Editions scientifiques du Muséum, Paris

BULLETIN DU MUSÉUM NATIONAL D’HISTOIRE NATURELLE

57, rue Cuvier, 75005 Paris

Section C : sciences de la terre

Rédacteur en chef ; Hervé Lelièvre

Secrétariat de rédaction : Florence Kerdoncuff

S'adresser

pour les échanges :

au Service des périodiques et des échanges de la Bibliothèque Centrale

du Muséum national d’Histoire naturelle

38, rue Geoffroy Saint-Hilaire, 75005 Paris

Tél. : 133] (1)40 79 36 41

pour les abonnements et achats au numéro :

aux Éditions Scientifiques du Muséum, Paris - Diffusion

57, rue Cuvier, 75005 Paris

Tél.: [33] (1)40 79 37 00

Fax : [33] (1)40 79 38 40

e-mail : dhenay@mnhn.fr

pour l’envoi de manuscrits :

aux Éditions Scientifiques du Muséum, Paris - Rédaction du Bulletin

57, rue Cuvier. 75005 Paris

Tél.: [33] (1)40 79 34 38

Fax: [33] (1)40 79 38 58

e-mail: bulletin@mnhn.fr

Abonnements pour l’année 1996 (prix HT) :

Abonnement général : 1800 FF

Section A : zoologie, biologie et écologie animales : 800 FF

Section B : botanique Adansonia : 500 FF

Section C : sciences de la Terre, paléontologie, géologie et minéralogie : 600 FF.

Numéro d'inscription à la Commission paritaire des publications et agences de presses : 1405 AD

BULLETIN DU MUSÉUM NATIONAL D'HISTOIRE NATURELLE, Paris

4' sér., 18, 1996, n°* 2-3, Section C, (Sciences de la Terre : Paléontologie, Minéralogie, Géologie)

SOMMAIRE

Francis Amédro, Georges Busson & Annie Cornée. — Révision des ammonites du

Cénomanien supérieur et du Turonien inférieur du Tinrhert (Sahara algérien) : im¬

plications biostratigraphiques . 179

Min Zhu. — The phylogeny of the Antiarcha (Placodermi, Pisces), with the description

of Early Devonian antiarchs from Qujing, Yunnan, China . 233

Dale A. RUSSEL. — Isolated dinosaur bones from tbe Middle Cretaceous of the Tafilalt,

Morocco. 349

Hélène David, Yannicke Dauphin, Martin Pickford & Brigitte Senut. — Conservation

de sucres dans les phases organiques d’os de bovidés fossiles . 403

Leandro O. SALLES. — Rooting ungulates within placental mammals: Late Cretaceous/

Paleocene fossil record and upper molar morphological trends. 417

Martin PiCKFORD. — Earth Expansion, Plate Tectonics and Gaia’s Puise . 451

Christian Marchal. — Earth’s polar displacements of large amplitude; a possible mech-

anism. 517

Bulletin du Muséum national d'Histoire naturelle, Paris, 4® sér., 18, 1996

Section C, 2-3 : 179-232

Révision des ammonites du Cénomanien supérieur et

du Türonien inférieur du Tinrhert

(Sahara algérien) : implications biostratigraphiques

par Francis AMÉDRO, Georges BUSSON et Annie CORNÉE

Résumé. — Le Tinrhert central et oriental, région du Sahara algérien située entre le Grand Erg oriental

au nord et le bassin d'Illizi au sud, fournit une succession du Cénomanien supérieur-Turonien inférieur particu¬

lièrement précieuse étant donné son isolement; les affleurements des régions plissées du Maghreb sont situées

près de 1 000 kni plus au noid et les ptemiers affleurcmcntç au sud du Massif central saharien (Niger, Nigeria)

à une distance du même ordre. En outre, la .série comparée à celle des plateaux qui prolongent vers l’ouest le

Tinrhert (Tademaït en particulier) est retnarquablenieni riche et se prêle donc à une zonation excellente. Enfin,

cette série qui a échappé à la doloinitisuuuri fournit des données pour dater les calcaires mas,sifs du Cénomano-

Turonien connus tant au Sahara algérien (a l'extrême est du Tinrhert. sut le Haut d’EI Biod. sur le Mzab), en

Libye (Nord-Tripoliiaine. hamada El Homni.) et sur une grande partie du Maghreb, de Textreme ,sud tuni.sien

(Dahar) jusqu’au littoral atlantique. La récolte de plu.s de 200 ammonites déterminables spécifiquement dans ces

formations cénomano-turoniennes du Tinrhert fSahara algérien) sert de base à une étude systématique et il la

révision de.s travaux de Cch.i.ign'on (1957- 1965). La mise eu synonymie d'un certain nombre de uixons décrits

par Coi.MGNON conduit à une réduction sensible du nombre des espèces qui pas.se de 3*1 a 15. La localisation

d’une grande partie du matériel sur de.s coupes métrées permet de distinguer six intervalles biostratigraphiques

successifs, soit du bas vers le haut ; I ) Neolohiiea yihrayeaniis (d'Orblgny. 1841) et CimningionwetDS linrhciiense

(Collignon. 1965); 2) N. vihrayeunuK, Farbakerus sp.. Ca/yeveems iCalycocenis) nmini/arc (.Mantell. 1822)

et Eticatycocerax penUimmum (Jukes-Brmvné. 1896); 3 ) NiK«ncerax nadmi (Chudeau, 1909); 4) Vasçoeerns go-

nuü (Choffat, 1898) et V. cainini iChudeau, 1909) a.ssociés vers le bas à M gadeni et au sommet à Pxeiidaxpi-

doceras giei'ul (Cnllignon, 1965), Eikaiies xuhtubenidatux fCollignon. 1965) et F. Utffirei (Collignon, 1965) ; 5)

Pxeudotixxoïia nigeriensii (Woods, 19)1); 6) P. mg^riensis, Choffcnkeràs gr. (pioaxi (Peron, 1904) -pavUlkri

(Pervinquière, 1907) et Choffatkerax sp. En lenani compte de la disparition de V. cauvini au sommet du 4' in¬

tervalle et de l apparitioii de P. ti'igeriensi.t à la ha,se du .5'', la limite Cénomanien Türonien esl placée entre les

intervalles 4 et 5. Ceci revient à remonter sentiihlcnienl la pqsition de la limite Cénoinaiiien-Turomen au Tinrhert

par rapport aux interprétations antérieures de Collignon (1957, 1965) et Busson (1965, 1972) où toutes les

faunes à Nigerkeras et les faunes plus récentes étaient attribuées au Türonien. Ces divisions nouvelles sont

particulièrement précieuses pour assurer des corrélations tant avec les séries du Maghreb qu’avec les séries du

Niger et du Nigeria.

Mot.s-clés. — Tinrhert, Sahara algérien. Cénomanien supérieur, Türonien inférieur. Ammonites, nouveaux

taxons, biostraligraphic.

Révision of the upper Cenonianian-Lower Tùronian ammonites of

Tinrhert (Algerian Sahara); biosiratigraphic implications

Abstract. — In central and eastern Tinrhert of the Algerian Sahai'a, located between ihe “Grand Erg Oriental”

in the north and the lllizi basin in lhe soiilh. an Llpper Cenomanian-Lower Turonian succession is exposed. This

occurrence is of parlicular inierest because of ils remolene.ss to other oulcrops of the saine âge: the outerops

from lhe Maugrabin folded areas in the north are about KKHI km away; the distance from the nearest outerops

of the Saharan Central Massif (Niger, Nigeria) in the souih is the same magnitude, [n comparison with the

plateaus wliich exiend from the Tinrhert westwards (particularly lhe Tademait), lhe sequence is rich in ammonites

and suitable for dctailcd zonation. Furthcrmorc. the rocks are not dolomitized and thus provide ineans of daiing

the extensive Cenomanian-Turonian limeslones from [lie Algerian .Sahara (in the easlerninost Tinrhert, High of

— 180

El Biod, Mzab), Libya (norihern Tripolitania, Hamada El Homra), and a large part ol the Magrab, ihe extreme

South of Tunisia (Dahar) lo the Atlantic coast. From this Cenomanian-Turonian sequence in Tinrhert more than

200 ammonites hâve heen collected, which are determinable to species. The material has permitted a systematic

study and révision of the work of Colugnon (1957, 1965). A number uf taxa descnbed by Collignon hâve been

placed in synonymy and the number of species has been reduced from 36 to 15 The positioning of a large part

of the material within measurcd sections allows the récognition of six successive biostratigraphie intervaks: interval

1 containing Neolohites vibrayeanux (d'Orbigny, 1841) and Cunningtoniceras tinrhenense (Collignon, 1969); in¬

terval 2 containing N. vihrayeanus, Forhesiceras .sp., Calycoceras (Calyioceras) mniculare (Mantell, 1822) and

Eucalycaceras pentagnnum (Jukes-Browne, 1896); intcrval 3 containing Nigericeras gadeni (Chudeau, 1909);

interval 4 containing Vascocents gamai (Choffat. 1898) and V, cauvini (Chudeau, 1909) associuted to N. gadeni

at the bottom and to Fseiidaspidticeras grecoi (Collignon, 1965), Fikaites subtuberculalus (Collignon, 1965) and

F. lajfilei (Collignon, 1965) at the top; interval 5 corttaining Pxeudntixsotk nigont’iisis (Woods, 1911); interval 6

containing F. nigeriensis. Choffmiceras gr. ipitiasi (Peron, 1904) -pavillieri (Pervinquiêre, 1907) and Choffaticeras

sp. On account of the disappearance uf V! cauvini al the top of intcrval 4 and the appearance of P. nigeriensis

at the base of interval 5, the Cenomanian-Turonian boundary is placed between intervals 4 and 5. This means

thaï the Cenomanian-Turonian boundary in the Tinrhen is now placed considerably higher than in previous studies

(Collignon, 1957, 1964; Busson, 1965, 1972), where the beds with Nigericeras and above were ascribed to the

Turonian. These new subdivisions allow corrélations with both the Maugrabin and the Niger-Nigeria successions.

Key-words. — Tinrhert, Algerian Sahara, Upper Cenomanian, Lower Turonian. ammonites, new taxon,

biostratigraphy.

F. Amédro. 26. rue de Notlingham. 62100 Calais.

G. Busson & A. Cornée, Muséum fuilionat d'Histoire nantreile. Laboratoire de Géologie, 43, rue de Buffon, F-7523I Paris Cedex 05.

INTRODUCTION (G. B. & A. C.)

L'intérêt de la série du Tinrhert oriental

Dans le Nord-Ouest africain, les dépôts marins du Cénomanien-Turonien correspondent à

la plus vaste transgression de Tliistoire post-carbonifère. Ces dépôts marins, particulièrement

étendus à partir du Cénomanien supérieur, se présentent souvent sous la forme de carbonates

massifs, d’une épaisseur de l'ordre d’une centaine de mètres, fréquemment dolomitisés et, de

ce fait, peu ou pas fossilifères. À peu près isopiques et isopaques, ils prennent en écharpe le

domaine nord-saharien et le domaine atlasiquc, de la hamada el Homra (Libye) jusqu’à l’Atlan¬

tique (Busson 1972. Fig. XI1I.2). Ces carbonates massifs, attribués globalement au Cénoma¬

nien-Turonien, occupent ain.si — pour s’en tenir au domaine nord-saharien - le Dahar tunisien,

le Tinrhert extrême-oriental, le Mzab (Fig. 1).

Au sud de 3]'’N, une intercalation manieuse se développe, pouvant atteindre une épaisseur

de 65 m. aux dépens de la partie moyenne de cette barre carbonatée, le total représentant toujours

une centaine de mètres d’épaisseur.

Dans cette «province argileuse méridionale» (Busson 1972), la série cénomano-turonienne

est alors occupée par une triade comprenant :

1) les calcaires inferieurs (c2-tl de la carte Fort-Flatters (BussON 1964));

2) des marnes médianes (t2);

3) des calcaires supérieurs (t3) (Fig. 2).

Or les «calcaires inférieurs», de même que les quelques bancs calcaires intercalés dans les

«marnes médianes» ont été remarquablement préservés de la dolomitisation et sont souvent ri-

— 181 —

Rg. 1. — Situation géographique du Tinrhert dans le nord de l’Afrique et localisation du secteur étudié.

chement fossilifères. De telles séries offrent donc une opportunité sérieuse et rare dans tout le

Nord-Ouest africain de dater plus précisément cet ensemble cénomano-turonien. Une telle datation

e.st d’autant plus précieuse que Tunilé sous-jacente aux «calcaires inférieurs» consiste, dans la

plus grande partie du domaine considéré, en argiles à gypse pratiquement azoïques (BU-SSON

1970) et que l’unité sus-jacente - souvent attribuée dubitativement au Sénonien - revêt le même

faciès et n’a jamais pu être datée.

Si l’on ajoute à ces faits que le Nord-Est du Sahara algérien étant occupé par les atterris¬

sements du Bas-Sahara et du Grand Erg oriental est dépourvu d’affleurements crétacés, la série

de la province argileuse mériodionale (Tademaït et Tinrhert) est particulièrement importante. Ces

— 182 —

Fig. 2. — a, cadre géographique, b, coupe géologique du secteur éuidié dans la hamada de rinrhert et locah.sation de.s gisements

d’ammonites dans les régions de Témassinine. c, Fort-Flatters et d, d’Ohanet.

— 183 —

affleurements en effet sont isolés, distants de près de mille kilomètres au sud des affleurements

plissés et fossilifères des régions atlasiques et à une distance du même ordre de grandeur des

affleurements du sud des massifs centraux sahariens (Niger et Nigeria). Cet isolement même en

fait des jalons particulièrement précieux.

Or, dans cette province, les affleurements du Tinrhert oriental offrent des séries relativement

épaisses comparées à celles du Tademaït et particulièrement fossilifères.

Les ammonites

La faune ici révisée a été récoltée par l’un de nous (G. B.) entre 1957 et 1965. Ces ammonites

ont déjà été décrites en détail sur le plan systématique par COLLIGNON en 1957, puis en 1965.

Trente-quatre espèces, réparties en dix-neuf genres, ont été citées ou figurées par cet auteur,

dont dix étaient nouvelles. Ces ammonites récoltées sur des coupes métrées (voir annexe 1) nous

ont paru mériter une révision systématique et biostratigraphique détaillée.

En effet, il y a une vingtaine d’années, l'ensemble des faunes à Nigericeras et à Vascoceras

était attribué au Turomen. Aujourd'hui, une part substantielle des Vascoceratidae est datée du

Cénomanien supérieur (BERTttoU et al. 1985; KENNEDY & COBBAN 1991). D’autre part, le

concept de l’espèce en paléontologie a considérablement évolué, passant d’une vision typologi¬

que, avec comme résultat un fractionnement des espèces, à une interprétation plus biologique,

acceptant un large spectre de variations morphologiques,

Les objectifs de ce travail sont d'abord d’ordre systématique. Ils incluent, outre de préciser

la répartition de ces faunes d’ammonites au sein des formations cénomano-turoniennes du Tinrhert

central et oriental, d’y établir des divisions biostratigraphiques fines. Grâce à elles, la limite

Cénomanien-Turonien se trouve placée avec une certaine précision.

HISTORIQUE DE TRAVAUX ANTÉRIEURS (G. B. & A. C.)

Nous ne ferons que citer quelques noms parmi les premiers explorateurs (CORTIER, Lavaudan,

Roche, Rolland, Basse & Lapparent) dont la mémoire, les travaux et les résultats - très

méritoires même quand ils n’étaient que ponctuels - ont récemment été fort bien évoqués par

Lefranc (1976). Le premier travail systématique a été celui de la mission BRP-Tinrhert 1955-

1956 (Rumeau et al. 1957). Ces auteurs donnaient, pour la première fois, une carte géologique

au 1/200000 d’une bonne partie de la hamada de Tinrhert. Pour ce qui est du Cénomano-Turonien,

ils reconnatssaient la triade lithologique caractéristique (calcaires inférieurs, marnes médianes et

calcaires supérieurs) dont ils attribuaient les termes respectivement au Turonien inférieur, au

Turonien moyen et au Turonien supérieur. Ils recueillaient des fossiles en particulier dans le

Cénomano-Turonien (c/. ci-dessous, CoLLiGNON 1957) et distinguaient cartographiquement le

Cénomanien supérieur (calcaires bruns dolomitiques, massifs) du bord du plateau, du Turonien

(calcaires blancs, crayeux, tendres, fossilifères) du bas de la cuesta turonienne. Il s’est avéré

que cette distinction ne pouvait être retenue car il ne s'agit là que de deux modes d’altération

dépendant des conditions d’affleurement. Les calcaires du bord du plateau affleurant depuis long¬

temps ont été soumis à de fortes actions tardi-diagénétiques; les affleurements apparaissant dans

— 184 —

les ravinements du plateau ou au pied de la falaise argileuse sont beaucoup plus frais et offrent

le faciès crayeux. Mais des couches représentées à la partie supérieure du bord du plateau se

retrouvent à la base des calcaires crayeux. Cette confusion a amené ces auteurs à exagérer l’é¬

paisseur de ce qu’ils imputaient au Cénomanien supérieur et au Turonien.

COLLIGNON* (1957) déterminait les fossiles ramassés au cours de cette mission. Selon les

conceptions alors en usage, au moins dans le domaine africain, il attribuait au Turonien inférieur

les faunes à Nigericeras et toutes les faunes sus-jacentes. Traduisant la grande variabilité mor¬

phologique des céphalopodes en terme.s spécifiques, il déterminait sur une cinquantaine d’exem¬

plaires dix-huit espèces ou variétés dont six nouvelles.

Cénomanien AngiiUthes flcuriausi, Ncolobites vibruyei, N. peroni, N. iaurteuui.

Turonien intérieur: cf. Nigericeras jacqueti, Manunites subcnnciliatus Choffai var., flattersi

nov. var., Mtimmites cf. pseudonodosoides, Vascoceras gainai, Paravaxcoceras nimeaui nov. sp.,

Discovascoceras cf. ainieirenxe, D. lesselhense nov. sp., D. defrennei nov, sp., Hnpiiloides aff.

ingens. Pseudotissotia gallienei var. injlata nov. var., Baucbioceras nigeriense, Furoniceras trum-

pyi nov. gen., nov. sp., Leonicerus pavillieh.

Busson (1960) précisait la lithologie de la triade de la partie centrale du Tinrhert oriental.

Il interprétait les mécanismes du passage de ces trois termes à la série carbonatée massive vers

l’est et vers l'ouest. Il mettait l'accent sur la zone intermédiaire - entre le domaine des marnes

médianes et le domaine d une série carbonatée massive unique - en particulier dans la région

de Gour ben Houilet où la partie supérieure des marnes médianes s’enrichit considérablement

en fossiles [Hemiaster semicavatus, Baucbioceras) permettant dès lors de confirmer l’âge turonien

de ces marnes, affirmé depuis KOCHb (1880) sans arguments paléontologiques. Il justifiait d’at¬

tribuer au Cénomanien inférieur et moyen (?) les argiles à gypse sous-jacentes aux calcaires

inférieurs, argiles à gypse qui auraient été trop souvent placées dans le Continental intercalaire.

Peu après. BUSSON publiait la carte géologique régulière au 1/500000 Fort-Flatters (1964)

puis une note (1965) introduisant l’étude paléontologique des premières faunes récoltées (cf.

COLI.IGNON 1965, ci-dessous). Par une étude de très nombreuses coupes et une cartographie

complète sur le terrain et avec l’aide des photos aériennes, il avait mis en évidence l’erreur

canographique ci-dessus imputée à Rumeau et al. La limite Cénomanicn-Turonien, coïncidant

à cette époque encore avec la base des couches à Nigericeras, .se situait dans un ensemble de

calcaires crayeux parfaitement monotones ou dissimulés sous les hamadas et regs séparant la

falaise cénomanienne au sud de la falaise turonienne au nord. Cartographiquement, il était pos¬

sible seulement de définir une unité Cénomanien supérieur-Turonien inférieur (c2-ll de la carte)

correspondant aux «calcaires inférieurs» ci-dessus évoqués et parfaitement individualisée par

les diagraphies dans les sondages. Les cartes et coupes fournies dès cette époque (Bus.son 1965,

fig. 2 A, B; fig. 3 A, B) correspondent aux documents de ba.se des figures ci-jointes (Figs 2-4).

Les aftleurements de Cénomanien supérieur et de Turonien inférieur des régions d’Ohanet-ln

Adaoui, de Fort-Ratters devenue Témassinine et de Gour ben Houilet principalement avaient

fourni des centaines de fossiles, en particulier de céphalopodes. Il apparaissait manifeste que

loin de s'agir de deux faunes globales (Tune cénomanienne, l'autre turonienne), une évolution

verticale et une zonation s’imposaient nettement. COLLiGNON (1965) étudiant les ammonites de

* Cette étude de fossile.s de la hamada de Tinrhert était située par son auteur au Fezzan. Cela s'écartait d’un usage qui

limite celte province au territoire de la Libye.

— 185 —

ces récoltes créait encore neuf formes nouvelles dans ces faunes où la diversité morphologique

n’est pas moins remarquable que l’extrême abondance des individus. L’inventaire des formes

déterminées consistait en :

- ammonites cénomaniennes : Neolobiies vibrayei, N. fourtaui, N. peroni, N. hussoni

nov. sp., Calycoceras grossoHvrei, C. boulei, Calycoceras sp., Euralycoceras pentagoniim, Pro-

tacanthoceras sp. ;

- ammonites turoniennes : Kamerunoceras tinrhertense nov. sp., Pseudaspidocerus footei

Stol. var. grecoi nov. var., Nigericeras gignoiLxi, N. lamberti, N. jacqueti, N. jacqueti Schneegans

var. crassecostara, N. (?) vel. Furoniceras (?), DisLovascoceras lesseHtense, D. defrennei, Pa-

ravascocerax aff. cauvini, Vascoceras gamai, Paratnammites luffitiei nov. sp., P. subtuberculatus

nov. sp., Neoptychitex sp. aff. taelingaeformis Solger var. discrepans, Pseudotissotia (Bauchio-

ceras) nigeriensis Woods var. egrediens nov. var., Pseudotissolia bussoni nov. sp., Leoniceras

pavillieri, Leoniceras luciae. L. segne, Hoplitoides aff. ingens, H. hourcqui nov. sp.

Les données géologiques et paléontologiques lui permettaient de proposer la zonation sui¬

vante, de bas en haut :

1) Neolobiies seul;

2) Neolobiies plus Calycoceras, Eucalycoceras, etc. ;

3) les mêmes plus Nigericeras-,

4) Nigericeras plus divers Vascoceratidae ;

5) divers Vascoceratidae plus Bauchioceras ;

6) Bauchioceras plus divers Choff'aticeras, Leoniceras, Hoplitoides.

Dans BUSSON (1965). une note ajoutée en cours d’impression faisait savoir qu’à la lumière

de nouvelles éludes de terrain, la zone 3 devait être amendée : une récolte plus fine montrait

qu’à quelques centimètres près les Nigericeras étaient toujours superposés aux Neolobiies et

Calycoceras. Les trois premières zones étaient attribuées au Cénomanien supérieur et terminal ;

les suivantes au Turonien inférieur. BussoN (1965) notait en outre quelques traits différenciant

la région d’Ohanet-In Adaoui de la région de Témassinine (e.x Forl-Flatters) : manque possible

des Calycoceras à Ohanet ; extrême amincissement de la série sus-jacente au niveau à Nigericeras

et tout spécialement de la série à Bauchioceras des coupes d'Ohanet à celles de Témassinine

{ex Fort-Flatters).

À la suite de nouvelles récoltes (dont les déterminations par Colugnon sont restées inédites)

et d’une synthèse générale sur le Mésozoïque saharien, BussON (1969, 1972) amendait légèrement

la succession proposée. On peut la schématiser avec un Cénomanien supérieur et terminal compre¬

nant :

1) une zone à Neolobiies divers;

2) une zone à Neolobiies et Calycoceras boulei, Eucalycoceras pentagonum ;

puis un Turonien inférieur avec :

1) un niveau à Nigericeras;

2) un niveau à Vascoceras gantai, Paravascoceras aff. cauvini, Paramammites lajfittei,

P. subtuberculatus, Pseudoaspidoceras footei var. grecoi. ;

3) un niveau à Pseudotissotia {Bauchioceras,) nigeriensis,

4) un niveau à Leoniceras luciae, L. segne, L. pavillieri, Discovascoceras defrennei, Ho¬

plitoides ingens et H. hourcqui..

— 186 —

Notons que le même travail imputait pour la première fois le banc calcaire de la partie

supérieure des marnes médianes de Fort-Flatters au Turonien inférieur.

Depuis 1972 aucune étude importante n'a été publiée sur le Cénomanien-Turonien de la

Hamada du Tinrhert oriental ou central. Toutefois, sur un plan général, on sait qu’à la suite des

études de Lauvl-rjat & Berthou (1974) menées au Portugal, confirmées par tous les travaux

postérieurs, les faunes à Vascoceras gamai et V. cauvini - et a fortiori les couches sous-jacentes -

ne doivent plus être datées du Turonien inférieur comme cela a été le cas dans tant de travaux

africains, mais du Cénomanien supérieur.

Une étude sur les échinides de ces niveaux et de cette région vient d’être publiée

(Néraudeau et al. 1993).

LE CADRE GÉOLOGIQUE (G. B., A. C. & F. A.)

La coupe du Crétacé moyen

Sur le glacis méridional du bassin nord-saharien, la hamada de Tinrhert est comprise entre

le Grand Erg oriental au nord et l’erg Isaouane et le bassin d’Illizi au sud, et s’allonge sur une

distance d'environ 300 km suivant une direction est-ouest (Fig. 2a).

Plus précisément, la région d'où proviennent les fossiles ici étudiés correspond à la partie

sud de la hamada de Tinrhert (ou Tinrhert oriental) couvert par la feuille au 1/500000 Fort-Flatters

et limitée par les parallèles 28 et 29”N et les méridiens 6 et 10”E.

Le modelé du paysage est marqué par la présence de comiches parallèles, couronnées par

les principales «barres» carbonatées ou gréso-carbonatées, soit du sud vers le nord (ou du bas

vers le haut dans l’ordre stratigraphique) : corniche du Cénomanien supérieur (partie inférieure

de l’ensemble c2-il du log liihologique et de la carte géologique au 1/500 000 de Fort-Flatters);

corniche du Turonien (t3); corniche du « Santonien-Campanien indifférencié»; comiche du

Maastrichtien. Ces corniches fournissent d’excellents repères visuels, à la fois sur le terrain et

lors de l’examen des photographies aériennes. La continuité des deux premières est à peu près

générale. La troisième est souvent beaucoup plus indécise. Les deux dernières, de toute façon,

ne sont nettement définissables qu'à l’est de l’oued Irharhar. Du bas vers le haut, la succession

des ensembles lilhologiques attribués au Cénomanien et au Turonien dans la hamada de Tinrhert

est la suivante (les index reprennent la classification des formations utilisées sur la carte géo¬

logique au 1/500000 de Fort-Flatters et reportées sur la figure 2b :

Albien, partie supérieure :

a : alternance de marnes sableuses et de bancs carbonatés.

Cénomano-Turonien :

c 1 : sur 130 à 140 m ; argiles à bancs de gypse massif avec trois unités ;

— les 30 à 50 m de base sont constitués de couches argileuses homogènes, en général rouges

ou brun rouge, à peu près dépourvues de gypse;

— les 25 à 35 m en position médiane sont particulièrement riches en gypse massif ;

— les 60 à 70 m derniers mètres sont constitués d’une alternance d’argiles vertes ou rouges,

parcoumes de filons de gypse, de bancs de gypse massif stratifié et, dans les 40 m sommitaux.

— 187 —

de bancs de dolomie blanchâtre, parfois pétrie de débris y compris, dans la région d’Alrar, un

oursin nain (BussoN 1960);

c2-il ; 25 m à 35 m : «calcaires inférieurs», avec deux unités ;

- à la base, sur 10 à 15 m; calcaires massifs, peu ou pas stratifiés, de teinte claire, riches

en ammonites (Nenlohiies et, au sommet, Calycacenis). Cette unité forme en général la plus

grande partie de la premiène corniche du Tinrhcrt qui s'étend depuis Ohanel jusqu'au-delà de

Témassinine (Fort-Flalters) et que l’on retrouve à l'ouest au Tinrhcrt central, occidental, puis

sur le pourtour du Tadémaït;

- au-dessus, sur 15 à 20 m ; bancs de calcaires crayeux, parfois intercalés de marnes.. Les

ammonites y abondent encore [Nigericeras, V(isc(?ceras, Pseiulotissotia, Choffaticeras) réparties

en plusieurs faunes successives.

t2 ; 40 à 60 m : «marnes médianes» constituées de marnes et d'argiles vertes, intercalées

de quelque.^ bancs calcaires à l'ouest de Takouazel et surtout à proximité du mole de la dorsale

d’El Biod, soit à l'est (Cour Ben Houilet), .soit à l’ouest (Chebka Tinrhert). Du gypse peut être

discrètement présent. Cet ensemble marneux forme la partie principale du front de la cuesta

turonienne. À Cour Ben Houilet, des Cliojfuticerus et Pseiidoiissotin du Turonien inférieur sont

encore présents à .3 m du sommet.

t3 : 10 à 15 m : «calcaires supérieurs»; calcaires du Turonien massifs, parfois dolomitiques,

et souvent abondamment silicifics en bancs durs, argileux, bien réglés. Malgré leur faible épais¬

seur et grâce au fait qu'ils sont intercalés entre deux masses argileuses également importantes,

ils constituent non seulement le couronnement de la cuesta turonienne, mais aussi la surface

d’un immen.se plateau rocheux qui peut-être considéré, à proprement parler comme étant véri¬

tablement la hamada de Tinrhert.

« Conidcien »-« Satuonien » :

SI : 50 m ; argiles rouges et vertes incluant du gypse et. plus rarement, des bancs dolimi-

tlques. A noter l’identité de faciès de ces argiles allribuées au Coniacien-Santonien sans aucun

argument paléontologique avec les argiles à gypse (cl) du Cénomanien.

Les calcaires inférieurs (c2-tl) et les marnes médianes (t 2) ont été datés grâce aux faunes

d’ammonites tc/i ci-dessous). Les argiles à gypse sous-jaccntc> n'ont pas livré d’ammonites :

leur attribution au Cénomanien paraît néanmoins extrêmement probable, ainsi que cela a été

admis de la Tripolitaine au Maroc (c/i BüS.SON 1960, 1972). Par contre, la datation des calcaires

supérieurs (t3) et des argiles à gypse sus-jacente.s («Coniacien-Santonien?») est très incertaine.

Des éléments de discussion se trouvent dans BussON (1960, 1972). La confirmation présente

de ratlribution au Turonien inférieur de la partie supérieure des marnes médianes pose plus que

jamais la question de l’âge des deux unités sus-jacentes (tj et S]).

Les ammonites considérées ici (225 .spécimens déterminés au total, cf. annexe 2) proviennent

principalement de trois séries de gisements répartis dans la région d'Ohanet, dans la région de

Témas.sininc (Fort-Flattcrs), et au licu-dit Gour Ben Houilet. localisé à une quarantaine de ki¬

lomètres à l’ouest de Témassinine (Fig. 2). La quasi-totalité du matériel a été récoltée dans les

«calcaires inférieurs» (ensemble c2-tl), les quelques spécimens restants provenant des calcaires

crayeux équivalents de la partie supérieure des «marnes médianes» (ensemble t2).

ypse (cl) |calc.lnf.(c2-t1) | marnes médianes

— 188

Calcaires grumalaux

Calcaires crayeux

Bancs calcaires

^thmés

O Bancs calcaires

massifs

|j-Z| Marnes

1^1 Argiles é gypse

Iffel Gastéropodes

e 1-2 ex. k

9 3é5ex. |Ammonites

6 et plus /

Intervalles fondés

sur les ammonites

^ Pseud nigeriensis

® CMtoticeroâ sp

vstfuO

fiiç&nceras godwn,

Nigçficeras gadertf

_ Iseul) _

Neohb. vibrayeanus

Eucdyc. pentogonum

C. {Colyc.)navicular0

Neohb. vibrayeanus

(seul)

Fig. 3. — Distribution verticale des ammonites de la série calcaire du Cénomanien supérieur-Turonien inférieur dans la région

de Témassinine, ex. Fort-Ratters, Sud algérien.

CENOMANIEN SUPERIEUR_: TURONIEN INFERIEUR

— 189 —

Les gisements des environs de Témassinine

La figure 2c présente le cadre géographique du secteur étudié. Tout un ensemble de gise¬

ments se localise à une dizaine de kilomètres au nord de Témassinine. Au total, quatre coupes

sont considérées. Les trois premières (Akba de Témassinine. est de l’Akba, et au nord de l'Akba)

exposent la base des «calcaires inférieurs», depuis le contact avec les «argiles à gypse» sous-

jacentes. Le calage vertical entre ces trois coupes est fondé sur des corrélations liihologiques.

Celles-ci sont facilitées par la quasi-horizontalité de ces terrains, mais en revanche compliquées

par l’extrême monotonie des faciès et par la discontinuité due aux regs et aux surfaces altérées.

Un banc à trous situé immédiatement au-dessus du gi.sement 143 fournit un niveau-repère local

(Fig. 3). La quatrième coupe, composite, est établie à partir de trois coupes ptutielles si.ses près

de la balise 14, Celte section est placée plus haut au sein des marnes et bancs calcaires formant

la seconde unité de l’ensemble c2-tl, mais en l’absence de repère lithologique commun, sa po¬

sition exacte par rapport aux trois coupes précédentes reste incertaine.

Au sein de ces quatre coupes, comme dans celles des secteurs d’Ohanet et de Gour Ben

Houilet, les récoltes d’ammonites .sont localisées par leurs numéros de gisements. Les gisements

situés le long de la coupe ou à proximité immédiate sont indiqués sur la gauche des logs

lithologiques par des numéros écrits en chiffres arabes. Les gisements un peu plus lointains,

mais corrélés avec l’une de ces coupes sont figurés par des numéros entre parenthèses.

Une remarque doit être formulée sur:

«la présence de nombreux gisements compréhensifs, c'est-à-dire insuffisamment localisés

dans le sens vertical et s'étalant de ce fait sur plus d’une zone straligraphique. Deux raisons

en rendent compte. D'une part, dans une série de calcaires crayeux très monotone, il est

parfois impossible de repérer à quelques décimètres près le niveau précis d'un affleurement

fossilifère isolé. D'autre part, les récoltes ont été effectuées il y a longtemps (de 1957 à

1964), à un stade de l'exploration où l'on visait à reconnaître le Cénomanien et le Turonien

et où l’on ne pressentait pas que les zones d'ammonites qui pourraient être définies un

jour seraient très minces - parfois de l'ordre de quelques décimètres, exigeant un ramassage

très finement différencié dans le .sens vertical» (NÉRAIJDEAU et al. 1993).

Malgré ces limites, la distribution verticale des ammonites recueillies dans la région de

Témassinine (Fort-Flatters) permet de définir six intervalles valables pour tout le Tinhert, soit

du bas vers le haut :

- 1 : intervalle à Neolobites vibrayeanus seul (sur 7 à 9 m d’épaisseur) ;

- 2 intervalle à N. vibrayeanus associé à Forbesiceras sp., Calycoceras (Calycoceras)

naviculare et Eucalycoceras pentagonum (1 à 2 m) ;

- 3 : intervalle à Nigericeras gadeni seul (2 m);

- 4 . intervalle à Vascoceras gamai et V. cauvini associés à N. gadeni (1 m);

- 5 ; intervalle à Pseudotissotia nigeriensis seul (1 à 2 m) ;

- 6 ; intervalle à P. nigeriensis associé à Choffaticeras gr. quaasi - pavillieri et Choffaticeras

sp. (2 m sommitaux des «calcaires inférieurs» et «marnes médianes», soit environ 50 m).

