Further Consideration of Placoderm Evolution (1983)

8/2/2019 Further Consideration of Placoderm Evolution (1983)

1/16

Further Consideration of Placoderm Evolution

Author(s): Robert H. DenisonSource: Journal of Vertebrate Paleontology, Vol. 3, No. 2 (Sep., 1983), pp. 69-83Published by: Taylor & Francis, Ltd. on behalf of The Society of Vertebrate PaleontologyStable URL: http://www.jstor.org/stable/4522935 .

Accessed: 25/10/2011 06:55

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

The Society of Vertebrate Paleontology and Taylor & Francis, Ltd. are collaborating with JSTOR to digitize,

preserve and extend access toJournal of Vertebrate Paleontology.
http://www.jstor.org/action/showPublisher?publisherCode=taylorfrancishttp://www.jstor.org/action/showPublisher?publisherCode=vertpaleohttp://www.jstor.org/stable/4522935?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/4522935?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=vertpaleohttp://www.jstor.org/action/showPublisher?publisherCode=taylorfrancis


2/16


3/16

Petalichthyida, Arthrodira, and Antiarcha, but givesno indication of regression of the dermal skeleton, ex-cept perhaps in the reduction of the superficial layerof dermal bones and in the loss of the scalation of thetail. The Acanthothoraci have been considered to pro-vide evidence for both skeletal assimilation (Gross,1958:26) and regression (Orvig, 1975:52), but the orderis known only from the Early Devonian and the evi-dence is inconclusive. Only in the Rhenanida can acase be made for regression of dermal bones (Stensi6,1969:90), for in the Late Devonian Jagorina (Fig. 3C)tesserae have replaced the suborbital plates of the EarlyDevonian Gemuendina (Fig. 3A, So). However, thisis not evidence of a general trend in Placodermi, butmay be rather an adaptation for increased flexibilityin a highly specialized group of fishes.Also in support of the theory that the ancestral plac-oderm was armored, Miles (1967a:66-67) has arguedthat the characteristic features of the group evolved asa result of the loss of flexibility of the trunk region dueto the presence of a rigid armor. If this were true, thearmor itself would have been a primary feature of plac-oderms, and could have led to the formation of thejoint between the occipital region and the anterior ver-tebrae, to the fusion of the latter into a synarcual, andto the crowding of the gills forward under the head.However, the anterior position of the gills may be aprimitive character, inherited from the ancestralgnathostome (Zangerl, 1981:39). The cranio-occipitaljoint occurs also in chimaeroids in which the synarcualsupports the large dorsal spine and fin, in rays in whichthe synarcual helps support the tremendously enlargedpectoral fins, and in certain teleosts in which the an-terior body is inflexible because of enlarged scales. Therecan be no doubt that in all of these the synarcual,whatever else its function, resulted in a stiffening ofthe anterior trunk. This is true also in placoderms, butthat it arose as a result of the enclosure of the trunk inarmor is improbable since it occurs in Stensioellida,Pseudopetalichthyida and Rhenanida, none of whichhas a trunk shield, but only a slender shoulder girdle.I suggest that the neck joint and synarcual of Placo-dermi may be simply an adaptation making it possiblefor these primarily benthic fishes to raise the head inbreathing, eating, and swimming without also raisingthe anterior thoracic region as in Osteichthyes. It couldbe argued that the neck joint was a prerequisite for theformation of the trunk shield rather than the converse.The above discussion shows that there is little sup-port for the view that the ancestral placoderm washeavily armored. That the ancestor was unarmored issuggested by negative evidence, namely the absence ofany trace of placoderms in Silurian rocks. [The occur-rence of Antiarcha in the Late Silurian of south Chinareport by Pan (1981:69) requires confirmation.] Bonesof a heavy placoderm shield would be easily preserv-able and very possibly identifiable even in small frag-ments if coated with the characteristic placoderm semi-dentine. Romer (1968:12) attributed the rarity ofvertebrate remains in the early Paleozoic to the dearth

of deposits of inland waters, in which he believed theancestors of Devonian fishes were evolving. But, withthe exception of most Arthrodira and possibly Antiar-cha, all Early Devonian Placodermi were marine, soit is safe to assume that some had marine ancestors inthe Silurian. Since there are abundant and varied ma-rine Silurian formations, I conclude that placodermslacked dermal bones at this time, or that those pos-sessing them were few in number and restricted togeographic areas or facies that have not been ade-quately sampled.Except for a few fragmentary remains in the LateSilurian, bones of Osteichthyes are not known beforethe Devonian, so the question arises whether bonesevolved independently in Placodermi and Osteichthy-es, or whether they shared an ancestor with dermalbones. Forey (1980:380) supports the second alterna-tive, identifying placoderms as a sister group of Os-teichthyes based on the shared possession of dermalbones on the skull roof, gill chamber, palate, and shoul-der girdles, as well as endochondral bone. His synapo-morphies do not all withstand detailed scrutiny, andthe grouping may also be criticized in that it is basedon a single complex of characters, unsupported by oth-er resemblances. Those who have chosen other char-acters have come to different conclusions. Thus Stensi6(1925:187-188) believed he had demonstrated a re-lationship of placoderms to elasmobranchs, basedmainly on the endocranium. Watson (1937:133)grouped placoderms with acanthodians, based on hisinterpretation of the jaw and gill structure. Schaeffer(1975:108) concluded on the evidence of the jaw mech-anism that placoderms were "a separate and distinctgroup, not more closely related to the Chondrichthyesthan to the Teleostomi." Zangerl (1981:37-38) alsoconsidered placoderms a sister group of Teleostomiplus Chondrichthyes, basing this on the manner oftooth replacement or non-replacement. Other rela-tionships have been proposed (see Denison, 1978:18)based on different characters or different interpreta-tions of the same characters. In my opinion, the con-flicting evidence supports the view that Placodermi arenot closely related to either Osteichthyes or Chondrich-thyes, and that dermal bones evolved independentlywithin Placodermi, Osteichthyes, and Acanthodii. Thisis supported by differences in bone patterns and ho-mologies, histology, tooth replacement, and the struc-ture of the endocranium and visceral arches. The Acan-thodii have significance in this matter, for if theirrelationship to Osteichthyes can be confirmed, it surelymeans that these two groups shared a micromericancestor.

TESSERAETesserae are here defined as small, thin, superficialdermal elements, without a basal or well developedmiddle layer, abutting on neighboring tesserae ratherthan overlapping them. It has been suggested by Gross(1962a:71) that the skeleton of the ancestral placodermconsisted oftesserae, but Stensi6 (1963:403) has argued

70 JVP 3(2), September 1983


4/16

that they are the result of fragmentationand thus area secondarydevelopment.It canhardlybe maintainedthat the tesserae of the body and fins of Stensioellida,Rhenanida,and Acanthothoraciare the resultof dis-integrationof largeplates,and sincetheyareessentiallyidentical to the tesserae of the head, it can be arguedthat the latter are also not the resultof fragmentation.If this is true, tesserae are either (1) a primitive plac-oderm character ost in most orders,or (2) indepen-dently acquiredin the three orders(as well as in thecheeks of Lunaspis),or (3) a derived characterof thecommon ancestorof two or three of the orders.I con-sider the second to be unlikelyas well as unparsimo-nious. Thethirdpossibilitycould besupportedbyotherevidence of relationshipbetween the tesserateorders,some of which will be consideredbelow. I favor thefirsthypothesisbecause tesseraeare presentin ordersthatI considerto be primitiveon the evidenceof othercharacters,and are absent or largelyabsent in ordersthatI considerto be advanced.However,this is merelyan assumption, unprovable on evidence presentlyavailable.

