Research in museum collections in theU.K. has Caithness and Sutherland, formerly included in shown that it is not possible to identify D. either Pentlandia or Dipterus. It has a different valenciennesi positively in sediments older than postcranial morphology and skull-roof pattern. the Achanarras-Sandwick fishbed horizon. 92 ].L.denBlaauwen,R.E.Barwick,K.S.W.Campbell CAITHNESS/ Fauna ORKNEY Fauna G SUTHERLAND ost—arthr—dipn. ost—arthr—dipn. I V John o' Groat Eday Flags E A T Subgr. -1 Ta '"] 1 I A Mey Subgr. Rousay Flags N 1 Mm. Mm. 1 1 Latheron Subgr./ Tp. 1 UpperStromness Tp 1 " Ham-Scarfsk.Subg. Flags E 1 I Dt. Gm Dt. F Achanarras Horizon Gm. Sandwick Fishbed E 1 1 L Robbery Head Oni. Dv. Lower Stromness Ga.= I Om. Dv. A Subgr. Flags N Pt Lybster Subgr. Cc. ost. Cc. Tni. Figure1 Biostratigraphic table of Caithness, Sutherland and Orkney indicating faunal elements which are of importance for correlation. Bars indicate the approximate range. Abbrcv.: arthr. arthrodire; Cc. Coccosteus cuspidatus; dipn. dipnoan; Dt. Dickosteus threiplandi; Dv. Dipterus valenciennesi; Ga. Gvroptychius agassizi;Gm. Gvroptychiusmilleri; Mm. Millerosteusminor, Pm. Pentlandiamacroptera; Ps. New Dipnoan Genus; Om. Osteolepismacrolepidotus', Osteolepispanderi; ost. osteolepid;Ta. Tristichopterusalatus; Tm. Thursiusmacrolepidotus',Wf. IVafsonosfeus. Specimens from the Lybster Subgroup which the dried up lake floors (Astin and Rogers 1991; belong to the osteolepid Thursius Rogers and Astin 1991). Sediments indicatingvery macrolepidotus are often misidentified as D. shallow lake deposits produce only disarticulated valenciennesi. The specimens from the Robbery fish remains, sometimes locally concentrated in Head Subgroup include a new genus being 'bonebeds'. described by Newman and den Blaauwen, and From extensive field work and mapping of fish thosefrom theJohn o' GroatSubgroup belongto remains a biostratigraphic pattern has been Pentlandiamacroptera. distilled. Naturally there are some difficulties in The lacustrine sediments above the Achanarras- correlationoffresh wafersequencesin theOrcadian Sandwick fishbed show climatically controlled Basin, where drying of parts of the lake and the cyclesresultingfrom long-term riseand falloflake prevalenceof desiccation features occur in contrast levels in an enclosed basin (Crampton and withwidespreadlakeextensionconditionsinwhich Carruthers 1914; Donovan et al. 1974; Donovan laminites were deposited. Details of the issues will 1980;TrewinandThirlwall2002).Thesebedsshow be discussed elsewherebydenBlaauwen etal.,but playa-lake conditions, though in places the water from the point of view of the dipnoans, D. may have been sufficiently deep to allow valenciennesi has been identified from the base of articulated fish skeletons to accumulate. The the Achanarras Horizon to the top of the Mey cyclicity of the sediments probably results from Subgroup on the mainland, and the equivalent Milankovitch periodicities. Some of the sediments Sandwick fishbed to the top of theRousay Flags in deposited in shallow water have polygonal Orkney. mudcracksand shrinkagecracks. Many secliments Specimensused inthisstudycomefrom theunits show structures the shape of gypsum crystals or indicated intheFigure1, abovetheAchanarrasand pseudomorphs showing gypsum crystal solution. Sandwick fishbeds. They are common in the These sedimentary structures are often preserved Latheron Subgroup and the Mey Subgroup in bysandinfill, introducedbywind transportacross Caithness and in the Upper Stromness Flags and ToothplatesofDevonianlungfish Dipterus 93 the Rousay Flags of Orkney. Also specimens from and part of the original median row resorbed. The Tynet Burn, one of the fishbeds from the nodule same specimen shows gaps for the occlusion with localitiesintheMorayFirtharea,havebeenstudied. themandibularteeth. Irregularityofrowsisshown Specimens of D. valenciennesi sampled in byG2004.10.5 which leavesspaces forthe insertion sediments indicating shallow lake conditions, are ofnew rows on the mediolateral parts ofthe teeth. disarticulated and are mostly mature or even fully Obviously the new rows were formed wherever a grown. Most specimens possess a well developed space exists because of irregular growth in old cosminecoating. rows, and we conclude that genetic control on the precise position of new teeth was limited. New rows occupy only a small part of the length of