Au total, six intervalles peuvent être différenciés dans les «calcaires inférieurs» et les

«marnes médianes» de la région de Témassinine. sur 65 m d’épaisseur.

Rappelons la correspondance entre les intervalles ici définis pour la description de l’am-

monitofaune et les unités utilisées, pour les mêmes coupes et les mêmes gisements, lors de la

— 190 —

description de l’échinofaune (NÉRAUDEAU et al. 1993) : intervalle 1 = Csl ; 2 = Cs2; 3 = Cs3;

4 = Cs4; 5 = Til ; 6 = Ti2.

LES GISEMENTS DES ENVIRONS D OllANET IN ADAOUI

Les coupes, indiquées sur la figure 2d, s’échelonnent du sud vers le nord sur une vingtaine

de kilomètres dans la dépression d‘Ohancl-In Adaoui L’échantillonnage le plus bas strati-

graphiquement se localise à 2 km et 8 km à l’est de l’Akba d’Ohanet sur le rebord de la corniche

du Cénomanien supérieur. Là aflleurent sur une dizaine de mètres d’épaisseur les calcaires mas¬

sifs, blancs, à Neolabîtes vibrayeanus qui constituent l'unité inférieure de l’ensemble c2-tl de

la carte géologique à 1/500000 de Fort-Flaltcrs. En terme d’ammonites, il s'agit du premier

intervalle à N. vibruyeuniis précédemment reconnu à Témassinine.

Nous n'avons pas relevé de coupes assurant une continuité stratigraphique entre cette pre¬

mière coupe Cl les suivantes» prise.s une dizaine de kilomètres plus au nord au pied de la corniche

du Turonicn. La lacune d’observation liée à cette discontinuité des affieurements peut être de

plusieurs mètres. La deuxième série de coupes (trois au total) porte sur les marnes et bancs

calcaires de la partie supérieure de l'ensemble Cénomanien-Turonien. La distribution des am¬

monites conduit à distinguer ici quatre intervalles, soit du bas vers le haut et en reprenant la

numérotation définie dans le secteur de Témassinine (Fig. 4) :

- 1 : intervalle à Neulubiles vibrayeanus^ associé à Cunningtoniceras tinrhertense

- 2 et 3 : non individualisés, certainement par la lacune d’observation;

- 4 ; inlervallc à Vusvocerus gainai, V. cuuvini, Pseuduspidoceras grecoi, Fikaites lajfitei

et F. subtubercukUiis (vu sur 3 m);

- 5 : intervalle à Pseudoli.walia nigerien.\is seul (S à 9 m);

- 6 : intervalle à P nigeriensis associé à Cbofjalicenix sp. (identifié sur 2 m).

Malgré la discontinuité des affieurements échantillonnés, on retrouve dans la région d'Ohanet,

au sein des «calcaires inférieurs» (ensemble c2-tl), les mêmes divisions fondées sur la distribution

verticale des ammonites que dans le secteur de Témassinine. À noter toutefois, l’augmentation d’é¬

paisseur dans la région d'Ohanet de l'intervalle 5 à Pseudatissotia nigeriensi.s qui passe de I à 2 m

dans les environs de Témassinine à 8 à 9 m près d'Ohanet. D’un autre côté, l’intervalle 4, dont

seule la partie supérieure a fait l'objet de recherches dans le secteur d’Ohanet. ne contient pas de

Nigericeras gadeni associé aux Vastocerai gainai et V. cauvini, mais des Fikaites et Psendaspidaceras

grecoi. Cette différence suggère qu'il existe peut-être deux faunes successives dans ce que nous

appelons aujourd'hui l’intervalle 4; l'association recueillie dans les cnviron.s d’Ohanet pouvant être

légèrement plus récente que celle de Témassinine. La mise en évidence au Nigeria où les successions

sont comparables à celle du Tinhrert d’un horizon à Fikaites aux confins de la limite Cénomanien-

Turonien, au-dessus du niveau de disparition des Nigericeras rend cette hypothèse très probable

(MeisTER 1989; Zaborskj 1993, 1995).

Cour Ben Houilet

La région des Gour Ben Houilet située à une quarantaine de kilomètres à l’ouest de Té¬

massinine (Fig. 2a) présente un double intérêt. D’une part, géographiquement, il s’agit d’une

zone de transition dans laquelle les «marnes médianes» tendent à passer latéralement dans leur

— 191 —

Fig. 4. — Distribution verticale des ammonites de la série calcaire Cénomanien supérieur-Turonien inférieur dans la région d’Ohanet,

Sud algérien.

— 192 —

partie supérieure au faciès carbonaté massif que l’on retrouve sur les hauts-fonds de la dorsale

d’El Biod. Cette transition se fait par l’intermédaire d’un enrichissement généralisé en bancs

calcaires riches en échinides (Néraudeau et al. 1993). D’autre part, la région des Gour Ben

Houilet est une des seules localités où la partie supérieure des «marnes médianes» passe à un

faciès ayant livré des ammonites déterminables spécifiquement, en nombre significatif. C’est,

avec le gisement 393 situé à la partie supérieure des «marnes médianes» dans les environs de

Témassinine (Fort-Flatters), le deuxième point où une attribution stratigraphique peut être ap¬

portée pour une partie élevée de la formation dans le Tinrhert.

La figure 5 présente la distribution verticale des ammonites recueillies. L’inventaire de l’am-

monito-faune qui inclut Pseudotissotia nigeriensis et Chojfaticeras sp., est très comparable à

celui de l’intervalle 6 de Témassinine (Fort-Flatters) et d’Ohanet.

Fig. 5. — Localisation des récoltes d’ammonites dans le Turonien inférieur de la coupe des Gour Ben Houilet, Sud algérien.

— 193 —

Synthèse BiosTRATtCRAPHiQUE

La détermination de deux cent ving cinq spécimens d’ammonites récoltés sur près de 300 km

dans le Tinrhert conduit à individualiser six intervalles biostratigraphiques successifs résumés

dans la figure 6, soit du bas vers le haut à :

- 1 : Neolohites vibrayeanus et Cunningtoniceras tinrhertense \

- 2 : N. vibrayeanus, Forbesiceras sp., Calycoceras (Calycoceras) naviculare et Eiicalyco-

ceras pentagonum ;

Fig. 6. — Synthèse biosiratigraphiquc dans les formations du Cénomanien supérieur-Turonien inférieur des régions de Cour Ben

Houüel, Tcmassinine et Ohunct au Tmriieri, Sahara algérien.

— 194 —

- 3 : Nigericeras gadeni ;

- 4 : Vascoceras gantai et V. cauvini associés vers le bas à N. gadeni et au sommet à

Pseudaspidoceras grecoi, Fikaites subtuberculatus et F. laffitei\

- 5 : Pseudotissotia nigeriensis ',

- 6 : P. nigeriensis, Chojfaticeras gr. quaasi-pavillieri et Choffaticeras sp.

Il apparaît qu’une succession identique existe à l'ouest de la dorsale d’El Biod, en particulier

dans le Tinrhert central (feuille Tilmas El Mra. à l’angle N.-E. de la feuille au 1500000 Amguid,

cf. annexe 1).

La LIMfTE CÉNOMANIHN-TURONIEN AU TlNRHERT

Tracer la limite Cénomanien-Turonien dans la hamada de Tinrhert n’est pas commode en

raison du provincialisme marqué de ses faunes d’ammonites. La plate-forme saharienne dont fait

partie le Tinrhert appartient au domaine téthysien alors dominé par le développement des Vas-

coceratidae. Comme les slratotypes des étages Cénomanien et Turonien se trouvent en domaine

boréal où les faunes d’Acanthoceratidac sont très dift'érentes, aucune corrélation directe n’est

possible dans l'état actuel des connaissances.

Suivant les recommandations émises par la Sous-commission internationale de stratigraphie

du Crétacé (BirkelUND et al. 1984), le Turonien pourrait débuter avec la zone d’ammonite à

Pseudaspidoceras flexuosum. Dans des travaux plus récents. KENNEDY & COBBAN fl991) ont

proposé de décaler légèrement cette limite vers le bas en la faisant coïncider avec la base de

la zone à Watinuceras devonense qui précède la zone à P. flexuosum, Quelle que soit l'opinion

retenue, ni W. devonense ni P. flexuosum ne sont connus sur la plate-forme saharienne et la

position de la limite Cénomanien-Turonien au Tinrhert ne peut être appréciée que de manière

indirecte en tenant compte des éléments suivants.

- Neolobites vibrayeanus, Calycoceras {C.} naviculare et Eucalycoceras pentagonum (in¬

tervalles 1 et 2 du Tinrhert) sont connus à la base du Cénomanien supérieur dans la zone à

C. (C.) naviculare du domaine boréal.

- Nigericeras gadeni est, dans la partie inférieure de son extension (intervalle 3 du Tinrhert),

associé à Metoicoveras gesHnianum dans les niveaux équivalents du Niger (Meister et al. 1992)

et du Nigeria (ZABORSKI 1990). M. gesHnianum est également un index de zone du Cénomanien

supérieur du domaine boréal.

- Vascoceras cf. gantai et V. cauvini (intervalle 4) sont connus aux États-Unis et en Israël dans

la zone à Neocardioceras fuddii ou ses équivalents et dans des niveaux immédiatement inférieurs

(Lewy et al. 1984; COBEAN et al. 1989). N. gadeni est également présent aux États-Unis dans la

Zone à N. fuddii toujours aitribitéc au Cénomanien supérieur (Kennedy et al 1989).

- L’apparition de Pseudotis.%otia nigeriensis (base de l’intervalle 5) coïncide au Nigeria avec

celle de Pseudaspidoc eras flexuosum (Meister 1989) à la base du Turonien ou se produit lé¬

gèrement avant (ZABORSK.1 1993, 1995).

- Enfin, Choffaticeras gr. quaasi-pavillieri (intervalle 6) apparaît en Israël immédiatement

avant Mammites nodosoides dans le Turonien inférieur (Freund & Raab 1969).

En tenant compte de ces informations, même si la corrélation est entachée d’une légère

imprécision, la limite Cénomanien-Turonien peut être raisonnablement placée au Tinrhert entre

les intervalles 4 et 5.

— 195 —

À noter que l’interprétation proposée a pour résultat de remonter sensiblement la position

de la limite Cénomanien-Turonien par rapport aux travaux antérieurs de COLLIGNON (1957, 1965)

et Busson (1965, 1972) où toutes les faunes à Nigericeras et à Vascoceras et les faunes plus

récentes étaient considérées comme turoniennes.

PALÉONTOLOGIE SYSTÉMATIQUE (F. A.)

Ordre AMMONOIDEA Zittel, 1884

Sous-ordre AMMONITINA Hyatt, 1889

Superfamille HOPLITACEAE Douvillé, 1890

Famille Engonoceratidae Hyatt, 1900

Genre NEOLOBITES Fischer, 1882

Espèce-type. — Ammonites vibrayeanus, d’Orbigny, 1841, par désignation originale.

Neolobites vibrayeanus (d’Orbigny, 1841)

(Fig. 7)

Ammonites vibrayeanus d’Orbigny, 1841 : 322, pl. 96, fig. 1-3.

Synonymes :

Neolobites choffati Hyatt, 1903 : 178, pl. 30, figs 1-4.

Neolobites peroni Hyatt, 1903 : 179.

Neolobites bussoni Collignon, 1965 : 171, pl. C, figs la-c

Autres références :

Neolobites vibrayei — COLLIGNON 1965 : 170, text.-fig. 1

Neolobites fourtaui Pervinquière — COLLIGNON 1965 : 170, text.-fig. 2

Neolobites peroni Hyatt — COLLIGNON 1965 : 179, text.-fig. 3

Neolobites vibrayeanus — KENNEDY & JUIGNET 1981 : 23, figs 3 a-c, 4 a-c, 5, 6 a (avec

synonymie). — Lefranc 1981 : 158, figs 1-3. — Moreau, Francis & Kennedy 1983 : 319,

figs 10 a-b. — LUGER & GrôSCHKE 1989 : 366, pl. 39, fig. 3, text.-fig. 5. — Meister,

Alzouma, L.ANG & Mathey 1992 : 60, pl. 1, figs 1-4, 6, text.-fig. 8. — Thomel 1992 ; 184,

pl. 80, fig. 4.

Holotype. — Mu.séum national d’Histoire naturelle Paris, n” 1896 -27. Suivant Kennedy & Juionet (1981),

le spécimen proviendrait d’une couche d’argile du Cénomanien supérieur passant latéralement aux sables à Ca-

topygus ohtusus.

Matériel. — Trente-cinq spécimens dont une vingtaine sont repérés stratigraphiquement par rapport aux

coupes décrites; à Témassinine, gi.sements 91, 93, 94, 143, 155, 269, 382, 390 et 5561 et, à Ohanet, gisements

329 et 330.

Distribution verticale. — En Europe, Israël et au Niger où les séries sont bien datées, N. vibrayeanus

est connu à la base du Cénomanien supérieur, juste sous le niveau d’apparition de Meioicoceras geslinianum

(Kennedy & Juignet 1981 ; Meister et al. 1992). Au Tinrhert, N. vibrayeanus est présent dans les intervalles 1

et 2, associé dans l’intervalle 2 à Eucalycoceraspentagonum et Calycoceras (Calycoceras) naviculare, c’est-à-dire

dans une position équivalente.

— 196 —

Distribution géographique. — France, Espagne, Portugal, Maroc, Algérie, Tunisie, Égypte, Niger, Israël,

Liban, Arabie, Pérou, Bolivie.

Description

Coquille très involute et comprimée avec des flancs légèrement convexes et une région

ventrale étroite et plane. L’épaisseur maximale de la section du tour est obtenue au tiers interne

du flanc. L’ornementation, peu accentuée, est constituée de côtes plutôt fines, qui naissent sur

la bordure ombilicale ou sont intercalaires. D'abord étroites et rectilignes, les côtes se renforcent

Fig, 7. — A-D, Neolobiles vibrayeanus (d’Orbigny, 1841). Les deux spécimens proviennent du gisement 390 au nord de Témas-

sinine. A-B, n° 390A ; C-D, n" 390B. Cénomanien supérieur, intervalle 1 du Tinrhert, (échelle : 1 cm).

— 197 —

au tiers interne du flanc, prennent un aspect sigmoïde, et enfin s’élargissent et s’atténuent sur

la partie la plus externe de la coquille. Les épaules ventro-latérales portent chacune une ligne

de petits tubercules allongés et serrés qui sont parfois indistincts. Enfin, la cloison est typiquement

pseudocératiforme avec des selles larges et des lobes plus étroits.

Discussion

Les études de COLLIGNON (1965), KENNEDY & JuiGNET (1981) et plus récemment Meister

et al. (1992) ont révélé une grande variabilité au sein des populations contemporaines de Neo-

lobites avec une série continue de transitions reliant N. vibrayeanus (d’Orbigny, 1841) (compri¬

mé), N. choffati (Hyatt. 1903) (forme moyenne) et N. peroni (Hyatt, 1903) (épais). Suivant la

proposition formulée par KENNEDY & JuiGNET (1981), les trois espèces sont maintenant mises

en synonymie, N. vibrayeanus restant la seule espèce valide par application de la règle d’anté¬

riorité.

Neolobites bussani, Collignon, 1965 a été créé à partir d’un spécimen unique provenant du

gisement 91 de la coupe de l’Akba de Témassinine. L’holotype. illustré par COLLIGNON (1965,

pl. C, Fig. 1 a-c) et réexaminé ici est un exemplaire de grande taille (17 cm de diamètre) dont

seule la moitié du dernier tour de spire est préservée. Selon COLLIGNON, les caractères spécifiques

de N. bussoni sont l’épaisseur de la coquille (36% du diamètre) et la largeur de la région ventrale

bordée de tubercules fins et serrés. La population de Neolobites vibrayeanus du Niger figurée

par Meister et al. (1992) illustre bien la variabilité de l’épaisseur du tour, de l’ornementation

et de l’ouverture de l’ombilic de l’espèce. Dans la mesure où les variants les plus ornés ont

une épaisseur du tour pouvant atteindre 35 à 38% du diamètre avec une région ventrale corré¬

lativement très large, N. bussoni est considéré ici comme un simple variant de N. vibrayeanu.s.

Neolobites fourtaui Pervinquière, 1907, possède des tubercules ombilicaux proéminents, de

fortes côtes concaves sur la partie externe du flanc et un ombilic plus ouvert, et constitue en

revanche une espèce séparée.

À noter enfin que la majorité des Neolobites du Tinrhert sont des formes assez comprimées,

involutes et relativement peu ornées.

Superfamille Acanthocerataceae de Grossouvre, 1894

Famille Forbesiceratidae Wright. 1952

Genre FORBESICERAS Kossmat, 1897

Espèce-type. —Ammonites largilliertianus d’Orbigny, 1841; par désignation subséquente de Diener, 1925.

Forbesiceras sp.

(Fig. 8)

Neoptychites sp. aff. taelingaeformis Solger var. discrepans Solger. — Collignon 1965 ; 188.

Matériel. — Deux fragments de moules internes provenant du gisement 380 situé dans le secteur de Té¬

massinine.

— 198 —

Fig. 8. — A-B, Forbesiceras sp. Gisement 380D Témassinine, Cénomanien supérieur, intervalle 2 du Tinrhert (échelle : 1 cm).

Distribution verticale. — Intervalle 2 du Tinrhert. Dans le gisement 380, les deux Forbesiceras décrits

ici sont associés à Calycoceras (Calycoceras) naviculare et Eucalycoceras pentagonum. Cénomanien supérieur.

Discussion

Les deux échantillons sont des portions de phragmocônes correspondant à des coquilles

discoïdales, lisses et très comprimées, d’assez grande taille (10 à 15 cm de diamètre). L’enrou¬

lement est involute. La section du tour, beaucoup plus haute que large, est subtriangulaire avec

des flancs légèrement convexes, convergents, et une région ventrale étroite et arrondie. La cloison

est très divisée.

Par leur section du tour, ces deux spécimens rappellent le genre Metengonoceras Hyatt,

1903 ; cependant, la comparaison des cloisons les en sépare nettement, les Metengonoceras ayant

une ligne de suture pseudocératiforme très simple et non très découpée comme ici. Ce même

critère de la cloison les rapproche beaucoup plus des Forbesiceras Kossmat, 1897, qui ont une

suture avec des lobes incisés et des selles phyllocératiformes (Kennedy et al. 1981 ; Wright

& Kennedy 1984). La morphologie générale de la coquille est aussi très comparable avec la

perte, à de grands diamètres, du méplat ventral et l’acquisition d'une région ventrale étroite et

arrondie. Le fragment de grande taille du Nouveau-Mexique (États-Unis) illustré par Cobban

et al. (1989, 77. fig. 66a-b) comme Forbesiceras sp. ressemble beaucoup aux spécimens du

Tinrhert (voir également la cloison dessinée par Cobban et al., 21, fig. 2). Sans l'observation

des tours internes, une détermination spécifique est toutefois impossible. À noter que COLLlGNON

(1965) avait déterminé ces spécimens comme Neoptychites sp. aff. taelingaeformis var, discrepans

Solger (espèce turonienne), malgré leur association à des ammonites typiques du Cénomanien

supérieur dans le gisement 380 : Calycoceras (C.) naviculare et Eucalycoceras pentagonum.

— 199

Famille ACANTHOCERATIDAE de Grossouvre, 1894

Sous-famille Acanthoceratinae de Grossouvre, 1894

Genre CUNNINGTONICERAS Collignon, 1937

Espèce-type. — Ammonites cunningtoni Sharpe, 1855, par désignation originale.

Cunningtoniceras tinrhertense (Collignon, 1965)

(Fig. 9)

Kamerunoceras tinrhertense Collignon, 1965 : 175, pl. D

non Kamerunoceras tinrhertense — Zaborski 1985 : 51, fig. 57-59 (= Romaniceras sp.)

Holotype. — Muséum national d’Histoire naturelle Paris, n° R53926, d’Ohanet, Algérie.

Matériel. — L'holotype provenant du gisement 329 dans le secteur d'Ohanet.

Distribution verticale. — Intervalle 1 du Tinrhert à Neolobites vibrayeanus. Cénomanien supérieur.

Distribution géographique. — Algérie.

Description

L’holotype (Fig. 9A) est un moule interne de 19 cm de diamètre. Seuls le flanc droit et

une partie de la région ventrale sont préservés. L’enroulement est très évolute, l’ombilic repré¬

sentant 43% du diamètre. La section du tour est sensiblement carrée avec un mur ombilical

arrondi, des flancs plats, parallèles jusqu’à l'épaule ventro-lalérale, puis convergents entre les

tubercules ventro-latéraux internes et externes. La région ventrale est plane.

Sur les tours internes, les côtes longues naissent seules ou occasionnellement par paires au

niveau de tubercules ombilicaux saillants et sont effacées à mi-flanc. Elles sont fréquemment

séparées par une courte intercalaire qui apparaît au tiers externe du flanc, La région ventrale

porte cinq rangées de tubercules : ventro-latéraux internes qui tendent à devenir épineux, ven¬

tro-latéraux externes et siphonal. Au diamètre de 127 mm, on compte 22 tubercules ombilicaux

par tour contre 34 tubercules ventro-latéraux et siphonaux.

Sur le dernier tour de spire, la densité costale diminue sensiblement tandis que toutes les

côtes longues deviennent simples et portent seules des tubercules ventro-latéraux internes sail¬

lants, les côtes intercalaires étant très effacées.

Discussion

Collignon (1965) en créant l’espèce Kamerunoceras tinrhertense était hésitant sur son

attribution générique. Les récoltes récentes effectuées aux États-Unis par Kennedy et al. (1987)

et Kennedy & Cobban (1991) ont clairement démontré que le genre Kamerunoceras (Reyment,

1954) dérive du genre Euomphaloceras (Spath, 1923), aux confins de la limite Cénomanien-

Turonien. Or le gisement 329 d’où provient le type de K. tinrhertense a également fourni plusieurs

exemplaires de Neolobites vibrayeanus (d’Orbigny, 1841), ce qui indique un niveau bas dans le

— 200 —

Fig. 9A. — Cunningtoniceras tinrhertense (Collignon, 1965). Holotype, du gisement 329A (échelle : 1 cm).

Cénomanien supérieur (sensiblement équivalent à la partie inférieure de la Zone à Calycoceras

naviculare du domaine boréal).

En fait, les Kamerunuceras ont le plus souvent des côtes légèrement concaves vers l'arrière,

des tubercules ombilicaux situés assez haut sur la bordure ombilicale et pincés radialement, et

des tubercules ventro-latéraux internes peu saillants à l'inverse de ce que l’on observe chez

K. tinrhertense. La morphologie de la coquille, la section du tour quadrangulaire et les tubercules

ventro-latéraux internes épineux rappellent beaucoup plus le genre Cunningtoniceras (Collignon,

1937). L’espèce la plus proche semble être Cunningtoniceras arizonense (Kirkland & Cobban,

1986) du Western Interior aux États-Unis qui a également une densité costale élevée et paraît

contemporaine. L’absence de côtes intercalaires et le développement de véritables cornes sur la

chambre d’habitation de C. arizonense pennettent cependant aisément de distinguer les deux espèces.

— 201 —

Fig. 9B-C. — Cunningtoniceras tinrhertense (CoUignon, 1965). Phragmocône de l’holotype, gisement 329A, Ohanet. Cénomanien

supérieur, intervalle 1 du Tinrhert, (échelle : 1 cm).

Genre et sous-genre CALYCOCERAS Hyatt, 1900

Espèce-type. — Ammonites navicularis Mantell, 1822, par désignation de la Commission internationale de

nomenclature zoologique (Opinion 557).

Calycoceras (Calycoceras) naviculare (Mantell, 1822)

(Fig. 10)

Ammonites navicularis Mantell, 1822: 198, pl. 22, fig. 5.

Autres références :

Calycoceras grossouvrei (Spath) — COLLIGNON 1965 : 172, pl. B, fig. 2 a, b.

Calycoceras boulei — COLLiGNON 1965 : 173, pl. B, figs 3, 4. (non Calycoceras (Meta-

calycoceras) houlei Collignon, 1937 : 43 (pars), pl. 5, fig. 2 seulement; pl. 8, figs 9-11 seule¬

ment (= Calycoceras {Calycoceras) bathyomphalum (Kossmat, 1895)).

Calycoceras naviculare — COBBAN 1971 : 13, pl. 1, figs 1-3; pl. 10, figs 1-8; pl. 11,

figs 1-5; pl. 12, figs 1, 2; pl. 13, figs 1-5; pl. 14, figs 1-3; pl. 15, figs 1, 2; pl. 16, figs 1, 2;

pl. 17; text.-figs 12-14.

— 202 —

Calycoceras (Calycoceras) naviculare — WRIGHT & KENNEDY 1981 : 34, pl. 4; pl. 5,

figs 1-3, text. figs 13, 14c-e (avec synonymie). — Wright & KENNEDY 1990; 236, pl. 61,

fig. 1; pl. 62, figs 1-6; pl. 63, figs 1-3; text figs 88E, 1; 89 D; 110 C (avec synonymie addi¬

tionnelle).

Holotype. — Natural History Muséum London, n° 5681, des Marnes à Aciinocamax ptenus à Offham,

Sus.sex (R.-U.).

Matériel. . — Neuf exemplaires provenant des environs de Témassinine, dans les gi.sements 143, 380, 382

et 5561.

Distribution verticale. — Intervalle 2 du Tinrhert, associé à Eucalycoceras pentagonum. Cénomanien su¬

périeur.

Fig. 10. —A-C, Calycoceras (Calycoceras) naviculare (Mantell, 1822). A-B, gisement 382A, Témassinine. C, gisement 5561.B,

Témassinine. Cénomanien supérieur, intervalle 2 du Tinrhert. x 1. D, Calycoceras (Calycoceras) naviculare (Mantell, 1822).

Gisement 5561 B, Témassinine. Cénomanien supérieur, intervalle 2 du Tinrhert (échelle: I cm).

— 203 —

Distribution géographique. — Angleterre, France, Allemagne, Espagne, Portugal, États-Unis (Western In-

terior), Californie, Afrique du Nord, Angola, Madagascar, Proche-Orient, Inde, Japon.

Discussion

L’espèce Calycoceras (Calycocenis) naviculare (Mantell, 1822) vient d’être révisée à plu¬

sieurs reprises par CoBBAN (1971), Kennedy (1971) et Wright & Kennedy (1981, 1990).

L’espèce est un Calycoceras à section du tour déprimée, à flancs arrondis sur les tours internes,

où l’épaisseur maximale de la coquille est observée au niveau des tubercules ombilicaux. Suivant

COBBAN (1971), on compte 12 à 25 côtes par demi-tour sur les tours internes, mais la densité

costale diminue quand la coquille s'agrandit. Les spécimens du Tinrhert s’intégrent bien dans

cette variation morphologique et deux exemplaires pris à des diamètres de 5 et 10 cm sont il¬

lustrés (Fig. lOA-B. C-D). COLLIGNON (1965) a figuré plusieurs Calycoceras provenant des

coupes de la région de Témassinine sous les noms de Calycoceras grossouvrei (Spath, 1926)

(COLLIGNON 1965; pl. B, Fig. 2 a-b) et Calycoceras bot4lei (COLLIGNON 1937, pl. B. figs 3,

4). Suivant Wright & Kennedy ( 1990). les deux espèces sont respectivement des synonymes

juniors de C. (C.) naviculare et de C. (C.) bathyomphaluw (Kossmat, 1895). Si la première

espèce n’appelle pas de remarque particulière, les spécimens déterminés par COLLIGNON (1965),

comme C. boulet, ne montrent pas la section plus comprimée, ni les flancs plats et la rétention

jusqu’au diamètre de 4 cm des tubercules ventro-laléraux internes épineux caractéristiques de

C. (C.) bathyomphalum. A notre avis, il s’agit également de C. (C.) naviculare.

Genre EUCALYCOCERAS Spath, 1923

Espèce-type. — Ammonites penlagonus Jukes-Browne, 1896, par dé.signation originale.

Eucalycoceras pentagonum (Jukes-Browne, 1896)

(Fig. 11)

Ammonites pentagonus Jukes-Browne, 1896 ; 156, pl. 5, fig. 1

Synonyme ;

Eucalycoceras pentagonum saharense Collignon, in Amard et ai, 1983 : 101, pl. 14; fig. 1.

Autres références :

Eucalycoceras aff. pentagonum — COLLIGNON 1965 : 174.

Protacanthoceras sp. — COLLIGNON 1965 : 174.

Eucalycoceras pentagonum — CoBBAN 1988: 9, pl. 3; text.-figs 6, 7 (avec synonymie).

- Wright & Kennedy 1990 • 282, pl. 78, figs l, 3; pl. 79, figs 1-5; text.-figs 89 E, 123 A,

B. - THOMEL 1992 : 180, pl. 78, figs I. 5; pl. 79, figs 1-5.

Holotype. — Geological Survev Muséum London, if 53481, du Bed C du Cenomanian Limestone à Humble

Point, Devon (R.-U.)

Matériel. — Cinq exemplaires de la région de Témassinine, dont quatre dans les gisements 91, 265, 380

et 5561.

Distribution verticale. — Intervalle 2 du Tinrhert, associé à C. (C.) naviculare et N. vihrayeanus. Céno¬

manien supérieur

Fig. 11. — Eucalycoceras pentagonum (Jukes-Browne,

1896). Gisement 556IA, Témassinine. Cénomanien

supérieur, intervalle 2 du Tinrhert (échelle : 1 cm).

Distribution géographique. — Angleterre, France, Espagne, Portugal, Roumanie, Tunisie, Algérie, Égypte,

Madagascar, Inde, Russie, Japon, États-Unis (Western Interior).

Discussion

Eucalycoceras pentagonum (Jukes-Browne, 1896) est une espèce d’Eucalycoceras aisément

identifiable par la pré.sence de cinq rangées de tubercules ventraux plus ou moins équidistants,

par ses flancs plats, parallèles et par la disparition fréquente des côtes à mi-flanc à la fin du

phragmocône. Un autre élément également caractéristique est la modification de l’ornementation

sur la deuxième moitié de la chambre d’habitation avec disparition des tubercules ventraux, ar¬

rondissement de la région ventrale et augmentation de la densité costale, les espaces intercostaux

devenant alors très étroits.

Le matériel du Tinrhert est identique aux spécimens recueillis dans les craies du bassin

anglo-parisien ou des formations du Cénomanien supérieur du Western Interior aux États-Unis

(COBBAN 1988; Wright & Kennedy 1990).

Eucalycoceras sp.

(Fig. 12)

Matériel. — Deux exemplaires de la région de Témassinine dans le gisement 380.

Distribution verticale. — Intervalle 2 du Tinrhert, associés dans le gisement 380 à C. (C) naviculare,

E. pentagonum et Forbesiceras sp.

Discussion

À côté des spécimens typiques d'Eucalycoceras pentagonum existent, dans le gisement 380

des environs de Témassinine, des Acanthoceratinae très voisins, mais à ornementation des tours

— 205 —

internes plus robuste. Ces exemplaires, déterminés comme Eucalycoceras sp., sont illustrés

(Fig. 12). La morphologie du phragmocône rappelle celle des Pseudocalycoceras et, en particu¬

lier, P. angolaense (Spath, 1931), espèce révisée par COOPER (1978) et Cobban (1988). Mais,

alors que les côtes restent fortes et largement espacées jusqu’au péristome chez P. angolaense,

les exemplaires du Tinrhert montrent le même changement d’ornementation qu' Eucalycoceras

pentagonum sur la deuxième moitié de la chambre d’habitation. Il s’agit d’un exemple de pro-

térogenèse et ces spécimens sont des intermédiaires entre £. pentagonum (espèce ancestrale

connue dans la zone à Calycoreras naviculare et la moitié inférieure de la Zone à Metoicoceras

geslinianum du domaine boréal) et P. angolaense (qui apparaît dans la Zone à M. geslinianum),

confirmant la filiation proposée par Cobban (1988).

Fig. 12. — A-B, Eucalycoceras sp. Gisement 380F, Témassinine. Cénomanien supérieur du Tinrhert, intervalle 2 du Tinrhert, x 1.

C-D, Eucalycoceras sp. Gisement 380G, Témassinine. Cénomanien supérieur, intervalle 2 du Tinrhert (échelle ; 1 cm).

— 206 —

Genre NIGERICERAS Reynient, 1955

Espèce-type. — Nigehceras gignotai Schneegans, 1943, par désignation subséquente de Reyment, 1955.

Le nom de genre Nigericeras ne satisfait pas aux dispositions de l’article 13b du code de

nomenclature, nonobstant la désignation subséquente d'une espèce-type.

Nigericeras gadeni (Chudeau, 1909)

(Fig. 13)

Acanthocerasl gadeni Chudeau, 1909 : 71, pl. 3, fig. 6.

Synonymes :

Nigericeras gignouxi Schneegans, 1943 : 119, pl. 5, figs 10-15, text.-figs 1,2.

Nigericeras lamherti Schneegans, 1943: 121, pl. 6, figs 1-5, 7, text.-figs 3-4.

Nigericeras jacqueti Schneegans, 1943 : 125, pl. 6, fig. 8, text.-fig. 7.

Nigericeras jacqueri var. crassecostata Collignon, 1965 : 178, pl. E, fig. 4 a-b.

Autres réferences :

Nigericeras gadeni — SCHNEEGANS 1943; 123. pl. 7, figs 3-4; text.-Fig. 5. - SCHÔBEL

1975 : 117, pl. 6. figs 1-3 (avec synonymie additionnelle). - KENNEDY, Cobban, Hancock &

Hook 1989 : 62. figs 9L. 11, O, P. - Zaborski 1990 : 4, figs 4-7. - Meister, Alzouma, Lang

& Mathey 1992 ; 67. pl. 3, figs 1-3, 5, 7; pl. 4. fig. I, text. fig. 13.

Nigericeras gignouxi — COLUGNON 1965 : 177, pl. E, fig. 2 a, b; pl. F, fig. 1.

Nigericeras lamherti — COLLIGNON 1965 : 178. — COLLiGNON in Amard et al. 1983 : 56,

pl. 8, fig. I a-b.

Nigericeras jacqueti — COLLIGNON 1965 : 178. pl. E, fig. 3 a-c. - Amard, Collignon &

Lefavr>\is-Henry, 1978 : pl. 1, fig. 2 u-b, - Meister 1989 ; II, pl. 2, figs 3-4; pl. 4, fig. 1.

- Meister, Alzouma, Lang & Mathey 1992: 68. pl. 3, figs 4, 6; pl. 4. fig. 2.

Nigericeras cf. gignouxi — WRIGHT & KENNEDY 1981 : 85, pl. 15, fig. 6.

non Nigericeras gadeni. N. lamherti — MEISTER 1989 : 10, pl. 3, figs 1-3 (= Vascoceras

caiivini Chudeau, 1909).

Holotvpe. — Muséum national d'Histoire naturelle Paris n° R 52466, coll. Chudeau, de Béréré, Damergou,

Niger.

M.WÉR 1 EL. — Vingt exemplaires dont huit sont situés sur les coupes retenues dans le secteur de Témassinine ;

gisements 121, 140, 145, 146 et 150.