Based on thisassumption,Stensioellamayrepresentthe most primitiveknown condition. Most of this fishis covered with small denticles(Fig. 1, de),possiblyofdentine,andthoughthe preservation s poor,there areplaceson the head,pectoralfins,and trunkwheretheyappearto be set on small superficialplates suggestiveof tesserae(Fig. 1, tes)and arrangedn diagonalrows.Onlyin the posteriorpartof the skull roofand markedby a radiatingstructureare a few dermal bones (Fig.1, Ce)that must have developed deeperin the dermisand probably ndependentlyof tesserae or denticles.IntheAcanthothoraciesseraehave beeninterpretedas secondary due to exoskeletal regression(Stensii6,1963:403; Orvig, 1975:52), or as primarystructures(Gross, 1958:26-27; Mark-Kurik,1973:329). In theearliestknowngenus,Kimaspis Fig.2C),theyare saidto overlie the dermal bones which completely coverthe cranial roof (Mark-Kurik,1973:325). But in thisgenustheyarefirmlyattached o the dermal bones anddifferfrom tuberclesonly by their outline which mayhave been modified by weathering. In other Acan-thothoraci tesserae are known only in Radotina andKosoraspis,and their occurrence is puzzling. In theRadotin Limestone there is one species of Radotina,R. kosorensis,whichmaybe tesserateornon-tesserate,and one species of Kosoraspis,K. peckai, which alsomay be tesserate or non-tesserate. In the Kon6prusyLimestone there are two species of Radotina, one tes-serate (R. tessellata) and one non-tesserate (R. prima).If one compares the two varieties of R. kosorensis, thecentral plates of non-tesserate individuals have centraloval areas (Fig. 2B, Ce,), presumably representing anearly growth stage, and these appear to correspond tothe entire central plates of tesserate forms (Fig. 2A,Ce). In the next growth stage, tesserae (Fig. 2A, tes)were formed in the latter, while a wide flange of tuber-culated bone (Fig. 2B, Ce,) developed in the former.According to Gross (1958:11), bone does not extend

. , r'I

IN 0 -tes

,.*

( of

41~'

FIGURE1. Stensioella eintzi,headandanterior ody ndorsalview,fromGross 1962b).Abbreviations:e,centralplate;de, denticle; es,tesserae.

underthe tesserae as suggestedby Westoll (1967:91),nor have tesseraebeen fused to the surface of dermalbones. Since similar-sized individuals may be eithertesserate or non-tesserate, t is apparent hat therearetwo differentmanners of growth.Whether or not thisis a specificdifference, t indicates,as Grosssuggested,that within Radotinathe "Bauplan"had not been sta-bilized, and this is supportedby the variationsin thepatternof cranial bones and the lateral ine canals.Aninterestingparallel o the situation n Radotina s foundin the psammosteid Drepanaspis,in which tesseraeappearbetween theplatesonlyaftertheearliestgrowthstages;however, in Psammosteidae the tesseraemaybecomeincorporatednto dermalbonesduringgrowth,and this apparentlydoes not happenin Acanthotho-raci.In the Rhenanida tesserae lie between the cranialroofingbones as in Acanthothoraci,but whethertheywereabsentin the earliestgrowthstages s not known.The most common rhenanid,the EarlyDevonian Ge-muendinastuertzi,is known from specimens rangingJVP 3(2), September 1983 71


5/16

/Ce

tesejMOP0

.k ,

FIGURE 2. Incompletecranialroofs of Acanthothoraci.A-B, Radotinakosorensis, pproximatelyhe samescale,from Gross(1959);A, tesserate orm;B, non-tesserate orm.C, Kimaspis ienshanica, rom Mark-Kurik1973).Abbreviations:Ce, centralplate;Ce,, early growthstageof Ce;Ce2, latergrowth stageof Ce; tes, tesserae.

in total length from 16 to 106 cm, and in absence ofword to the contrary, it must be assumed that thesmallest known individual is tesserate. ComparingGross' (1963, fig. 1) restoration of a 23 cm long indi-vidual with Broili's (1933, fig. 1) drawing of the 106cm long specimen (Fig. 3A), it appears that individualtesserae have enlarged with age, but in most areas notas much proportionately as the whole fish, nor as thesubmarginal and suborbital plates. Numerous tesseraehave been added between the submarginal and sub-orbital plates, but the number between the submarginaland the cranial roof or the gill opening is not signifi-cantly different. However, it seems that tesserae havebeen incorporated into the sides of the cranial roof(Fig. 3A, tel), especially at its postero-lateral corners,and also to the margins of the suborbital and submar-ginal plates. Behind the head tesserae overlie, and areattached to, the anterior lateral and spinal plates, buton the head itself tesserae form a flexible but protectivecovering of the gill chamber and neck joint.The Middle Devonian Asterosteus (Fig. 3B) is knownlargely from the endocranium and its dermal coveringwhich, in contrast to the situation in small and me-dium-sized individuals of Gemuendina studied byGross, consists of well developed bones. The rest ofthe cranial covering is unknown except for isolatedplatelets described mostly as Ohioaspis. These are notthin, superficial structures as in the tesserae of Ge-imuendina, but thick dermal elements with a well de-veloped basal layer, a thin middle layer, and a super-

ficial layer consisting of two or three overgrowngenerations of semidentine tubercles. The series ofplatelets carrying the main lateral line and the infraor-bital canal (Fig. 3B, tef) have become incorporated inthe cranial roof by attachment to a deep flange of bonefrom the lateral dermal bones of the roof. If thin, su-perficial tesserae occur in Asterosteus, they have notbeen recognized or described.The Late Devonian Jagorina (Fig. 3C) has been usedas evidence of a trend to fragmentation of dermal bones(Stensi6, 1963:403; Westoll, 1967:92), but the evi-dence consists largely of the replacement of the sub-orbitals of Gemuendina by tesserae. However, the lossof the suborbitals may be related to the expansion ofthe pectoral fins to the anterior end of the head andthe resulting need for flexibility in this region. Thetesserate borders of the submarginals (Fig. 3C, Sm)and of the cranial roof may not have resulted fromfragmentation but from the incorporation of tesseraeduring growth, as probably happened also in the largestindividual of Gemuendina.To summarize, I agree with Gross (1958, 1962a,1963) that superficial denticles or tesserae (odontodesor odontodia of Orvig, 1977:69-70) were the dermalcovering of ancestralplacoderms, and that dermal bonesarose later and deeper in the dermis. The superficialtesserae were retained in a few, mostly early and prim-itive placoderms, as a protective covering, a flexibleone in Rhenanida and in the cheek of Lunaspis. Thetesserae disappeared as independent structures in most

72 JVP3(2), September 1983


6/16


7/16

A B0 'oc

m0l-- "orb"m p l - - -

t -- p 3pC p Pppp- 0- -Nu

FIGURE 4. A, Stensioella heintzi; B, Pseudopetalichthysproblematica.Redrawn and simplifiedfrom Gross (1962b)to show dermalbones. Abbreviations:Ce, centralplates;Ic,mainlateral-line anal;mpl,middlepitline;Nu, nuchalplate;orb, orbit;Pp, postpineal plate; ppl, posteriorpit line; soc,supraorbital ensoryline.