the SPECIMENSEXAMINED head. All the specimens examined have come from The parasphenoid is well defined, is up to three Caithness and the Orkneys. They have been taken times thelengthofthetooth plates, andhasa well- from the collections of den Blaauwen, Michael defined buccohyphophysial foramen. The nasal Newman and Jack Saxon, and they have been capsules occupy about two thirds of the length of placed in the National Museum ofScotland (NMS) the plates. Most of the posterior buccal cavity is collections.Thenewnumbersareasfollows: thereforenotroofedbythedentalplates.This point G2004.10.1 From Clardon Haven, Caithness. is emphasized bythemandiblein which the dental Latheron Subgroup. Posterior end eroded in situ. plates are relatively small in relation to the whole Palataltoothplateswelldeveloped. structure. G2004.10.2 From Clardon Haven, Caithness. Latheron Subgroup. Palate sectioned to show the StructureoftheMandible 'cosmine'betweentheteeth. The best specimens we have are of individuals G2004.10.3 From Clardon Haven, Caithness. which are a little above half grown, and show LatheronSubgroup.Mandiblewithtoothplates. features which we consider significant. The tooth G2004.10.4 From Thurso East, (the slates), platesare0.33-0.40thelengthofthejaw(Figures3 Caithness. Latheron Subgroup. Mandible with left A,B), and the distance between the two tooth toothplatelost. plates is large in comparison with the Early G2004.10.5 From Thurso East, Caithness. Latheron Devonian genera Dipnorhynchus and Subgroup. Right palatal tooth plate. Sectioned Speonesydrion (Campbell and Barwick 1984). The horizontally,andvertically. ratioofthemedian lengthtothetotal lengthofthe G2004.10.6 From Clardon Haven, Caithness. mandible, is only about one third. Note also that LatheronSubgroup.Pectoralgirdle. the mandibular dental plates have a short median G2004. 10.7 From Thurso East, Caithness, Latheron length in comparison with the posterior length. Subgroup. This is different from the shape of the palatal G2004.10.8 From Clardon Haven, Caithness, dental plates, suggesting that the contactbetween LatheronSubgroup. the mandibular and palatal plates was not one-to- G2004.10.9 From Buckquoy west of Aikerness, one. This interpretation is supported by the fact MainlandOrkney,RousayFlags. that the inner face of the mandibular plate is G2004.10.10Sameas2004.10.9. turned ventrally, and could not have met the G2004.10.il From Thurso East, Caithness, Latheron palatal plate on full closure of the jaw. This is Subgroup. standard for the assembled Late Devonian species G2004.10.12 to G2004.10.16 From Clardon Haven, (Barwick and Campbell 1996; Campbell and Caithness, LatheronSubgroup. Barwick 1998). A second point is that the anterior gap between the two mandibular plates is very large, and the GROSSFEATURESOFTHEDENTALSYSTEM unencumbered space for the tongue pad would be not only widebutalso long and deep.The point of ThePalatalToothPlates originofthetoothrowshavebeen resorbed,andon New rows of teeth are introduced between the G2004.10.3 restorative dentine has been added to anterior sets of rows as spaces become available the anteromedian side ofthe tooth plate.Thetooth (Figures2, 8A). Somespecimensshow symmetrical plates of Dipterus platycephalus from a Scottish insertionsofthetwoplatesoftheonespecimen,but specimen in the Manchester Museum, and figured others do not. The specimen figured by White byWatsonandGill (1923,figure34),alsoshowsthe (1965, plate 1, figure 1) shows small teeth inserted reduced tooth plates similar to those described between the first and second rows, and in places above. The cavities for the cartilage forming the these teeth are more closely spaced. Specimen articulationwith thequadratearedeepand slightly G2004.10.8(Figure2A) isremarkablein thatithasa doubled (Figure 3C), thus limiting the lateral new irregularrow ofteeth anteromedially inserted. movementofthemandible. 