Di.stribution verticale. — N. gadeni est connu au Niger et Nigeria en association avec Meloicnceras ges-

linianum. index de ïone du Cénomanien supérieur (Mki.ster et al. 1992; Zaborski 1990). Aux États-UnÊs, l'espèce

est également présente dans la zone suivante à Neocardiaceni^ juddii, toutours du Cénomanien supérieur (Ke.snedy

et al. 1989). Au Tinrhcrt, Nigericeras gadeni est recueilli dans l'intervalle 3 à Ai. gadeni seul, et dans l’intervalle 4

à N. gadeni. Va.scoceras gamai et V! cainnni. Cénomanien supérieur.

DisiRiBuiiüN GÈooRAfHiouE. — Angleterre. France, Algérie, Niger. Nigeria, Égypte, Angola, Israël, Inde,

Japon, États-L^nis (Western Interior).

Description

Coquille modérément évolute, dont l’épaisseur varie depuis des formes relativement compri¬

mées (rapport H/E de la hauteur du tour sur l’épaisseur = 1,6) jusqu’à des formes déprimées

(H/E = 0,75). La section du tour, subovale à subtrapézoïdale, montre une région ventrale arrondie

et des flancs convergents, plats ou légèrement convexes. L’épaisseur maximale de la coquille

est observée près du bord ombilical. L’ornementation des tours internes est typiquement «acan-

thocératiforme », en particulier chez les variants épais, avec des côtes droites, simples ou naissant

Fig. 13. — A-B, Nigericeras gadeni (Chudeau. 1909). Variant épais du gisement 145, Témassinine. Cénomanien supérieur, inter¬

valle 3 du Tinrhert, x 1. Nigericeras gadeni (Chudeau, 1909), C-D gisement 150D et E-F, gisement 121 A; tous deux dans

le secteur de Témassinine. Cénomanien supérieur, intervalles 3 et 4 du Tinrhert (échelle : 1 cm).

— 208 —

par paires et qui portent des tubercules ombilicaux, ventro-latéraux internes et externes, et si-

phonaux. Entre les côtes primaires s’intercalent fréquemment une ou deux côtes secondaires qui

naissent au tiers interne ou à mi-flanc. Sur les tours externes, l’ornementation disparaît très

rapidement et la morphologie de la coquille devient alors homéomorphe de celle de certains

Vascoceras. A noter que chez les variants comprimés, la perte de l’ornementation s’effectue très

tôt et la coquille est alors rapidement lisse, des nodosités péri-ombilicales persistant seulement

chez certains individus.

Discussion

SCHÔBEL (1975) a clairement démontré à l’aide d’une étude statistique des Nigericeras du

Niger que N. gignouxi Schneegans, 1943, N. lamberti Schneegans, 1943 et N. jacqueti Schnee-

gans, 1943, sont des synonymes juniors de N. gadeni (Chudeau 1909), et sont de simples mor¬

photypes illustrant un large spectre de variation d’une seule espèce biologique. Les Nigericeras

gadeni du Tinrhert sont totalement identiques à ceux du Niger et montrent la même variabilité

morphologique avec des formes comprimées et d’autres plus épaisses, et une perte rapide de

l’omementalion. Ce dernier caractère permet d'ailleurs de différencier facilement N. gadeni de

Nigericeras scotti (COBBAN 1971) du Western Interior aux États-Unis, qui garde son ornemen¬

tation « acanthocératiforme » à un diamètre beaucoup plus grand que N. gadeni.

À noter que les Nigericeras ont été longtemps considérés comme le lien phylétique unissant

les Acanthoceras (ou les Proiacanihoceras) du Cénomanien supérieur aux Vascoceras cénomano-

turoniens. Ils étaient de plus inclus comme premiers représentants de la famille des Vascocera-

tidae. En fait, il est maintenant clair que les Vascoceras dérivent directement des Protacantho-

ceras à un niveau bas du Cénomanien supérieur (WRIGHT & KENNEDY 1980; KENNEDY &

WRtGHT 1994) et que leur apparition est antérieure à celle des Nigericeras. C’est la raison pour

laquelle les Nigericeras sont traités aujourd’hui comme des Acanthoceratidae chez lesquels la

perte de l’ornementation au stade adulte conduit simplement à une morphologie homéomorphe

de celle des Vascoceratidae.

Genre FIKAITES Zaborski, 1993

Espèce-type. — Fikaites varicostatus Zaborski, 1993.

Fikaites laffitei (Collignon, 1965)

(Fig. 14)

Paramammites laffitei Collignon, 1965 : 186, pl. A, fig. 2.

Holotype. — Muséum national d’Histoire naturelle Paris, n” R53920, d’Ohanet, Algérie.

Matériel. — L’holotype du gisement 283; secteur d’Ohanet.

Dlstribution verticale. — Intervalle 4 du Tinrhert, associé à Vascoceras gamai et V. cauvini. Cénomanien

supérieur. Au Nigeria, les Fikaites coexistent avec les derniers Pseudaspidoceras pseudonodosoides, Vascoceras

aff. gamai, V! nigeriense et les premiers Thomasites gongilensis aux confins de la limite Cénomanien-Turonien.

Distribution géographique. — Algérie.

— 209 —

Description de l holotype.

Il s’agit d’un spécimen de 82 mm de diamètre montrant le dernier tour de spire du phrag-

mocône et une portion de la chambre d’habitation. La coquille est modérément évolute et globuleuse.

La section du tour, légèrement déprimée (rapport H/E = 0,80), est d'abord subtrapézoïdale .sur

la première partie du phragmocône, avec des flancs convergents, légèrement conve.\es, et un

ventre faiblement arqué, puis devient plus ou moins arrondie vers la fin du phragmocône et sur

la chambre d’habitation. L'épaisseur maximale de la section est observée près du bord ombilical.

L’ornementation, atténuée, est constituée de fines côtes longues ou courtes qui traversent sans

interruption la région ventrale. Les côtes primaires naissent sur la bordure ombilicale au niveau

d’un pelit tubercule allongé radialement, puis s’infléchissent progressivement vers l’avant. Entre

les côtes primaires s'intercalent fréquemment une, parfois deux côtes secondaires. Toutes les

côtes longues et la plupart des courtes porteni sur l'épaule ventro-latérale de faibles tubercules

qui sont plutôt des renflements des côtes. Sur le dernier demi-tour de phragmocône visible, on

compte dix-huit tubercules ventro-laléraux contre huit tubercules ombilicaux.

Discussion

Fikaites Iqffitei a été considéré lors de sa création par COLLlGNON (1965) comme un Pa-

ramammites. Le genre Paramammites Furon, 1935, avec comme espèce-type P. polymorphus

(Pervinquière, 1907) est caractérisé par des côtes saillantes, alternativement longues et courtes,

pourvues sur les tours internes de tubercules épineux. Les côtes primaires portent des tubercules

ombilicaux, latéraux, ventro-latéraux internes et externes, tandis que les côtes secondaires pos¬

sèdent seulement les tubercules latéraux. Les spécimens-types de Paramammites polymorphus

Fig. 14. — Fikaites lajfiîei (Collignon, 1965). Holotype du gisement 283A, Ohanel. Cénomanien .supérieur, intervalle 4 du Tinrhert

(échelle : 1 cm).

— 210 —

viennent d’être refigurés et discutés par Chancellor et al. (1994). La comparaison du matériel

montre que l’on ne retrouve pas l’ornementation très vigoureuse de P. polymorphus chez les

« Paramammites » décrits au Nigeria par Barber (1957) puis Meister (1989), et au Tinrhert

par COLLIGNON (1965). En revanche, la cosiulation relativement dense et parfois irrégulière avec

des côtes longues séparées par une à trois côtes intercalaires est beaucoup plus caractéristique

du genre Fikaites récemment créé par ZABORSKf (1993).

Comme COLEIGNON (1965) le note, Fikaites lajfitei se rapproche des «Paramammites» à

ornementation réduite décrits par Barber (1957) au Nigeria. L’espèce la plus proche paraît être

«P.» inflatus (Barber, 1957) qui est cependant beaucoup plus épaisse, avec un ombilic plus

profond et une ornementation plus atténuée. Fikaites varicostatus (Zaborski, 1993) est également

très voisin, mais avec des flancs plus plats et des côtes légèrement flexueuses et non concaves

vers l’avant.

Fikaites subtuberculatus (Collignon, 1965)

(Fig. 15)

Paramammites subtuberculatus Collignon, 1965 ; 187, pl. A, fig. 3.

Holotype. — Muséum national d’Histoire naturelle Paris, n” R53921, d’Ohanet, Algérie.

Matériel. — L'holotype du gisement 38IX) dans les environs d'OhaneL

Distribution verticale. — La position du gisement 3800 est incertaine mais correspond vraisemblablement

à l’intervalle 4 du Tinrhert (voir la discussion dans le paragraphe consacré à Pseudaspidoceras grecoi). Céno¬

manien supérieur élevé. Au Nigeria, Fikaites varicostatus (très voisin, sinon identique à F. subtuberculatus) est

connu aux confins de la limite Cénomanien-Turonien (Zaborski 1993, 1995).

Distribution géographique. — Algérie et, peut-être, Nigeria.

Description de l’holotype.

Coquille modérément évolute. Section du tour subrectangulaire avec des flancs plats, pa¬

rallèles jusqu’à l’épaule ventro-latérale. puis convergents entre les tubercules ventro-latéraux

internes et externes Jusqu'à la région ventrale légèrement convexe. Bien que le flanc gauche

soit seul préservé, la hauteur et l’épaisseur de la section du tour semblent très comparables. Sur

le phragmocône, les côtes atténuées naissent par paires au niveau d’un tubercule ombilical rond

et mousse. D'abord radiales ou légèrement convexes vers l’avant, elles deviennent franchement

obliques vers l’avant entre les tubercules ventro-latéraux internes et externes, puis traversent la

région ventrale en étant presque droites. Quelques intercalaires apparaissent très haut sur les

flancs, au niveau de l’épaule ventro-latérale. Sur la chambre d’habitation, malgré la conservation

partielle du spécimen, les côtes semblent ne plus naître par faisceaux de deux, mais devien¬

nent alternativement longues et courtes. Les côtes primaires portent au total six rangées de

tubercules bien visibles : ombilicaux, ventro-latéraux internes et externes. Un léger renflement

des côtes, décelable sur la ligne siphonale, suggère cependant l’existence possible d’une sep¬

tième rangée de tubercules siphonaux sur les tours internes. La ligne du suture, bien visible,

présente une première selle latérale large et un premier lobe latéral large avec de nombreuses

denticulations.

— 211 —

Fig. 15. — Fikaites subtuberculatus (Collignon, 1965). Holotype du gisement 3800B, Ohanet. Cénomanien supérieur, vraisemblable¬

ment de l’intervalle 4 du Tinrhert (échelle: I cm).

Discussion

Morphologiquement, Fikaites subtuberculatus Collignon, 1965 est très voisin de Fikaites

varicostatus Zaborski, 1993. Le paratype illustré par ZABORSKI (1993 : 363, fig. 2-P-Q) comme

variant épais de l’espèce est en particulier quasiment identique à F. subtuberculatus. Il n’est pas

impossible qu’une comparaison ultérieure du matériel conduise à la mise en synonymie des es¬

pèces, mais pour l’instant une position conservatrice est adoptée.

Sous-famille Euomphaloceratinae Cooper, 1978

Genre PSEUDASPIDOCERAS Hyatt, 1903

Espèce-type. — Ammonites fooleanus Stoliczka, 1864, par désignation originale.

Pseudaspidoceras footei “var.” grecoi Collignon, 1965‘

(Fig. 16)

Mammites (Pseudaspidoceras) footeanus (?) Stol. — GreCO 1915 : 208, pl. 17, fig. 5.

Pseudaspidoceras footei var. grecoi Collignon, 1965 : 176, pl. E, fig. la, b.

I. Collignon (1965) a décrit sous le nom Pseudaspidoceras footei var. grecoi une ammonite que nous considérons spéci¬

fiquement di.stinete de P. footeanus. En application de l'art. 16 du Code de Nomenclature, le nom grecoi n’est pas dis¬

ponible. Cependant, compte tenu de la similitude de cette variété grecoi avec P. pseudonodosoides Choffat, 1898 nous

nous abstenons de valider formellement le nom grecoi et remettons à une publication ultérieure l’évaluation de ces différents

noms.

— 212 —

IPseudaspidoceras footei van grecoi — COLLIGNON in AMARD et al. 1983 : 50, pl. 5, fig. la-b.

Holotype. — Muséum national d’Histoire naturelle Paris, n° R53927, d’Ohanet, Algérie.

Matériel. — L'holotype et unique exemplaire provenant du gisement 3800 dans le secteur d'Ohanet.

DtsTRiBi'TioN VERTICALE. — Le gisement 3800 est un affleurement isolé situé à 16,5 km au N.B. de la balise

d’Ohanet. Toute conélation liihologique avec les coupe.s décrites est impossible. La récolte dans le même gisement de

Vascoceras afF. ciiuviin Chudeau, 1909 et Fiktûre.'i .^iihiubeivulaïus Collignon, 1965 suggère cependant que l’on se trouve

dans un équivalent de l'intervalle 4 ù Vasioceras gainai, V. cauvini et Nigericem.i gadeni. Au Nigeria, Pseudaspidoceras

pseudnnodo.wide.s (très voisin, sinon identique à P. grenii) est également as.socié h Miscoceiw cauvini., dans des niveaux

attribués au Cénomanien supérieur élevé (Popofe et al. 1986; Meistur 1989 ; Zaborski 1990). Aux ÉtaLs-Unis. P. pseudo-

nodosoides et K cauvini sont présents dans la Zone ü Neocanlioceim juddii (Xennedy et al. 1989).

DtsTRiBUTiON géographique. — Algérie, Égypte.

Description de l’holotype

Il s’agit d’une chambre d’habitation montrant un demi-tour de spire. La coquille est évolute.

La section du tour, presque carrée, est légèrement déprimée (rapport H/E = 0,90) avec des flancs

Fig. 16. — A-B, Pseudaspidoceras grecoi Collignon, !965. Holotype du gisement 3800, Ohanet. Cénomanien supérieur, équivalent

probable de l’intervalle 4 du Tinrhert. C, Pseudaspidoceras grecoi Collignon. 1965 (holotype prise au niveau de la chambre

d’habitation, au diamètre approximatif de 90 mm) (échelle : 1 cm).

— 213 —

faiblement convexes et une région ventrale arrondie. L’épaisseur maximale de la section est

observée au niveau des tubercules ventro-latéraux internes (Fig. 16C). L’ornementation est constituée

de côtes simples, droites, assez serrées, qui naissent au niveau d’un tubercule ombilical allongé ra-

dialement. s'effacent à mi-flanc, et redeviennent saillantes sur l’épaule ventro-latérale où elles portent

un tubercule ventro-latéral interne qui tend à se développer en corne. Entre les côtes longues s'in¬

tercale parfois une côte courte non luberculée. La région ventrale, large et légèrement arrondie, est

en relief par rapport aux tubercules ventro-latéraux externes en nombre identique aux précédents.

On compte douze côtes longues et courtes sur le demi-tour de chambre d'habitation préservé.

Discussion

Pseudaspidoceras grecoi Collignon, 1963 diffère des espèces ultérieures comme P. fiexuosiim

Powell, 1963 par l’absence de côtes flexueuses, bouclées, et comme du génotype: P. footeaiium

Stoliczka. 1864 par ses côtes droites et non convexes vers l'avant, et par le développement des

tubercules ventro-latéraux internes .saillants. P paganuin Reyment, 1954, récemment révisé par Za-

BORSKi (1995) a des côtes moins saillantes, souvent plus denses et des lianes plus convergents. La

morphologie de P. grecoi le rapproche en revanche beaucoup de Pseudaspidoceras psendontMiosoidcs

(Choffat 1898) qui semble contemporain et ce n’est pas sans hésitation qu’une séparation spécifique

est maintenue pour l'instant. Les larges collections de P. pseudonodosoides illustrées par COBBAN et

al. (1989) du Nouveau-Mexique montrent en effet des spécimens très comparables - cf. fig. 83 1-J.

L’holotype de P. grecoi s'en sépare cependant par la rétention des tubercules ventro-latéraux internes

et externes qui restent séparés et bien marqués jusque sur la chambre d’habitation. De plus, la région

ventrale du type de P. grecoi est large et convexe tandis que chez les spécimens figurés de P. pseu-

donodosoidex, celle-ci fomie plutôt une dépression étroite au niveau de la chambre, séparant les ter¬

minaisons ventro-latérales des côtes. Il n’est pas impossible que des récoltes futures pemiettent d’inclure

P. grecoi dans le spectre de variation de P. pseudonodosoides mais une position d’attente est adoptée

pour le moment.

Famille Vascoceratidae Douvillé, 1912

Sous-famille Vascoceratinae Douvillé, 1912

Genre VASCOCERAS Choffat, 1898

(= Paravascoceras Furon, 1935; Pachyvascoceras Furon, 1935; Paracanthoceras Furon,

1935; Broggiieeras Benavides-Caceres, 1956; Discovascoceras Collignon, 1957; Greenhorno-

ceras Cobban & Scott, 1972 et Provascoceras Cooper, 1979). Voir pour la discussion de la

synonymie Berthou et al. (1985)).

Espèce-type — Vascoceras gamai Choffat, 1898 par désignation subséquente de Roman, 1938.

Vascoceras gamai Choffat, 1898

(Fig. 17)

Vascoceras gamai Choffat, 1898 : 54, pl. 7, figs 1-4; pl. 8, fig. 1; pl. 10, fig. 2; pl. 21,

figs 1-4.

Autres références ;

Vascoceras gantai — COLLlGNON 1965 : 183, figs 5-7. - Berthou, Chancellor &

Lauverjat 1985 : 66, pl. 2, figs 1-12; pl. 3, figs 1-3, 5-7, 10, 13, 14 (avec synonymie). - LUGER

& GrOSCHKE 1989 ; 378, pl. 40, figs 5-7; text-fig. 6c. - Thomel 1992: 320, pl. 121, figs 1, 2.

Lectotype. — Service géologique du Portugal, coll. Choffat, n” 808-2, de Meirinhas Baixo, Portugal.

Matériel. — Dix exemplaires dont sept sont repérés par rapport aux coupes décrites, soit dans la région

de Témassinine les gisements 150, et 5566 et, dans le secteur d'Ohanet, les gisements 283, 302 et 309.

Distribution. — Intervalle 4 du Tinrhep, associé à V. cauvini et Nigericeras gadeni. Cénomanien supérieur.

Au Portugal et au Nouveau Mexique aux États-Unis, V. gamai occupe une position similaire dans la zone à

Neocardioceras juddii et dans la zone précédente à Burroceras clydense (Cobban et al. 1989).

Distribution géographique. — France, Portugal, Espagne, Nigeria, Maroc, Algérie, Tunisie, Égypte, Brésil,

Nouveau-Mexique.

Fig. 17. — Vascoceras gamai Choffat, 1898. Tous deux du gisement 302 dans la région d’Ohanet. A-B, n° 302C. C-D, n° 302B.

Cénomanien supérieur, intervalle 4 du Tinrheit (échelle : 1 cm).

Discussion

Vascoceras gamai Choffat, 1898, récemment révisé au Portugal par Berthou et al. (1985),

est un Vascoceras typiquement évolute, modérément comprimé, qui possède une chambre d’ha¬

bitation arrondie, lisse ou très faiblement costulée, et des tours internes pourvus de forts tubercules

ombilicaux au nombre de huit à dix par tour. Les spécimens du Tinrhert, dont certains ont déjà été

illustrés par COLLIGNON (1965 : 183, figs 5-7) sont comparables aux exemplaires de la série type.

Vascoceras cauvini Chudeau, 1909

(Fig. 18)

Vascoceras cauvini Chudeau, 1909 : 68, pl. 1, figs 1,2; pl. 2, figs 1-3; pl. 3, figs 1, 2, 4.

Synonyme ;

Paravascoceras rumeaui Collignon, 1957 : 122, pl. 1, fig. 2.

Autres références :

Paravascoceras aff. cauvini — COLLIGNON 1965 : 183.

Paravascoceras cauvini — SCHÔBEL 1975 ; 119, pl. 4, figs 1-3; pl. 5, figs 1-4 (avec sy¬

nonymie).

Vascoceras cauvini — KENNEDY et al. 1989 : 82, figs 96, 20 c-g. - Zaborski 1990 : figs 8a-b,

12-15.

Vascoceras (Paravascoceras) cauvini — Meister et al. 1992 ; 71 ; pl. 4, fig. 6; pl. 5, fig. 1-

3; pl. 6, figs 1-4; pl. 7, fig. 1 (avec synonymie additionnelle).

Nigericeras gadeni, N. lamberti — MEISTER 1989 : 10, pl. 3, figs 1-3.

Lectotype. — Muséum national d’Histoire naturelle Paris, coll. Chudeau, n“ R52488 de Gjadjidouna, Da-

mergou, Niger.

Matériel. — Sept exemplaires dont cinq des gisements 150 (secteur de Témassinine) et 283 (environs

d’Ohanet).

Distribution verticai.e, — Vascoceras cauvini est présent au Tinrhert dans rimervalle 4 avec V. gamai et

Nigericeras gadeni. L'espèce est également connue en Israël, associée à Metoicoceras geslinianum (Lewv et al.

1984) et son extension monte jusque dans la zone à Neocardioceras juddii et ses équivalents, en particulier en

Israël et aux États-Unis (Kennedy et al. 1989). Cénomanien supérieur.

Distribution gécwraphique. — Israël, Égypte. Algérie, Niger, Nigeria, Soudan, Pérou, Texas. Nouveau-Mexique.

Discussion

Vascoceras cauvini Chudeau est aujourd’hui bien connu grâce à l’excellent travail de

SCHÔBEL (1975), réalisé sur les populations du Niger d’où provient le type. Les tours internes

sont presque lisses avec des flancs aplatis et convergents, mais l’espèce est surtout aisément

reconnaissable si la chambre d’habitation est préservée, avec sa section du tour qui tend à devenir

triangulaire et sa costulation caractéristique constituée de côtes arrondies et larges. La compa¬

raison du matériel du Tinrhert avec les spécimens du Niger figurés par Schneegans (1943) et

SCHÔBEL (1975) montre que les deux populations sont identiques.

— 216 —

À noter que tous les « Vaseoceras», «Paravascoceras» et «Pachyvascoceras» décrits dans

l’ouest du Tinrhert et au Tademaït par COLLIGNON & Roman in Amard et al. (1983) sont, selon

Berthou et al. (1985), dont l’opinion est acceptée ici, spécifiquement indéterminables. Tous les

spécimens concernés sont en effet de grande taille et très usés.

Fig. 18. — A-B, Vascoceras cauvini Chudeau, 1909. Gisement 283, Ohanet. Cénomanien supérieur, intervalle 4 du Tinrhert, x

1. C-D, Vascoceras cauvini Chudeau, 1909. Gisement 2831, Ohanet. Cénomanien supérieur, intervalle 4 (échelle: 1 cm).

Sous-famille PSEUDOTISSOTIINAE Hyatt, 1903

Genre PSEUDOTISSOTIA Peron, 1897

(= Bauchioceras Reyment, 1954; Furoniceras Collignon, 1957)

Espèce-type. — Ammonites gaUiennei d'Orbigny, I8.S0, par désignation originale.

— 217 —

Pseudotissotia nigeriensis (Woods, 1911)

(Figs 19-22)

Hoplitoides nigeriensis Woods, 1911 : pl. 23, fig. 3; pl. 24, figs 1-5.

Synonymes ;

Pseudotissotia (Bauchioceras) nigeriensis macrocarinata Barber, 1957 : 49, pl. 21, fig. 5ab;

pl. 34, fig. 7, 15;

Pseudotissotia (Bauchioceras) nigeriensis plana Barber, 1957 : 49, pl. 20, fig. 2a-b; pl. 21,

fig. 2a-b; pl. 22, figs 5, 6a-b; pl. 23, fig. la-b; pl. 34, figs 2. 8-10.

Pseudotissotia (Bauchioceras) nigeriensis bicarinata Barber, 1957 : 49, pl. 22, figs la-b,

2a-b, 3a-b. 4a-b; pl. 34, figs 4, 12.

Furoniceras trumpyi Collignon, 1957 : 17, pl. 3, fig. l-la.

Pseudotissotia (Bauchioceras) nigeriensis var. egrediens Collignon, 1965 : 188, pl. H, fig. la-c.

Pseudotissotia (Bauchioceras) hussoni Collignon, 1965 : 190, pl. H, fig. 2a-b.

Autres références ;

Pseudotissotia (Bauchioceras) nigeriensis nigeriensis — Barber 1957 : 47, pl. 21, fig. 3a-

b; pl. 22, fig. 6a-b; pl. 34. figs 3, 16.

Pseudotissotia (Bauchioceras) nigeriensis tricarinata (Reyment) — Barber 1957 : 47,

pl. 20, fig. la-b, pl. 21, fig. 4a-b; pl. 34, figs 6, 14.

Discovascoceras lesselitense — COI.LIGNON 1965 : 181, pl. G, fig. la-b. (non Discovasco-

ceras lesselitense Collignon, 1957: 125, pl. 1, fig. l-la (= Vascoceras indéterminé).)

Discovascoceras defrennei — COLLIGNON 1965 : 182, fig. 4a-b. (non Discovascoceras de-

frennei Collignon, 1957 : 125, pl. 2, fig. 3-3a (= Vascoceras indéterminé).)

Pseudotissotia nigeriensis — HlRANO 1983 : 46, pl. 1, figs 1-10; pl. 2, fig. 110; pl. 3,

figs 1-18; pl. 4, figs 1-6 (avec synonymie additionnelle). — COLiRVtLLE 1992 : pl. 12, fig. 4;

pl. 13, figs 1-3. — ZABORSKt 1993 : 369, fig, 6 J.L.

«Pseudotissotia nigeriensis — MEISTER 1989 ; 44, pl. 21, figs 4-6; pl. 22, figs 1-4; pl. 23,

figs 1-5, pl. 24, fig. 1; pl. 25, figs 1-7; pl. 26, figs 1-3; pl. 27, figs 1-3.

Thomasites nigeriensis — Meister et al. 1992; 78, pl. 10, figs 4-6; pl. 11, figs 3, 4;

text.-fig. 19.

Holotype. — Sedgwick Muséum Cambridge, eoll. Woods, du N.-E. du Nigeria.

Matériel. — Une centaine de spécimens dont près de la moitié sont situés sur les coupes décrites ici, soit

dans la région de Témassinine des eisements 125, 393 et 5581, et dans le .secteur d’Ohanet des gisements 279,

280, 281, 286, 287, 292, 293, 298, 314, 316, 319, 3802, 5299 et 5361.

Distribution verticale. — Intervalle 5 à Pseudotissotia nigeriensis (seulj et intervalle 6 à R nigeriensis

et Choffaticeras sp. du Tinrhert. Turonien inférieur. En Israël, P. nigeriensis est connu dans la zone à Choffaticeras

securiforme à la base du Turonien inferieur (Freund & Raab 1969). Au Nigeria, l'espèce occupe une position

équivalente dans les niveaux les plus bas du Turonien (Popoff e/ al. 1986; Meister 1989; Zaborski 1990, 1993,

1995).

Distribution géographique. — Niger, Nigeria, Algérie, Israël, Mexique, Brésil.

— 218 —

15 n

Nombre de

spécimens

Fig. 19. — Variation du rapport H/E (hauteur du tour

sur l’épaisseur mesuré sur 42 exemplaires de Pseu-

dotissoria nigeriensis (Woods, 1911) du Tinrhert.

La majorité des spécimens sont modérément com¬

primés (H/E compris entre I.IO et 1,30, mais le

spectre de variation s'étend de spécimens épais

(H/E = 0,59 à des formes très comprimées (H/E

= 1 , 68 ).

-I-1-1-1-1-1-1-r-

0,5 0,7 0,9 1,10 1,30 1,50 1,70 H/E

DE.SCRIPTION

Coquille discoïdale, involute à très involute, avec un ombilic étroit et profond. La section

du tour, subtrapézoïdale, présente le maximum d’épaisseur au tiers interne du flanc, parfois même

plus bas, avec des flancs légèrement arrondis qui convergent vers la région ventrale plane ou

légèrement tectiforme, modérément large et ornée de trois carènes. La population du Tinrhert

montre une variation considérable dans l’épaisseur du tour, allant de formes involutes et compri¬

mées à des formes évolutes et très rentlées. Le spectre de variation apparaît bien dans l’histo-

gramme de la figure 19 qui clas.se les spécimens en fonction du rapport de la hauteur sur

l’épaisseur du tour (H/E). Si la majorité de,s exemplaires ont un rapport H/E compris entre 1,10

et 1,30 et correspondent à des coquilles modérément comprimées, les deux extrémités du spectre

vont de 1,68 (variants les plus comprimés) à 0,59 (variants déprimés). A noter que chez les

variants comprimés, la région ventrale tend parfois à devenir plane ou même plus rarement lé¬

gèrement concave, et la carène siphonalc peut disparaître plus ou moins.

Au diamètre de préservation du matériel qui est généralement supérieur à 8 cm l’ornemen¬

tation est inexistante ou très atténuée. Sur les tours internes des variants les plus épais, on dis¬

tingue cependant des côtes primaires larges et mousses qui naissent sur un renflement ombilical,

traversent le flanc en étant radiales ou légèrement concaves vers l’avant, et se terminent sur

l’épaule venlro-latérale. Entre les côtes longues s’intercalent fréquemment deux ou trois côtes

courtes. Cette ornementation disparaît cependant très vite, en particulier chez les variants compri¬

més et le stade adulte est entièrement lisse. Les carènes ventrales sont le plus souvent entières

et saillantes, mais dans quelques spécimens les carènes ventro-latérales et, plus rarement, la

carène siphonale apparaissent onduleuses, formées de clavi pincés dans le sens de l’enroulement,

en correspondance avec les terminaisons ventrales des côtes et en nombre identique sur les dif¬

férentes carènes (Fig. 22C).

— 219 —

Fig. 20. — Pseudotissotia nigeriensis (Woods, 1911). Variant épais du gisement 5361F. Ohanet. Turonien inférieur, intervalle 5

du Tinrhert (échelle : 1 cm).

La ligne de suture possède un lobe externe étroit et profond, une première selle latérale

haute et large denticulée et un premier lobe latéral souvent bifide.

Discussion

Barber (1957) et Hirano (1983), ont décrit en détail la grande variabilité morphologique

de Pseudotissotia nigeriensis (Woods, 1911). Les formes comprimées, faiblement ornementées,

portent trois carènes bien individualisées; tandis qu’à l’autre extrémité du spectre de variation,

les variants épais sont costulés et les carènes développent des clavi qui correspondent à la ter¬

minaison ventrale des côtes, ce fait étant particulièrement net sur les tours internes du phrag-

mocône. Le matériel du Tinrhert illustre bien cette large diversité morphologique et tous les

représentants du spectre de variation sont récoltés à travers Pensemble des gisements, les formes

modérément comprimées étant néanmmoins les plus communes.

Pseudotissotia nigeriensis var. egrediens Collignon, 1965 est un variant comprimé à flancs

plats pour lequel une séparation sous-spécifique est superflue (HiRANO 1983). De la même façon,

Pseudotissotia bussoni Collignon, 1965 à coquille comprimée, ombilic étroit, carène siphonale

saillante et flancs modérément convexes, entre dans le spectre de variation de P. nigeriensis

dont il constitue un synonyme junior. Quant aux spécimens épais figurés par Cûllignon (1965)

sous les noms de Discovascoceras tesselitense Collignon. 1957 (COLLIONON, 1965 : 181, pl. 6,

fig. la-b) et Discovascoceras defrennei Collignon, 1957 (COLLIGNON 1965 ; 182, fig. 4a-b), ce

sont des variants déprimés et évolutes proches de Pseudotissotia nigeriensis tricarinata Reyment,

1954 {cf. l’exemplaire du Nigeria illustré par Barber 1957, pl. 20, fig. la-b). À noter en re¬

vanche que les types de D. tesselitense et D. defrennei de Collignon (1957) sont vraisemblable-

— 220 —

Fig. 21. — Pseudotissotia nigeriensis (Woods. 1911). A-B. variant moyennement épais du gisement 314C, Ohanet. C-D, variant

comprimé du gisement 3802F, Ohanet. Turonicn inférieur, intervalles 5 et 6 du Tinrheri (échelle : 1 cm).

ment, selon l’avis de Berthou et al. (1985) des Vascoceras usés indéterminables spécifiquement.

Enfin, Furoniceras trumpyi Collignon, 1957 est également un synonyme junior de P. nigeriensis

dont la cloison aberrante «semble liée à l’usure éolienne du fossile» (Reyment 1978).

P. nigeriensis diffère du génotype Pseudotissotia gallienei (d’Orbigny, 1850) par son en¬

roulement plus involute, la perte rapide de toute ornementation et son stade adulte entièrement

lisse (Kennedy et al. 1979). La tuberculation des carènes ventrales sur les tours internes des

formes épaisses rappelle dans une moindre mesure ce que l’on observe chez les Thomasites

Pervinquière, 1907, et en particulier Thomasites gongilensis (Woods 1911). Ceci a conduit

Meister et al. (1992) à proposer un changement d’attribution générique de P. nigeriensis sous

le nom de Thortmsites nigeriensis. Dans la mesure où l’espyèce-type du genre Pseudotissotia,

P. gallienei, possède également des clavi et des carènes onduleuses sur les tours internes et

puisque, dans tous les variants de P. nigeriensis, le dernier tour de spire est entièrement lisse

— 221

Fig. 22. — Pseudotïssotia nigeriensis (Woods,

1911). A-B. variant comprimé à région ven¬

trale légèrement concave, du gisement 461,

Goût Ben Mouilel. Turonien inférieur, inter¬

valles 5 et 6 du Vinriiert, x 1. C. Pseudotis-

solia nigeriensix (Woods. 1911). Vue ventrale

d’un variant donl les carènes portent des

clavi, gisement 5299, Ohanel. Turonien in¬

férieur, intervalle 5 du Tinrherl (echeile : ! cm).

et les trois carènes ne sont pas tuberculées, cette proposition n’est pas suivie, ce qui est également

l’avis de Zaborski (1990, 1993) et de Chancellor et al. (1994).

Genre CHOFFATICERAS Hyatt, 1903

(= Leoniceras Douvillé, 1912)

Chqffaticeras gr. quaasi (Peron, 1904) - pavillieri Pervinquière, 1907 pl. 10, fig. 1.

Espèce-type — Pseudotissotia meslei Peron, 1897, par désignation originale.

Matériel. — Un spécimen du gisement 125 situé à l'otiesi de la balise 14 dans les enviroms de Témassinine.

Distribution verticale. — Intervalle 6 à Pseudotissotia nigeriensis et Choffaticeras sp. du Tinrhert. Turonien

inférieur. En Israël, Ch. quaasi et Ch. pavillieri sont connus au sommet du Turonien inférieur dans la zone à

Choffaticeras quaasi et, pour la première d'entre elles, dans la zone suivante à Choffaticeras luciae trisellatum

qui est équivalente à la zone à Mummites nodosoldes du domaine boréal (Freund & R.aab 1969). En France,

Ch. pavillieri est présent dans la zone à Kamenmoceras turoniense à la base du Turonien moyen (Amédro &

Hancock 1985).

DiSTRUiuriON géographique. — Ch. quaasi est connu actuellement en Égypte, Israël et Tunisie. La dispersion

de Ch. pavillieri est plus vaste et comprend Israël, l’Égypte, la Tunisie, la Roumanie, la France, le Zaïre, Ma¬

dagascar et les États-Unis (Western Interior).