(Fig. 3B-C, Pp). In Acanthothoraci the postpineal maybe absent, or it may be a large bone carrying the pos-terior ends of the supraorbital canals (Fig. 5A, C, Pp).In Ptyctodontida this bone is also an important ele-ment of the skull roof (Fig. 6D, Pp), is traversed byboth the supraorbital canals and posterior pit lines,and extends posteriorly between the centrals, some-times to the posterior margin of the cranial roof. Thepostpineal is absent, at least as a separate element, inPetalichthyida and Phyllolepida, but is retained in An-tiarcha (Fig. 6F, Pp) and in an occasional primitivearthrodire.Nuchal: The nuchal is absent in Stensioella, but oneor two small median elements at the posterior end ofthe skull roof Pseudopetalichthys (Fig. 4B, Nu) may beprovisionally identified as nuchals. In Rhenanida thenuchal is very small (Fig. 3B, Nu), and in Ptyctodon-tida (Fig. 6D, Nu) it is small or absent. In Acanthotho-raci (Fig. 5, Nu) the nuchal, where known, is moderate-sized or long; it is very long in Petalichthyida (Fig. 6E,Nu), and long and broad in Phyllolepida (Fig. 6C, Nu)where it covers much of the cranial roof. It is an im-portant element in the skull of Antiarcha (Fig. 6F, Nu),while in Arthrodira it shows considerable variation insize and shape (Fig. 6A, Nu). I find no evidence tosupport the view that a long nuchal was a primitiveplacoderm character (Miles, 1973b: 111; Young, 1980:52), nor a primitive arthrodire character (Van Valen,1963:260). In fact, the very long nuchal of some Ac-tinolepidae resulted in part from the fusion of the post-pineal to its anterior end, as suggested for Kujdanow-iaspis by Stensi6 (1954:42), for Baringaspis by Miles(1973b:111), and for Aethaspis by Denison (MS). Thevery large nuchal of Phyllolepis could have resultedfrom fusions or crowding out of adjacent plates, and

in Petalichthyida the shape of the nuchal and the pat-tern of ornamentation in some specimens of Lunaspis(Gross, 1937, figs. 25, 26B, pl. 9, fig. 1) could be in-terpreted as indicating a compound origin for this plate.Rostral Region: This region is poorly known in mostorders of Placodermi. In Stensioellida it appears to lackdermal bones and in Pseudopetalichthyida availablespecimens do not reveal its structure, though bothgroups probably had antero-ventrally placed nostrils.In Rhenanida the nostrils are dorsally placed near themidline, are surrounded by small plates or tesserae,and the rostral region is tesserate, lacking dermal bones.In Ptyctodontida the anterior part of the head is poorlyknown due to the absence of dermal bones and theslight or no ossification of the endocranium; the po-sition of the nostrils is unknown, though Orvig (1960,fig. 4B) restored them in a ventral position just in frontof the mouth, much as in holocephalians. In Acan-

, .-nanaPtn Ro

PiPiPt orb

,'cena---4-Ce -01NuNu Pnp

C Ro ,ncaCe

o a

Ce~a Itt 9NuOb

N NuCee

FIGURE 5. Dermalskullbones of Acanthothoraci,dorsalview. A, Radotinaprima, after Gross (1958) and Westoll(1967). B, Brindabellaspis tensioi, after Young (1980). C,Romundinastellina,afterOrvig(1975). D, Kosoraspispeck-ai, afterGross(1959). Abbreviations:Ce, centralplates;nca,position of nasal capsule;Nu, nuchal plate; orb, orbit; Pi,pineal plate; Pna, anteriorparanuchalplate; Pnp, posteriorparanuchalplate; Pp, postpineal plate;Ptn, postnasal plate;Ro, rostralplate.



8/16

RB n Ro? ,o-orborbro Pi PtoPtoPtbco Mg

CeCe

Nu

CP

M

Nu Pnp np N-

~or,"t M-C

n:: :uP. ..:...... . . :P;... -.:

FIGURE 6. Dorsalviews of cranialroofsof differentordersof Placodermi.A, ArthrodiraCoccosteus uspidatus),afterMilesand Westoll(1968).B, AcanthothoraciRadotinaprima),basedon Gross(1958)and Westoll(1967).C, Phyllolepida Phyllolepisorvini),afterStensi6(1936). D, Ptyctodontida Rhamphodopsishreiplandi), fter Miles(1967b).E, Petalichthyida Wijdeaspisarctica),afterOrvig(1957). F, Antiarcha Bothriolepis anadensis),after Stensi6(1948). Abbreviations: c, centralsensoryline;Ce, central plate; ioc, infraorbitalsensory line; La, lateralplate; Ic, main lateral-linecanal; Mg, marginalplate; no, nasalopening; Nu, nuchal plate; orb, orbit; Pi, pineal plate; Pmng,postmarginal plate; Pn, paranuchal plate; Pna, anterior paranuchalplate; Pnp, posterior paranuchal plate; Pp, postpineal plate; ppl, posterior pit line; Pro, preorbital plate; Pto, postorbital plate;Ro, rostral plate; soc, supraorbital sensory line.

thothoraci the rostral region is covered by a dermalbone commonly identified as a premedian (Fig. 6B,Ro?), and the nostrils typically have moved to a dorsalposition between or just in front of the orbits. Thenostrils are not known in Petalichthyida, but presum-ably were placed anteriorly and ventrally, and wereincompletely roofed by a rostral plate. In Phyllolepida,the typical genus, Phyllolepis, lacks dermal bones inthe rostral region, and the nostrils were presumablyanterior and ventral. Only in Arthrodira is the structureof the anterior part of the head reasonably well known.The rostral region is covered by rostral, pineal, inter-nasal, and paired postnasal dermal bones beneath whichis the endoskeletal rhinocapsular ossification enclosingthe nasal capsules which open ventrally or antero-ven-trally; primitively, this rostral unit is separate, but moreusually it is fused to the rest of the skull. In Antiarchathe nostrils are placed close together dorsally between

the eyes, all lying in an orbital fenestra in the cranialroof; the rostral is a small bone between the orbits, andthe large median bone forming the anterior part of thecranial roof is called a premedian (Fig. 6F, Prm).The main problem in the rostral region involves thepremedian and rostral plates of the Acanthothoraciand Antiarcha. In the Acanthothoraci (Fig. 5) the largeanterior median element was identified as the rostralby Gross (1958:10) in Radotina. Westoll (1967:87) be-lieved that as the nostrils migrated to a dorsal position,the rostral plate maintained its original morphologicalrelations to them, and so identified a "rostral plate"in Radotina immediately behind the nostrils and an-terior to the pineal plate. However, this region is nota plate but tesserae in Radotina according to Orvig(1975:45). The large median plate at the front of theskull was considered to be fused supragnathals by Wes-toll, as fused postnasals by Miles (1971:114), and as

JVP 3(2), September 1983 75


9/16

A ,

Ptn /Ro

P t n n c a

' /...' .