94 J.L.denBlaauwen,R.E.Barwick,K.S.W.Campbell 10mm Figure2 A, palateofG4004.10.8. Specimen antero-posteriorlycompressed, 'cosminc' removed;boneon theposterior ofthetooth plates. B,G2004.10.], 'cosmine' on the palate; first rowofteeth partly covered by'cosmine'. C, palatal view ofthespecimenG2004.10.2; 'cosmine'scctioned from right palatal tooth plate. D, thespecimen fromwhichFigure2Bwasprepared;squashedantero-posteriorly.Scale=10mm. ToothplatesofDevonianlungfishDipterus 95 the reality of Westoll lines will be considered later inthispaper.Thepresenceofenamelonthesurface of this tissue is the most important point to be consideredhere. SignificanceofTheseGrossFeatures The functional significance of these features is largely related to air breathing. We consider these points under the following headings; extant air breathers; palatal plates and parasphenoid; tongue padspace;andbrachiallaminae Extantairbreathers Extant dipnoans fall into two groups - Lepidosiren and Protopteruswhich areobligate air breathers, and Neoceratoduswhich is a facultative airbreather.Thesetwogroupshavebeendiscussed by Thomson (1969), who has also compared them withtheMiddleDevonianDipterus. In Lepidosiren and Protopterus (Bishop and Foxon 1968) the tongue fits between the pterygoid tooth plates and makes a closing valve when air is depressed into the lungs. This is done by the anterior rotation of both the ceratohyal and the pectoral girdles. The air is stored in the parabranchial cavity partly roofed by the elongate parasphenoid, the teeth are small with respect to thesizeoftheheadand theyareseparated to leave a space for thetonguetocloseoffthebuccal cavity when air is forced into the lungs. Associated with this procedure is the increase in spacebetw^een the mandibular tooth plates which allows the tongue pad to expand forwards.Theventral surface ofthe head is also able to expand the buccal cavity to permit more space for the retention of air. Neoceratodushasadifferentarrangementbasedon an opercular pump, and Thomson (1969, figure 5) showsthemovementoftheopercularfold duringa breathing phase. In addition Neoceratodus has a massive ceratohyal which takes part in breathing Figure3 A-C, two small mandibles G2004.10.3 and G2004.10.4; A and B are dorsal views show- movements. ingtooth-plates.Cisamoreposteriorviewof B showing the articulation cavities and the PalatalPlatesandParasphenoid positionoftheadductorfossae. D, G2004.10.7 The palatal plates in Dipterus are situated well showingbothtoothplates. anterior in the mouth, and they are well separated from each other. In comparison with such Early Devonian genera as Dipnorhynchus or the Late Devonian Chirodipterus the plates are very short. 'Cosmine'ontheMedialPartsofthePalatal The parasphenoid has a long posterior projection, Tooth White (1965, plate2) and Denison (1974, figure4) aptnedryegxotiednsd.s bBaoctkhotvheersea fleoantgurdeisstamnackeebefhoirndlotnhge figured a thin layer of tissue occupying the space between thetwo palatal tooth plates. Itwas termed orobranchialandparabranchialcavities. 'cosmine' by both the above authors, because this layer has a shiny surface, it often contains a large SpacefortheTonguePad number of pores, and superficially it has a The gap left for the tongue between the resemblance to cosmine. But pores are not always prearticulars is large and deep in comparison with present, and where pores are present, no pore that of Early Devonian genera Dipnorhynchus and canals can be found beneath them. This point and Speonesydrion. The gap between the palatal tooth 96 J.L.denBlaauwen,R.E.Barwick,K.S.W.Campbell plates is covered with 'cosmine', and this shows morphological features which would indicate that that the large tongue pad had ample room to lie theywereairbreathers. betweentheseplateswhenthemouthisclosed.The enamel surface on the 'cosmine' shows that the epidermis was in contact with this surface. This is DESCRIBEDHISTOLOGYOFTHETOOTH the ideal arrangement for the stop valve when the PLATES air was being forced into the lungs from the ThehistologicalstructureofD. valenciennesiwas orobranchialandparabranchialcavities. not described from Scottish material until recently, because like all the bones at the fossiliferous PectoralGirdleofD.valenciennesi localities, the teeth were deeply stained by organic We have access to several specimens of D. carbon. White (1966, plate 1, figure 2) published a valenciennesiwhich arebetter preserved than any figureofasection ofa tooth plate inwhich theend specimens previously described. The four tooth was sectioned medially and showed a specimens are now labeled G2004.10.6, translucent core. The figured adjacent teeth G2004.10.13,G2004.10.14andG2004.10.15. Wewill apparently showed a bony core, presumably describe this material in a separate paper. The because they were from marginal sections of the pattern of the branchial laminae and the teeth. Denison (1974: 39) commented