Discussion

Choffaticeras quaasi (Pérou, 1904) et Choffaticeras pavillieri (Pervinquière, 1907) font par¬

tie selon FreUND & Raab (1969) du groupe des Choffaticeras comprimés à ombilic étroit.

Contrairement à Choffaticeras sinaiticum (Douvillé, 1928) où la section devient rapidement tran¬

chante, Ch. quaasi garde une section du tour ogivale et un ventre arrondi ; tandis que Ch. pavillieri

— 222 —

Fig. 23. — Chojfaticeras gr. quaasi (Peron, 1904) - pavillieri Pervinquière, 1907. Gisement 125C, Témassinine. Turonien inférieur,

intervalle 6 du Tinrhert (échelle : 1 cm).

possède une région ventrale tectiforme avec une carène siphonale qui persiste même sur la cham¬

bre. Un exemplaire du gisement 125, malheureusement assez mal conservé, rassemble plusieurs

caractères communs à Ch. quaani et Ch. pavillieri : coquille comprimée, ombilic étroit et région

ventrale non tranchante. L’altération de la région ventrale ne permet pas de choisir entre les

deux espèces même si l'aspect général de la section du tour avec des flancs sub-parallèles, peu

convergents, rappelle plus Ch. pavillieri {cf. les spécimens illustrés par Pervinquière (1907),

pl. 23, fig. 6a-b, et par AmÉDRO & HANCOCK 1985 : 30, fig. lOa-b).

Choffaticeras sp.

(Fig. 24)

Matériel. — Une quinzaine de spécimens dont la plupart proviennent de gisements intégrés dans les coupes

décrites, soit les gisements 279, 318 (six exemplaires) et 5346 (région d’Ohanet); 393, 5576, 5577 et 5578

(secteur de Témassinine) et, 461, 467, 5597 et 5598 (Cour Ben Houilet).

Distribution verticale. — Intervalle 6 à Pseudotissotia nigeriensis et Choffaticeras sp. du Tinrhert. Turonien

inférieur.

Discussion

Un certain nombre d’ammonites de 15 à 20 cm de diamètre, à coquille discoïdale, mo¬

dérément comprimée à comprimée, ombilic étroit et profond avec un mur anguleux, et section

— 223 —

tranchante, sont considérés comme Choffaticeras indéterminés. La plupart sont des phragmocônes

légèrement écrasés, ce qui renforce le caractère comprimé de la section du tour. Certains ont

des flancs légèrement convexes jusqu’au niveau de la carène siphonale tandis que chez d’autres,

les flancs deviennent concaves sur la partie la plus externe.

Le caractère comprimé de la coquille, l'ombilic étroit et la section lancéolée rappellent les

formes d’Égypte et du Sinaï décrites par DOUVILLÉ (1928) et, en particulier, Choffaticeras si-

naiticum (Douvillé, 1928) (dont Ch. luciae var. .stricto Pervinquière, 1907 pourrait être un sy¬

nonyme antérieur selon Freund & Raab 1969). La différence de diamètre (7 cm pour le type

contre 15 à 20 cm pour les spécimens du Tinrhert) et l’écrasement du matériel algérien rendent

cependant la comparaison difficile. Certains Choffaticeras pavillieri Pervinquière, 1907 acquièrent

par exemple également une section lancéolée à des diamètres supérieurs à 10 cm (Freund &

Raab 1969, fig. 11b),

COLLiGNON (1965 : 29, text-fig. 8a-b) a figuré sous le nom de Leoniceras luciae Perv. un

spécimen plus épais provenant du gisement 3827. Cette détermination, actualisée sous le nom

de Choffaticeras luciae semble exacte, malheureusement la position du gisement qui est isolé

ne peut être précisée par rapport aux coupes décrites.

Hoplitoides hourcqui Collignon, 1965 (193, pl. F, fig. 3a-b), dont le type provient du gi¬

sement 318 dans le secteur d’Ohanet, possède enfin une morphologie comparable à certains spé¬

cimens considérés ici comme Choffaticeras sp. Selon Collignon (1965), H. hourcqui est

distingué par sa cloison, caractérisée par un très fin découpage des selles et des lobes. Si certains

Fig. 24. — Choffaticeras sp. Gisement 3I8C, Ohanet. Turonien inférieur, intervalle 6 du Tinrhert (échelle: I cm).

— 224 —

Tableau 1. — Comparaison des déterminations anciennes (COLLIGNON 1957, 1965) et actualisées des ammonites cénomano-

turoniennes du Tinrhert (Sahara algérien).

COLLIGNON, 1957, 1965

Neohbites vlbrayel d'ORBIGNY

Neolobiles faurtaui PERVINQUIÈRE

Neolobites peraniHYATT

Neolobiles busson/COLUGNON

Kamerunoceras (in/Pertense COLLIGNON

Calycoceras grossoovre/(SPATH)

Calycoceras boute/COLLIGNON

Eucalycoceras pe/lfago/ium JUKES-BROWNE & HILL

Protacanthoceras sp.

Nigericerss gignouxl SCHNEEGANS

Nigericeras lambe/t SCHNEEGANS

Nigericeras jacqueti SCHNEEGANS

NIgericeias jacquetivar. crassecostata COLLIGNON

Pseudaspidoceras toote/var. grecoi COLLIGNON

Paramammites lairitel COLLIGNON

Paramammites subtuberculatus COLLIGNON

Mammilesd. psaudonodosoides CHOFFAT

Vascoceras pamai CHOFFAT

ParavascoceiasaH. cauwn/CHUDEAU

Paravascoœms mmeau/COLLIGNON

Neoptychilessp. ait. laellngaalormisvai. discrepans

SOLGER

Discovascoceras fesse/Zteose COLLIGNON. 1957

Discovascoceras def/enne/COLLIGNON, 1957

Discovascoceras tesseWense COLLIGNON, sensu 1965

Discovascoceras defrenne/COLLIGNON, sensu 1965

Furoniceras Irurvpyi COLLIGNON

Pseudotissotia nlgerienslsvat. egred/ens COLLIGNON

Pseudotissotia busson/COLLIGNON

Leoniceras luclae PERVINQUIÈRE

Leoniceras pavillleri PERVINQUIÈRE

Leoniceras sepneSOLGER

Hoplitoides bourcgu/COLLIGNON

Ce travail

Neolobites vibrayeanus (d'ORBIGNY)

Cunningtoniceras tinrhertense (COLLIGNON)

Calycoceras (C.) naMCutere(MANTELL)

Eucalycoceras pentagonum (JUKES-BROWNE)

Nigericeras gaden/(CHUDEAU)

Pseudaspidoceras grecoi COLLIGNON

Fikaites laffilel (COLLIGNON)

Fikaites subtuberculatus (COLLIGNON)

? *ParamammitBs» sp.

Vascoœras gamai CHOFFAT

Vascoceras cauvini CHUDEAU

Forbesiceras sp.

Vascoceras sp. ind.

Pseudotissotia nigeriensis (WOODS)

Chotfatrceras tec/ae PERVINQUIÈRE

Choffaticeras gr. quaasi (PERON)-paw//ien

PERVINQUIÈRE

Choffaticeras sp.

? Choffaticeras sp. ou ? Hoplitoides sp.

— 225 —

Hoplitoides montrent également une cloison très découpée {cf. la ligne de suture d’W. cf. wohlt-

manni (von Koenen) illustrée par COBBAN & HOOK 1980), la plupart, tel Hoplitoides ingens

(von Koenen, 1897), ont une cloison beaucoup plus pseudocératiforme (Reyment 1955). Selon

COBBAN & Hook (1980), la suture des Choffaticeras rappelle celle des Hoplitoides. D’un autre

côté, les Hoplitoides ont un ombilic très fermé et un ventre plat sur les tours internes, deux

caractères qui n’apparaissent pas sur les deux spécimens de la série type d'H. hourcqiü conservés

dans les collections du Muséum national d'Hisloirc naturelle à Paris. Aussi n’est-il pas impossible

qu’Hoplitoides hourcqui puisse être interprété dans le futur comme étant un Choffaticeras, niais des

récoltes complémentaires dans la localité-type seraient nécessaires pour soutenir l'option.

CONCLUSION

La révision d’une collection de plusieurs centaines d'ammonites recueillies entre 1957 et

1965 dans les fomiations du Cénomanien supérieur et du Turonien inférieur du Tinrhert (Sahara

algérien) conduit à inventorier quinze espèces, ce qui constitue une réduction sensible par rapport

aux études antérieures de Collignon ( 1957, 1965) qui admettait la présence de trente-quaü'e taxons.

La différence tient à la mise en synonymie d’un certain nombre d’espèces définies dans un

concept typologique où la moindre variation d’un caractère morphologique externe était le pré¬

texte pour la création d’une «espèce» n’ayant probablement pas une valeur, ni un sens biolo¬

giques. La comparaison des déterminations anciennes et actualisées est résumée dans le tableau 1.

D’un autre côté, malgré la relative monotonie des faciès, la localisation des spécimens sur

des coupes pennet de distinguer six intervalles bioslraligraphiques fondés sur la répartition des

espèces. La limite Cénomanien-Turonien, interprétée jusqu'alors à l'apparition du genre

Nigericeras est placée maintenant au niveau de disparition de Vascoceras canvini, ce qui s’ac¬

corde mieux avec l'usage actuel dans le domaine boréal. Cet ajustement revient à remonter sur

le terrain de 2 à 5 m la limite au sein de la première masse calcaire, c’est-à-dire des «calcaires

inférieurs» (c2-tl). Enfin, la position stratigraphique de plusieurs espèces .significatives est pré¬

cisée. Tel est le cas par exemple de « Kamerunoceras.» tinrhertense (COLLlGNON 1965) interprété

à l’origine comme un taxon turonien et associé en réalité à Neolohites vihrayeanus à la base

du Cénomanien supérieur.

Manuscrit soumis pour publication ie 7 février 1995 ; accepté le 20 octobre 1995.

RÉFÉRENCES

Amard b., Collignon M. & Lefavrais-Henry M. 1978. — Le Cénomanien d’El Goléa (Tademaït N., Sahara

algérien) : coexistence de Calycoceras avec Nigericeras et implications stratigraphiques. Cahiers de Mi-

croiioléoiitoiogie 4 ■ 29-39, pis 1-5.

Amard B., Collignon M. & Roman J. 1983. — Étude stratigraphique et paléontologique du Crétacé supérieur

et Paléocéne du Tinrhen-W et Tademaït-E (Sahara algérien). Documents du Uiboraioire de Géoiogie de

Lyon. H. S. 6: 15-173.

226 —

AmÉDRO F. & Hancock J. M. 1985. — Les ammonites de l’autoroute «L’Aquitaine», France (Turonien et

Santonien). Cretaceniis Research 6 : 15-32.

Barber W. 1957. — Lower Turonian ammonites from north-eastern Nigeria. Bulletin of the Geological Survey

of Nigeria 26. 86 p., 34 pis.

BENAVlDE.S-CACF.Rrs V. E. 1956. — Cretaceous System in northem Peru. Bulletin of the American Muséum of

naiural History 108: 353-494, pis 31-66.

BerthOU P. Y., CHANCHL1.ÜR G. R. & LAUVERJAT J. 1985. — Révision of the Cenotnanian-Turonian Ammonite

Vascoceras Chof'fal, 1898, from Portugal. Cnmunicaçoes dos Servtcos Geologicos de Portugal 71 (1 ) : 55-79,

6 pli.

Birkelund T.. Hancock J, M., Hart M. B., Rawson P. F., Remane J.. Robaszynski F., Schmid F. &

SURLYK K. 1984. — Cretaceous .stage boundaries - Proposais. Bulletin of the Geological Society of Denmark

33: 3-20,

Burollet P.F. 1956. — Contribution à l’étude straitgraphique de la Tunisie centrale. Annales des Mines et de

la Géologie. Tunis 18, 345 p.

BuS.SON O. I960 — Sur la coupe du Crétacé supérieur et de l’Éocène inférieur du Tinrhert central (Sahara

algérien). Travaux de l'Institut de Recherches sahariennes 19 : 141-149, 4 pis.

— 1964. — Carie géologique de l’Algérie au 1/500000. Feuille Forl-Flatiers. Publication du Centre de Re¬

cherches sur les Zones .Mides. Paris (CNRS).

— 1965. — Sur les gisements de céphalopodes crétacés sahariens. Annales de Paléontologie (Invertébrés)

51 (2) : 153-161.

— 1969. — Sédimentation, transgression et paléogéographie sur les grandes plates-formes du Mésozoïque :

l'exemple du Cénomanien-Turonien du nord-e.sl de la plate-forme saharienne et de Berbérie. Bulletin de la

Société géologique de France 7 (Il 1: 687-703.

— 1970. — Le Mésozoïque saharien : 2^ partie. Essai de synlhè.sc des données des sondages algéro-tunisiens.

Publication du Centre de Recherches sur les Zones Arides, Paris (CNRS), 811 p„ 5 pis.

— 1972. — Principes, méthodes el résultats d'une élude slraligruphtque du Mésozoïque saharien. Mémoire du

Muséum national d'Histoire naturelle, Paris (Sciences de la Terre) N.S. C 26 . 1-442.

Chancellor g.. R.. Ke.NNEDY W. J. & Hancock J. M. 1994. — Turonian ammonite faunas from Central

Tunisia. Spécial Papers in Palacontology 50, 118 p., 37 pis.

CHOFF.AT P. 1898, — Recueil d’études paléontologiques sur lu faune crétacique du Portugal. 1. Espèces nouvelles

ou peu connues. Deuxième .série. Les ammonées du Bellasien, des couches à Neolohites vibrayeanus, du

Turonien et du Sénonien. Travaux de géologie du Portugal : 41-86, pl. 3-22.

ChudeaC R. 1909. — Ammonites du Damergou (Sahara méridional). Bulletin de la Société géologique de France

4 (9) : 67-71, pl. 1-3,

COBBAN W. A. 1971. — New and litile-known ammonite.s from the Upper Cretaceous (Cenomunian and Turonian)

of the Western Intcrior of Ihc United States. Prvfessional Paper U.S. Geological Survey 699: 1-24, 18 pis.

— 1988. — Tarrantoceras Stephenson and related Amnionoid généra from Cenomunian (Upper Cretaceous)

rocks in Texas and the Western Interior of the United States. Professional Paper U.S. Geological Survey

1473, 30 p.. 10 pis.

COBBAN W. A. & Hook s. C. 1980. — The Upper Cretaceous (Turonian) ammonite family Coilopoceratidae

Hyatt in the Western Interior of the United States. Profe.tsional Paper U.S. Ceidogical Survey 1192, 28 p.,

21 pis.

CoBBAN W. A.. HtïOK S. C. & Kennedy W. 3. 1989. — Upper Cretaceous rocks and ammonites faunas of

southwestern New Mexico. New Mexico Butvau of Mines c6 Minerai Resources 45, 137 p.

COBBAN W. A. & Scott g. R. 1972, — Suatlgraphy and ammonite faunas of the Oraneros Shale and Creenhorn

Limeslone near Pueblo, Colorado. Professional Paper U.S. Geological S/irvey 645, 108 p., 41 pis.

COLUONON M. 1937. — Ammonites cénomaniennes du Sud-Ouest de Madagascar. Annales géologiques du Service

des Mines de Madaga.scar 8 : 29-72, pl. 1-11.

— 1957. — Céphalopodes néocrétacés du Tinrhert (Fezzan). Annales de Paléontologie (Invertébrés) 43 : 113-

136, pl. 16-18.

— 1965. — Nouvelles ammonites néocrétacées sahariennes. Annales de Paléontologie (Invertébrés) 5 : 165-202,

pl. A-H.

— 227 —

COOPER M. R. 1978, — Uppermost Cenomanian-basal Turonian ammonites from Salinas, Angola. Armais of the

South African Muséum 75: 51-152.

— 1979. — Ammonite évolution and its bearing on the Cenomanian-Turonian boundary problem. Palaontologie

Zeitschrift 53 : 120-128.

COURVILLK Ph. 1992. — I.cs Vascoceratinae et les Pseudoli.ssotiinac (Ammonitina) d’Ashaka (N.-E. Nigeria):

relations avec leur environnement bioscdimentaire. Bulletin du Centre de Recherches Exploration Production

Elf Aquitaine 16 : 407-457. 14 pis.

DiENER C. 1925. — Ammonoideu neocretacea. Eossililim Cat., 244 p.

DouvillÉ h 1890. — Sur la classification des cératites de la Craie. Bulletin de la Société géologique de France

3 (18); 275-292.

— 1912. — Evolution et classification des Pulchelliidés. Bulletin de la Société géologique de France 4 (11) :

285-320.

■— 1928. — Les ammonites de la Craie supérieure en Egypte et au Sinaï. Mémoire de l'Académie des Sciences,

Institut de France 60, 44 p., 7 pl.

Fischer p. 1882. — Manuel de Conchyliologie et de Paléontologie conchyliologique. Masson (ed.). Paris, 1369

p., 23 pis.

FREUND R. & Ra.'VB M. 1969. — Lower Turonian ammonites from Israël. Spécial Papers in Palaeontology 4,

83 pp., 10 pl.

FURON R. 1935. — Le Crétacé et le Tertiaire du Sahara soudanais (Soudan. Niger, Tchad). Archives du Muséum

national d'Histoire naturelle de Paris 6 (13). 96 p.. 7 pl.

Greco b. 1915. — Faiina cretacea delF Egitto raccolta dal Figari Bey, Palaeontographica lialica 21: 189-231,

pis 17-22.

Grossouvre a. de 1894. — Recherches sur la Craie supérieure, 2 ; Paléontologie. Les Ammonites de la Craie

supérieure. Mémoire du Service de la Carte géologique détaillée de France. 264 p.. 39 pis.

Hirano h 1983. — Révision of iwo Vascoceraiid Ammonites from the Upper Crelaccous of Nigeria. Bulletin

of Science and Engineering Research Laboralory, Wuseda Universilv 105: 44-79, 5 pis.

HyatT a. 1889. — Genesis of the Arielidac. Smilhsonian Contribution lo Knowledge, 238 p., 14 pis.

— 1900. — Cephalopoda. In K. A. von Zl i ri'i. (cd.). Texhook of Palaeontology. Eastman, London : 502-604.

— 1903. — Pseudoceraliles of the Cretaceous. Monography U.S. Geological Survey 44, 351 p., 47 pis.

Jukes-Browne a. J, 1896. — Critical remarks on some of the fossil.s_y« A. 1 Jiikes-Browne & Hill (eds).

A délimitation of the Cenomanian ; being a comparison of the correspunding beds in South-Western England

and Western France. Quaterly .lournal of the Geological Society of London 52 : 99-178, pl. 5.

Kennedy W. J. 1971. — Cenomanian ammonites from Southern England. Spécial Papers in Palaeontology 8,

133 p., 64 pis.

Kennedy W. J. & Cobb.an w. .A. I991. —Siraiigraphy and interregional corrélation of the Cenomanian-Turonian

transition in the Western Interior of the United States near Pueblo. Colorado, a potential boundary stratotype

for the base of the Turonian stage. Newsletrers in Stratigraphy 24; 1-33.

Kennedy W. J, Cübban W. A., Hancock 1. M. & Hook S. C. 1989. — Biostratigraphy of the Chispa Sunimit

Formation at its type locality ; a Cenomanian through Turonian reference section for Trans-Pecos Te-xas.

Bulletin of the Geological Institutions of the University of Uppsukt 15: 39-119.

Kennedy W. J., CoOPER M, R. & Wricht C. W. 1979. — On Ammonites galliennei d'Orbigny, 1850. Bulletin

of the Geological Institutions of the University of Uppsala 8: 5-15.

Kennedy W. J. & Juignet P. 1981. — Upper Cenomanian Ammonites from the Environs of Saumur, and the

provenance of the types of Ammonites vibraveanus and Ammonites geslinianus. Cretaceous Research 2:

19-49.

Kennedy W. J., Juignet P. & Hancock J. M. 1981. — Upper Cenomani.an Ammonites from Anjou and the

Vendée. Western France. Palaeontology 24 (I): 25-84. pis 3-17.

Kennedy W. J. & Wright C. W. 1994. — The affmities of Nigericeras Schneegans, 1943 Cretaceous Am-

monoidca. Geobios 27 (5); 583-589, 2 pl.

Kennedy W. J., Wright C. W. & Hancock J. M. 1987. — Basal Turonian ammonites from West Texas.

Palaeontology 30: 27-74, pis l-IO.

— 228

Kirkland J. I. & COBBAN W. A. 1986. — Cunningtoniceras arizonense n. sp., a large acanthoceratid ammonite

from the iipper Cenomanian (Cretaceous of eastern central Arizona), Htmteria 1; 1-14.

KOENEN A. VON 1897-1898. — Ueber Fossilien der unteren Kreide am Ufer des Mungo in Kamerun. Abhan-

dlungen der Kônigliscbe Geselischafi der Wtssenscimften zu Gottingen. 65 p., 7 pis.

KOSSMAT P- 1895-1898- — Llntersuchiingen liber die südinUische Kreideformaiion. Beitrage zur Palaontologie

imd Geotogie Ô.steneich Uiigiirnx iitid des Orients. 217 p., 19 pis.

Lauverjat J. BerthOE p. y. 1974. — Le Cénomanien-Turoiiien de l'embouchure du Rio Mondego, Beira

lilloral. Portugal. Cnminunicinnes dos Servicos gerilogicos de Porliigid 57 : 263-301.

LeI'Ranc J. Ph, 1976 — Ktai îles connaissances actuelles sur les zonations biostratigraphiques du milieu du

Crétacé (Albien à Turonien au .Sahara. Annules du Muséum d'Ilisloire naturelle de Nice 4, 19 : 1-19.

— 1981. — Élude de Neolohiies vihrayeanus. ammonite cénomanienne du Sahara algérien. I06‘' Congrès

national des Sociétés savantes. Perpignan ; 155-166, 2 pis.

Lewy Z., Kenneoy W. J- & CllANCELLOR G. 1984. — Co-occurence of Metoicoceras gesiinianum d'Orbigny,

and Vascoceras cauvini Chudeau (Cretaceous Animonoidea in the Southern Negev Israël and its stratigraphie

Implications. Newsletrers in Straligrapliy 13 (2): 67-76, 4 figs, 1 pl.

LUGER P- & ürOSCHKE M. 1989. — Laie Cretaceous ammonites Irom the Wadi Qena area in the Egyptian

eastern désert. Palaeontology 32; 355-407, pis 38-49.

ManTBLL g. a. 1822 — The fossils of the South Oowns: or illustrations of the geology of Sussex, London,

327 p.. 42 pl.

Mei.STER C- 1989. — Les ammonites du Crétacé supérieur d’Ashaka, Nigeria. Bulletin Centres de Recherches

E.sploralion Production Elf Aquitaine 13. 84 p., 28 pl.

Meister C.. ALZOUM.A K., Lang J. & Matuey B. 1992. — Les ammonites du Niger (Afrique occidentale) et

la transgression Transsaharienne au cours du Cénomanien-Turonicn. Geobios 25 ; 55-IO(), Il pis,

MoreaI! p., Francis I. 11. & Kennedy w. J. 1983. — Cenomanian ammonites from Northern Aquitaine.

Cretaceous Research 4: 317-339.

NéraL'DEaU D., BiiS-SON G. St Cornée A 1993. — Les échinides du Cénomanien supérieur et du Turonien

inférieur du Tinrherl oriental et central (Sahara algérien). Annules de Paléontologie (invertébrés) 4 : 273-313.

ORBIONY a. d" 1840-1842 — Paléontologie française Description des Mollusc/ues rayonnés fossiles. Terrains

Crétacés, 1 — Céphalopodes. Masson, Paris. 662 p.. 151 pis.

— 1x50. — Prodrome Ue Paléontologie stratigraphique universelle des animawi mollusques et rayonnés. (2),

Masson, Paris. 428 p.

PÉRON A. 1896-1897. — Les ammonites du Crétacé supérieur de l'Algérie. Mémoire de la Société géologique

de France, Paléontologie 17, 88 p.. 18 pl.

— 1904. — In PoURi'AU. Contribution à l’étude de la faune crétacique d'Égypte. Bulletin de ITnstitut égyptien

4, 255 p., pl. 1.

PervinoliiÊRE L. 1907. — Études de paléontologie tunisienne. I. Céphalopodes des terrains secondaires. Carte

géologique de la Tunisie, 438 p.. 27 pis.

POPOEF M., WiEDM.A-NN J. & DE Klasz I. 1986. — The Upper Cretaceous Gcmgila and Pindiga formations,

Northern Nigeria : subdivisions, âge. stratigraphie corrélations and paleogeographic implications. Eclogae

gcnlogica Heh't'Uae 79 (2): .343-363.

PowELL J. D. 1963. — Ccnomanian-Ttironian (Cretaceous ammonites from Trans-Pecos Texas and North-Eastern

Chihunhua, Mexico. Journal of Paleontology 37: 309-322, pl. 31-34.

Reyment R. A. 1954. — Some new Upper Cretaceous Ammonites from Nigeria. Colonial Geological Minera!

Resources 4: 149-164.

— 1955. — The Cretaceous Animonoidea of Southern Nigeria and the Southern Cameroons. Bulletin of Geo-

logicitl Survey of Nigeria 25, 112 p., 25 pis.

— 1978. — Analyse qtianiiiative des Vascoeeraiidês à carènes. Cahiers de Micropaléontologie 4 ; 57-64.

ROBASZYNSKI E. CARON M.. DUPLIS C.. AMEDRO E. GONZALEZ DONOSO J. M., LiNARES D. H.ARDENBOL J.,

Gartner S., Calandra R & DEI-OEPRE R. 1990. — A tentative integralcd stratigraphy in the Turonian

of Central Tunisia : formations, zones and scquential stratigraphy in the Kalaat Senan area. Bulletin des

Centres de Recherche Exploration, Production Elf-Aquitaine 14 (I) : 213-384, 44 pis.

Roche J. I880, — Sur la géologie du Sahara septentrional. Comptes Rendus Je IWcadémie des Sciences. Paris

91 ; 890-893

— 229 —

Roman F. 1938. — Les ammonites jurassiques et crétacées. Essai de généra. Mas.son, Paris, 554 p.. 53 pl.s.

RUMEAU J. L., Debrenne P. & DECREMPS P. 1957. — Mission BRP Tinrhert. Rapport de fin de campagne

1955-1956. Publication Institut française du Pétrole, réf. 1241, 33 p.

SCHNEEGANS D. 1943, — Invertébrés du Crétacé supérieur du Damergou (Territoire du Niger). Bulletin de la

Division des Mines d'Afrique occidentale française 7 : 87-150. 8 pl.

SCHOBEL J. 1975. — Ammoniten der Familie Vascoceratidae aus dem Unterturon des Damergou-Gebietes, Ré¬

publique du Niger. Publications from the Palaeontological Institution of the University of Uppsala Spécial

Volume 3, 136 p.. 6 pis.

SHARPE D, 1853-1857. — Description of the fossil rernains of Mollusca found in the Chalk of England. I.,

Cephalopoda. Monograph of the Palaeontugraphical Society, 68 p., 27 pis.

Spath L. F. 1923. — On the ammonite horizons of the Gault and contiguous deposits. Summaty of Progress

geological Sursey of Créât Brilain. London: 139-149.

— 1926. — On new ammonites from the English Chalk. Geological Magazine 63; 77-83.

— 1931. — A monograph of the Ammonoidea of the Gault, Part 8. Monograph of the Palaeontographical

Society 83: 313-378. pl. 31-36.

Stoliczka F. 1863-1866. — The fossil cephalopoda of the Cretaceous rocks of Southern india : Ammonitidae,

with révision of the Nautilidae. Memoir of the Geological Survey of India. Palaeontolagica Indien, 216 p.,

94 pis.

THOMEL g. 1992. — Ammonites du Cénomanien et du Turonien du Sud-Est de la France. Serre. Nice, 2, 381 p.,

130 pis.

WOODS Fl. 1911. — The palaeontology of the Upper Cretaceous deposits of Northern Nigeria. In J. D. Falconer

( ed.). The Geology and Geography of Northern Nigeria. London: 273-286, pl. 19-24.

Wright C. W. 1952. — A classification of the Cretaceous ammonites. Journal of Paleontology 26: 213-222.

Wright C. W. & Kennedy w. J. 1980. — Origin, ev'olution and systematics of the dwarf acanthcKcratid Pro-

tacanthoceras Spath. 1923 (Cretaceous Ammonoidea, Bulletin of the British Muséum (Natural History)

Geology sériés 34: 65-107.

— 1981. — The Ammonoidea of the Plenus Maris and the Middle Chalk. Monograph of the Palaeontographical

Society, 148 p.. 32 pis.

— 1984-1990. — The Ammonoidea of the Lower Chalk (part 1-3). Monograph of the Palaeontographical

Society, 294 p., 86 pis.

Zaborski p. M. p. 1985. — Upper Cretaceous ammonites from the Calabar région, South-East Nigeria. Bulletin

of the British Muséum ! Natural History) Geology sériés 39, 72 p.

— 1990. —The Cenomanian and Turonian (mid-Cretaceous ammonite biostratigraphy of North-Eastern Nigeria.

Bulletin of the British Muséum (Natural History) Geology sériés 46: 1-18.

— 1993. — Some new and rare Upper Cretaceous ammonites from North-Eastern Nigeria. Journal of African

Earth Sciences 17 (3): 359-371.

— 1995. — The Upper Cretaceous ammonite Pseudaspidoceras Hyatt, 1903, in North-Eastern Nigeria. Bulletin

of the Natural History Muséum London (Geology) SI (1): 53-72.

ZITTEL K. A, 1884. — Handbuch der Palaeontologie, Cephalopoda. Munich & Leipzig: 329-522.

— 230 —

Annexe I. — Situation stratigraphique des principaux gisements fossilifères de la hamada de Tinrhert. Un certain nombre de

gisements n’ont pas fourni d’ammonites décrites dans ce texte, mais leur examen a permis leur localisation stratigraphique.

Il a paru important de les localiser dans cette annnexe car la plupart fournira la matière à l’étude d'autres taxons.

1 INTERVALLES (ammorvies)

NOM

CES

Localisation

sur caries

al coupes

1 CENOMANIEN SUPÉRIEUR

■aiRON. INF.

GISEMENTS

Akba de Tèmaaemine (ex Fon-Flatters)

fig. le

Vallée de la roule de Fort Saint

lig. 1d

AKba de Témassinir>e

figs 1c et 2

Akba de Témassmirve

figs 1c et 2

Akba de Témassmine

figs le et 2

121

W. Balise 14

ligs 1c et 2

125

W. Baltse 14

figs 1c et 2

133

14 km N N W Tèmaasinine

figs 1c et 2

137

14 km N.N.W. Témaesinine

figs le et 2

140

N. Akba de Tèmassinine

figs 1c et 2

143

N. Akba de Témaasinine

(igs 1c et 2

145

N. Akba de Témaasinine

figs 1c et 2

146

N. Akba de Tèmaasinine

figs 1c et 2

N. Akba de Témaaainine

figs 1c et 2

150

W. Bahse 14

figs 1c et 2

155

W Baliss 14

figs 1c et 2

265

Akba de Témassinine

figs le et 2

269

Akba de Tèmassinine

figs le et 2

270

15 km W. Balise d'Ohanet

fig. ld

271

Berge de l'oued au puits-d'Ohanel

lig. ld

279

Gara brune (liane nord)

figs ld et 3

260

Gara brune (Hanc nord)

figs ld et 3

281

Gara brune (liane nord)

figs ld et 3

263

Gara brune (liane nord)

ligit ld et 3

285

Environs de Gara brune

figs ld et 3

286

Flanc W Gara brune

figs ld et 3

287

N. de gara brurve

figs ld et 3

291

Vallée route de Fort Saint. Ranc W.

fig. ld

298

Vallée route de Fort Saint. Flanc \N.

fig. 3

Vallée roule de Fort Saint. Flanc W.

figs 1d et 3

306

15 km N W. de la Balise d'Ohanel

309

Valide route de Fort Saint Rive nord

ligs ld et 3

312

Vallée route de Fort Saint Riva sud

lig. ld

314

Vallee route de Fort Saint. Rive sud

figs ld et 3

315

Nord de Gara brune

lig. ld

316

Nord de Gara brune

figs ld et 3

318

Nord de Gara brune

figs ld et 3

319

Nord de Gara brune

figs ld et 3

320

4kmES.E Akba d'Ohanet

fig, ld

329

8 km S E. Akba d'Ohanet

figs ld et 3

330

8 km S.E. Akba d'Ohanel

figs ld et 3

331

6 km S.E Akba d'Ohanel

tig. ld

360

Est Akba Témassinine

ligs 1c et 2

362

Est Akba Témassinine

fig. 1c

Est Akba Témassinine

lig. le

368

Est Akba Témassinine

lig le

389

Est Akba Témassinine

lig. IC

390

Esl Akba Témassinine

figs 1c et 2

393

Nord de Témassinine

figs 1c et 2

455

Akba d'Amguid

— 232 —

Annexe 2. — Liste des gisements d'ammonites cités dans le texte.

TÉMASSININE (FORT-FLATTERS)

Akba: 91, 93-94, 265, 269

Est Akba ; 380, 382, 390

Nord Akba: 140, 143, 145, 146, 147, 5561, 5566

W. Balise 14: 121, 125, 150, 155, 5576, 5577, 5578, 5581

sommet des Marnes médianes sur le log. synthétique : 393

OHANET

Gara brune : 279, 280, 281, 283, 286, 287, 3800, 3802, 5361

N. Gara brune : 316, 318, 319, 5346

Vallée Fort-Saint-Flanc W : 292, 293, 298, 302, 309, 314, 5299

Bord de la corniche cénomanienne : 329, 330

GOUR BEN HOUILET

461, 467, 5597, 5598.

Bulletin du Muséum naiionol d’Hisfoire naturelle. Paris. 4^ sér., 18. 1996

Section C. 2-3 : 233-347

The phylogeny of the Antiarcha (Placodermi, Pisces),

with the description of Early Devonian antiarchs

from Qujing, Yunnan, China

by Min ZHU

Abstract. — Five ncw uniiiiahs {Yunnannhi^is (mrij'em n. sp.. Mi:.hi hinglmovnsh n gcn., n. sp., Phvniohpis

guoruii n. sp.. Cliiicliinnlt’pis rnhusta n. sp. and C. su/nihi n. sp.) are deseribcd on rhe basis of the ncw niaterial

from the Furly Devi>nian of Qujing, Yunnan Province, south-we.slem China. K thii is reesaniined to show lhe

CHANa’.s apparatus and latcroventral fossa of ihe irunk-Khield. New material of Qujing comprises aiso fl^uroy-

umumolepix (/u/;nt;e«.if.v. F. sp. from lhe Xitun Formation. P. aii/eilftsliam'nsis and Zliaiijiti’pi.'i tixpralilis from

the Xishancnn Formation, gnuilix and C. ifUjinÿtnsh from lhe .Xtshancun and Xilun Fonnaiions, and an

unnamed untiareh from lhe Xitun Fommiion. !l is proposed that Priif.niulyloUpix. as wel) as C,)ujinijhpix. |s the

junior synonym of Chuchiiiolepis. The pecioraf fin uriiculation of lhe Chuchinolepididae is argucd lo havc only

one dermal joint. Il rs .suggesied that the pectoral fin of ihe Chuchinolepididae is laterally compre.ssed and that

the fossa on the parabrachial process is lhe allachmeni .area for lhe ahducior muscle of the fin. A cladistic

analysis of antiarchs is prcscmed. Thiriy two equally pursimomous Irees are foiind, and a .strict con.sensus tree

is construcicd. The monophyly of the Yunnunolcpidoidei is torruboraied m the cladogram. with an etnended

définition of lhe group. The Chuchinolepididae. Vumhiettolcpis. and an unnamed aniiarch are meluded in lhe

Yunnanolepidoidei. Minicrunûi (Zitu fi- Ivrsvit R 1906) is lhe sisier-group of lhe anti.nchs possessing the brachial

process and funncl pit. The Bothriolepidoidei. as previousiy defined. lurri oui to be a paraphyletic group. The

Microhrachiidae are at the hase of the Euantiarcha. NdWiigUispix is as.sumed lo be the nearest sister-group of the

Asterolepidoidei, and Hunantilepis is placed ai lhe base of lhe Asierolepidoidei.