. ...

FIGURE 7. Dorsal views of cranial roofs. A, Bothriolepiscanadensis,fromStensi6(1948). B, Holonemawestolli, romMiles (1971). Abbreviations:nca,probablepositionof nasalcapsule;no, nasalopening;Pi, pineal plate;Ptn, fused post-nasal plates;Ro, rostralplate.

an internasal by Young (1980:54). Orvig (1975:45)identified a rostral plate in Romundina, lying behindand between the nares, and a premedian ("medianprerostral plate") on the snout. The recently describedBrindabellaspis (Young, 1980) throws important lighton the structure and evolution of this part of the skullin Acanthothoraci. In this genus the nasal cavities (Fig.5B, nca) are not dorsal, but occupy the antero-dorsalcorners of the orbits. They lost their primitive antero-ventral openings and presumably opened in the ante-rior corners of the orbits. Though the external openingshave moved, the position of the nasal cavities is stillclose to the presumed primitive position, as exempli-fied by that of primitive Arthrodira. This leads to twoimportant conclusions. First, since Brindabellaspis isin most respects clearly an acanthothoracan, the rel-atively primitive position of its nasal sacs indicatesthat dorsal nares were independently acquired in this

order, and hence are no indication of relationship toRhenanida as suggested by Westoll (1967:89), Ob-ruchev (1967:185), Gross (1963:68), Miles and Young(1977:134-135), and Young (1980:65). Secondly, sincethe nasal sacs have not migrated far, it is unlikely thatthe rostral plate has moved much or at all from itsoriginal anterior position. The dermal bones of thesnout are unknown in this genus, but I suggest that theantero-median plate is the rostral (Fig. 5B, Ro), not apremedian as restored by Young (1980, fig. 16), andthat the wide plate between the orbits is a pineal, nota rostropineal as identified by Young. If the rostral hasnot moved in this genus, it is probable that it hasretained its primitive anterior position in other Acan-thothoraci. Dermal bones do not necessarily move withthe sensory organs they surround, witness the migra-tion of the orbits dorsally between the preorbitals andpostorbitals in Homosteidae (Heintz, 1934, fig. 51). InAcanthothoraci it is possible that the nostrils migratedposteriorly between the rostral and postnasals andeventually came to be bounded posteriorly by the pi-neal plate. Radotina can be interpreted in this fashion(Fig. 5A). In Romundina, Orvig (1975) has identifieda premedian ("median prerostral") and a rostral plate.There can be no question that his "rostral" platebounded the external nasal openings medially and pos-teriorly, but it should be noted that he failed to find apineal plate, and that his "rostral" plate has near theposterior edge of its visceral surface an oval depression(Orvig, 1975, pl. 5, fig. 3) that may well have beenoccupied by the pineal organ. Therefore, it is possiblethat the anterior plate of Romundina is the rostral (Fig.5C, Ro), not a premedian, and that Orvig's "rostral"is the pineal, bounding the nostrils behind.In Antiarcha (Fig. 7A) there is also an anterior me-dian bone called the premedian, but in this order thereare definitely two median bones behind it in the orbitalfenestra, the rostral bounding the nares, and the pinealbehind it. There are no known primitive antiarchs toindicate how this situation arose, and the earliest knowngenus, Yunnanolepis, differs in this area only in thatthe preorbital region and premedian plate are short.Among Arthrodira, the Holonematidae are adaptivelyconvergent to antiarchs and suggest how the rostralregion of the latter may have evolved. In Holonema(Miles, 1971:110-116) the paired postnasal plates (Figs.7B, 8D, Ptn) have enlarged, met, and fused in themidline in front of the rostral plate which has beenreduced in size and crowded posteriorly. The positionof the nasal capsules (Fig. 7B, nca) and external naresis not known in this genus, but both have probablymoved posteriorly, and the external nares may be inthe anterior part of the orbit, as in Brindabellaspis. Itis possible that in Antiarcha the premedian is the en-larged and fused postnasals, and that the rostral hasbeen crowded back and reduced in size as in Holo-nema. The antiarchs differ, however, in that the ros-trals remain separate and retain their original relationsto the external nares.Postnasals: The postnasals are generally minor, in-



10/16

Ro

.....,

A / -- P"

O e ~ N u ,...

IPi

Ptnn

C 'Ro

eOPror

- /t

.."e"'..PPn,oD.......... . t ..O

Ptnt

-Pi r ....--- - .....'....t..n\

FIGURE 8. Postnasalplates in variousarthrodires.A, Bryantolepis p., basedon Field MuseumPF 158, 329, 1542, 1544.B, Wuttagoonaspisfletcheri,romRitchie(1973). C, Anarthraspis p., PrincetonPU 13659. D, Holonema westolli,from Miles(1971). E, Coccosteuscuspidatus, rom Miles and Westoll (1968). Abbreviations:cc, centralsensory line; Ce, central plate;Mg, marginalplate;Nu, nuchalplate; pfc, profundussensoryline; Pi, pineal plate;Pn, paranuchalplate;Pro, preorbitalplate;Pso, postsuborbitalplate;Ptn, postnasal plate;Pto, postorbitalplate;Ro, rostralplate;So, suborbitalplate;soc, supraorbitalsensoryline.

conspicuous elements of the placoderm cranial roof.They are unknown in Pseudopetalichthys and ptycto-donts, and this region is tesserate in Stensioella andRhenanida. They have been identified in two generaof Acanthothoraci, but in Petalichthyida this region isknown only in Lunaspis where it is covered with skinand tesserae. I have identified postnasals in Phyllolepis(Fig. 6C, Ptn), and have suggested above that the pre-median of Antiarcha may be fused postnasals. Theyare best known in Arthrodira (Fig. 8E, Ptn) where theylie lateral to the rostral, bound the external nasal open-ings, and carry the anterior ends of the supraorbitalcanals. Typically they are small elements, occasionallyfused to the rostral or preorbitals, and are sometimesreduced or lost, but they are strongly developed insome genera. In the actinolepid Bryantolepis (Fig. 8A,Ptn) they are much enlarged and form an importantpart of the cranial roof. Heintz (1962:29-30) ques-tioned this, but their extent is clearly indicated by su-tures and pattern of ornament in two Field Museumspecimens (PF 158, 1544). They are also large bonesin Anarthraspis (Fig. 8C, Ptn). In Holonema (Fig. 8D)

they have enlarged to meet and fuse in the midline,displacing posteriorly the small rostral plate. They haveprobably fused with the rostral in Groenlandaspis andArctolepis, where the broad anterior plate has generallybeen identified as the rostral (Heintz, 1962:25-27;Goujet, 1972:6). Finally, in Wuttagoonaspis, Ritchie(1973:64) identified as postnasals long, narrow platesthat form the margin of the cranial roof for half itslength and meet the postsuborbitals posteriorly (Fig.8B, Ptn). The relations of these plates anteriorly to therostral and preorbital plates and to the supraorbitalcanals support this identification, though their positionlateral to the suborbitals and their contact with thepostsuborbitals are unique. Miles and Young (1977:142) identify these lateral plates as suborbitals largelyon their interpretation of the lateral line canals, butthis leaves Wuttagoonaspis without postnasals and withnew bones, "x", which are Ritchie's suborbitals. Ritch-ie's identification of the latter is supported by the factthat they bound the orbits laterally and carry the in-fraorbital canals from the postorbitals to the post-nasals.