on these scapulocoracoid are very similar to those on structures,buthisworkhasnotbeenconfirmed. Chirodipterus australis (Campbell and Barwick, Smith (1984) described teeth from the specimens 1987)a marineform from theLateDevonianGogo described by White (BMNH P44691), and later Formation, Western Australia. The branchial another specimen, BMNH P53537, from Caithness laminae would have operated in the same way in (Smith 1989). These papers give no details of the thetwospecies. pallial dentine,and their structureofthetooth core is obscure. Comments will be made on this work Summary laterinthispaper. Dipterus has a large opercular plate as well as Kemp (2001) in her paper on petrodentine does smallsuboperculars,themovementofwhichwould not describe the histological structure of D. haveproducedalargeexpansionandcontractionof valenciennesi, and most ancient forms dealt with theparabranchial chamberasdetailed byThomson are of Carboniferous age. Because of this lack of (1969). Wenotethat Neoceratodususesa opercular direct analysis of Dipterus valenciennesi tooth parabranchial pump when breathing air, and uses plates, the discovery of well-preserved specimens the very large ceratohyals to push the air from the nowgivesusanopportunitytoplacetheseplatesin orobranchial and parabranchial chambers into the the primitive position which their stratigraphic lung. On the other hand, thebranchial laminae are positionaccordsthem. so large and the ceratohyal so short, that Dipterus could not have used the methods of breathing adopted by Lepidosiren and Protopterus. We have INTERPRETATION OFDENTINEINNEW concluded that Dipterus was a facultative air MATERIAL breather,thoughthestructurescouldnothavebeen asefficientasthoseofNeoceratodus. There is nothing more contentious than the WenotethatSchultzeand Chorn (1997) consider terminology of dentine in dipnoan teeth. For thatlungswereafeatureofprimitiveosteichthyans, present purposes Smith (1984, 1985, Table 1) has provided the basis on which subsequent work has quoting the fact that lungs are present in primitive developed.Furtherworkon thistopiccanbe found actinopterygians (e.g., Polypterus), actinistians, lungfish, and tetrapods. Campbell and Barwick in Lison(1941), Barwicketal. (1997),Campbell and (1999: 137-138) have commented on their Smith (1987), Lund et al. (1992), Kemp (2001) and arguments, and these will not be repeated here. Reiszetal. (2004). Incidentally they offer no mechanism supporting theirviews.Neoceratodusisalsogillbreathing, and VerticalSectionsofTeeth spendsmostofitslifesubmerged.Comparisonwith The considerable advantage we have is the Dipterus valenciennesisuggests that this Devonian availabilityofgrowthstagesoftheteeth.Growthof form alsohad thecapacity tousegill respiration as the layers in the dentine can be outlined by a wellasaerialrespiration. number of specimens. In the first instance we Our arguments are based on morphology of describeanumberofteethfromasinglesection. Dipterus and the extant dipnoans, and not on cladistics or the range of air breathing in some Sectionsthroughsedhnentwithplacodermplates extant animals. Contrary to the argument of Some specimens have structures sufficiently well Schultze and Chorn (1997), we still maintain the preserved to show histological detail throughout marine Devonian dipnoans lacked the theteeth.SectionG2004.10.9showsthebestvertical ToothplatesofDevonianlungfishDipterus 97 towardsthebase.Theedgescontainsomebranched tubules which arise directly from the pulp canals, and these areeither simpleor branched. At higher magnifications, the tubules subdivide extensively towards their outer edges (Figure 5A) making a meshwork-like pattern. In the lower half of the tooth the pallial dentine shows very fine tubules, butatthebaseofthetooth,thepallialdentineturns inwards(Figure5B)andhascrenellateddendriform pattern. This is best known as pedestal dentine. Pallial dentine is well shown on the lateral tooth and the tangential tooth (Figure 6). It forms a dark layerwhich isalsopenetratedbytubules. In places the boundary between the pallial dentine and the core dentine is sharp, but in other areas the boundaryisgradational. On G2004.10.9, interstitial dentine is clearly exposed (Figures 4, 6). Dark columns of interstitial dentine extend into the basal pulp cavity where their outlines are clear. In the basal pulp cavity at highmagnification theinterstitialdentineshowsan