Key-words. — Antiarcha. Early Devonian. morphology. phylogeny. China.

Phylogénie des Aniiarches (Placodermi, Pisces) et description

d’antiarches du Dévonien inférieur de Qujing, Yunnan. Chine

Résumé. — Cinq nouveau.v antiarches O'uniumnitpif. porifera n sp., Mizui hitgiwaensix n. gen., n. sp.,

Phyiiwlepis gtianiii n.sp., CImcIniwU'pix rohtixui n. sp. et C. xhIcciui n. sp.) sont décrits sur lu base de nouveau

matériel du Dévonien inférieur de Qujing (province de Yunnan), dans le sud-ouest de la Chine. Y. ebii est réétudié,

afin de montrer l'appareil de Ciiano et la fosse venlrolalérale de la cuira.sse thoracique. Le nouveau matériel

comprend aussi Heleroyminiinolepix i/ujingenxix et T. sp. de la Formation de .Xilun, P. ciiifengxluiiK'itxix et /.Imn-

iilepix axpnitilix de la Formation de Xishancnn. C gnicilix et C. (/n/in.ee/n/s des Formations de Xishancnn et

Xitun. et un anliarchc non nomme de la Formation de Xilun. Il est paiposé que Pniccndylolepix, ainsi que

Qujinnlepix, e.st le synonyme plus récent de Cluieliinolepix. L'articulation de la nageoire peeiorale des Chuchi¬

nolepididae est considérée comme n'ayant qu'une seule articulation dermique II est suggéré que la nageoire

peeiorale des Ctiiichinolcpididae est comprimée latéraleineni et que la fosse sur le processus parabrachial sert à

l'insertion du mu.scle abducteur de nageoire Une analyse cladisiique des aniiarches est proposée. Trenie-deu.x ar¬

bres égaienieni parsimonieux sont obtenus, et un arbre de strict consensus esl consiruit La monophylie îles

Yunnunolcpidoidei est confirmée dans le cladogrammc, avec une définition émendéc. Les Chuchinolepididae,

Vanchiennl.pix et un antianrhe non nommé soni inclus dans les Yunnanolepidoidei. Minirniniii (Zui & Janvier

1996) esl le groupc-frcie des anti.trches possédant un processus brachial pourvu d'un canal en entonnoir Les

Bothriolepidoidei définis auparavant se trouvent être un groupe paraphylétique. Les Microbrachiidae sont à la

racine des Kiianliarcha. Niiwagiaxpix est supposé être le groupe-frère le plus proche des Astciolcpidoidei, et

Hiinanolepix est placé à la racine des Asterolepidoiilei.

Mots-clés. — Antiarcha, Dévonien inférieur, morphologie, phylogénie, Chine.

M. Zhu, Inxiitule of Vertebrate Paleonlology and Palenanlhmpolngy, Chinexe Acadeiny of Sciences, PO. Box 643, Beijing 100044,

China.

— 234 —

Abbreviations

a anterior angle of AMD plate

aa anterior angle of PMD plate

adg anterodorsal sensory-line groove of

trunk-shield

ADL anterior dorsolateral plate

AL anterior latéral plate

air postlevator thickening of AMD plate

AMD anterior médian dorsal plate

ar3v extemal articular area of Cvl plate

art.v ventral articular dépréssion for dennal

process of pectoral fin

AVL anterior ventmiateral plate

Cl corner betwcen anterior and middle

divisions of mesial margin of ventral

lamina of AVL plate.

C.Chang cavity of Chano's apparatus

c.al antérolatéral corner of siihcephalic

division of ventral lamina of AVL plate

Ca.o opening of Ciiano’s apparatus

Cdl-4 dorsal central plate 1-4

cf.ADL area overlapping ADL plate

cf.AMD area overlapping AMD plate

cf.AVL area overlapping AVL plate

cf.MV area overlapping MV plate

cf.PDL area overlapping PDL plate

cf.PVL area overlapping PVL plate

cf.PL area overlapping PL plate

cit crista iransvenalis interna anterior

cr.tp crista iransversalis interna posterior

Cvl-4 ventral central plate 1-4

d dorsal corner of PDL plate

dig posterior oblique dorsal sensory-line groove

dim dorsal lamina of ADL and PDL plates

dlr dorsolateral ridge of trunk-shield

dma tergal angle of trunk-shield

dm.ppbr dorsal margin of parahrachial process

dmr médian dorsal ridge of trunk-shield

f.ab fossa of AVL plaie for abductor mu.scle of fin

f,ad fossa of AVL plate for adductor muscle of fin

f.art articular fossa of ADL plate

fe.orb orbital fenesba

f.lv lateroventral fossa of trunk-shield

f.pec pectoral fenestra

f. retr ievator fossa of AMD plate

g. Chang groove of trunk-shield caused by

Chang's apparatus

gpbr parahrachial groove

grm médian ventral groove of dorsal wall

of trunk-shield

L latéral plate

l.Chang inner lamina of Chang’s apparatus

le latéral corner of AMD plate

Icg main lateral-line groove

llm latéral lamina of ADL. AVL, PDL and

PVL plates

l.pbr postbranchial lamina or ridge

M12-4 latéral marginal late 2-4

Mml-4 mesial marginal plate 1-4

MV médian ventral plate

mvr médian ventral ridge of dorsal wall of

trunk-shield

m.SL margin of AVL plate in contact with SL

plate

mx sedimentary matrix of fossil

MxL mixilaieral plate

Nu ntichal plate

n node formed by tubercics

oa.ADL area overlapped by ADL plate

oa.AMD area overlapped by AMD piale

oa.PDL area overlapped by PDL plate

oa.PL area overlapped by PL plate

oa,PMD area overlapped by PMD plate

oa.PVL area overlapped by PVL plate

pa posterior angle of PMD plate

p.AL process of AL plate

pda po.slerior dorsal angle of trunk-shield

pdg postenxlorsal sensory-line groove of

trunk-shield

PDL posterior dorsolctteral plate

PL posterior latéral plate

pic postérolatéral corner of PMD plate

pig po.sterolaieral groove of trunk-shield

plr postérolatéral ridge of trunk-shield

PM poslrnarginal plate

pma posterior marginal area of PMD plate

PMD posterior médian dorsal plate

PNu paranuehal plate

Pp postpi neal plate

p.PMD pit on cxternal surface of PMD plate

ppbr parahrachial process

pre prepecioral corner

pr.p posterior process of PMD plate

prvl anterior ventral process of dorsal wall of

trunk-shield

prv2 posterior ventral process of dorsal wall

of trunk-shield

ptl anterior ventral pit of dorsal wall of

trunk-shield

pt2 posterior ventral pit of dorsal wall of

trunk-shield

PVL posterior ventmiateral plate

r.Chang ridge caused by Chang’s apparatus

r.ibr infrabrachial ridge

r.obi oblique ridge on dorsal wall of trunk-

shield

r.semi semicirçiilar ridge of skull-roof

scap scapulocoracoid

SL semilunar plate

Sp spinal plate

vlm ventral lamina of AVL and PVL plates

vlr ventmiateral ridge of trunk-shield

vm.ppbr ventral margin of parahrachial process

— 235 —

INTRODUCTION

The Paleozoic early vertebrates of China were first reported from Panxi (Po-Si), Huaning

County, Yunnan by Mansuy (1907). However, the remains (an “ichthyodorulite”) were impossible

to be ideniificd by Mansuy and no f'urther description or photo was given. From the same

locality, three brachythoracid arthrodires and sonie acanthodian spines were collceted laier from

the Middle Devonian deposits (J.-Q. WANG 1979. 1982. 1991c). The first figured early vertebrate

from China (MANSUY 1912) was an anliarcli from neai Xundian. north-east of Kunming (Yun-

Nan-Foii), the capital of Yunn.m Province. Mansuy assigned this fossil (part of a irunk-shield)

to a placoderm similar to Bothriolepis, but he refcrred it to the “Silurian” (in the sense of this

time, that is. including most of the Lower Devonian). According to the pholograph. lhe only

information we hâve about this specimen. M.ANSUY 's fossil is possibly to be referred to as

Dianulepis, which bas later been found in the same région. If it is irue. the âge of this antiarch

should be Middle Devonian. This began the discoveries of antiarchs in China.

The forerunner of antiarch research in China is Y.-S. CHt (1940). who dcscribed a species

of Bothriolepis {B. sinen.sis), from rhe Middle Devonian of Chansha, Hunan. Later, Cm (1942)

and H.-C. WANG (1942) reported the occurrence of Bothriolepis from the Devonian of Yunnan,

and PlEN (1948) noted the finding of Bothriolepis in Guangdong. By this time, ail antiarchs

found in China were euantiarchs. no older than Middle Devonian in âge ; that, is consistent with

lhe âge of antiarchs in Europe, North America and Australia.

The subséquent studios of Chinese antiarchs are rclated to the di.scoveries of new antiarch

groups, which hâve a major significance in lhe underslanding of antiarch évolution and biogeo-

graphy. Moreover, lhe antiarchs found in China hâve extended the history of the group back to

Early Silurian (J.-Q. Wang 1991b). since before 1963. when Yunnanolepis was formally described

by Y.-H. Liu, antiarchs were restricted to Middle and Late Devonian. The exception ts the remains

described by Man.su Y (1915) in North Vietnam, near die frontier between China and Vietnam.

These fossils. originally referred to as an ostracoderm, 1 Homostius and Asterolepis, turned out

to be yunnanolepidoids or galcaspids (Tong-Dzuy & Janvier 1987), which are the major com-

ponents of the highly endémie Early Devonian fish fauna in South China and North Vietnam.

The first new antiarch group found in China is the Sinolepididae (LtU & P’AN 1958).

Sinolepis, the type genus of lhe family. was discovered in the Upper Devonian Wutung Sériés

near Nanjing, in association with a bothriolcpidoid Jiang.xilepis siiiensis (LtU & P’AN 1958;

P’AN 1964; Zhang & LlU 1991) which was originally referred to as Asterolepis. Sinolepis is

the first indication of the highly endemic feature of the early vertebrate fauna in the Silurian

and Devonian of China (including South China, Tarim and Ningxia) and Vietnam. The further

research on the Sinolepididae (RlTCHtE et al. 1992), which arc now known to be unique to

China and Australia, indicaies a close paleogeographic afl'inity between these two régions, and

shows that the Sinolepididae are lhe closesl relatives of lhe Euanliarcha.

The second new group is the Yunnanolepidoidei (Y.-H. Liu 1.963; CiROSS 1965; Miles 1968;

G.-R. Zhang 1978; M.-M. Zhang 1980), exemplified by the Yunnanolepididae. The Yun-

nanolepididae is a distinctive élément of the Early Devonian endemic vertebrate fauna of South

China and North Vietnam, and was assumed to be the most primitive antiarch group because

of its plesiomorphies shared with other placoderms, and thus having bearings on lhe problems

— 236

of antiarch relationships and interrclalionships. The yunnanolepidoid Shimenaxpis (J.-Q. Wang

1991b) found in the Early Silurian (Llandovery) of Lixian, Ilunan (Fig. 2) is lhe earliest pla-

coderm record. However, since the Yunnanolepidoidei, as previnu.sly defined, was mainly based

on symplesiomorphies. its monophyly had been doubted (JANVIER & P'AN 1982), and the suborder

was generally referred to as lhe stemgroup of antiarchs. The relationships among the varions

généra of the Yunnanolepidoidej are still vague. Therefore, a furiher study and strict définition

of the Yunnanolepidoidei is essential to research on antiarch évolution.

Tlie third new group is the Chuclnnolepididae (G.-H. ZHANG 1984; YOUNG & ZHANG 1992),

which is characterizcd by its peculiar pectoral fin articulation. This group was found in the

Early Devonian of Yunnan and Guangxi, China, and was discovered latcr in the Early Dcvonian

of North Vietnam (ToNG-D/HY & JANVIER 1990). G.-R. ZHANG (1984) and Youno & Zhang

(1992) proposed a hypothesis where the Chuchinolepididae are intemiediate between the Yun¬

nanolepidoidei and the other advanced antiarchs (the Siiiolepididae and Euaniiarcha), as to the

pectoral fin articulation. In contrasî. Zitii &. JANVIER (1996) suggest thaï the Chuchinolepididae

and the Sinolepididae + Euantiarcha represent two distinct evolutionary paths towards the for¬

mation of the pectoral fin articulation.

Thc.se discoveries hâve established the Chinese antiarchs, especially those from the Early

Devonian, as being of crucial importance in Lhe undersianding of antiarch évolution. The first

aim of this work is to describe new antiarch material from the Early Devonian of Qujing (Yunnan,

China), and clarify certain aspects of yunnanolepidoid comparative morphology. The second aim

is to discuss the structure of the pectoral fin articulation of the Chuchinolepididae. which has

been an interesting subjecl as to its origin and fatc. A new hypothesis about the plaeement of

the pectoral fin in this bizarre fish is proposed. and conlrasts with tho.se of G.-R. Zhano (1984)

and Young & Zhang ( 1992). The third aim of this work is to explore the phylogeny of antiarchs

in the light of character analysis. Since lhe Chinese antiarchs werc taken inlo considération, the

phylogeny of antiarchs has been discussed in depih, in the framework of either Cladistics or

Evolutionary Systematics (Janvier & Pan 1982; G.-R. ZHANG 1984; YüUNG 1984c. 1988; Pan

et al. 1987; YoUNG & ZHANG 1992; RiTCHIE ei al. 1992; Ztlü & Janvier 1996). Wilh the aid

of the information provided by the Yunnanolepidoidei. the phylogeny of the Sinolepididae and

Euantiarcha is now belter corroborated (RlTCHlE et al. 1992). even though Yumtanulepis as the

root of lhe antiarch cladogram sumetimes biases the détermination of character polarily. In fact,

Yunnanolepis is a relatively derived antiarch (Zhu & Janvier 1996). However, turning to the

early évolution of antiarchs. the schenic bccomes mttre or less obscure, and this ambiguity affects

the undersianding of the phylogeny of the advanced antiarchs. The obslaclc is the deficiency of

the character analysis on the Yunnanolepidoidei, including the Chuchinolepididae, and the fact

that lhe relationships among the various primitive gênera are unavailable by now. Moreover. the

previous cladugranis of antiarchs. except thaï of bolhriolepidoids (Zhang & Young 1992), were

somewhat empirical. and were nol strictly based on the principle of parsimony. In addition, the

cladogram of bolhnolepidoids included ihe assumplion of iheir monophyly, which will be refuted

below. The phylogenetic analysis in this work includes ail well-understood antiarch taxa, and is

achieved using Hennig 86 (Farris 1988).

— 237 —

REVIEW OF THE EARLY DEVONIAN ANTIARCHS FROM CHINA

Eleven localities yielding Early Devonian antiarchs in China are reviewed below (Fig. 1).

I) The first Early Devonian aniiarch de.scribed in China is Yuniuinolepis cliii (Y.-H. Liu

1963), which was represented by a skull-rooF in viscéral view, and wa.s found in ihe Xi.shancun

Formation (Y.-H. Liu pers. comm.) of the Ciiit'engshan Group of Qujing, Yunnan. This antiarch

was originally regarded as belonging to the Pterichthyodidae (=thc Asterolepidoidei), and was

later placed in the Bothriolepididae by Gross (196.5). At the Paris meeting on vertebrate paleon-

tology (1966), M,-M. CHANC reported on the research advances on the Early Devonian antiarchs

of Qujing, and indicated that the Yunnanolepididae are very primitive in some features of the

trunk-shieid. Since a close relationship lo the Bothriolepidoidei was uniikely. Miles (1968)

erected a new suborder Yunnanolepidoidei, of cqual rank to the Bothriolepidoidei.

Deiailed descriptions of lhe Early Devonian antiarchs from Qujing, Yunnan (mainly from

the Xitun Formation of the Cuifengshan Group) hâve not been published until 1978. The works

of K. -J. Chanc (1978), P’AN & Wang (1978), G.-R. Zhang (1978, 1979) and M.-M. Zhang

(1980) showed that the pectoral fin and fin articulation of the Yunnanolepididae fill the mor-

Fig. 1. — Sketch inap showing the localitie.<; of Early Devonian antiarchs in China. 1. Qujing, Yunnan; 2. Zhaotong, Yunnan;

3. Wuding. Yunnan; 4. Luquan, Yunnan; 5. Wcnshan, Yunnan; 6. Liujing. Hengxian. Guangxi; 7. Dale, Xiangzhou. Guangxi;

8. Piiiglc, Guangxi; 9. Xindu, Hexian. Guangxi; 10. Wudang, Guiyang, Guizhou; 11. Longmenshan, Sichuan.

— 238 —

phological gap between euanliarchs and other placoderms. YOUNG (1984c) stated that the Early

Devonian antiarchs of South China play a very important rôle in the discussion of the antiarch

phylogeny and paleobiogeography. G.-R. ZHANG (1984) describcd an antiarch, Procondylolepis,

from the Xitun Formation of Qiijing, which beats a more derived pectoral fin articulation than

the Yunnanolepididae- The taxon Procondylolepiformes was erected to distinguish il from the

Yunnanolepiformes. It is shown below that, Pmcondyloleins. as well as Qujtnolepis, is the junior

synonym of ChuMmdepis.

Janvihr (1995) described iwo antiarchs, referred to as sinolepids, from the Xitun Formation

of Qujing which were collected during the field trip of the Farly Vertebrate Symposium held

in Bcijing (1987). These two antiarchs beat the large rectangular fenestra in the ventral wall of

the irunk-.shicld, like the Sinolepididac, however, lheir pectoral fin articulations are most sug¬

gestive of thaï of the Yunnanolepididae.

The Farly Devonian antiarchs of Qujing are best preserved in the Xitun Formation of the

Cuifengshan Group, but ZHU et al. (1994) pointed out that the antiarchs were aiso abundant in

the other three formations of the group (Xishancun, Guijiatun and Xujiachong Formations), and

recorded )'. ebii and Chmiiiiwlepis IProcondylolt'pi.'!) qujingen.ii.'i from the Guijiatun Formation.

Z.-S. Wang (1994) describcd Hetemyunnanolepis qujingensis from the Xishancun Formation.

Minicrania lirouyii (Zhu & JANVIER 1996). a very tiny yunnanolepidoid-like antiarch, is also

found in this formation.

Il is suggesled by Zhu & Wang (in press) that the Chuandong Formation of Qujing is

mainly of Emsian in âge. and Wudinulepis has never been reported in this horizon.

2) From the Pusongehong Formation of Zhaotong, Yunnan, were described galeaspids (Y.-H.

Liu 1973; Pan & Wang 1981; Wang & ZHU 1994) and an onychodontiform (ZHU & Janvier

1994). A yunnanolepidoid-like imnk-shield wiih pectoral fins attached, as well as a porolepiform

and a possible petalichthyid, was discovered in 1991 and is under study (Wang ei al. in prep.).

3) Il is argued that the Jiucheng Formation of Wiidtng. Yunnan. is of Emsian âge (Wang

& ZHU 1995). Wudinolepis (K.-J. CHANG 1965), in addition to abundant arthrodires (Liu &

Wang 1981; Wang & Wang 1983, 1984), has been described from this formation. More an¬

tiarchs, including two new forms, hâve been collected during the pasi years (S.-F. Liu in prep.).

4) An Early Devonian euantiarch Luquanolepis was described from Luquan, Yunnan (ZHANG

& Young 1992). In 1991, Wudimdepis was also found in the same locality, and it is not yet

clear whether il is from the same horizon as Luquanolepis.

5) Many Early Devonian antiarchs hâve been collected from the Pusongehong Formation

of Wenshan, Yunnan (J.-Q. WANG pers. comm.). These specimens hâve not been described.

6) P’ AN (1973) described an Early Devonian antiarch, Kwang.silepis kwangsiensis, from the

Lianhuashan Formation of Liujing, Hengxian, Guangxi. This was the first report of Early

Devonian antiarchs from the Guangxi Province. The referred material included an internai mould

and several detached plates of the Irunk-shield, which probably belong to Yimnanolepis chii

(S.-F. Liu 1992). S,-F. Liu (1974) reported Yunnannlepis sp, from the Lianhuashan Formation

of Liujing, and compared the Lianhuashan Formation to the Xitun Formation of Qujing, Yunnan.

In 1978, P'an & Wang described two new antiarchs (Orieniolepis iieokwangsien.sis and

Lianhuashanolepis liukiangen.'ii.s) from the same horizon and locality. G.-R. Zhang (1979) and

— 239 —

S.-F. Liu (1992) later poinled out lhat Lianhuaslianolepis was a junior synonym of Yunnanulepis,

whereas Orientolepis was the Junior synonym of Chuchinolepis. S.-F. Liu (1992) described and

reviewed the antiarchs from the Lianhuashan Formation of Liujing, i.e. Y. chii, C. gracilis and

Zhanjilepis sp., as in the Xitun Formation of Qujing, Yunnan.

7) S.-T. Wang (1987) described an anliarch, Liujiaiigolepis suni, from Dale, Xiangzhou,

Guangxi. The fish-bearing horizon is referred to as the “Nagaoling Formation” (aiso the Xiaoshan

Sandstone) of Latc Gedinnian or Early Siegenian âge by S.-T. Wang ( 1987), and is about 100 me-

ters below the Tonggeng Formation, which roughly corresponds to the Pojiao Formation with

its typical “Euryspirifer tonkinensis-Ûicnelostrophia punctata" fauna. This fish-bearing bed is

aIso incorporated into the Dayaoshan Group (Hou et al. 198Sb). Liujiangolepis turns out to be

a sinolepid (RiTCHIE et ai 1992 362); although it shows many similarities with the Yun-

nanolepidoidei (S.-T. WANG 1987).

8) S.-T. Wang in Ritchib et ai. (1992) described another sinolepid, Dayaoshania youngi,

from Pingle, Guangxi. The horizon is the upper part of the Dayaoshan Group, for which RirCHlE

et ai. (1992) provisionally proposed an Emsian-Eifelian âge. Httwever, by comparison to the

horizon of Liujiungoiepi.'i from the neighbouring région, which is aiso at the top of the Dayaoshan

Group, it is reasonable to as.sume the âge of Dayaoshania is no younger than the Emsian.

9) P'AN in P'AN & Wang (1978) described Hohsienoiepis hsintuensis from Xindu, Hexian,

Guangxi. The horizon was referred to as the Xindu Formation, of Middle Devonian âge (P’AN

& Wang 1978, 331; aiso see P’AN 1981; Pan & DlNF.t.EY 1988). In the same volume, P’AN et

ai. (1978, 250) assigned the fish-bearing horizon to the “Yujiang Formation”, of Early Devonian

âge. Even though their “Yujiang Formation” is not the exactiy like the type Yujiang Formation

and may include part of Middle. Devonian sédiments, the âge of Hvhsienoiepis is dcfinitely no

younger than the Emsian, since il occurs at the bottom of the “Yujiang Formation” which overlies

directly the Nagaoling Formation.

10) P’AN et al. (1978) reported Early Devonian antiarchs from the Wudang Formation of

Guiyang. Guizhou Province, which are very suggestive of Yunnanolepis. Associated are the

galeaspid Neodnyatuispis minuta, the arthrodire Kueichowiepis sinensis and the petalichthyid

Sinopetalkhthys kueiyangensis (P’AN et al. 1975; P'AN & WANG 1978).

11) The Early Devonian vertebrates from the Pingyipu Formation of Longmenshan, Sichuan,

were first reported by Y.-H. LtU (1973), including the galeaspid Sanqiaspis rostrata and the

petalichthyid Neopetalichthys yenmenpaensis. P’AN et al. (1975) and P’AN & WANG (1978) de¬

scribed two other galeaspids (Lungmeiishanaspis kiangyouen.si.s and Sinoszechuanaspis yanmen-

paensis) and a petalichthyid Xinanpetuiichîliys shendaowanensis. P’AN et al. ( 1978 ; 246) reported

on the detached plates of antiarchs found in the same horizon. The formai description of the

antiarchs from the Early Devonian of Longmenshan has been made by S.-T. Wang in HOü et

al. (1988a). who described two antiarch généra (Chuanbeiolepis and Yunlongolepis). Comments

on these two généra will be made below.

MATERIAL AND METHODS

As generally acknowledged. the Early Devonian deposits of Qujing (Yunnan, China. Fig. 2)

are relcrred to as lhe Cuifengshan Group and can be siibdivided into ihe Xishancun. Xitun,

Guijialun and Xujiachong Formations, in ascending order (ZHU et al. 1994; Fig. 2Cj. Zhu (in

prep.) argued thaï ihc overlying Chuandong Formation is mainly Emsian (Fig 20, insTead of

Middie Devonian. Ail lhe matcrial used in the préparation of ihis work cornes from the Cuifeng¬

shan Group of Qtijing, mainly the Xishancun and Xitun Formations, and is housed in the Ipstitule

of Vertcbratc Palcontology and Paleoanthropology, Chinese Academy of Sciences, Beijing. Some

specimens of ihc Xitun Formation were collected by Prof. M.-M. Chang, whose contributions

to this work arc acknowledged.

The spécimens hâve been prepared by the standard mechanical methods. In some cases,

the acelic acid technique has been used on specimens from the Xitun Formation. The material

from the Xishancun Formation was preserved as internai and/or extemal nioulds, and elastomer

casts were made after the mechanical préparation.

The line drawings in the test were made by means of the caméra-lucida. The abbreviations

used in the figures and text are mainly after the Works of Miles (1968) and YoUNG (1988).

TERMINOLOGY

The placoderm classification of Denison (1978) is used here, i.e. lhe Antiarcha is ranked

as one of the orders of the class Placodermi.

As 10 the classification of lhe Antiarcha. ihe Works of Mn.ES (1968), Young (1984c, 1988)

and Ritchte e! al (1992) hâve been used as a référencé, and the major subgroups are ranked

as suborders. The monophyly of the suborder Bothriolepidoidei sensu Zhang & YOUNG (1992)

was provisionally acknow ledged before the phylogenelic analysis of antiarchs in this paper. The

Microbrachiidae and Bothriolepididae are iwo widely recognized families of the Bothrio¬

lepidoidei. The family Asterolepididae is a repre.sentative ol the suborder Asterolepidoidei. The

Sinolepididae (Mil. ES 1968) is a Justified emendalion of the Sinolepidae (LlU & P’AN 1958;

Ritchie et al. 1992) which is an incorrect original spelling in accordance with the International

Code of Zoological nomenclature (Ride et al. 1985, Art. 29A, Art. 32c-d. Art. 33b). However,

according to the Code (Art. Ilf, Art. 33b). the namc thus corrected retains the author and date

of the original spelling (Sinolepididae Lit.' et P'AN 1958).

Since the Antiarcha is ranked as an order, the order Yunnanolepiformes (a subgroup of the

Antiarcha), commonly used in the recent lilerature, is replaced hcre by the suborder Yun-

nanolepidoidci Miles, 1968. In this work. lhe Yunnanolcpidoidei comprise the Yunnanolepididae,

Chuchinolepididae, and related forms. The Yunnanolepididae takes lhe authorship and date of

Grüss 1965. According to the Code (Ride et al. 1985, An. 5()c). the change in lhe rank of a

taxon wilhin the family group (subfamily and family) does not affect the authorship of the name

of the taxon. In the same way. the Chuchinolepididae is altributed to K.-.I. Chang 1978.

— 241 —

Fig. 2. — A. hkelch map of China, .showing the position of Qujing, where ihe malerial under study was tollecled; B, geological

sketch map of the Qujing area in south-easlern Yunnan (modified front Fa.ng et ni. 1985, fig. I); C, synthetie log of the

Early Devonian in Qujing. Yunnan. China. Sjg - Guandi Formation, Sun Miaogao Formation, Sjv - Yulongsi Formation,

Dixs - Xishancun Formation. Dixt - Xitun Formation. Dig - Guijiatun Formation. Dixj - Xujiachong Formation. Dic -

Chuandong Formation, Rrh Haikou Formation. Di.i- Chuandong and Haikou Formations.

200 m

— 242 —

As to the terminology for niorphological characlers. SXENSlô’s (1948) nomenclature, as mod-

ified by Miles (1968) and YOUNG (1988), is adopted here with minor changes. The terminology

of the dermal plates of the pectoral fin in the Chuchinolepididae is discussed below.

SYSTEMATIC DESCRIPTIONS

Class PLACODERMI M’Coy, 1848

Order ANTIARCHA Cope, 1885

Suborder YUNNANOLEPIDOIDEI Miles, 1968

EMENfÆD DEFINITION. — Antiarcha in which lhe skull-roof is broad and .short, with a broad latéral plate;

anterior margin of the AMD plate pointed.

Remarks. — Since the characters such as the absence of the brachial process, the large preorbital dépréssion,

are plesiomorphic for antiarchs, they are not included in the diagnosis of the suborder. These symplesiomorphies

are aiso présent in Minicrania (Ziiu & Janvier 1996). None of the characters in Ihis définition are unique to

this taxon. Ail appear at varions levels of lhe Euantiarcha, as homoplasies.

Family Yunnanolepididae Gross, 1965

Emended diagnosis. — Yunnanolepidoidei in which the crima tnmsversatis interna posterior turns forward

on the viscéral surface of the PMD plate, and lies in front of the posterior ventral process and pit.

Type genus Yimnanolepis Y.-H. Liti, 1963

Referred GENERA. — Phyniotepis G.-R. Zhang, 1978; Mizia n. gen.

Rr.MARKS — Many diagnostic characters of lhe Yunnanolepididae (K.-J. Ciiano 1978; G -R. Zhang 1978)

turn OUI to be the symplesiomorphies. such as lhe preorbital dépréssion, and lhe large posipineal plate which

excludes the tiuchal plate from the orbital fenestra. Therefore. a more restricted diagnosis based on unique

characters is given here.

The t rixia transveixalis imernu poxterior of antiarchs W'ould hc homologous to the Iransverse ventral crest

of the médian dorsal plate of olher placoderms, if we accept Y.-H Ltr's suggestion (1991) thaï the PMD plate

of anliai'clis is homologous to the médian dorsal plate of other placoderms. In outgroups of antiarchs, e.g., ar-

throdires and peiaJichlhyids. lhe Iransverse ventral crest generally lic.s latéral to the médian ventral process, which

could be a.sîigned to the plesiomorphic state for antiarchs, Among antiarchs, this ancestral State is seen in Zhaii-

Jilepix (G.-R. ZiiaNu 1978), Minkrumt tZiii,‘ & J.anvier |996), and ChiuUinoU'pix. In the Sinolcpididae, ex-

emplificd hy Crenfellaxpix and Xichoimlepix (G.-R. Zhang 1980; Ritciiie et al. 1992). lhe trixui tranxverxalix

interna posferiur lies slightly in front of the posterior ventral proce.s.s. However, lhe condition in lhe .Sinotepididae

is different from Ihat in the Yunnanolepididae. The crisla Irnnxverxalix interna poxterior in lhe ,Sinolepididae

runs domally wilhom a twist ami lias lhe sanie palh as thaï in Zhanjilepix, Minicrania and Chuchinoirpix, w'hereas

that in lhe Yunnanolepididae iiiriis forward on the viscéral surface of the PMD plate.

In the Yunnantdcpidoidei, the lalcroveniral fossa on lhe viscéral surface of lhe Irtink-shield is found to lie

at the junclion of ihree piales (the AVL, PL and PVL plates). The portions of lhe trunk-shicld which encircle

the fossa are thickened. implying lhe aitaehmeni of some mii.scle.s in lhe fussu. This dépression was meniioned

by G.-R. Zhang (1978) in tlie AVL pLate of Y. chii. ,'siiice then, it had been overlooked until Tong-dzuy &

jANVir.R ( 1990) described the so-called “latéral thoracic recess " m the internai Irunk-shield mould of Yimnanolepis

cf. Y. parvus. However, this tossa is aIso found in Zhanjilepix. The Yunnanolepidoidei Heteroyunnanolepis lacks

this unique structure.

Genus YUNNANOLEPIS Y.-H. Liu, 1963

Yunnanolepis Y.-H. Liu, 1963: 39, fig. 1, pl. I,

Synonyms:

Eoantiarchilepis P’an & Wang. 1978: 319, fig. 10, pl. 31.1-2.

Tsuifengshanolepis P’an & Wang, 1978: 320, pl. 30.2.

Lianhuashanolepix P’an & Wang, 1978: 323, pl. 32.1-5.

Olhers référencés:

Yunnanolepis K.-J. ChanG 1978: 293, pl. 25.1-4. — G. -R. ZHANG 1978: 148, figs 1-7, pis 1-

IV, V.3-6. — Tong-Dzuy & Janvier 1990: 159, pis II-IV. - S.F. Liu 1992: 212, pl. I. — Zhu,

Wang & F.AN 1994: 3, pl. 1.1-3.

Emendf.u niAGNo.si.s. — Yunnanolepididae in wliich the opening of the Ciianc’s apparatus is visible in latéral

aspect.

Type specjes, — Yimnanolepis chii Y.-H. Liu, 1963.

Remarks. — The opening of the Cha.no’s apparatus (a name erected here after Prof. M.-M. Chang) was

first figured by M.-M. Zhang (1980. Fig. 3), and neither the de.scription nor the explanalion of this strange

opening was given. As described below. the Cha.vo’s apparatus consist.s of an internai cavity and au external

opening winch straddles the suture between the ADL and AVL plates Since the CnANfi’$ apparatus is compleiely

separated front the inlerior of the inink-shield. tl cannoi funclion as a mechaiioreceplor System, hke the sensory-

line System which pénétrâtes the derinal plates. However, Ihis structure may function as an espanded ampullary

electoreccptnr and its opening is covered hy the epidermis. Another e.xplanation is that the Cn tsxi's apparatus

was glandular and had a rôle in mucus .sécrétion. The gohlct cells and compound mucus glands would hâve been

inlegrated in the cavity of the Chang's apparatus. Ainong the Yunnanolepididae, only Yunmmolcpis and Phvmolfpis

share the Cha.ng’s apparatus. The différences bctwecn theni am following:

1) PhymoU’pis retains the AL plate which hides the opening of the Chang's apparatus front the outside;

2) In Yunnunalepis the postérolatéral ntargin of the AMD plate is a.s long as the antérolatéral margin,

whereas in Phymolepis the postérolatéral ntargin is obviously longer the antérolatéral margin;

3) the PMD plate of Phymolepis bettrs a strong posterior process.

Yunnanolepis chii Liu, 1963

(Fig. 3)

Yunnanolepis chii Y.-H. Liu, 1963: 39, fig. 1, pl. I.

Synonyms:

Eoanliarchilepis xiiunensis P’an & Wang, 1978: 319, fig. 10, pl. 31.1-2.

Tsuifengshanolepis dianlungensis P’an & Wang, 1978: 320, pl. 30.2.