11/16

Paranuchals: Westoll (1967:96) divided placodermsinto two groups, largely on their possession of one ortwo pairs of paranuchals, which he believed was cor-related with the possession of a short or long occipitalregion. This has been followed by Miles and Young(1977:134) and Young (1980:69), who have consideredtwo pairs to be a derived condition. Two pairs areclearly present in Acanthothoraci (Figs. 5, 6B, Pna.Pnp) and Petalichthyida (Fig. 6E). As for Rhenanida,Westoll's identification of two pairs in Gemuendinamust be questioned, and in Asterosteus and Jagorina.if paranuchalsare present, they are representedby someof the numerous small plates lateral to the centrals thatcarry the main lateral-line canal. The plates in Pseu-dopetalichthys that Westoll identified as paranuchals,I consider to be centrals. Although it cannot be dem-onstrated at present. I have assumed that two or morepairs of paranuchals were developed in primitive plac-oderms, and that the single pair in Ptyctodontida, Phyl-lolepida, Arthrodira, and Antiarcha is a derived state.If this is so, two pairs of paranuchals cannot demon-strate a relationship between Acanthothoraci and Pe-talichthyida. Finally, the correlation of occipital lengthwith the number of paranuchals is not established, aspointed out by Miles and Young (1977:133).

DERMAL SHOULDER GIRDLERecent theories about the phyletic history of Plac-odermi have been based largely on assumptions aboutthe primitive condition of their dermal shoulder girdle.Gross (1954) once believed that primitive vertebrateswere protected by a trunk armor which was later re-duced to adermal shoulder girdle in many groups. Miles(1967:66) concluded from his functional analysis thatthe ancestral placoderm had a well developed trunkarmor. Miles and Young (1977:130-131) were morespecific and listed the plates they believed to be presentin primitive placoderms: except for the presence ofboth intero-laterals and anterior ventro-laterals, theseare the same as those of the ptyctodonts, a group which,at an earlier date, Miles (in Moy-Thomas and Miles,1971:178) considered to have a greatly reduced trunkarmor. In support of this they present a table showingthe dermal trunk bones developed in each placodermorder. How this indicates the primitive condition isnot specified, but apparently they selected those platespresent in a majority (5) of the nine placoderm orders.I believe, on the contrary, that a trunk shield is a

derived and specialized condition in Placodermi andin other gnathostomes, and that primitive gnatho-stomes had only an endoskeletal shoulder girdle orscapulocoracoid. This may have evolved at the sametime as the pectoral fins which, together with the pelvicfins,were a common heritageof all gnathostome groups;or. conceivably, it may have arisen before the originof the paired fins to enclose the gill region posteriorly,to furnish an anterior attachment for the body muscleswhere they are interrupted by the gill chamber, and toserve as an origin for hypobranchial and branchial

MMddAdiA

Al A- A

Md7Md C Md

Pd' Ad,PI

Al

- ------ -- -- -- --

FIGURE 9. Lateral views of shouldergirdles and trunkshields, with scapulocoracoiddotted. A, Jagorinapandora.simplifiedafter Stensi6 (1959). B, Brindabellaspistensioi.after Young (1980). C, Kujdanowiaspis p., after Stensi6(1969). D. Brachyosteus dietrichi, after Stensi6 (1959). Ab-breviations: Adl, anterior dorso-lateral; AI, anterior lateral;AvI, anterior ventro-lateral; Md. median dorsal; Pdl, pos-teriordorso-lateral,PI, posterior ateral;Pvl, posteriorven-tro-lateral;Sp, spinal.

muscles. This primitive shoulder girdle is retained bythe Chondrichthyes, where it serves all of the abovefunctions, as well as acting as an articulation for thepectoral fin skeleton and as an origin for its muscles.The scapulocoracoid is the primary shoulder girdle ofAcanthodii also, though Climatiida added, I believeas a derived condition, dermal plates that strengthenand protect the shoulder region mostly ventrally. InOsteichthyes and Placodermi dermal sholulder girdleshave evolved probably independently in each group,and have taken over most of the functions of the shoul-der girdle, the scapulocoracoid becoming reduced usu-ally to a small element serving mostly for the attach-ment of the pectoral fin and its muscles (Fig. 9C).However, a primitive condition is retained in a fewPlacodermi. notably in the acanthothoracan Brinda-bellaspis, where there is a high scapular blade (Fig. 9B);a smaller scapular process is preserved in Pseudope-talichthyida (Gross, 1962b, fig. 8A) and in the ptyc-todonts Cienurella (Miles and Young, 1977, fig. 29)and Rhynchodus (Stensio, 1959, fig. 75). The scapu-locoracoids of Rhenanida (Fig. 9A) are large and havebeen restored by Stensid (1969, figs. 179, 195) withhigh scapular blades; clearly the exoskeletal girdle isnot dominant in this order, but much of the shoulderstructure is specialized in relation to the tremendousenlargement of the pectoral fins. The scapulocoracoidsare enlarged, especially in a posterior direction, in many

78 JVP 3(2). September 1983


12/16

B?Avl

ScAl

Md -

AdII Avi

Al--IPvI\Sp Sp

MdPdl AdI MdAmd

(AlG Pd AdIPvl Al

SpSPvi AlFIGURE 10. Dermal shouldergirdles, A-B ventral, C-H lateral views. A, Pseudopetalichthys roblematica,after Gross(1962b), simplified. B, Gemuendina stuertizi, after Gross (1963). C, Rhamphodopsis threiplandi, after Miles (1967b). D,Palaeacanthaspis vasta, after Stensi6 (1944). E, Phy//olepis orvini, after Stensid (1959). F, Lunaspis heroldi, after Stensib (1963).G, Phylactaenium acadicus, after Heintz (1934). H, Pterichthyodes milleri, after Traquair (1914). Abbreviations: Adl, anteriordorso-lateral;Al, anteriorlateral;Amd, anteriormedio-dorsal;Amy, anteriormedio-ventral;Avl, anteriorventro-lateral; I,intero-lateral;Md, mediandorsal; Mxl, mixilateral;Pdl, posteriordorso-lateral;PI, posteriorlateral;Pmd, posteriormedio-dorsal;Pvl, posterior ventro-lateral;Sc, scapulocoracoid;Sp, spinal.

advanced Arthrodira (Fig. 9D), a specialization relatedto the lengthening of the fins.I have suggested (Denison, 1975) that a relativelyslight development of the dermal shoulder girdle isprimitive in Placodermi, and that the stages of its evo-lution within the group were somewhat as follows:1) Primitively the dermal shoulder girdle consistedonly of paired anterior laterals and anterior ventro-laterals or intero-laterals; separate spinals may not havebeen present. This is the probable condition in Sten-sioellida and Pseudopetalichthyida (Fig. 10OA).2) The girdle became braced dorsally by a mediandorsal and paired anterior dorso-laterals; spinal plateswere present. This stage is exemplified by Rhenanida(Fig. O10B)nd Ptyctodontida (Fig. 10C).