open-work structure, which is seen on all three teeth (Figures 5B; 6B; 14E). The implication is that the growing edge of the material in the basal pulp cavity is made of crenellated dendriform material, which has the same appearance as the pedestal dentine, and laterally this material joins the pallial dentine (Figure 5B). In the central part ofFigure 4, and the distal parts ofFigures 6A,C, the interstitial dentinebecomes vaguely outlined, notbecause the section is marginal to the dentine layer but the dentine is partly transformed into translucent dentine. The marginal tooth is very informative with respect to the formation of the interstitial dentine. Figure4 G2004.10.9. Vertical section in single AsshownonFigure6A, inthecoreofthe tooththe polarized light; specimen was incompletely interstitial dentine becomes rough in its outline grown; enamel is largely destroyed; towards the basal pulp cavity. On the left interstitial dentinegrey in colour; translucent ventrolateral margin the interstitial dentine grades dentinearoundpulpcanals. into the pallial dentine and ventrally into the pedestal dentine as shown on Figure 6B. The toothsections wehaveseen (Figures4-6; 14E). The tangential tooth (Figure 6C) also shows the largest tooth in the section has a large basal pulp interstitial dentine extending to the basal pulp cavity, and the smaller teeth are in process of cavity. formation. These show the development of the The translucent columns are the most striking histologicalstructures. part of the core, and show up in plain light as a Thelargesttoothhaslostitsapex,butonthesame clear translucent structure. Near the crest of the sectiontwoothersmallteeth,oneatthelateraledge largetooththestructureisclear(Figure5A)andthe of the plate (referred to as the lateral tooth below) translucent layer carries many tubules which are and the other being cut tangentially to the axis of derived from the pulp canals, and in places these thetooth (referred toasthetangential toothbelow. run through the layer into the pallial dentine. This Figures 6; 14E). The enamel is present around the is in an early stage of the tooth formation. In the marginofthelateraltooth,butitispartlydestroyed central part of the tooth, the fine structure of the by decomposition on all sides of the larger tooth. translucent layers is not clear, but it does contain The core dentine consists of two different types of small openings and vague lines. The mode of structure, clear translucent columns, and formation of this tissue is clearly demonstrated by interstitialdentinecolumns(Figure14E). this section (Figures 4; 5B; 14E). It was deposited As Figure 4 shows, the pallial dentine is thickest against the interstitial dentine by cells in the pulp towards the apex but fades away somewhat canals. Thebase ofthe large tooth shows a layer of 98 J.L.denBlaauwen,R.E. Barwick,K.S.W.Campbell pulpcanals pallial dentine interstitial dentine tubules translucent dentine translucent dentine interstitial dentine crenellated dendriform dentine pedestal dentine Figxu-e5 A, apex of Figure 4; tubules penetrating the translucent dentine and connecting with the pallial dentine; translucentdentinehasreplacedmostoftheinterstitialdentine.B,enlargementofthebottomrightofFigure 4; interstitialdentinearoundthepulp canalsandtranslucentdentinedeposited fromthepulp canalsonthe interstitial dentine;crenellatededgeofthepallialdentinejoined by acolumnofverticalinterstitialdentine, andbasallyjoiningthepedestaldentine. ToothplatesofDevonianlungfishDipterus 99 Figure6 G2004.10.9asonFigure14E.A,thelateraltoothrotated;interstitialdentinerunsfrompulptopallialdentine; pallial dentinewithtubules;translucentdentinearound pulp canals. B,bottom leftofFigure6A with pallial dentinejoiningwiththepedestal dentine.C,thetangentialtooth. Figure14E;contactbetweeninterstitialand translucentdentineintergrading;translucentdentineoftenbulbous. 100 J.L.denBlaauwen,R.E.Barwick,K.S.W.Campbell 10mm dentine mm 0.1mm-- •interstitwi 0.1 .*j .'I .V i dentinl carbonate gulp I in pulp canals ' canals. V w. Figure7 A,G2004.10.5C,slightlyetched inaceticacid;coreofeachtoothextendsintothebasalbone.BandC,optical sections cut from the opposite face of Figure 7A; baseof pulp canals with CaCOj; interstitial dentine near margins but replaced medially by translucent dentine. D, enlargement of the area outlined in A; pallial dentineonleftjoinedwithpedestaldentineofrighttooth.