Lianhuashanolepis liukiangensis P’an & Wang, 1978: 323, pl. 32.1-5.

Others référencés:

Yunnanolepis chii K.-J. CHANG, 1978: 293, pl. 25.1-4. — G. -R. ZHANG 1978: 148, figs x 1-7,

pis I-IV, V.3-6. — S. F. Liu 1992: 212, pl. I. — Zhu, Wang & Fan 1994: 3, pl. 1.1-3.

Emended diagnosis. — A Yiiniumolepis with a head-shield length of at least 3 cm; AMD plate with a

central élévation; .semilunar plate rectangular and broad.

— 244 —

Hoi.otype. — A complété head-shield preserved ventrally and its niould, V2960.1 and V2960.2 (Y.-H. Liu

1963, Fig. I, pl. I).

Loc ality and horizon. — Qujing (Chutsing). Yunnan. Cuifengshan Group, Early Devonian.

OiHUR coNcfcRw.ti MAIEKIAU - ADL plaie, V442.3.U) (G.-R. Ziiano 1978, pl. III, 1-2); AVL plate. V4423.I8

(ihid, pl. IV, 3-6); PL plate. V4423.23 Uhul.. pl. III, 3-4); PVL, V4463.30 {ihitl.. pl. III. 6).

RrMARK.t. — Muny diagniisiic characters of Yimruiiioiriiis chii (Y-H. I.io 1963; K.-J. Chano 1978; G.-R.

Zhang 1978) turn oui lu be plesioitiorphic. Some of Ihein, such as Ihc position ttf the crisUi Ininsversalis interna

posterior and the he.vagonal head-shield are synapomorphie.s of the higher taxonomie units, wherea.s those like

the preorhiial dépréssion and the simple pectoral joint are .sympicsiomorphies even for antiarchs. This species

has been described in detail by Y.-H. Lm |1963) and G.-R. Zhang (1978). However. since Y. chii is the type

spccies of Yunnanolepis. a supplementary description is noleworthy, even if mainly made on the basis of new

observations on the specimens figured by G.-R. Ziiano (1978). >'. chii differs front the other Yunnanolepis species

by its size. (he large MV plate and an élévation at lhe centre of the AMD plate.

Description

ClIANCi's appanitus (Fig. 3A-D)

The ChanG's apparatus is poorly preserved in V4423.8. a fairly complété irunk-shield (G.-R.

Zh.\nü 1978, pl, I, 6-7), However, lhe delached a\DL and AVL plates described by G.-R. Zhang

(1978) clearly exhibit the CliANG's apparatus.

The upper margin of lhe opening of lhe Chang’s apparatus (Ca.o, Fig. 3C, D) is seen on

the venlral margin of the ADL plate (V4423.I6). slighlly behind the anterior cxlreinity of the

plate. The ridge causcd by ihe Ch.ang’s apparatus (r.Chang, Fig. 3C, D) lies above the opening

in cxternal view. Intcrnally, an additional lamina (I.Chang, Fig. 3D) lies just behind lhe crista

trunsverxiilis intenui anterior (cit. Fig 3D) This lamina extends ventrally and forms a relatively

large fossa (C-Chang, Fig. 3D). together with the latéral lamina of the plate,

The dorsal margin of lhe AVL plate (V4423.I8, Fig. 3A, B) shows an conspicuous notch

which represents lhe lowcr margin of the opening of the Chang's apparatus (Ca.o. Fig. 3B).

G.-R. Zhang (1978) described prominenl dorsal and anlerodorsal proces.ses on lhe dorsal margin

of lhe AVL plate, and compared them to those of Remignlepis. In fact, thèse iwo corners are

due to lhe opening of lhe Chang’s apparatus, and arc différent from those processes in Re-

migolepis. A ridge (r.Chang. Fig. 3B) is siluated below lhe notch in external view. As for the

ADL plate, an extra lamina (I.Chang, Fig. 3A) is seen behind the tristu ininsversalis interna

anterior (cit) and boiinds off a fossa internnlly (C.Chang, Fig. 3A). This fossa. together with

the corresponding fossa of the ADL plate (C.Chang, Fig. 3D), constituies a large, enclosed cavity

which communicates only with the outside by the opening of the Chang's apparatus and is

completely isolated from the interior of the trunk-shield.

Lite lovent ral fossa of the trunk-shield (flv, Fig. 3A. E)

The lateroventral fossa is an internai structure of the trunk-shield. and lies at the junction

of ihree plates (the AVL, PL and PVL plates). Part of this fossa has been mentioned by G. R.

Zhang (1978, 163, Fig. 6B) ou the AVL piale, and has also been figured on the PVL plate.

The lateroventral fossa consisl of three dépréssions on lhe AVL. PL and PVL plates, and thaï

of the AVL plate is the largest (f.lv, Fig. 3A, E). The plate portions encircling the fossa are

thickened, presumably for lhe atiachment of some muscles. The same fossa is found in Yun¬

nanolepis cf. Y. chii (Tong-Dzuy & Janvier 1990. Fig. 15).

— 245 —

Fig. 3. — Yunnanolepis chii Liu, Xilun Formiüion, Qujing. A-B» leti AVL piale (V4423.27) in vi.scerni (A) and latéral (B) views;

C-D. left ADL plate (V4423.23) in dorsal (C) and latéral (D) views; E, right PVL plate (V4423.30) in vi.sceral view. (Scale

bar 10 mm.)

Yunnanolepis porifera n. sp.

(Figs 4-7; pis I, 1-7; II, 1-7; III, 1-8)

Yunnanolepis parvus pars G.-R. Zhang, 1978; 164, pl. V, 3-6

“forni probably akin to Y. chii” (pars) V4424.5, V4424.20, M.-M. Zhang, 1980: 179, 185,

fig. 3, pis 111, 2-4, IV, 2-4, V2-3.

Diaonosis, — Yunnanolepis in which the PMD piale has a cone-shaped posterior dorsal angle.

Etymoi.ogy. — From parus (Lai.), “pore", by référencé to the opening of the Ciiano's apparatus.

Holotype. — V4424.20 (see M.-M. Zhang 1980, pl. III, 2-4, fig. 3).

Locality and horizon. — Qujing, Yunnan, Xishancun Formation and Xilun Formation, Early Devonian.

— 246 —

New material. — VI0499.1-3, irunk-shields; VI0499.4-20. AMD plates; V10499.21-23, PMD plates;

V10499.24-28, ADL plates; V 10499.29-34. AVL plates; V 10499.3.3-37, PDL plates; VI0499.38-42. PL plates;

V10499.43-44. PVL plates; V 10499.45, MV plate. The spccimens who.se number begins with V 10499 are front

the Xishanciin Formation, and V 10.507 (V10507.I-7) from the Xitun Formation.

Remarks. — By now, four species hâve heen referred to ïiiiimmolef’is ()' chii, Y. hachoensis. Y. deprati

and K porifera). The Cii.wi's appuraliis has iiol bcen dcscribcd in the Iwo Vietnam species K hachoensix and

Y. depraii (Tosti-DztiY & Janvier 1990). howevet. it can be inferred from the notch and lhe anterior vertical

ridge of the ADL plate. Y pnrifent resembles hoth K, dtproti and )' hacboeiisis hy its .small MV plate. As lo

the plaie proportion. T porifera is more sugge.slive of f'. drpraii. Flowever, lhe trunk-shicld of T depruti is quite

diffcrenl from lhal of Y. porifera in lacking ridges.

y. poriftra frop) lhe .Xilun Formaljon has heen dcscribcd under lhe namc of F. paniix (C.-R, Zhang 1978;

M.-.M. ZiiANti 1980). Il IS fomid lhal lhe hololype of Y parvux (V4424.3) definiiely differs from the ofher spéci¬

mens which weie leferred to as V. /mnn.ï by G.-R. Zhang (1978). V4424.3 bas a diflercni skiill-roof front lhal

of K chii. and ils Irunk sliield Ls mosl suggeslivc of lhal of Miria puni n. g., n. sp.. therefore Ihis hololype

should be more reasomihly referred lo as Mi'Ja parvus ruther iltan K parvici .Siibseqiienlly. Ihose spccimens

removed from Mi:io parvux should be referred to a nevv species of Yiainanalcpix. Uniil now. the head-shield of

Y. porifera. as well as lhal of K depraii or Y. hachoenxix, is slill unknown lo us. The specimens of Y. porifera

are firsi de.scribed here from the Xisltuncun Formation, providing us wiih more information about the shape of

its semilunar plate.

Description

Material from the Xitun Formation (Figs 4, 5A, 6; pis I, 1-7; II, 5-7)

The material of }' porifera n. sp. from the Xitun Formation has heen descrihed by G.-R.

Zfiang (I97S) and M -M. Zhang (1980). Complementary notes are given here in order to clarify

several features.

K porifera is a small-sized yimnanolepid (Fig. 4), which is characterized by its sharp post-

erior dorsal angle of the PMD plate (pda, Fig. 4B). The skull-roof of K porifera cannot be

ideniified from the material of the Xitun Formation. The trunk-shield of Y. porifera is somewhal

higher than that (tf K chii. However. its relative height is variable. V10507.I and Vl()507.2

(pl. I, .3-5) hâve a relalively low trunk-shield in comparison to lhe holotype. The AMD plate is

gencrally ornamented w'ith eight faim ridges radiating from the tergal angle. The tubercles on

ridges are slighlly larger lhan lhe olhers. V 10507.1 (pl. F 3) has ils dorsal wall partiy eroded

to expose lhe internai mould. Il is observed that lhe anterior ventral process and pit are situated

jusl beneaih lhe tergal angle, and the crista tranxversalix interna posterior lies in front of the

posterior ventral process and pit. V 10507.3 (pl. I. 1-2) is more or less distinctive from other

specimens. It is simihu to the oiher spccimens of Y. porifera in size, high Irunk-shield and

structure of the ChaNG's apparalus. However, this trunk-shield is devoid of radiating ridges on

the A.MD plate and probably lacks lhe sharp posterior dorsal angle of lhe PMD plate, as inferred

front lhe visible portion. Sincc only this spccimen is availablc, V 10507.3 is provisionally thoughi

to represenl an individual varialion of K porifera. V10507.7 is a fragment of ihc AMD and PDL

piales (Fig. 5A). Us AMD plate is genlly arched and lacks ridges in exlernal view, as in VI0507.3.

In viscéral view (pl. Il, 5), lhe médian ventral ridge inivr, Fig. 5A) extends from lhe anterior

médian proce.ss l’orwards and backwards.

The opening of lhe CiiANG’s apparatus (Ca.o. Figs 4A-C, 6A1, 3C), which lies close to

the anterior extremily of the trunk-shield and al the suture between the ADL and AVL plates,

has been sketchcd by M.-M. ZHANG ( 1980). Several specimens of K porifera hâve been prepared

and examined in order lo cxplain the funclion of lhe Chang’s apparatus. It has been shown that

— 247 —

Fig. 4. — Yunnanoîepis porifeni n. sp.. Xitun Formation. Qujing. Restoralion of the Iriink-shicld in anterior (A), latéral (B),

dorsal (C) and ventral (D) views (moditicd from M.-M. Zhang 1980, fig. 3) (Scale bar 5 mm.)

the cavity ot' the Chang’s apparatus (C. Chang, Fig. 6A3) is a blind tube which extends

downwards and slightly upwards. In external view, this tube produced a vertical ridge (r.Chang,

Fig. 6A1, B) along the anterior margin of the trunk-shield. Behind this ridge is a vertical groove.

In comparison to that of K chii, the cavity of the Chang’s apparatus of Y. porifera is smaller

and narrower. somewhat like that of Phymalepis. In contrast to Phymolepis, the opening of the

Chang’s apparatus of Y, porifera is directiy exposed to the oiitside and there is no anterior

latéral plate.

The laterovcntral fossa of the trunk-shield, at the junction of the AVI., PL, PVI. plates, is

as in Y. chii. In V 10507.5, a portion of the lateroventral fossa is clearly seen at the posterior

end of the AVL plate in viscéral view (f.lv, Fig. 6A2).

As to the pectoral fenestra, K porifera has a similar structure to other yunnanolepids and

Minicrania (Zhu & JANVIER 1996). Noteworthy is an AVL plate, probably belonging to

Y. porifera. which has its pectoral fenestra filled by the perichondrally ossified scapulocoracoid

— 248 —

(scap, Fig. 6B-D. pl. I, 6-7). This scapulocoracoid. extending from the bottom of the pectoral

fenestra, is conical in sliape and lias its outmost end oval in .section. In contrast to the dermal

skeleton, its outer surface is smooth and devoid of ornamentation. Its mesial surface is closely

in contact with the latéral wall of the plate.

Material from the Xislumcun Formation (Figs 5B-G, 7, pis II, 1-4; III, 1-8)

Numcrous spccimen.s of Y. porifera, including three articulated trunk-shicids. hâve been

collected from the Xishancun Formation. Forty five of them arc numbered for the study. and

the isolalcd plates arc idcntilïcd hy comparison with the articulated trunk-shields, in addition to

their si/.e and ornamentation. The material from the Xishancun Formation is very similar to that

from the Xitun Formation, cither in shapc or in si/e. li bcars the vcntrolateral l'ossa and Chang's

apparatus as thaï from the Xitun Formation. The ornamenl is composed ol stiiall. closely sel

tubercles. Its dorsal wall is slightiy narrower lhan the ventral wall. and the latéral wall is relatively

high. The dorsal wall bears a sharp postenot dorsal angle on the PMD plate, and the médian

ventral plate is fairly small.

The ,MV1D plate (Fig. 5B-G) is roughiy rhombic in sbape, with a pointed anterior margin

(a, Fig. 5C-D). Us posterior margin (oa.PMD, Fig. 5C-D), which is by the PMD plate, is fairly

narrow. The plate shows apparent latéral corners (le, Fig. 5C-D), and its antérolatéral margin is

slightiy longer lhan the postérolatéral one. In external view, the plate is gently arched at the

tergal angle (dma. Fig. 30, which is almost haliway along the mid-line. Eight fainl ridges radiale

from the tergal angle to the margins, and the small tubencles are getling a little laiger on the

ridges The short posterior oblique abdominal pit-line grooves (dig, Fig. 50 are somelimes vis¬

ible behind the tergal angle. The overlap rclalionships with the adjoining plates are normal, as

in olhcr yunnanolcpids. In viscéral view, therc exisl many individual variations as to the develop¬

ment of the anterior ventral process (prvi, Fig. 5B. E. G), médian ventral groove (grm. Fig.

5B, F) and ridge (mvr. Fig. 5G). In general, the AMD plate of Y. porifera bcars ihc anterior

ventral process and pit (prvl, ptl. Fig. 5B) as in F chii. jitsi beneath the tergal angle. Ilowever,

in some spccimens, e.f>. V 10499.6 (Fig. 5F), the plate is devoid of the anterior ventral process

and pit. Behind the anterior ventral process, or in the corresponding position relative to the

tergal angle, the plate generally displays a médian ventral gioove (grm, Fig. 5F), contrary to the

AMD plate of y. chii (G.-R. ZHANG 1978). The exception is scen in V 10499.12, which lacks the

médian ventral groove and has the médian ventral ridge extending to the posterior end of the plate

(mvr. Fig. 3G), like that of Y. chii. In front of the anterior ventral ptocess. or in the corresponding

position relative to ilic tergal angle, the plate is generally devoid of any ridge or groove. Ilowever,

a short médian ventral ridge is found in some .spécimens, such as VI0499.II (pl. 111. 8).

The PMD plate is characterized hy its dorsally projecting posterior dorsal angle (pda.

Fig. 7B). In viscenil view, the cm/n iransverxalis interna posterior (cr.tp) is developed in front

of the posterior ventral process and pit (prv2, Fig. 7F,).

The ADL plate (pl. III, 4) is subdivided into the dorsal and latéral laminae by the dorsolateral

ridge of the trunk-shield. On the dorsal lamina, the articular fossa lies beneath its broad anterior

margin. On the latéral lamina, the ridge and groove caused by the Chang's apparatus are seen

next to its anterior margin. Along the ventral margin there is the notch of the opening of the

Chang’s apparatus, as in Y. chii, The main lateral-line groove runs through the plate below the

dorsolateral ridge.

— 249 —

Fig. 5. — Yuniwiioleiiis porifera n. .sp., Xitun Formation (A) and Xishancun Formation (B-G), Qujing. A, incomplète AMD and

PDL piales in viscéral view. VI0507.7; B-G. AMD piales (elaslomer casls). B. VK)499.18 in viscéral view; C, V10499.16

in dorsal view; D. VI0499.4 in dorsal view; E, 10499.7 in viscéral view; F. VIl)499.6 in viscéral view; G. V10499.12 in

viscéral view. (Scale bar 5 mm )

The PDL plate (Fig. 7D, pl. III, 5) aiso comprises dorsal and latéral laminae. The dorsal

lamina has a conspiciious dorsal corner. The latéral lamina is fairly low and, along its ventral

margin. it overlaps the Pl. plate anteriorly and is overlapped by the PL plate posteriorly. The

area which is overlapped by the PL plate (oa.PL. Fig, 7D) extends ventrally lo form a process

near its posterior end.

The PL plate (F'ig. 7F) is elongated, low', and posteriorly descending. The crixta trcmxversalis

interna posterior (cr.tp, Fig. 7F) is situated next lo the posterior end of the plate. In viscéral

view, al the anierovenlral corner of the plate, therc is a shallow dépression which is pan of the

laterovenlral fossa of the irunk-shicid (f.lv, Fig. 7F).

The AVL plate consists of the latéral and ventral laminae. Since therc is no detached semi-

lunar plate which could be dctlnilely assigned to Y. porifera. the shape of the SL plate is inferred

from the corresponding notch in the anteromesial margin of the AVL plate. As the AVL plates

of Y. porifera from the Xitun Formation are generally broken at their anterior end. those from

the Xishancun Formation are worthwhile in their completeness. The latéral lamina is relatively

low. Anteriorly, the cavity of the Chang’s apparatus can be inferred from the internai mould

of the plate, The plate has a conspicuous prepecloral corner, which is continuons with the

— 250 —

Fio. 6. — Yunnaiiolepis porifera n. sp., Xilun Fonnalion, Qujing. A. righl ADL and AVL plaies in latéral (Al) and poslerior (A2)

views, A3 represenis Ihc section through the Chang’s apparatus, V 10507.5; B-D. lefi AVL plate In latéral (B), ventral (C)

and po.sterolateral (D) views, VIO.507.4; E, incomplète right AVL plate in dorsal view, V 10507.6. (Scale bar 5 mm.)

independent spinal plate. The anterior extremity of the plate is the antérolatéral corner (c.al,

Fig. 7A), from which the anterior division of the mesial margin extends posteromesially and

straight. According to this portion of margin, it is reasonable to assume that the semilunar plate

of Y. porifera is more or less triangular in shape, unlike that of Y. chii. The posterior division

of the mesial margin, along which the plate overlaps the MV plate (cf. MV, Fig. 7G), is com-

— 251 —

Fig. 7. — Yunnanolepis porifera n. sp.. Xishancun Formation, Qujing. A. irunk-shield in dorsal view (inlcrnal moiild of Ihc dorsal

and luicrui walls. and cxicmal mould of tlie ventral wall), V 10499.3; B, trunk-shield in ventral view (interna) mould of lhe

ventral wall and external mould in dorsal wall). V 10499,1; C, trunk-shield in venüal view (internai mould of ventral and

latéral walls), V 10499.2; D, PDL plaie in exiemal view (elastomer casi). V 10499.35; E, PMD plate in viscéral view (elastomer

casl), V1Ü499.23; F, PL piale in viscéral view (elastomer casl), V10499.38; G-H. lefl AVL and Sp plates in viscéral (G) and

external (H) views, V 10499.37; I. MV plate in external view (elastomer casl). V 10499.45; J, lefl PVL plate in viscéral view

(elastomer cast), V 10499.43; K. left AVL and Sp plates in external view (elastomer casl). V10499.36 (Seule bar 5 mm.)

paratively short. In viscéral view, there is a posterior dépréssion which is part of the lateroventral

fossa (f.lv, Fig. 7C, G). The postbranchial lamina (l.pbr, Fig. 7B-C, G) extends slightly anterome-

sially from near the small pectoral fenestra, and behind it is the transversely directed crista

transversaux interna anterior (cit, Fig. 7B).

— 252 —

An independent Sp plate is présent in K porifera. In general, this plate is fairly small but

bears a conspicuous suture with the AVL plate as in V 10499.37 (Fig. 7H). In V 10499.36, this

plate is somewhat enlarged, more or less similar to the small spinal plate of some arthrodires

(Fig. 7K. pl. II, 4).

The PVL plate (Fig. 7B-C) bas a low latéral lamina and relatively narrow ventral lamina,

whose contact margin with the médian ventral plate is fairly short. In viscéral view, the component

of the lateroventral fossa (f.lv, Fig. 7.1) is seen at its antérolatéral corner. The crista transversalis

interna posterinr (cr.tp, Fig. 7J) is situated nexl lo the posterior end of the plate.

The MV plate (Fig. 71) is fairly small and rhombic in shape. It is overlapped by the AVL

and PVL plates.

Yunnanolepis sp.

(Figs 8-9, pl. II, 11)

Vanchienolepis sp. Janvier, 1995: 153, MNHN-CHDOl, Fig. 6.

New matf.riai.. — A trunk-shield a.ssociated with part of skull-roof, V10514.

Locality anu horizon. — Qujing, Yunnan. Xitun Formation. Early Devonian.

Remarrs. — Janvier (IVV.S) dcscrihcd an antiarch (MNHN-CHDUI) refened to as VuncMvnoh'iûs sp. or

‘'Vanchientilfi)i\-\\kc speeinien from Qujing" (p. 157). MNHN-CHDOl eaine from the Xitun Formation of Qujing,

and is e.xeellenl in llie préservation of the pectoral fin which had the natural po.sition to the tmnk-shield. However,

its assignment a.s Vanrhiennlepis is doiibifui, since its brachial articulation is poorly preserved, and i.s more

suggestive of the .simple ont* of the Yunnanolepididae ratherihan that of Vanchienolepis. Moreover. MNHN-CHDOl

has a head-shield which is jiisi same a.s thaï of Yunnanolcpi.s, which i.s confirmed by my new speciineii and was

implied by Janvier tl9*J5). The new material dcscrihcd here came from the same bed as MNHN-CHDOl, and

was prepured to show the viscéral surface of the shield.

Thi.s antiarch is quite different fmm Vanchienolfiiis (Tonci-Dzi'v & Janvier PJyO) e.xcepi for the large

rectangular fenesira On the ventral wall of the trunk-shield. They hâve different pectoral fin structures, AMD

and ADL piales. As far as wc know, the head-.shield, the dorsal trunk-shield wall and lhe pectoral fin structure

of MNHN-CHDOl and VI05I4 are just same as ihose of Yuimunotepis. Il is teferred to Yunnanolepis by its

typical CiiANu's apparalus, whose opening, uniike that of Phymolvpis, is visible in latéral aspect.

Description

V10514 is a small individual, whose trunk-shield length i.s shorter than 15 mm (Fig. 8).

The preserved skull-roof plates include the nuchal, paranuchal and latéral plates (Nu, PNu, L,

Fig. 8). which are just same as those of other Yunnanolepis spccies. The semicircular ridges

(r.semi, Ftg. 8A) are seen on the exlernal surface of the nuchal plate.

The trunk-shield has lost its posterior margin. and pari of the ventral wall was damaged

during the préparation of the viscéral surface of the shield (Fig. 8A, B). However, it is quite

certain that its ventral wall is similar to that described by Janvier (1995), i.e. possessing a

large opening instead of the MV plaie.

Externally, the Chang’s apparatus is well-preserved. The opening of the Chanü’s apparatus

(Ca.o. Figs 8A, 9) is fairly elongated in shape and lies on a vertical ridge (r.Chang) next to the

anterior extremity of the trunk-shield. The suture between the ADL and AVL plates delineates

its upper margin. The cavity of the Chang’s apparatus is narrowed, a.s in Y. porifera. The crista

— 254 —

transversalis interna anterior (cit, Figs 8A, 9) is in front of the Chang’s apparatus and beneath

the skull-roof.

Inlernally, there is no anterior ventral process and pit on the AMD plate. Instead, there is

a short médian ventral ridge (mvr, Fig. 8B) which corresponds in position to the tergal angle

(dma, Fig. 8A). The ridge reappears next to the posterior end of the AMD plate, yet it is relatively

faint.

Genus MIZIA n. g.

Diaonosis. — Yunnanolepididae in which small tubercles on the dorsal and latéral watts of the trunk-shield

are arrangcd into regular ridges, with tiny tubercles between ridges,

Etymoloüy. — After the Chinese character (Mi, in the Chinese phonctic alphabet). Zi = word,

character; indicating the ridges on the dorsal wall of the trunk-shield looks like the Chinese character (Mi).

Type species. — Miziu Umghimenxis n. sp. (Fig. 10; pl. II, 8-10).

Reperred sprriES. — Mizui pairux (G.-R. Zhang 1978).

Remarks. Micin is referred to the Yunnanolepididae becau.se of its crisla iranxversalis interna posterior

in front of the posterior ventral process. M. langhaaensis bears an indépendant AL plaie in front of the suture

between the ADL and AVL plates and is devoid of Chanc's apparatus. As stated above, the specimen previousiy

described as the holotype of Vuniuiiwlepis parvtis (V4424.:L G.-R. Zhang 1978; also see M.-M Zhang 1980),

which has lost the antenor eAtreiuiiy of the dorsal wall of ils linnk-shield, differs front the other material assigned

to this specics by G..-R. Zhang iI978i. and ils skull-roof cicarly differs front that of L chii in the pallem of

sensory-line groove.s. ,Since ils ornamentation i.s quite suggestive of thaï of M. longiiuaensis. Y. parvus is referred

here to Miua. The anicrior région of the latéral wall of ihe Irunk-shield is poorly preserved in V4424.3, and

the available cvidence .suggests it may he similar to that in M. Itmghiiac'nsis.

Mizia longhuaensis n. sp.

(Fig. 10; pl. II, 8-10)

Diagnosis. — Mizia in which the médian dorsal ridge branches into two parallel ridges behind the tergal

angle.

Etymology. — After the fosstl locality of Longhua hill, Qujing, Yunnan.

Holotype. — A complété trunk-shield, V10515.

Locality and horizon. — Longhua hill. Qujing, Yunnan, Xujiachong Formation, Early Devonian.

Remarks. — The new species differs from M. (Yunnanolepis) parvus mainly by its médian dorsal ridge,

which branches into two parallel ridges behind the tergal angle.

Description

The holotype is a small-sized trunk-shield, characterized by its regular ridges on the slightly

arched dorsal and latéral walls. In general, the ridges converge toward the growth centre of the

plate, and Iransect the suiurcs. For example, the ridges radiate from the tergal angle (dma,

Fig. lOA, C) on the AMD plate. The tubercle.s on the ridges are small, however, they are ap-

parently larger than tho.se between ridges. The latéral wall is relatively low. Most of the bone

on the ventral wall was eroded to leave the internai mould. The ventral wall is Hat and has the

— 255 —

same breadth as the dorsal wall. It is a little longer than the dorsal wall, since its subcephalic

portion is fairly short and its poslerior end is situated slightly in front of the level of the posterior

extremity of the dorsal wall.

The AMD plate (AMD, Fig. lOA) bas a length/breadth index of about 125, and is pentagonal

in shape. The latéral corner is distinct. The antérolatéral margin is slightly longer than the post¬

érolatéral margin. The posterior rnargin is relatively broad and about two fifth of the plate breadth.

The plate is somewhat arched with the tergal angle at the level of the latéral corners. In extemal

view, the plate is adorned with radiating ridges. which are fairly high and delimit grooves. The

development of the médian dorsal ridge (dmr, Fig. lOA) in M. longhuaensis is not typical as

in other yunnanolepids. In front of the tergal angle, the ridge becomes broader toward the anterior

edge of the trunk-shield. Behind the tergal angle, the médian dorsal ridge diverges into two

Fig. 10. — Miva longhuaensis n. g., n. sp.„ Xujiachong Formation, Qujing. Trunk-shield (holotype, VI0515) in dorsal (A), ventral

(B) and latéral (C) views. (Scale bar 5 mm.)

— 256 —

parallel ridges extending to the PMD plate (part of the ridges were damaged during the pré¬

paration, however lhey siill left the impressions). Bclween the parallel ridges is a conspicuous

groove. At the tergal angle, the ridge bounds off three small, roundcd dépréssions. The ridges

of M. Itmgliuaeiisix are different from those of Yunnanolepis. which has only eight ridges radiating

from the tergal angle towards the margins and not forming grooves in between. In M. long-

huaensisy there are at least len ridges radiating from the tergal angle, and some of them branch

outwards. The ridges form almosi right angles with plate margins. Eilher the antérolatéral margin

or postérolatéral margin is transected by three ridges. Four ridge.s, among whieh two ridges are

derived from the médian dorsal ridge, traverse the posterior margin. The viscéral surface cannot

be observed in the .spccimen,

The PMD plate (PMD. Fig. lOA, C) is rclatively broad, with a length/breadth index of

about 78, and its posterior margin is slightiy protruding. On the cxiemal surface, there is no

angle or hump as generally seen in Yiinnanolepis. There are four ridges extending from the

AMD plate. A ridge proceeds to the PDL plate in front of the faint postérolatéral lidge (pir.

Fig, lOA). Silice part of the bone was crushed, the internai mould of the plate is expo.scd, showing

that the crisut Iransversalix interna piistennr (crtp, Fig. lOA) arises from the PDL piales towards

anterior portion of the plate, close to its anterior edge, as in other yunnanolepids. The posterior

ventral process is not exposed in the specimen, and must hâve been siluaied far behind the

crista iransversalis interna posterior.

The ADL plate (ADL. Fig. lüA, C) is composed of dorsal and latéral laminae. The dorsal

lamina has a veiy broad anterior margin, below which the articular fossa and crista transyersalis

interna anterior (cit, Fig. lOA) are visible in anterior view. Except for those four ridge.s coming

from the AMD plate, there is a ridge above the dorsolateral ridge extending towards the PDL

plate. The latéral lamina contacts the AVL plate vcntrally. PDL and PL plates posteriorly and

AL piale anteriorly. There are two, somewhal parallel ridges extending posteriorly. The main

lateral-line groove is not observed in the specimen.

The PDL plate (PDL. Fig. lOA, C) is similar to that of other yunnanolepids in having a

very low latéral lamina and a distinct dor.sal corner. On the dorsal lamina, three ridges extend

from the AMD plate, onc in addition to the faim postérolatéral ridge from the PMD plate and

one from the ADL plate. Iniernally, the crista transversalis interna posterior (cr.ip. Fig. lOA)

is situated ncar the posterior margin of the plate.

The AL plate of M. longhitaensis (AL, Fig. lOA, C) is most inleresling. It is a small plate

on the anterior extremity of the latéral wall. It shows distinct sutures with the ADL and AVL

plates, and is remarkahie by ils posteriorly extending process (p.AL, Fig. lOA. C) which is easily

confused with the piepectoial corner, a structure of the AVL plate. This AL process has a relatively

dorsal position and is siluaied just below the dorsolateral ridge. With the rest of the plate, it

delimils a blind fossa.

The PL plate (PL. Fig. lOA. C) is a long and low plate ornamented with ridges. Internally,

the crista transversalis interna po.sterior lies nexl to its posterior end.

The AVL plate (AVL, Fig. lOB, C) has latéral and ventral laminae. The latéral lamina is

very low and extends behind the AL plate. The pectoral région is poorly preserved. Part of the

prepectoral corner is preserved. and it .seems that M. longhitaensis has a simple pectoral joint

like other yunnanolepids. The ventral lamina is somewhat similar to that of Y. poriferu, since

— 257 —

it has narrow contact margin with the MV plate. However, as to the notch for the semilunar

plate, M. longhuaeiisis, as well as M. parviis (M.-M. Zhang 1980, pl. 1, 2), is more suggestive

of Y. chii (G.-R. ZHANG 1978, Fig. 6). The crista tmnsversalis interna anterior (cit, Fig. lOB)

extends from the level of the prepectoral corner to the mesial margin, and, halfway bends some-

what backwards. The ventrolateral ros.sa (f.lv, Fig. lOB, C) is clearly visible in the specimen,

as in Yunnanolepis.

The PVL plate (PVL, Fig. lOB, C) is very similar to that of Y. porifera. It shows part of

the ventrolateral fossa (f.lv) at its anterior end and the crista transversalis interna posterior

lying close to its posterior margin.

The MV plate (MV, Fig. lOB) is a very small rhombic plate. Its breadth is less than two

fifth of the trunk-shield breadth, as in Y. porifera.

Genus PHYMOLEPIS K.J. Chang, 1978

Phymolepis K.-J. Chang, 1978: 295, pl. 25, 5-7.

Other refcrence:

Phymolepis G.-R. ZHANG 1978: 168, figs 10-12, pl. VI.

EMKNDtD uiAUNO.sis. — Yuiinanolepidiclae in which (here is a strong posterior process* of the PMD plate;

AMD piale with anterior division longer than posterior division: anierior ventral process and pit situated slightly

behind the level of the latéral corners of the AMD plate; conspicuous médian dorsal ridge between the tergal

and posterior dorsal angles.

Typp sPtciES. — Phymolepis i'Hifeii);shaiien.sis K.-J. Chang, 1978.

Remarks. — Phymolepis is more suggestive of Yumumolepis by its Ciianc's apparatus. The other siniilarities

are due to characters of the Yunnanolepididae or the Yunnanolepidoidei. They differ by the fact that the former

retains the AL plate which hides the opening of the CtiANo’s apparatus from the outside.

Phymolepis cuifengshanensis K.-J. Chang, 1978

(Figs 11-12; pl. I, 8-10,. IV. 1-9)

Phymolepis cuifengshanensis K.-J. Chang, 1978: 295, pl. 25, 5-7.

Other reference:

Phymolepis cuifengshanensis G.-R. ZHANG 1978: 168, figs 10-12, pl. VI.

Emended DiAtiNOSis. — Pliyniolepis species in w'hich the strong po.sterior process of the PMD plate is rounded

in shape, reaching one third ot the plate length.

Holotype. — A trunk-shield, V442.‘t.3 (K.-J. Chanc: 1978, pl. 25, 5)

New material. — .Several detached trunk-shield plates front the Xishancun Formation, VI0500; several

PMD plates from the Xilun Formation, VI0508.

Locality and horizon. — Qujing. Yunnan, Xitun Formation, Early Devonian.

Remarks. — P. ciiifeng.shunen.sis. characterized by its ball-shaped posterior process of the PMD plate, was

erected by K.-J. Ciiano (1978) on the basis of the material from the Xitun Formation, Qujing, Yunnan. This

species is aiso found in the underlying Xishancun Formation (see below'). The material of P. cuifengshanensis

— 258 —

from tht Xiliin Formation has alruady been studied in detail by G R. Zhang (1978), Hnwever, sincc the posterior

proces.s of thc PMD plate is the defining character t'or the species, its variations should be deseribed.

Sincc the posterior process of Fhymolepis is a fairly variable structure as seen in R cuifennshunensis, the

length/breadth index of the PMD plate is easily biased and üiis index is hard to he used for comparisons. The

length/breadlh index of the PMD plate proper is proposed here as u substitule, The portion of the PMD plate

propcr is thc estent in front of thc level of the postérolatéral corners. This région corresponds to the plate portion

which is in contact with the adjoining plates.