3) The evolution of a trunk shield from the shouldergirdle is commenced by the addition of posterior dor-so-laterals in Acanthothoraci (Fig. 10OD)and Petal-ichthyida (Fig. 10OF),nd by the addition of posteriorventro-laterals in Phyllolepida (Fig. 10E) and possiblyin some Petalichthyida (Fig. 11E).4) The trunk shield is completed by the additionlaterally of posterior laterals enclosing pectoral fenes-trae in Arthrodira (Fig. 10G) and Antiarcha (Fig. 10OH).5) Still further lengthening of the trunk shield is ac-complished in Antiarcha by the addition of a posteriormedian dorsal plate (Fig. 1OH).Although the above may represent in a general waythe stages in the evolution of the placoderm trunk shield,certain facts indicate that it was not necessarily a sim-



13/16

A B-? A v i A I

Amiy ?AvI

Av Amym

PmFAPv' PvlFIGURE 11. Dermalshouldergirdle,ventral views. A, Pseudopetalichthysroblematica,after Gross (1962b), simplified.B,Gemuendina stuertzi, after Gross (1963), simplified. C, Rhamphodopsis trispinatus, after Watson (1938). D, Romundina stellina,modified from Orvig (1975). E, Lunaspis broilii, after Gross (1961). F, Phyllolepis orvini, after Stensi6 (1959). G, Aethaspismajor,after Denison (1958). H, Phlyctaeniusacadicus,after Heintz (1934), modified. Abbreviations:Amy, anteriormedio-ventral;Av, antero-ventral;Avl, anteriorventro-lateral;I, intero-lateral;Pmv, posteriormedio-ventral;Pvl, posteriorventro-lateral;Sp, spinal.

ple progression from one stage to another, and surelynot from one order to another. A number of problemsrequire resolution.Anterior Ventro-laterals (AVL) and Intero-laterals(IL): Where both are present, the AVL's are typicallyrelatively broad, flat bones attached to the medial mar-gins of the spinals. The IL's are folded to form twolaminae, one ventral and the other ascending to formpart of the postbranchial wall; they attach to the an-terior ends of the spinals, to the AVL's and to theanterior laterals. When only a single pair of plates ispresent in the ventral shoulder girdle, their identifi-cation is difficult. Thus in Ptyctodontida (Figs. 10C,11C), Stensi6 (1959:208-210) assumed that both platesare present but fused, the individual components beingidentified by their topographic position. This has beenfollowed by Orvig (1960:238), but in the absence ofevidence of fusion, I do not accept the presence of morethan a single pair. Watson (1938:403) called them IL's,while Gross (1933:60), Miles (1967b:108), and Milesand Young (1977:151, footnote) identified them as

AVL's. In all ptyctodonts there is a well developedascending lamina, as well as a ventral lamina, whichsupports their identification as IL's.In Rhenanida (Figs. 10B, 11B) the identification ismore difficult. The ventral plates are broad and flat,meeting in a long median suture and attached laterallyto the spinals. In these respects they agree well withAVL's, with which they have been homologized byGross (1963:58) and Stensi6 (1959:178, 182), thoughthe latter considered the scales on their anterior bordersin Jagorina to represent part of the IL's. However,Watson (1937:139) identified the ventral plates as IL's,probably because of their connection with the laminadorsoventralis which extends from them to the anteriordorso-laterals or anterior laterals, and bounds the bran-chial chamber posteriorly and medially. It thus cor-responds to the postbranchial wall of other placodermsin function, if not in position. Whether it is a specialformation of Rhenanida, as Gross believed, or a dis-placed and modified postbranchial wall, as apparentlyWatson thought, is not possible to determine; thus the



14/16

homology of the ventral plates is uncertain in this or-der.In Stensioellida and Pseudopetalichthyida (Figs. 10A,11A) there is no evidence of an ascending lamina onthe ventral shoulder girdle, though because of poorpreservation its presence cannot be excluded. The ho-mology of the plates is thus uncertain.In Acanthothoraci both AVL's and IL's are presentin Romundina (Orvig, 1975:59-61). The IL's have welldeveloped postbranchial laminae, but the AVL's aresmall and widely separated from each other and themidline (Fig. 11D). In the other placoderm orders (Fig.1IE-H) both IL's and AVL's are present, the latterbeing broad and either meeting in the midline (Petal-ichthyida, Phyllolepida, Antiarcha), or separated byanterior and posterior medio-ventrals (Arthrodira). InAntiarcha the IL's are known only in the primitiveYunnanolepis (Zhang Mi-man, 1980, fig. 4), while inother genera they are lost or fused to the AVL's.Antero-ventrals (AV) and Anterior Medio-ventral(AMV) Plates: These are absent in Stensioellida, Pseu-dopetalichthyida, and Rhenanida (Fig. 11A-B), a con-dition I consider to be primitive. An AMV is presentin Ptyctodontida (Fig. II1C, Amv), lying between theIL's, and Orvig (1975:62) assumed the presence of AV'sfused to the anterio-medial corners of the IL's, thoughthere is no indication of them. In Acanthothoraci, Or-vig (1975:61) identified AV's bounded by the IL's andAVL's of Romundina (Fig. 11D). He did not find anAMV, and assumed it was absent and that the rightAV overlapped the left one in the midline. However,his material did not include a left AV; it is possible torestore the ventral shield so that its anterior border ismore nearly transverse as in other placoderms, leavingspace between the AV's for an AMV (Fig. 11D, Amy).The petalichthyid Lunaspis broilii has a T-shapedAMV whose lateral wings are identified by Orvigas fused AV's, and whose slender posterior stalk as theAMV proper (Fig. IIE, Amy). Only in the arthrodirefamily Actinolepidae have separate AV's and AMVbeen found (Fig. IIlG, Av, Amv), and as I will showelsewhere, it is probable that the AV's were lost inmore advanced arthrodires (Fig. 11H) by fusion withthe AMV, IL's, with each other, or by failure to ossify.In Antiarcha, which presumably evolved from prim-itive Arthrodira, the AV's are preserved as pairedsemilunars in Asterolepidae and Yunnanolepidae, whilethe AMV is retained as a median semilunar in Both-riolepidae and Gerdalepis.Spinal Plates: These were considered "the best plac-oderm trademark" by Romer (1968:29), but this is notquite correct because they are probably absent prim-itively in Stensioellida and Pseudopetalichthyida, andare lost secondarily in advanced Arthrodira and inmost Antiarcha. In the latter the spinal is recognizableonly in the Early Devonian Yunnanolepis (Zhang Mi-man, 1980:188), where it forms a small lateral projec-tion from the anterior edge of the AVL, and forms thelateral boundary of a pectoral fossa, fundamentally as