Description

Maicrial fmm the XishancHn Formation (Fig. 11, pl, IV. 1-9)

The new material inclinles detached PMD, ADL, AVL, PDL, PL uitd PVL plates, The as-

signment t>f the PMD plate is unequivocal since it bears the typical posterior process and distinct

médian dorsal ridge. The other plates are more or less similar to those of K porij'era from the

same horizon, however, the plates referred to P. ctiifen^sltancnsis are evidently larger than those

of Y. porifeni which is a small-sized antiarch as seen ahove. At least AVL and PVL plates could

not belong to Y chii which has a large MV plate. They are distingitished from the plates of

Liuo^uolepis by the presence of the ClIANG's apparatus and ventrolateral fossa.

The PMD plate (Fig. 11 B. C, pl. IV, 2-4) bears a distinct médian ilorsat ridge (dmr, Fig. IIC)

and postérolatéral ridges (plr, Fig. IIC), like that of /’. cuifen^shanensis from thc Xitun Forma¬

tion. The médian dorsal ridge ends at the po.sterior dorsal angle (pda. Fig. IIC), which forms

a conspiciious dorne. The postérolatéral ridges and grooves (plr, plg. Fig. IIC) extcnd laterally

from the posterior dorsal angle lo thc latéral margins. Behind thc posterior dorsal angle and

postérolatéral grooves, therc is a well-dcvcloped posterior process (pr.p. Fig. IIC), which is more

or less rounded, about one ihird the length of thc plate, and apparcntly larger than that of Yun-

nanoh'pis and Chitdiinoîepia. The latéral corners arc poorly preserved. However. the post¬

érolatéral corners (pie, Fig. IIB, C) are conspicuous. The length/breadth index of the proper

portion is about 100, a little larger than that of P. citifengshanensis in the Xitun Formation,

however smaller than that of P. gnontii.

In viscéral view, the posterior ventral process and pit (prv2, pl2, Fig, IIB) are well-

developed in the centre. The aa-a in front of thc process is thickened, as it represents the position

of the erista transversalis interna posterior (cr.ip, Fig. ! IB). Along the anterior and latéral mar¬

gins arc the overlap areas for the AMD and PDL plates (cf.AMD. cf.PDL. Fig. IIB).

The ADL plate (Fig. Il A, pl. IV, 1) consists of dorsal and latéral laminae (dim, llm,

Fig. 11 A), formiiig an angle of about 160^ at the dor.solateral ridge (dlr. Fig. IIA). The dorsal

lamina has u broad anterior margin, which is about three quarters of the plate length and three

times the posterior margin breadth. In anterior view, the transverse articular fossa (Fart. Fig. I IA)

is situated beneath the anterior margin of the dorsal lamina. The plate is overlapped by the

AMD plate along its mesial margin (oa.AMD. Fig. Il A). The main lateral-line groove (Icg,

Fig. IIA) extends jusi helow the doesolateral ridge. The latéral lamina becomes deeper towards

the posterior end. The ridge causcd by the OUANG's apparatus (r.Chang. Fig. Il A) is seen on

the surface, near the anterior exiremiiy. The notch for the opening of the Chang's apparatus

(Ca.o, Fig. 11 A) is situated at the ventral margin of thc latéral lamina.

The PDL plate (Fig. IID, pl. IV, 5) is represented by an internai mould. which has the

dorsal and latéral laminae with a fairly blunt angle (dlm, llm, Fig. IID). The plate has a

— 259 —

Fig.

Il _ Ph\molepis cuifengshanensis K.-J. Chang, Xishancun Formation, Qujing. Ail specimens are elastomer casis. A. left

ADL plate in dorsolateral view. Vl()500.2; B-C, PMD plate in viscéral (,B) and dorsal (C) views, VIOSOO.I; D, left PDL

plate in viscéral view, V 10500.3; E, right PL plate in extcrnal view, V10500.4; F. lefi AVL plate in viscéral view, V 10500.5;

G, right PVL plate in viscéral view, V.10500.6. (Scale bar 5 mm.)

— 260 —

conspicuous dorsal corner (d, Fig, IID), in front of which is the overlap area for the AMD

plate (oa.AMD, Fig. IID). It overlaps the PL plate along its ventral margin. The crista trans-

versalis interna posterior (cr.tp, Fig. IID) is seen to extend from the latéral lamina, where it

is Sharp, to the area near the dorsal corner where it becomes less distinct.

The PL plate (Fig. Il F. pl. IV, 6) is the same as thaï of cuifengxhanensis from the Xitun

Formation (G.-R. ZliANG 1978). It is relatively high in comparison with thaï of Rernigolepis and

Stegolepis, and the anterior portion is higher than the posterior portion. It is overlapped by the

PDL plate along its dorsal margin (oa.PDL, Fig. I lE). The crista transversalis interna posterior

(cr.tp. Fig. I lE) is developed close to the posterior end of the plate.

The AVL plate (Fig. IIF, pl. IV, 7-8) is exemplified by an internai mould. The plate is

composed of ventral and latéral laminae, forming an almost right angle in viscéral view. Post-

eriorly, the plate thickens to form a dépréssion which is part of the ventrolateral fossa (f.lv,

Fig. I IF). The crista transversalis interna anterior (cit, Fig. I IF) extends mesially on the rela¬

tively broad ventral lamina. The oblique mesial margin behind the antérolatéral corner (c.al,

Fig. 11F) is relatively long and straight. The overlap area for the MV plate (cf.MV, Fig. 11F)

is comparably short. Since the plate under description is a left AVL plate, il bears the overlap

area along its mesial margin for the right AVL plate (cf.AVL. Fig. IIF). The anterior part of

the latéral lamina displays the wall of the CHANG's apparatus (C. Chang, Fig. IIF). The overlap

areas for the PL and PVL plate arc seen along ils posterior margin (er PL, ef PVL, Fig. 11F).

The PVL plate (Fig. IIG, pl. IV, 9) is also an internai mould, showing ils viscéral surface.

Its anterior extremity bears a dépression which forms part of the ventrolateral fossa (f.lv,

Fig. IIG). The latéral lamina (llm, Fig. IIG) is fairly low at ils anterior end. and gets slightiy

higher posteriorly. Along its dorsal margin is the overlap area for the PL plate (cf.PL, Fig. 1IG).

The ventral lamina is relatively broad. as that of the AVL plate, and the overlap area for the

MV plate (cf.MV, Fig. 1IG) is short. The crista transversalis interna posterior (cr.tp, Fig. IIG)

lies very close to the posterior extremity of the plate.

Muieriul l'rom the Xitun Fonnation (Fig. 12, pl. 1. 8-10)

Some detached PMD plates from the Xitun Formation are described to show variation in

the posterior proce.ss. In external view, the PMD plate bears a distinctive médian dorsal ridge

and postérolatéral ridges (dmr. plr. Fig. 12). The médian dorsal ridge forms a relatively large

angle with the contact margin plane of the plate. In V10508.2, this angle rcaches about 45“

(Fig. 12B). From the anterior margin of the plate, the médian dorsal ridge strelches posteriorly

to the posterior dorsal angle (pda. Fig. 12), whicli in general forms a conspicuous dôme. The

postérolatéral ridges, as well as the postérolatéral groove (pig, Fig. 12) behind them, extend

laterally from the posterior dorsal angle to the latéral margins, slightiy in front of the postéro¬

latéral corners. The length/breadth index of the proper portion of the plate is less than 100.

Fig. 12. — Phymolepis cuifengshanensix K.-J. Chang, Xitun Formation. Qujing. A-C. PMD plate in dorsal (A), latéral (B) and

viscéral (C) views. V10508.2; D-E, PMD plate in latéral (D) and dorsal (E) views, V10508.3; F, PMD plate in latéral view,

VI0508-1. (Scale bar 5 mm.)

— 262 —

Behind lhe posterior dorsal angle is the well-developed posterior process (pr.p, Fig. 12), which

is quite variable in shape as described below.

In viscéral view, the overlap areas for PDL plates (cf.PDL, Fig. i2C) extend along the

latéral margins, which form a contact margin plane together with the anterior margin. The post¬

erior ventral proccss and pii (prv2. pt2, Fig. I2C) arc strongly dcveloped in the centre, and

elongated in shapc. The rrisiu iransversalis interna posterior (cr.tp. Fig. 12C) lies in front of

the posterior ventral process.

The po.stcrior process is quite variable, in both size and shape. Its length is about one third

to two fifths of the plate length. In general, it has a rounded arched dorsal face and a fiat ventral

face (G.-R. ZtlANG 1978. Fig. 10). The tubercles on the ventral face, i.e. of the posterior marginal

area (pma, Fig. I2B. C, F), aie niiich smaller than lho.se on the dorsal face. In V 10508.4, the

dorsal lamina with larger tubercles bends toward the ventral face to form three small triangular

lobes al lhe posterior end of lhe process. Vl()508.1 has the posterior end of the process raised

(Fig. 12F). V10508.2 is rcmarkable by its swolicn posterior process (pr.p. Fig. 12A, B), which

extends posteriorly along the direction of the médian dorsal ridge, and the lack of distinct limit

belween its dorsal and ventral faces. In contrasl, V10508.3 has a cylindrical, and more or less

slender posterior process (pr.p, Fig. 12D), which forms a conspicuous bend with the médian

dorsal ridge.

Phymolepis guoruii n. sp.

(Figs 13-14; pl. V)

Diaünosis. — Phymolepis in which the tergal angle of the AMD plate forms a low cockscomb-shaped

ridge; proper portion of the PMD plate with LAV index belween 130 and 150; médian dorsal ridge behind the

tergal angle sharp.

Etvmology. — Aftcr Dr G.-R. Zhang, Beijing.

Holotype. — A trunk-shield, V10509.I (Fig. 13, pl. V, 1-3).

Other .MATERIAU. — A trunk-shicld (VI0509.2) and several AMD or PMD plates, V10509.3-9.

Locauty ani> horizon. — Qujing, Yunnan. Xitun Formation, Early Devonian.

Rcmark.s. — This new specie.v resembles very itiiich P. ciiifcngshanensis by its strong posterior process of

the PMD plate and the conspicuous médian dorsal ridge between the tergal and posterior dorsal angles. It is

disiinguished froiu P. niifengshaïu'i/sis mainly by the following three characters ;

1 ) the AMD plate of P. gwnuii form.s a low cockscomb-shaped crest at its tergal angle, whereas the tergal

angle of P- eiitjengsluinensis is jiist a simple rise;

2) the médian dorsal ridge of this new species is fairly .sharp:

3) the proper portion of the PMD plate in P. guoruii is relatively long and narrow, with the L/W index

between 130 and 150, whereas it is less than 100 in P, cuifeiigshcmensis.

Description

The available .specimen.s of P. guoruii include two trunk-shields and several detached AMD

or PMD plates. No skull-roof plate is referred to this new species. The holotype is a relatively

complété trunk-shield. although ils left half is distorted and the PMD plate is missing. V10509.2

(pl. V, 4) is an almosi undisturted trunk-shicld. but unfortunately, its bones in the ventral and

latéral walls hâve been partly eroded in the internai mould. In overall aspect, the trunk-shield

— 263 —

of P. guoruii resembles very much lhat of P. aiifengshcinemis. Its dorsal wall is slightiy narrower

than its ventral wall, and the latéral wall is comparatively high. The dorsolateral and ventrolateral

ridges are fairly conspicuous. Moreover. it retains the separate AL plate which shelters the open-

ing of the Chang’s apparatus. Mowever, it differs from P. cuifengslKwensi.s in possessing a low

cockscomb-shaped crest ai its tergal angle. Between the tergal and posterior dorsal angles, its

médian dorsal ridge is sharpcr than lhat ol' P. cuifengshanensix. Although the PMD plate is absent

in these two articulated trunk-shicids. several detached PMD plates are assigned |o this new

species by the sharp médian dorsal ridge and the long contael margin for lhe PMD plate on the

PDL plate. P. gunruii bears the saine devclopcd posterior proeess ol'lhe PMD plate as P. t tiifeng-

shanensis, howevcr, its proper portion of the PMD plate is relatively long and narrow in eontrast

with that of the I aller,

The AMD plate is wcll prescrved in V 10509.2 (pl. V, 4), whereas in the holotype (Fig. I3A,

pl. V, I) il.s posterior portion is partly damaged. The AMD plate has the similar shape as that

of P. cuifengsham’nxis. Us anlcrior margin is pointed and the posterior margin is tairly narrow.

The latéral corners are con.spicuous and slightiy posteriorly located, the anterior poition of lhe

plate being longer than lhe posterior portion, as in P. cuifengsluinensis. In dorsal view, the plate

is arched by its elevaied tergal angle which is situaled slightiy behind lhe level of the latéral

corners. The very peculiar tergal angle (dma. Figs 13A. I4D-F). is a low cockscomb-shaped

crest which is hordered by an elongaied ll-shaped ridge. This ridge is ornamenlcd by slightiy

larger tuhercles and encircics a central spacc of the tergal angle which is almost devoid of

tubercles. This central space is fiat in ils anterior part, which is continuous with the area anterior

lo it. and is slightiy expanded lalerally in lhe posterior direction. This space is a liltle arched

in its posterior part to form a summit. The U-shaped ridge of the tergal angle prolongs posteriorly

to form the sharp médian dorsal ridge (dmr, Fig. 14F. pl. V, 4) on the AMD plate. Two AMD

fragments of P. guoruii hâve bcen found in lhe collection, characterized by their specialized

tergal angles (dma. Fig. 14D, F). They correspond lo lhe posterior half of the cockscomb-shaped

crest seen in lhe holotype and VI0509.2. In viscéral view, the elongated anterior médian proeess

and pit (prvl, pli, Fig. 14E, G), and part of lhe médian ventral ridge (mvr, Fig. I4G) in front

of lhem are visible.

The PMD plate is represented by several detached specimens. As in P. ciiifeugxltanensis,

it has the conspicuous posterior dorsal angle (pda, Fig. 14A-B), and the devclopcd posterior

proeess (pr.p, Fig. I4A-B) behind lhe angle. From the posterior dorsal angle radiale (hree ndges,

one is lhe médian dorsal ridge (dmr, Fig. I4A-B) prolonged from lhe AMD plate, while the

others are the postérolatéral ridges (pir, Fig. I4A-B) exlcnding to lhe latéral margins near the

postérolatéral corners. The posterior proeess is arched on lhe dorsal surface and somewhat flat-

tened venlrally (pma. Fig. 14C), and occupies about one Ihird of the plate Icngth. The proper

portion of the plate is different from thaï in P. l uifengshanensis. Firstly. il is somewhat narrow

and elongated in shape. The L/W index is between 130 and 150, whereas thaï in P. cuifeng-

shunensis is less than 100. Secondly, lhe médian dorsal ridge in the new species is sharpcr than

that in the laller. In viscéral view, the overlap areas for PDL piales are observed along the

latéral margins. The anterior margin is fractured off. In the middle, the elongated posterior ventral

proeess and pit (prv2, pl2, Fig. 14C) are developed as in P. cuifengsbunensis. In front of them,

lhe plate is thickened to form the crista trausversuHs interna posterior (cr.tp, Fig. 14C).

— 264 —

The ADL plate has a distinct dorsolateral ridge (dlr, Fig. 13A). As in othcr yunnanolepids,

its dorsal lamina has a broad anterior margin. In anterior view, the articrdar fossa and crisla

transversalis interna anterior arc situated beneath the anterior margin. The latéral lamina is fairly

high, and the posterior portion is higher than the anterior one. The main latcral-line groove (Icg,

Fig. 13. — Phymolepis guoruii n. sp., Xitun Formation. Qujing. Trunk-shield (holotype, V10509.1) in dorsal (A), ventral (B) and

latéral (C) views. (Scale bar 10 mm.)

— 265 —

Fig. I3A. C) runs just below the dorsolateral ridge. A short pii line (adg. Fig. 13C) branches

off dorsally from the main lateral-line groove near its anterior end, as in Y. porifera (M.-M.

Zhang 1980, Fig. 3). The région of the latéral lamina near its anterior extremity is complicated

by the ovcrlapping AL plate. The ADL portion dorsal or posterodorsal to the AL plate seems

to bc ihickened, An oblique ridge lies between the dorsolateral ridge and the AL plate, and

diminishes posleriorly. The opening of the Chang's apparatus (Ca.o. Fig. I3C) is hidden by the

overlying AL plate and eannoi be direclly seen from the oulside.

The AL plate (Fig. I3A, C) is a small plate, adjacent to the anterior extremity of the latéral

wall of the trunk-shield. However, it does not fomi the anterior margin of the tnmk-shield. as

in Mizia. Il looks like a large node on the latéral walL across the suture between the ADL and

AVL plates, but il is indeed an independent élément of the trunk-shield and belongs ncilher to

the ADL plate, nor to the AVL plate. A large part of the AL plate is situalcd above the ADL

plate, and posleriorly forms a fossa between the AL and ADL plates. In anterior view, an even

larger fossa is visible between the AL and ADL plates. The opening of the CilANG’s apparatus

(Ca.o, Fig. 13C) is located in this fossa and is invisible in latéral view.

The AVL plate forms an ainiost right angle at the ventrolateral ridge. The dorsal margin

of the latéral lamina descends posleriorly and the ridge caused by the CHANG's apparatus is

faintly seen near ils anterior margin (Fig. 13C. pl. V, 3). It bears the lypical yunnanolepid pre-

pectoral corner (pre, Fig. 13A) and pectoral fenestra. The ventral lamina has a relativcly short

contact margin for the MV plate, as in P. aiifenf’shauensis. The postbranchial lamina (l.pbr,

Fig. 13A) bears many parallel striations, and is a rod-shaped structure which extends anterome-

sially from near the pectoral fenestra, Immedialely behind the postbranchial lamina is the mesially

directed crista iransversalis interna anterior.

The SL plate (Fig. I3A-B) is partiy preserved in the hololypc. Il is broad and frapezoid in

shape, intermediate in shape between the SL plates of Y. cliii and Y. porifera.

The PDL plate (Fig. 13A) is well preserved in ihc right half of the holotype. The dorsal

lamina has a distinct dorsal corner, behind which is the relalively long area overlappcd by the

PMD plate (oa.PMD. Fig. I3A) On the latéral lamina, the main lateral-line groove runs below

the well-defined dorsolateral ridge. VH)509.2 has preserved the internai mould of the plate, from

which the crista transversaux interna posterior is inferred to be siluated near the posterior end

of the plate and extends dorsally close to the suture between the AMD and PMD plate.

The PL plate (Fig. 13C) is relatively high. Along its ventral margin, it is overlappcd by

the PVL plate. In viscéral view, the crista transversaux interna posterior lies almost at its post¬

erior end.

The PVL plate (Fig. 13B-C) has a fairly low latéral lamina. Its ventral lamina is similar

to that of P. citifeng.shanensis in having a short contact margin for the MV plate.

The MV plate (Fig. I3B) is a relatively small. rhombic plate. As in Y. porifera and

P. cuifengshanensis, its breadth is smaller than two fifth of the ventral wall breadth of the trunk-

shield.

The ornamentation of the trunk-shield consists of fme-grained and closely set tubercles.

Along the ridges, and on the région anterior to the pectoral fenestra, the tubercles are slightly

larger.

^ «’i. i

9 ..". %2Sî

Fig. 14. — Phymolepis guoruii n. sp., Xitun Formation, Qujing. A-C, PMD plate in dorsal (A), latéral (B) and viscéral views

(C). V10509.5; D-E^ incomplète AMD plate in dorsal (D) and viscéral (E) views, V10509.4; F-G, incomplète AMD plate in

dorsal (F) and viscéral (G) views, V10509.3. (Scale bar 5 mm.)

Family indet.

Genus ZHANJILEPIS G.-R. Zhang, 1978

Zhanjilepis G.-R. Zhang, 1978; 178: figs 18-21, pl. VIII.

Other référencé:

Zhanjilepis S.-F. LlU 1992: 216, pl. II, 8-9.

Emended diacnosi.s, — Same as for the type species (monotype).

Typh spcrTr;.s. — Zhanjilepis aspralilis G.-R. Zhang, 1978.

Remarks. The genu.s was erecied on ihe hasts of several deiached trnnk-shield plate,s (G.-R. Zhang 1978).

The best diagnostic character is ils ornamentation. Zhanjilepis resembles the Yunnanolepididae in the shape of

its AMD plate, the position of the anlerior ventral process and pit, and the presence of the ventrolateral fo.ssa.

But Zhanjilepis differs in the position of the crista iransversalis interna pasierior, which is suggestive of Chu-

chinolepis and Minicrania (Zho & Janvier 1996). However, as discussed below, the latéral position of the crista

iransversalis interna poslerior, relativcly to the posterior ventral process, as in Chuchinolepis, is a plesiomorphic

feature.

Zhanjilepis aspralilis G.-R. Zhang, 1978

(Fig. 15, pl. III, 9-11, IV, 10)

Zhanjilepis aspralilis G.-R. Zhang, 1978: 178. figs 18-21, pl. VIII.

Emended diagnosis. — Yunnanolepidoidei in which the ornamentation is composed of coarse, scattered

tubercles.

Holotype. — An AMD plate (V4427.I) from the Xitun Formation.

New material. — Detached trunk-shield plates from the Xishancun Formation, including four AMD, an

ADL, three PDL and a PVL plates, VI0501.1-10.

Locality and horizon. — Qiijing, Yiinnan, Xishancun Fonnation and Xitun Formation, Early Devonian.

Remarks. — Since Z. aspralilis was described from the Xitun Formation, Qujing, Yunnan (G.-R. Zhang

1978), the distribution of Zhanjilepis, either stratigraphie or paleogeographic, has not been expanded until S.-F. Liu

(1992) reporled Z. sp. front the Lianhuash.an Fonnation (Early Devonian) of Liujing, Guang.xi. Here, Z aspralilis

is described from the Xishancun Formation (Lochkovian), the stratigraphie unit underlying the Xitun Formation.

Description

Ail the .specimens referred to as Z. aspralilis in the Xishancun Formation are characterized

by their coarse tubercle ornamentation, like those in the type horizon, i.e., the Xitun Formation.

However, the individual size of the specimens in the Xishancun Formation is definitely smaller

than in those from the Xitun Formation.

The AMD plate is similar to that of the Yunnanolepididae and Chuchinolepididae with the

pointed anterior end (a, Fig. 15A), Us posterior margin is relatively narrow, and about one fifth

of the plate breadth Us latéral corners (le. Fig. 15A-B) are noticeable, and the anterior portion

of the plate is slighlly longer than its posterior portion. When compared with the holotype of

Z. aspralilis, which is an AMD plate from the Xitun Formation, the AMD plates referred to

here are slightly narrower, as in Y. porifera. V 10501.1 (Fig. 15A-B) has a length/breadth index

of about 145, and the index between the anterior and posterior portions is about 128. In extemal

— 268 —

view, ihe plate is gently arched with the tergal angle at the level of the latéral corners. The

médian dorsal ridge is faintly developed behind the tergal angle. In viscéral view, the anterior

ventral process (prvl, Fig. I5B) is small and situaied at the level of the latéral corners, like the

tergal angle. Behind it, there is a shallow médian ventral groove (grm, Fig. 15B) which diminishes

posterioriy. No slnicturc i.s developed in front of the anterior ventral process. The overlap rela-

tionship with adjoining plates is typical of the Yunnanolepidoidei.

The PMD plate is not found in the Xishancun Formation. The re-examination of the plesi-

otypc of Z. aspratilis (V4427.3. Fig. I5C). that is a PMD plate found in the Xitun Formation,

shows that the crista transversalis interna posterior (cr.tp, Fig. I5C) lies latéral to the posterior

ventral process and pit (prv2, pl2, Fig. 15C), as in Clmchinolepis.

The ADL plate (Fig. 15E) has the same shape as that of Z aspratilis from the Xitun For¬

mation (G.-R. Zhang 1978). The plate is suhdivided into the dorsal and latéral laminae by the

dorsolateral ridge (dir. Fig. I5E). The dorsal lamina has the broad anterior margin, below which

there is the transverse articular fossa. Along its dorsal margin is the area overlapped by the

AMD plate (oa.AMD, Fig. I5E). The latéral lamina descends posterioriy. The main lateral-line

groove runs just below the dorsolateral ridge.

Fig. 15. — ZhanjUepis aspratilis G.-R. Zhang. Xishancun Formation (A-B, D-E) and Xitun Formation (C), Qujing. A-B, AMD

plate in dorsal (A) and visceraJ (B) view.s (elaslorner casts). VI0501.1; C. PMD piale in viscéral view, V4427.3; D, right

PDL plate in dorsal view (elastomer cast), V10501.7; E, right ADL plate in dorsal view (elaslorner casi). V10501.6; F, right

PVL plate in vi.sceral view (elaslorner casi), Vl()501.5. (Seule bar 5 mm.)

— 269 —

The PDL plate (Fig, 15 D, pl. 111, 9) aiso has dorsal and latéral laminae. The dorsal lamina

(dlm, Fig. 15D) is relatively long and narrow, with a dorsal corner (d, Fig. 15D) which séparâtes

the dorsal margin inlo Iwo portions. Along the anlerior portion of the dorsal margin, the plate

is overlapped by the AMD plate anleriorly (oa.AMD, Fig, I5D) and overlaps the AMD plate

posteriorly. The plate is overlapped by the PMD plate (oa.PMD, Fig. 15D) behind the dorsal

corner. The latéral lamina is fairly long and low, with the main lateral-line groove running below

the dorsolateral ridge (dir. Fig. I5D). Along its ventral margin, there is an area which is over¬

lapped by the PL plate. This overlap area is less developed in its anlerior part. Along its anterior

margin, the PDL plate is overlapped by the ADL plate.

The latéral lamina of the PVL plate (Fig. 15F, pl. TV, 10) is partly observed in the specimen,

showing a granular ornamentation. The ventral lamina has ils anterior margin extending anter-

olaterally. In viscéral view, the overlap area for the MV plate (cf.MV. Fig. 15F) is relatively

short. At the anterior end of the ventrolateral ridge, the plate has a dépréssion which is part of

the ventrolateral fossa of the trunk-shield (f.lv, Fig. 15F). The crista tninsveraalis intenici postedor

(cr.tp, Fig. 15F) lies close to the posterior extremity of the plate and becomes higher laterally.

Genus HETEROYUNNANOLEPIS Z.-S. Wang, 1994

Heteroyuntuinolepix Z.-S. Wang, 1994: 21, fig. I, pis 1, II,

Emrndf.d DiAONissiS. — Yunnanolepidoîtlei in which the heuü-.shield has a large semicircular preorbital recess

instead of the preorbital dépréssion; infraorbital .sensory canal passing through the postmarginal plate; ADL plates

of both sides with a shon suitire in front of the AMD plate; large MV plate with the anterior portion shorter

than the posterior portion; irunk-.shield fairly bmad and low.

Type specids. — Heteroymiwiwltp'n tiujingrnsii Wang. 1994.

Remarks. — Heteroyunnancilepis differs from the Yunnanolepididae in its large semicircular preorbital

recess, the absence of the preorbital dépréssion, the path of the infraorbital sen.sory canal through the po.stmarginal

plate, and the position of the crista iransversalis inlerrta posterior, which is latéral to the posterior ventral process

and pit. The diagnostic character of lhe Yunnanolepididae is that the crista transversalis interna posterior bends

anteriorly on the viscéral surface of the PMD plate, and passes clearly in front of lhe posterior vetilral process

and pit. As to the relative size ol lhe MV plate. Heteroyunnanoiepis resemhles >' rlili. Hnwever. a large MV

plate is most likely to bc plesiomorphic. Moreover, the MV plaie of Heteroyunnanolepis has lhe postérolatéral

margin longer than the antérolatéral margin. whereas the MV plate of the Yunnanolepididae. including K cMi,

is generally rhombic in shape. Heteroyunnanolepis resemblcs T. deprali and Zhanjilepis by the lack of crests,

ridges or hunips usually seen in the external surface of the PMD plate in other yunnanolepidoids Heternyun-

nanolepis is suggestive of Vanchienolepis (Tong-Dzoy & Ja.mvidr 1990) by the joinuig of the ADL plates of

both sides in front ol lhe AMD plate. Howevei, these Iwo généra differ definilcly in niany olhei characlers.

Vanchienolepis lack.s the MV plate and bas a derived pectoral joint structure. The AMD plate of Heteroyun¬

nanolepis is similar to that of the Yunnanolepididae and Zluinjilepis in shape, whereas the AMD plate of

Vanchienolepis has a relatively short antérolatéral margin and ils tergal angle is anteriorly placed.

Heteroyunnanolepis qujingensis Z.-S. Wang, 1994

(Figs 16-17, pis VI, VII)

Diagnosis. — Heteroyunnanolepis in which the dorsolateral ridge of the trunk-shield is less developed; a

round pit is présent on the surface of the PMD plate; the posterior dorsal sensory-line groove extends from the

PDL plate to the PMD plate.

— 270 —

Holotype. — External mould of a head-shield, VI0II3 (Z.-S. Wang 1994, pl. IA).

New MATERiAL. — A complété trunk-shield, VI0502.I (Fig. 16A, pis VI, 1-2, VII, 1-2), an AMD (VI0502.2),

an AVL (VI0502.3), a PVL (VI0502.4) and a MV (V10502.5) plates.

Locai.ity and horizon. — Oujing. Yunnan, Xishancun Formation, Early Devonian.

Rkmarks. — Il i.s likely that Yuimrmoiriiis me<‘numtiiie (Tong-Dziiy à Janvier 1994.) should be referred to

Hetemyumianolepis. This antiarch (lhe only specinien) was found frorn the lowcr part of lhe Bac Bun Formation

of Vietnam (Pragian in âge). Il resembles H. qujingensh in the relaiively low and broad trunk-shield and the

PMD plate which i.s devoid of crests, ridge.s and humps. Both species hâve the po.sterior margin of the trunk-shield

moderalely cun'ed and lack the pronounccd postcrior dorsal angle. According to personal observations, lhe PMD

plate of Yuniianolepis meemannae lias lhe crisia tranxversalis interna pnsterior situaled latéral to the posterior

ventral process, as in H. cjujingensis and Zhanjilepis, that is qiiile different from the condition in lhe Yunnanolepi-

didac. As far as the available information, yunnaiwlepis meemannae differs from H. qujingensix in the following

characters •

1) y, meenumnae ha.s a conspicuous dorsolateral ridge whereas H. cpijmgenxis lacks this ridge, like Y. deprati;

2) the posterodorsal sensory-line groovc in K meemannae i.s a short groove reslricled to lhe PDI. plate as

in Yunnanolepix and Chuchinniepis (Tong-D/.uy & Janvier 199(1), whereas thaï of H. tjujingenxis exiends dorsally

to the PMD plate;

3) H. qujingensis has a rounded pit on the surface of the PMD plate.

Description

VI 0502.1 (Figs 16A, 17) is a medium-sized trunk-shield, almost completely preserved as

an internai mould (pl. Vil, 2). The external mould shows only the dorsal and latéral walls (pl. VII,

1), from which ihc bonc fragments hâve bcen removed for latexing (pl VI, 1). The specimen

was tlatlened during fossilizalion, however, the low trunk-shield can be inferred from the posi¬

tions of the dorsolateral and ventrolaleral ridges. The dorsal wall of the irunk-shield is gently

arched. more or Içss square-shaped, wtth bolh the length and width of about 5 cm. There is no

médian dorsal ridge. The ADL piales of both sides join in front of lhe AMD plate, which is

situaled about 5 mm behind the anterior margin of the dorsal wall. The posterior margin of the

dorsal wall does not foim a conspicuous posterior process as in Yunnanolepis chii, resembling

somewhat K deprati (TONG-DZUY & Janvier 1990, Fig. 10). The fiat ventral wall is just a little

longer than lhe dorsal wall. The MV plate occupies a fairly large area. The ornamentation consists

of dcnsely di.stribuicd. small lubercles.

The AMD plate of H. qujingenxis (Fig. 16A) is similar to that of K chii (G.R. ZHANG

1978) in shape and size. The AMD plate of V10501 has a length/width index of about 124,

whereas lhe isolaled AMD plate (VI0.502.2, Fig. 16B) is a little smaller and has a length/width

index of about 150, as the AMD plate of Y. chii. The plate has a very narrow anterior margin.

The latéral corners are conspicuous, and the antérolatéral and postérolatéral margins are of about

same length. The overlap relationships with the adjoining plates are just the same as tho.se of

the Yiinnanolepididac. The posterior margin is straighl and not as narrow as that of lhe Yun-

nanolepididae. In external view, the plate is gently arched and does nol show the médian dorsal

ridge or the médian élévation as in Y. chii. The arched point is at the Icvcl of the latéral corners,

resembling Ihose of the Ytinnanolepididac and Zhanjilepis Iri viscéral view, the anterior ventral

process (prvl, Fig. 16A) and pit (ptl, Fig. 16B) are situated just beneath the arched point of

the plate. In front of the anterior ventral process, a conspicuous médian ventral ridge (mvr,

Fig. 16A,. B) extends forwards to the anterior margin of the plate. In V10501.1 (Fig. 16A), the

less pronounccd médian ventral ridge, as in }' chii. extends from the anterior ventral process

— 271 —

c.al

Fig. 16. — Hetemyunnanoiepis qujin^ensis Z.-S. Wang, Xishancun Formation, Qujing. A, dorsal and latéral walls of the trunk-shield

(holotypc, V10502.1) in viscéral view (elastomer cast); B, AMD plate in vi.sceral view (elastomer cast), V10502.2; C-D, left

AVL plate in external (C) and viscéral (D) views (elastomer casts), V 10502.3. (Scalc bar 10 mm.)

to the posterior ventral process of the PMD plate. However, individual variation exists in

VI 0502.2. In this AMD plate, the médian ventral ridge behind the anterior ventral process is

very short, and behind the ventral ridge is a long médian ventral groove (grm, Fig. 16B), as in

Bothriolc/Jis. In the anterior portion of the plate, the levator fossa (f.retr, Fig. 16B) is subdivided

into two parts by the médian ventral ridge. Behind the levator fossa are the posllevator thick-

enings. The levator fossa and postlevator thickenings are widely distributed in euantiarchs, how¬

ever, they are generally less developed in yunnanolepidoids (G.-R. Zhano 1978).

The PMD plate (Figs I6A, 17) is more or less trapezoid in shape since its posterior margin

does not project posleriorly to form a prominent posterior corner. The plate is relatively broad,

and has a length/breadth index smaller than 100. The anterior margin, ovetlapping the AMD

plate, is slightly convex, The latéral margin is slightly concave and overlaps the ADL plate. On

its external surface, there is no ridge or élévation, but instead, a pit (p.PMD, Fig. 17A. C) is

seen in the mid-line, corresponding to the position of the posterior ventral process. Behind this

pit there is a pair of sensory-line grooves (pdg. Fig. 17A, C) extending laterally to the PDL

plates. The groove is parallel to the posterior margin of the dorsal wall. In viscéral view, the

posterior ventral process and pit (pt2. Fig. 16A) are relatively large and occupy a fairly posterior

position. In front, there is a low médian ventral ridge (mvr, Fig. 16A). The crislu transversalis

— 272 —

interna posterior (cr.ip. Fig. 16A) is latéral to the poslerior ventral process and pit, as in Zhan-

jilepis and Chuchinolepis.