in primitive Arthrodira; in other Antiarcha this regionis much modified and no identifiable trace of the spinalremains. In Arthrodira the spinal is absent in a numberof families, a condition considered primitive by Sten-si6 (1959), and used by him to define his "super orderAspinothoracidi," but it is generally agreed today thatthe loss of the spinal is secondary in Arthrodira.Posterior Ventro-lateral Plates (PVL): The additionof PVL's was an important step in the formation of atrunk shield from a shoulder girdle, and they take ad-ditional significance with their selection by Miles andYoung (1977:136) as a synapomorphy uniting Phyl-lolepida, Anthrodira, and Antiarcha in the superorderDolichothoracamorpha. However, Phyllolepida arecharacterized not only by the presence of PVL's butalso by the absence of posterior dorso-laterals. Thiscombination of characters could be the result of a num-ber of evolutionary histories, which are not discussedsince there is at present no sound basis for determiningwhich is correct. Moreover, the validity of the Doli-chothoracamorpha must be questioned for two rea-sons. First, it is based on a single character and needsto be supported by other synapomorphies. Secondly,PVL's apparently are not restricted to the three ordersincluded. They have been identified in Rhenanida byStensi6 (1969, figs. 195, 196), though this identificationis questionable as Gross (1963:58) did not find themin Gemnuendina.Orvig (1975:61) suggested their pres-ence in Acanthothoraci on the evidence of possibleoverlap areas on the posterior edges of the AVL's;however, these may have been for skin or scales (Young,1980:48). In the Petalichthyida, a single specimen ofLunaspis broilii has what appears to be two pairs ofPVL's (Fig. IlIE, Pv/). Young (1980:63) showed howthese could be interpreted as a single pair, and Milesand Young (1977:130) considered them to be enlargedtrunk scales, though the distinction seems to be largelysemantic. However, PVL's are unknown in other pe-talichthyids, and are absent in another specimen of thesame species that is ventrally exposed and preservesthe scales of the body and tail (Gross, 1961, fig. 10).Thus PVL's are not characteristic of the order or evenof the species, but their presence in one specimen showsthat the potentiality for their development exists with-in the Petalichthyida.

CONCLUSIONSAccording to my hypothesis, ancestral placodermshad denticles or tesserae that were retained in someDevonian orders and lost in others. Retention of the

primitive state is no evidence for relationship betweenRhenanida, Acanthothoraci, Stensioellida, or Petalich-thyida as believed by Gross (1963:68-70), Obruchev(1967:185), and Miles and Young (1977:144).Dermal bones probably appeared in Placodermi inthe Late Silurian and acquired a definite and stablepattern gradually. Early stages of this acquisition areretained in Stensioellida, Pseudopetalichthyida, Rhe-



15/16

nanida, and Acanthothoraci. Primitively there weremore numerous, smaller bones, for example two ormore pairs of centrals, a series of small paranuchals,a number of postpineals. Thus the presence of two ormore pairs of paranuchals is primitive rather than evi-dence for a relationship between Petalichthyida, Acan-thothoraci, Rhenanida, and Pseudopetalichthyida asbelieved by Westoll (1967:96) and Miles and Young(1977:134, 144).The placoderm dermal cranial-bone pattern wasprobably acquired by a common ancestor of Acan-thothoraci, Ptyctodontida, Petalichthyida, Phyllole-pida, Arthrodira, and Antiarcha, and included mediannuchal, postpineal, and pineal, as well as paired para-nuchals, centrals, marginals, postorbitals, and preor-bitals. If a rostral was present in the ancestral form, itwas secondarily lost in Ptyctodontida and Phyllole-pida. Postnasals are known only in Arthrodira, An-tiarcha, and probably Acanthothoraci, and may alsooccur in Phyllolepida. Differences in the pattern ofthese dermal bones are commonly used to distinguishthese orders, though there is still disagreement abouthomologies of some bones.It is probable that the nasal openings were primi-tively anterior and ventral in Placodermi, and thattheir dorsal position in Rhenanida, Antiarcha, and mostAcanthothoraci is a derived condition attained inde-pendently in the three orders. In Antiarcha the rostralplate migrated posteriorly with the nasal openings, andit is probable that the postnasals enlarged and fused inthe midline in front of them. In most Acanthothoracithe nasal openings are dorsally placed near the midline,but in Brindabellaspis the nasal capsules have notmoved far from their primitive postion and the rostralplate remains at the front of the skull. This indicatesthat the dorsal position of the nasal openings was in-dependently acquired in this order and does not con-stitute a proof of relationship to Rhenanida as claimedby Gross (1963:68-70), Westoll (1967:89-90), andMiles and Young (1977:135).I believe that scapulocoracoids formed the ancestralshoulder girdle, and that dermal plates were acquiredgradually, partly replacing the endoskeletal girdle. Firstto develop were anterior laterals and either intero-lat-erals or anterior ventro-laterals. I do not accept thatprimitive placoderms had MD, ADL, AL, IL, SP, AVL,and AMV as proposed by Miles and Young (1977:131).Their hypothesis requires the loss of MD, ADL, AMV,either IL or AVL, and possibly SP in Stensioellida andPseudopetalichthyida, the loss of AMV and IL or AVLin Rhenanida, and the loss of AVL in Ptyctodontida.I have not discussed the relationships of Ptyctodon-tida and plan to consider this at another time. How-ever, I cannot accept that pelvic and prepelvic clasperswere primitively present in placoderms, retained inptyctodonts, and lost in all other orders, as proposedby Miles and Young (1977:137-138).My analysis of currently available facts does not sup-port a close relationship between any of the orders ofPlacodermi except Arthrodira and Antiarcha.

REFERENCES

Broili,F. 1933. WeitereFischresteausdenHunsrtickschie-fern. Sitzungsberichteder mathematisch-naturwissen-schaftliche Abteiling der BayerischenAkademie derWissenschaftenzu Miinchen 1933:269-313.Denison, R. H. 1975. Evolutionand classificationofplac-oderm fishes. HarvardMuseum of ComparativeZool-ogy, Breviora432:1-24.1978. Placodermi. In Schultze,H.-P. (ed.), Hand-book of Paleoichthyology2:vi + 128 pp.Forey, P. L. 1980. Latimeria,a paradoxical ish. Proceed-ings of the Royal Societyof London(B) 208:369-384.Goujet,D. 1972. Nouvelles observationssurlajoue d'Arc-tolepis(Eastman) t d'autresDolichothoraci.AnnalesdePalkontologie,Vertebr6s58:3-11.Gross, W. 1933. Die Wirbeltieredes rheinischenDevon.Abhandlungender Preussischen eologischenLandesan-stalt (N.F.) 154:1-83.1937. Die Wirbeltieredes rheinischenDevons. TeilII. Abhandlungender PreussischengeologischenLan-desanstalt(N.F.) 176:1-83.1954. ZurPhylogeniedes Schultergiirtels.Paldion-tologischeZeitschrift28:20-40.1958. Uber die tiltesteArthrodiren-Gattung.No-tizblatt des Hessischen Landesamtes fir Bondenfor-schungzu Wiesbaden86:7-30.1961. Lunaspis broilii und Lunaspis heroldi ausdem HunsrtickschieferUnterdevon,Rheinland).Notiz-blatt des Hessischen LandesamtesfUirBodenforschungzu Wiesbaden89:17-43.1962a. Peut-onhomologuer es os des Arthrodireset des T6l6ostomes?Colloques nternationauxdu CentreNational de la RechercheScientifiqueno. 104:69-74.1962b. Neuuntersuchungder Stensioellida (Ar-throdira,Unterdevon).Notizblattdes HessischenLan-desamtesftirBodenforschung u Wiesbaden90:48-86.1963. Gemuendina stuertizi Traquair. Neuunter-suchung. Notizblatt des Hessischen LandesamtesfUirBodenforschung u Wiesbaden91:36-73.1965. Ubereinen neuenSchdidelrest on Stensioel-la heintzi und Schuppen von Machaeracanthus sp. indet.aus dem Hunsrtickschiefer.Notizblatt des HessischenLandesamtesfuirBodenforschungzu Wiesbaden93:7-18.Heintz,A. 1934. Revision of the EstonianArthrodira.PartI. FamilyHomostiidaeJaekel.Arkivftirdie NaturkundeEstlands,ser. 1, 10, Lief 4:1-114.1962. New investigationon the structureof Arc-tolepisfromthe Devonian of Spitsbergen.Norsk Polar-institutt, Arbok 1961:23-40.Mark-Kurik,E. 1973. Kimaspis,a new palaeacanthaspidfrom the Early Devonian of central Asia. Eesti NSVTeaduste Akadeemia Toimetised, 22, Koide Keemia *Geoloogia, 1973, nr. 4:322-330.Miles, R. S. 1967a. The cervical joint and some aspects ofthe origin of the Placodermi. Colloques internationauxdu Centre National de la Recherche Scientifique no. 163:49-7 1.1967b. Observations on the ptyctodont fish, Rham-phodopsis Watson. Journal of the Linnean Society ofLondon (Zoology) 47:99-120.1971. The Holonematidae (placoderm fishes), areview based on new specimens of Holonema from theUpper Devonian of Western Australia. Philosophical



16/16

Transactions of the Royal Society of London (B) 263:101-234.1973a. Articulated acanthodian fishes from the OldRed Sandstone of England, with a review of the structureand evolution of the acanthodian shoulder girdle. Bul-letin of the British Museum (Natural History), Geology24:113-213.1973b. An actinolepid arthrodire from the LowerDevonian Peel Sound Formation, Prince of Wales Is-land. Palaeontographica (A) 143:109-118.& Young, G. C. 1977. Placoderm interrelationshipsreconsidered in the light of new ptyctodonts from Gogo,Western Australia. Linnean Society Symposium Series4:123-198.

Moy-Thomas, J. A. & Miles, R. S. 1971. Palaeozoic Fishes.W. B. Saunders Company, Philadelphia, Toronto, xi +259 pp.Nelson, G. J. 1970. Pharyngeal denticles (placoid scales)of sharks, with notes on the dermal skeleton of verte-brates. American Museum Novitates no. 2415:1-26.Obruchev, D. V. 1967. Class Placodermi. In Fundamentalsof Paleontology, Volume 11, Agnatha, Pisces. Translat-ed from Russian, Israel Program for Scientific Trans-lations, Jerusalem, pp. 168-259.Orvig, T. 1960. New finds of acanthodians, arthrodires,crossopterygians, ganoids and dipnoans in the upperMiddle Devonian calcareous flags (oberer Plattenkalk)of the Bergisch Gladbach-Paffrath trough. Paldiontolo-gische Zeitschrift 34:295-335.1968. The dermal skeleton: general considerations.In Orvig, T. (ed.), Current Problems of Lower VertebratePhylogeny. Proceedings of the Fourth Nobel Sympo-sium, Stockholm, 1967:373-397.1975. Description, with special reference to thedermal skeleton, of a new radotinid arthrodire from theGedinnian of Arctic Canada. Colloques internationauxdu Centre National de la Recherche Scientifique no. 218:41-71.1977. A survey of odontodes ('dermal teeth') from

developmental, structural, functional, and phyletic pointsof view. Linnean Society Symposium Series 4:53-75.Pan Jiang 1981. Devonian antiarch biostratigraphyof China.Geological Magazine 118:69-75.Ritchie, A. 1973. Wuttagoonaspis gen. nov., an unusualarthrodire from the Devonian of western New SouthWales, Australia. Palaeontographica (A) 143:58-72.

Romer, A. S. 1963. The "ancient history" of bone. Annalsof the New York Academy of Sciences 109:168-176.1968. Notes and Comments on Vertebrate Paleon-tology. University of Chicago Press, Chicago and Lon-don, viii + 304 pp.Schaeffer, B. 1975. Comments on the origin and basic ra-diation of the gnathostome fishes with particular refer-ence to the feeding mechanism. Colloques internation-aux du Centre National de la Recherche Scientifique no.218:101-109.Stensi6, E. A. 1925. On the head of the macropetalichthy-ids. Field Museum of Natural History, Geological Series4(4):197 pp.1945. On the heads of certain arthrodires. II. Onthe cranium and cervical joint of the Dolichothoraci(Acanthaspida). Kungliga Svenska Vetenskapsakade-miens Handlingar (3) 22(1):70 pp.1959. On the pectoral fin and shoulder girdle ofthe arthrodires. Kungliga Svenska Vetenskapsakade-miens Handlingar (4) 8(1):229 pp.1963. Anatomical studies on the arthrodiran head.Part I. Kungliga Svenska Vetenskapsakademiens Hand-lingar (4) 9(2):419 pp.1969. Elasmobranchiomorphi. Placodermata. Ar-throdires. In Piveteau, J. (ed.), Trait6 de Paleontologie,4(2):71-692.Van Valen, L. 1963. The head shield of Macropetalichthys(Arthrodira). Journal of Paleontology 37(1):257-263.Watson, D. M. S. 1937. The acanthodian fishes. Philo-sophical Transactions of the Royal Society of London(B) 228:49-146.1938. On Rhamphodopsis, a ptyctodont from theMiddle Old Red Sandstone of Scotland. Transactions ofthe Royal Society of Edinburgh 59:397-410.Westoll, T. S. 1967. Radotina and other tesserate fishes.Journal of the Linnean Society of London (Zoology) 47:83-98.

Young, G. C. 1980. A new Early Devonian placoderm fromNew South Wales, Australia, with a discussion of plac-oderm phylogeny. Palaeontographica (A) 167:10-76.Zangerl, R. 1981. Chondrichthyes I. Paleozoic Elasmo-branchii. In Schultze, H.-P. (ed.), Handbook of Pa-leoichthyology 3A:ii + 115 pp.Zhang Mi-man 1980. Preliminary note on a Lower De-vonian antiarch from Yunnan, China. VertebrataPalAsiatica 18(3):179-190.


Further Consideration of Placoderm Evolution (1983)

Documents