The much flattened ADL plates of the holotype is relatively short and broad, similar to

that of K hachoensis (TONO-DZUY & Janvier 1990). The main lateral-line groove (Icg, Fig. 17A)

runs through the plate rostrocaudally, but disappears in the posterior portion of the plate. The

dorsolateral ridge is invisible, and its approximate position could be inferred from the main

lateral-line groove (Icg. Fig. I7A), which generally lies close to the ridge. The plate lias a rela¬

tively broad dorsal lamina and a low latéral lamina, as infened from the relatively low position

of the groove. In Yunnaiinlepis, the dorsolateral ridge of the ADL plaie, as well as the main

lateral-line groove. has a dorsal or middle position, corresponding to the comparatively high

trunk-shield. A short pit-line groove (adg, Fig. I7A) diverges from the lateral-line groove in the

anterior part of the plate, as in Y. porifera.

The dorsal lamina of the ADL plate has a fairly broad anterior margin. The ariicular fossa

and crisla transversalis iiilenia anterior are noi prc.served in the spccimen. Sincc the ADL plates

of both sides join in front of the AMD plate by a short suture, lhe plate has an additional

anteroniesial margin which was aiso seen in Vanehieiiolepis (TONG-DZ.UY & JANVIER 1990). The

ADL plate is overlapped hy the AMD plate, and overlaps the PDL and PL plates. The ventral

margin of the low latéral lamina is more or less parallel to the main lateral-line groove.

The PDL plate is as flattened as the ADL plaie. The faini dorsolateral ridge is seen in the

posterior portion of the plate. As in Ynnnanolepis, its dorsal lamina is l'airly broad and its latéral

lamina very low. The main lateral-line groove (Icg, Fig. 17A, C) reappears in the posterior portion

of lhe plate, and lies closely below lhe faded dorsolateral ridge. linmediaiely in front of lhe

posterior margin of the plate, an additional sensory-line groove (pdg. Fig. I7A. C) extends from

the main sensory-line groove to the PMD plate. The dorsal lamina has a distinct dorsal corner,

where the plate gcts its maximum breadth. The anterior margin of the plate is somewhat concave

and lhe area overlapped by the ADL plate is clearly visible in lhe holotype. The anierodorsal

margin in contact with the AMD plate is slighlly longer lhan the posterodorsal margin. where

the plate is overlapped by lhe PMD plate. The plate is overlapped by lhe PL plate along its

ventral lamina, which bcars a shallow notch close to lhe posterior extremity. as in Y. chii (G.-R.

Zhano 1978). The crisla transversalis interna posterior (cr.tp, Fig. I6A) is as m other yun-

nanolepidoids.

The PL plate (Fig. 17A) is a very low and long plate. The area overlapping lhe PDL plate

is clearly observed in the holotype, whereas the other contact margins arc invisible.

The AVL plate has ventral and latéral laminae scparaied by lhe venlrolateral ridge. VI 0502.3

(Fig. 16C-D) is an almost complété detached AVL plate, and with the aid of the mould, its

detailed moiphology can be studied. As in lhe Yunnanolepididae, the plate has a conspicuous

prepecloral corner (pre, Figs 16C. I7A). The simple pectoral fenesira can be observed in the

external mould of V 10502.3 (Fig. I6C). The prepecloral corner is siluated at the same level as

the corner belween the anterior and middle divisions of the mesial margin of lhe ventral lamina

(ci, Fig. I6C). At Üie anterior extremity of the plate is lhe antérolatéral corner (c.al, Fig, 16C).

From c.al to C| extends a long, oblique margin in contact with the semilunar plate (m.SL,

Fig. 16C). Even though the semilunar plate is not preserved in the collection, its general shape

can be estimated from the AVL plate to be triangular, as in }'. porifera, and quite different from

— 273 —

Fig. 17. — Heteroytmminolepis qujingensis Z.-S. Wang, Xishancun Formation, Qujing. Restoralions of the trunk-shield (mainly

after the hololype, VI050I.1) in dorsal (A), ventral (B) and latéral (C) views. (Scale bar 10 mm.)

the rectangular semilunar plate of K chii. The mesial margin in contact with the AVL plate of

the opposite side is relatively short, about two thirds of the length of the whole plate. Front

VI0502.3 is inferred thaï the righi AVL plate overlaps the left one, like in mosi antiarchs except

for Y. chii (G.-R. Zhang, 1978, Fig. 6). The mesial margin in contact with the MV plate is

very long, and exlends laterally almost to the ventrolateral ridge, as in Y. chii- Along this margin,

the area ovcrlapping the MV plate is visible (cf.MV, Fig. I6D). The latéral lamina (llm. Fig. 16C,

D) is relatively low. Anteriorly, the margin overlapping the ADL plate fornts a corner, behind

which the margin descends posteriorly. Internally, the crisla transversalis interna anterior (cit,

Fig. 16C) is at the same level as the prepectoral corner. Thcrc is no ventrolateral fossa of the

trunk-shield along the posterior margin of the plate, unlike in Yunnanolepis.

— 274 —

The PVL plate (Fig. 17B, C) is longer than the AVL plate, and also comprises iwo laminae.

The latéral lamina is low, as in the Yunnanolepididae. The ventral lamina has a very long margin

in contact with the MV plate, which occupies nearly the two thirds of the plate length. The

PVL plates of both sides meet on a relatively short suture. The crista iransversalis interna post-

eriar lies close to the posterior margin of the plate.

The MV plate (Fig. I7B) is a fairly large plate, like that of Y. chii. It occupies about three

fifih of the trunk-shield breadlh and about two third of the ventral wall length. The plate is

overlapped by the AVL plates anteriorly and the PVL plates posteriorly. Since its antérolatéral

margin is obviously shorter lhan ils postérolatéral margin, the MV plate of Heteroyunnanolepis

qujingensis is not rhombic in shape, which is typical for the MV plate of Yunnanolepis.

Family Chuchinolepididae K.-J. Chang, 1978

[= Qujingolepidae G.-R. Zhang. 1978; Procondylolcpidac G.-R. Zhang, 1978]

Emevded diaonosis. — Yunnanolepidoidci in which the Irilohate, perichondrally ossified scapidocoracoid

is exposed in ihe bottom of ihe pectoral tenestra. three toramiiiii for the neiirovascular canals around ihe .scapulo-

coracoid; single dermal articulation (parabrachial condyle and ventral arlicular fo.s.sa) between the unjointed pec¬

toral tin and the irunk-shield; pectoral fin triangular in Iransverse section; two fossae for the abductor and adductor

muscles of the lin on the tnmk-shield: anterior médian dorsal plate long and tiarrow. B/L index about 50-65,

with the anteriorly placed tergal angle and anterior ventral process; no anterior ventral pit.

Remark.s. — This définition is modified from K.-J. Chanc; (1978) and G.-R. Zha.no (1984). As stated below,

Procondylolepis and Qujinolepis. are in fact junior .synonyms of Chuchinotepis. Therefore, both the Procondy-

lolepidae and Qujinolcpidae arc the Junior synonyms of the Chiichinolepidae. According to the International Code

of Zoolügical Nomenclature (Ride et al. 1985, Art 29a), the Chuchinolepidae is revised as the Chuchinolepididae.

Genus CHUCHINOLEPIS K.-J. Chang, 1978

Chuchinolepis K.-J. Chang, 1978: 296, pl. 26.

Synonyms;

Orientcilepis P'an & Wang, 1978: 322, pl. 28, 2.

Qujinolepis G.-R. Zhang. 1978: 173, figs 1317, pl. VIL

Prncnndylolepis G.-R. Zhang, 1984: 82, figs 14, pis III.

Others references:

Chuchinolepis Tonc-Dzuy & Janvier 1990: 176, figs 18-22, pis V, VI, 1-3.

Qujinolepis TONG-DZUY & JANVIER 1987; 12, figs 6A-C, 7A, pl. I, 3-4. — S.-F. Liu, 1992:

214, pl. Il, 1-7.

Procondylolepis YOUNO & ZHANG 1992: 445, figs 2F-G, 3B-D, 4-6. — F. Zhu, Wang &

Fan 1994: 4, llg. 2, pl. 1, 4-7.

Emendeü diagnosis. — As for the family (monogeneric).

Type species. — Chuchinolepis graciUs K.-J. Chang, 1978.

Remarks. — K.-J. Chang (1978) and Tonü-Dzuy & Janvier (1990) defined Chuchinolepis as “chuchinolepid

with very fine tubercular ornamentation, which is largely invisible to the naked eye”. As seen below.

— 275 —

Procondylolepis qujingensis (G.-R. Zhang 1984) is quite siinilar to C. gnwilix. cxcepi for the size uf tubercles,

and il seems better to place R gujingensis in the genus Chuchinok'pis to make the classification more simple.

This problem i.s aiso related to the assignment of two other chuchinolepids from Qiijing, Yunnan, the type locality

of C. gracilis and P. gujingensis. One is a chuchinolepid with fairly large lubercles (C. robusm n. sp,). With

regard to the pectoral fin articulation, P. giijingensi.'i and C. gracilis are more similar thnn cither is m C. mbiista.

If Procondylolepis was reiained as an indcpendcnt genus, C. robusut shoiild bc refcrred to as a new genus,

theteby making the classification unnecessurily complicatcd. The other is a chuchinolepid (C sulcuta n. sp.) in

which the fine-gramed lubercles (of the sanie relative size as in P. gujingensis) arc arranged in regular ridges

on the pectoral fin plates and the dorsal wall of the irunk-shield. and very fine-grained lubercles (of the same

relative size as in C. gracilis) are densely distributed between the ridges. Ils assignment is greatly dépendent of

the eharacler polarity of the tubercles. Thcrcfore, for .simplicity, wc propose a more general définition of Cliu-

chinolepis to include P. gujingensis, C. robusiu and C. sulcatu.

Chuchinolepis gracilis K.-J. Chang, 1978

(Figs 18-20, pis VIII, 1-5; IX, 4-5)

Chuchinolepis gracilis K.-J. Chang, 1978: 296, pl. 26.

Synonyms;

Orientolepis neokwangsiensis P’an & Wang, 1978: 322, pl. 28-2.

Qujinulepis gracilis G.-R. Zhang, 1978: 173, figs 13-17, pl. VII.

Others référencés:

Qujinolepis gracilis S.-F. LlU 1992: 214, pl. II, 1-7.

Emended diagnosis. — Chuchinolepis in which the length/width index of the PMD plate is about 110.

Holotypr. — A detached AMD plate, V4426.1.‘i.

Plesiotypp. — A detached PMD plate, V4426.6.

New material. — VI05I0.I-I2. material from the Xiliin Formation; V10503.I-5, matcrial from the Xi-

shancun Formation,

Remarks. — This species is the type species of Chuchinolepis (K.-J. Chang 1978), and was assumed lo

beat a simple pectoral fin articulation, like the Yunnanolepididae (G.-R. Zhang 1978). This had nol been questioned

uniil 1990 when Tong Dzuv & Janvier regarded it as a procondylolepiform. How'ever. the paiposal of Tong-Dzoy

& Janvier (1990) still awaits confirmation by finding the AVL plate of C. gracilis in the original locality and

horizon (the Xilun Formation in Qujing, Yunn.'tn), where "Procondylolepis" gujingensis was found together. The

diagnostic feature of C gracilis is Us very fine-gramed ornamentation, by which K.-J. Chang (1978) and O.R.

Zhang (1978) attributed the detached skull-roof and trunk-shield plates to this species. In this work, several AVL

plates, which could be definitcly refen-ed to C. gracilis, are dcscribcd from ihc Xitun Fonnaiion of Qujing, and

it i.s shown that C. gracilis is indeed a procondylolepiform.

The discovery of C. gracilis in the Xi.shancun Formation shows that Chuchinolepis has an earlier distribution

which might hâve a bcaring on the biostraiigraphie corrélation.

This species shares the very fine-grained ornamentation (invisible to the naked eye) with C. dongmoensis

(Tong-Dzuy & Janvier 1990). They differ in the length/width index of PMD plate, that of C. dongmoensis being

about 150.

Description

Material from the Xitun Formation (Fig. 18, pis Vlll, 5; IX, 4-6)

C. gracilis was originally described from this horizon, and ail of the specimens were

detached plates, including the Nu, L, AMD, PMD, ADL and PVL plates. Later, it was recorded

— 276 —

from the Lianhuashan Formation of Liujing, Guangxi, but no further anatomical character bas

been added (S.-F. Liu 1992).

The AVL plate of C. gracilis bears a brachial articulation, and has the very fine-grained

tubercular ornamentation, like the other plates referred to C. gracilis which were found in the

saine localiiy and horizon (K.-J. Chang 1978; G.-R. Zhang 1978). As in C. dongmoensis (TONG-

DzuY & Janvier 1990), the lubercles on the area antérolatéral to the parabrachial process (ppbr,

Fig. I8A) and on ihe oblique ridges of the trunk-shield (r.obl, Fig. 18A) are slightly larger than

those on the rest of the plate,

The ventrolaleral ridge (vlr, Fig. 18C) divides the plate into ventral and latéral laminae

(11m, vlm, Fig. 18), which meet at an almost right angle. The latéral lamina is relatively short

and high, as in C. dongmoensis (TONG-DZUY & JANVIER 1990). Considering the latéral lamina

of the PVL piale (G.-R. Zhang 1978), which is very long and low, it could be inferred that

vlm

Fig. 18. — Chuchinolepis gracilis K.-J. Chang, Xitun Formation, Qujing. Right AVL plate in ventral (A), dorsal (B), latéral (C)

and postérolatéral (D) views. (Scale bar 5 mm.)

— 277 —

the PVL plate of C. gracilis is as elongated as in C. dongmoensis (ToNG-DzuY & Janvier

1990, Fig. 18). The AVL plate overlap.s the ADL plate along its posteriorly slanting dorsal margin.

The overlap area is very narrow. In C. dongmoensis, the latéral lamina ol the AVL plate seems

to ascend posteriorly. The posterior margin of the latéral lamina, which overlaps the PL and

PVL plates, is nol emhayed, as in C. dongmoensis. On the external surface of the latéral lamina,

there are two oblique ridges (r.obi, Fig. I8C, D), as in C. dongnu/ensis and C. qujingensis.

The slightly posteriorly exlended parabrachial process (ppbr. Fig 18) is antérolatéral to the

latéral lamina. Sinee the dorsal margin of the parabrachial process (dm.ppbr. Fig. 18B) forms

an acute angle with the latéral lamina, the brachial dépréssion between ihcm is more or less

closed. The ventral margin of this process (vm.ppbr, Fig. I8D) prolongs lowards the venirolatcral

ridge by a gently curved ridge. On the lip of the parabrachial proce.ss, there is an oval fossa

for the attachment of lhe abductor muscle of fin (f.ab. Fig. 18A. C. D). On the inner surface

of the parabrachial process, there is a groove (gpbr, Fig. ISA) which was tetmed by Younc; &

Zhang (1992) as the “funnel groove". The perichondrally ossified scapulocoracoid (.scap,

Fig. 18D) extends outwards from the bottom of the pectoral fenestra. The distal end of the

scapulocoracoid is trilobate, as in C. tjujingensis. In the brachial dépression, two other dépréssions

or fossae are visible. One is the ventral articular dépréssion for the dermal process of the pectoral

fin (art.v, Fig. 18A, C, D), which lies ventrally to the pectoral fenestra. In many spccimens, the

siebknochen texture is visible on the surface of this ventral articular dépréssion. The other is

the dépréssion for the attachment of the adductor muscle of the fin (f.ad, Fig. 18C, D), which

lies mesiodorsally to the pectoral fenestra.

Anteriur médian dorsal plates from lhe Xishancun Formation (Fig. 19, pl. VIII, 1-4)

The AMD plate resembles ver>' much that of C. gracilis from the Xitun Formation (K.-J.

Chang 1978; G.-R. Zhang 1978). Only one external mould of the .AMD plate (V 10503.4) was

discovered, and its ornamentation is found to be made up of very fine tubercles The spccimens

of C. gracilis from the Xishancun Formation are smaller than those from the Xitun Formation.

The largest one (V 10503.2. Fig, I9C) is about 12 mm long. Some specimens .should bclong to

juvénile individuals, however. it is possible that in the earlier period. C. gracilis is represented

by a smaller form.

In general, the AMD plate is long and narrow, with a length/width index ranging from

149 to 176. Both the anterior and posterior margins are narrow. The overlap relationships

with the adjoining plates are constant in ail specimens. In viscéral view, the anterior ventral

process (prvl, Fig. I9A-C, E-F) lies far in front of ihc level of the latéral corners. This

process is more or less elongated in shape, and seerns to be devoid of the anterior ventral

pit. Behind this process, there develops a médian ventral groove (grm. Fig. I9A-C. E-F),

which is prolonged to the po.sterior margin of the plate by a faim médian ventral ridge (mvr,

Fig. I9B). The samc is found in the holotype of C. gracilis and in C. dongmoensis (TONG-

Dzuy & Janvier 1990). K.-J. Chang (1978) and G.-R. Zhang (1978, Fig. 14) assumed that

there is a short médian ventral ridge immediately behind the anterior médian ventral process.

However, our examination of the type specimen shows that there is not such a subdivision

between the process and supposed médian ventral ridge, and the elongated process is followed

posteriorly by the médian ventral groove.

— 278 —

FtG. J9. — Chuchinolepis gracilis K.-J. Chang, Xishancun Formation, Qujing. Ail spccimens Hgured are elaslomer casts. A, AMD

plate in viscéral view, VI0503.3; B, AMD plate in viscéral view. V10503.5: C. AMD plate in viscéral view, V10503.2; D,

AMD plate in cxlemal view. V10503.3; E, AMD plate in viscéral view, VI0503.6; F, AMD plate in viscéral view, VI0503.4.

(Scale bar 5 mm.)

Juvénile specimen of Chuchinolepis gracilis from the Xishancun Formation (Fig. 20,

pl. VIII, I).

This .specimen belongs to a juvénile individual, and consists of a small, articulated skull-roof

and trunk-shield, preserved as an internai mould.

The skull-roof is incompletely preserved. and lhe premedian, left latéral, postmarginal and

paranuchal plates are missing. When restored. the Juvénile skull-roof of C. gracilis appears as

relativcly long and narrow, like that of Minicrania aniiqua (Zhu & Janvier 1996). If we assume

that the skull-roof of the adult C. gracilis had the similar morphology as that of C. dongmoensis

(TonG-Dzuv & Janvier 1990), it can be inferred thaï, during the growth, the skull-roof was

expanding laterally. Therefore. it is reasonable to regard the short and broad skull-roof as apo-

morphic for antiarchs. The latéral plate (L, Fig. 20) is poorly preserved, with its anterior extremity

missing. The orbital notch, the margin for the postpineal plate, and the groove for the infraorbital

canal are visible. This plate lias the same shape as thaï of the adult (C.-R. Zhang 1978). The

postpineal plate (Pp, Fig. 20) is roughly oval and transverse in shape, and excludes the nuchal

plate (Nu, Fig. 20) from the orbital fene.stra (fe.orb, Fig. 20). Its anterior margin is fairly curved.

As to the relative proportions, the juvénile postpineal plate seems shorter than that of the adult.

The nuchal plate (Nu, Fig. 20) is long and narrow with a shallow anterior notch for the postpineal

plate, similar to that of C. dongmoensis. Proportionally, the juvénile nuchal plate is even longer

— 280

and narrower than lhat of the adull. and its latéral corner is less conspicuous. The postmarginal

plate (PM, Fig. 20) is relatively large. However, contrary to the adult one, it does not extend

laterally to form a distinct angle. Therefore, it has a triangular shape, unlike the rhombic shape

of the postmarginal plate of the adult. The paranuchal plate (PNu, Fig. 20) is relatively small,

and its anterior portion is narrow. The infraorbital groove passes through the plate lengthwise

and, as in the nuehal plate, the obtected nuchal zone extends along ils posterior margin. The

obstantic margin is slightiy inclincd anteriorly, whereas in lhe adult of both Yunnunolepis and

Chuchinolepis, it has a more anterior oblique orientation.

The tnmk-shield is fairly long and narrow, as in C. dongmoensis (TONC-DZUY & Janvier

1990). The AMD plate (AMD, Fig. 20), with miich narrowed anterior and posterior margins, is

similar to the isolated one.s described above. The PMD plate is not preserved. Flowever, from

the short posterior dorsal margin of the PDL plate, it is inferred thaï lhe juvénile PMD plate of

C. gracilis is not as long as that of C. dongmoensis. The ADL plate (ADL, Fig. 20) is relatively

short, with a dorsolateral ridge. In dorsal view, its anterior margin is fairly broad, and lhe plate

is ovcrlapped by lhe AMD plate. Posteriorly. il overlaps lhe PDL and PL plates. On the latéral

lamina of the ADL plate, there is another faint oblique ridge whieh slretches posteriorly to the

PL plate. The PDL plate (PDL, Fig. 20) is much elongated, as in C. dongmoensis (TONG-Dzuy

& Janvier 1990). Its dorsal corner is posteriorly placed, wiih a long conlaci margin with the

AMD plate. Its latéral lamina is comparatively low. The groove corresponding to the cri.sla trans-

versalis interna posterior (cr.lp, Fig. 2()A, C) appears in the posterior portion of the plate. The

PL plate (PL, Fig. 20) is very similar to that of C. dongmoensis. It is low and long, extending

anteriorly as far as below lhe posterior part of the ADL plate, and bears a faint longitudinal

ridge. Along its posterior margin, ihcre is a shallow groove corresponding to the crista trans-

versalis interna posterior in the internai mould.

Chuchinolepis qujingensis G.-R. Zhang, 1984

(Figs 21-22, pis VIII, 6-12; X, 8-13)

Procondylolepis qujingensis G.R. Zhang, 1984: 82, figs 1-4, pis I-II.

,Synonym:

Procondylolepis YOUNG & ZHANG 1992: 445, figs 2F-G, 3B-D, 4-6.

Others references:

Procondylolepis qujingensis Zhu, WANG & FAN 1994: 4, fig. 2, pl. I, 4-7.

Diaonosis. — Chuchinolepis in which the ornamentation is composée) of finc-grained tuberclcs.

Holotvpe. — An incomplète AVL plate (V6y4LI).

New MATERIAl.. — V 1051)4.1-3 front the Xishancun Formation ; V10511.1-20 from the Xitun Formation.

Rr.MARKS. — This species tiiffers from C. gracilis and C. dongmoensis mainly by its tuberclcs, which are

much larger than those of C gracilis and C dongmoensis. C, tpijingensis is more suggestive of C. dongmoensis

than of C. gracilis in the latéral lamina of lhe ADL plate, which decrea.ses in heighl posteriorly. Since the

omamenlalion of C. qujingensis is similar to that of Yimnanolepis and Phymolepis from the saine locality, it is

difficuli to détermine detached plates excepi the AVL plate with the fin articulation structures, ihc AMD plate

and pectoral fin plates. Other plates which can be assigned to C. qujingeii.sis with lillle doubi include three PMD

plates from lhe Xitun Formation and two ADL platc.s from the Xishancun Formation, These PMD piales hâve

— 281 —

the same ornamentation as the AVL plate of C. qujingensis, as well as those of Yunnanolepis and Phymolepis.

However, their crisia transversaiis interna posterior is situated latéral to the posterior ventral process and pit,

in contrast. the PMD plate of Yunnanolepis and Phymolepis is specialized. wilh the crista transversalis interna

posterior passing clearly in front of the posterior ventral process and pit. Another possibility, as to the assignment

of these PMD plates, is Zhanjilepis (G.-R. Zhang 1978). However, the tubercles of Zhanjilepis are coarser, and

its PMD plate seems to be devoid of the médian ventral ridge. Therefore. il is most likely that these PMD plates

from the Xitun Formation belong to C. qujinnensis. Two ADL plaiç.s front the Xishancun Formation are specialized

by the shape of their latéral lamina. As fai as we know, the ventral murgin of latéral lamina of ADL plate in

antiarchs, except lot C. dongiiioetisis, descends posteriorly or is ut leust parallel to the dorsolaterul ridge. The

ADL plates from the Xishancun Formation hâve the maiching size and sanie ornamentation as the AMD plate

of C qujingensis from the sume site, Moreover, lhey possess the ascending ventral margin of the latéral lamina

as in C. dongmoensis. Willi refereiice to C. dongmoensis, the only .species of Chuchinolepis in which the ADL

plate is described, these two ADL plates from the Xishancun Formation, which hâve the same kind of ventral

margin as that of C. dongmoensis, might be assigned to Chuchinolepis. By the ornamentation, they are referred

to C. qujingensis.

Description

Material from the Xishancun Formation (Fig. 21, pl. VllI, 6-12)

The AMD plate (V 10504.1, Fig. 21) is roughly rhombic and elongated in shape, as in

C. gracilis and C. dongmoensis. U has the length of 24.9 mm and width of 14.7 mm, and a

length/width index of about 170. The plate has very narrow anterior and posterior margins. The

Fig. 21. — Chuchinolepis qujingensis (K.-J. Chang), Xishancun Formation, Qujing. AMD plate (V10504.1) in dorsal (A) and

viscéral (B) views (elastomer casts). (Scalc bar 5 mm.)

— 282 —

antérolatéral and postérolatéral margins are relatively long and more or less convex. In extemal

view, the plate is considerably arched with an anteriorly placed tergal angle (dma, Fig, 21A).

The mid-line length in front of the tergal angle is about 38.4% of the plate length. The médian

dorsal ridge (dmr, Fig. 21 A) is situated behind the tergal angle and extends backwards to the

posterior margin of the plate. The tubercles are fine-grained, and densely distributed, but clearly

visible to the naked eye, unlike C. gracilis and C. dongmnensin where a hand lens or microscope

is needed to see the tubercle ornamenl. In viscéral view, the elongated anterior médian ventral

process (prvi, Fig. 21B) lies beneath the tergal angle, and is prolonged posteriorly by a médian

ventral groove (gmt, Fig 21 B), and a faint médian ventral ridge (mvr. Fig. 21 B). The levator

fossa (f.retr, Fig. 21 B) and postlevator thickenings (air, Fig. 21 B) lie on each side of the elon¬

gated anterior ventral process. The overlap relationships with the adjoining plates are like those

in other Chuchinolepis species.

The ADL plate (pl. VIH, 4-12) has its dorsal lamina relatively long and narrow, and with

a broad anterior margin. Its mesial margin is overlapped by the AMD plate. The latéral lamina

is remarkable by its posteriorly ascending ventral margin. The deepest position of the lamina is

at its anterior end. The main lateral-Iine groove lies just below the dorsolateral ridge and traverses

the plate. Along the ventral and posterior margins, the plate overlaps the AVL and PDL plates.

The crista transversalis interna anterior is visible in the anterior edge of the plate.

scap

Fig. 22. — Chuchinolepis qujingensis (K.-J. Chang). Xilun Formation, Qujing. Incomplète left AVL plate in dorsal (A), ventral

(B) and posterior (B) views, VI0511.L (Scale bar 5 mm.)

— 283 —

Maierial front the Xitun Formation (Fig. 22, pl. X, 8-13)

The AVL plates of C. qujingensis from the Xitun Formation were described in detail by

G.-R. Zhang (1984) and YOUNG & ZHANG (1992). More material was collected and prepared

for the serial grinding sludy (Zhu in prep.), VI0511.1 show.s that the po.sibranchial lamina (I.pbr,

Fig. 22A) is similar lo that of C. gradlis and Yunnanolepis, which extends anleromesially and

lies horizontally onto the floor of the plate. The crista irùnsversalis interna anterior (cit. Fig. 22A,

C) is fairly high and clearly behind the poslbranchial lamina. The fossa (f.ab, Fig. 22B, C) on

the parabrachial process (ppbr, Fig. 22) is kidney-shaped and fairly deep. A low ridge is some-

times visible on the dorsolatcral surface of the fossa.

The PMD plates described here hâve the samc ornamentation as the AVL plates of C. qujin¬

gensis. In external view (pl. X, 13), the plate is gently arched and is devoid of any ndge and

process, as in C. gradlis (K.-J. Chang 1978; G.-R. Zhang 1978). The plate lias its maximum

breadth at the level of the latéral corners. In viscéral view (pl. X, 12), the médian ventral ndge

extends from the anterior end of the plate to the posterior ventral process and pit. which are

conspicuous and hâve a relatively posterior position. The crista transversalis interna posterior

lies laterally to the posterior ventral process and pit, as in C. gradlis (G.-R. Zhang 1978, pl. Vil, 3).

Chuchinolepis rohusta n. sp.

(Fig. 23, pl. IX, 1-3)

Diaonosis. — Chuchinolepis in which the ornamentation consists of coarse tubercles; pectoral fin articulation

région of the AVL plate fairly open.

Etymolocy. — From robustus (Lat.), “robust”, by reference to the coarse tubercles of the new species.

Holotype. — V10512, a detached AVL plate (the only material).

Locality and horizon. — Xitun Formation, Cuifengshan Group, Early Devonian, Qujing. Yunnan, China.

Remarks. — This specimen could be assigned lo Chuchinolepis by ils unique pectoral fin articulation struc¬

ture. It differs from othei species of Chuchinolepis by its fairly open pectoral fin articulation area of the AVL

plate and coar.se ornamentation.

Description

This specimen is a medium-sized AVL plate with a length of 38 mm. The plate is similar

to that of C. gradlis, C. qujingensis and C. dongmoensis with its typical pectoral fin articulation

structures, i.e., the trilobate, perichondrally ossified scapulocoracoid (scap, Fig. 23B, C) extend-

ing from the bottom of the pectoral fenestra, and the parabrachial process (ppbr, Fig. 23) with

a fossa for the attachment of the fin muscle.

The AVL plate is dividcd into two perpendicular latéral and ventral laminae, by the ven-

trolateral ridge. On the latéral lamina, the tubercles are fairly coarse and loosely distributed,

whereas on the ventral lamina, lhe tubercles are somewhat smaller. On the area anterior to the

parabrachial process, the tubercles are a little larger than those on the ventral lamina, however,

they are clearly smaller than those on the latéral lamina.

In dorsal view, the latéral lamina (11m, Fig. 23A) extends anteriorly far beyond the dorsal

margin of the parabrachial process (dm.ppbr, Fig. 23A). At the level of the parabrachial process,

— 284 —

Fig. 23. — Chuchinolepis wbusta n. .sp., Xitun Fonnation. Qujing, Left AVL plate in dorsal (A), ventral (B) and latéral (C)

views, V10512. (Scale bar 10 mm.)

it turns mesially to form the crista transversalis interna anterior (cit, Fig. 23A) which extends

to the mesial margin of the plate and becomes lower mesially. The postbranchial lamina (l.pbr,

Fig. 23A) has a ridged ornamentation, as in other yunnanolepidoids, and attaches on the anterior

face of the crista transversalis interna anterior. In C. gracilis and C. qujingensis, the post-

branchial lamina is siluated in front of the crista transversalis interna anterior, and has a more

oblique direction. Unlike that of C. qujingensis and C. gracilis, the parabrachial process of this

new species has a more anterior and latéral position. The dorsal margin of the parabrachial

process (dm.ppbr, Fig. 23A, C) extends more or less posteriorly and forms an oblique angle

with the latéral lamina.

In latéral view, the fin articulation structures are clearly visible. Due to the anterior and

latéral position of the parabrachial process, the articulation area of the plate is fairly open. On

— 285 —

the tip of the parabrachial process, there is a small, oval fossa for the altachment of the abductor

muscle of fin (f.ab, Fig. 23). As in the other Chuchinolepis species, a groove (gpbr, Fig. 23B-C)

descends from the parabrachial process toward the pectoral fenestra. The trilobate, perichon-

drally-lined .scapulocoracoid (scap, Fig. 23C) is relatively large and extends outwards from the

bottom of the pectoral fenestra. Since the ventral and middle lobes are very close to the ventral

lamina of the plate, the infrabrachial ridge and ventral articular dépréssion for the dcrmal process

of the pectoral tin. which are présent in the other Chuchinotepis species, arc not prescrved in

the specimen. Dorsally to the middle lobe of the scapulocoracoid, there is a relatively large

foramen for the nerves and vessels of the fin. The fossa for the adductor fin muscle (f.ad,

Fig. 23B,-C) is large and shallow, and situated behind the postbrachial ridge. Behind the fossa

is the omamented latéral lamina of the plate, which descends backwards. The lamina has a

sinuous posicrior margin, due to the anteriorly intruding PL plate, as in C. dongmuensis (ToNG-

Dzuy & JANVIER 1990).

In ventral view, only the mesial margin of the plate is somewhat damaged. The plate is

broadest al the level of the parabrachial process, and the estimated breadlh is more than 30 mm.

Behind lhe parabrachial process, the plate is about 23 mm in breadth. The ventral margin of

the parabrachial process (vm.ppbr, Fig. 23B, C) forms an obtuse angle with lhe ventrolateral

ridge. The contact margin for the semilunar plate extends posteriorly in a straight oblique line,

and does not form a notch. as in Yunnanolepia (G,-R. Zhang 1978, Fig. 6). The contact margin

with the MV plate indicates that the MV plate of C. rohusta is relatively small, as in Y. porifera.

Two or three growth lines are visible near the plate margins.

Chuchinolepis sulcata n. sp.

(Figs 24-25, pl. X, 1-7)

Procondylolepis qujingensis G.-R. Zhang, 1984 (in part); V6941.5, fig. 2a-c, pl. I, 3.

Diaonosis. — Chuchinolepis in which the rme-grained tubercles are arrangée! into the regular ridges on

the dorsal wall of the trunk-shield and the pectoral fin plates; very fine-grained tubercles distributed between

the ridges.

Etymology. — From sulcus (Lat.), groove, by reference to the shallow grooves between the ridges on the

dorsal wall of the trunk-shield.

Holotype. — Vl()513.l. a trunkshield (pl. X, 1-7).

Other material. —Three PMD plates (VI0513.2-4), an incomplète pectoral fin (VI0513.5), and a fragment

of AMD plate (VOS 13.6).

Locauity and horizon. — Xitun Formation, Cuifengshan Group, Early Devonian, Qujing, Yunnan, China.

Remarks. — This new species could bc assigned to Chuchinolepis by its typical long and narrow AMD

plate, which has its anterior ventral process far anterior to the level of the latéral corners. The différence between

C. sulcata and the other Chuchinolepis species lies maiiily in the ornamentation, tlie latter having tubercles,

either fine or coarse, deprived of any ridges. By the very tiny and densely distributed tubercles between the

ridges, the new species is more similar to C. gracilis and C. dangmoensis than to C. qujingensis and C. robusta.

The detached incomplète pectoral fin V694I 5 had been a.ssigncd to bc the paratype of C (Procondylolepis)

qujingensis (G.-R. Zhang 1984). However, this pectoral fin was definitely different from the holotype and other

pectoral fins of C. qujingensis by its regular ndges. Since the .saine kind of the regular ridges are found in the

trunk-shield of C. sulcata, this ridged pectoral fin .should be removed from the species which arc devoid of the

ridges.

— 286 —

Description

Trunk-shield (Figs 24, 25C-E; pl. X, 1-6)

This is a small to medium-sized antiarch. As in the other Chuchinolepis species, the trunk-

shield of C. sulcata (Fig. 24) is relatively long and low, with a less developed médian dorsal

ridge. Its diagnostic character is the regular ridges formed by the tubercles on the dorsal wall

of the trunk-shield.

The AMD plate (Figs 24, 25C) is very similar to that of the other Chuchinolepis species

(G.-R. Zhang 1978; Tong-Dzuy & Janvier 1990) except for the ridged ornamentation. It is

long and narrow, with narrow anterior and posterior margins. The breadth/length index is about

57, just as that of C. gracilis (G.-R. ZHANG 1978). The latéral corner is less conspicuous, sub-

dividing the latéral margin into two portions, which are roughly équivalent in length. Along the

antérolatéral margin, the AMD plate overlaps the ADL plate over the entire suture. The overlap

relationship between the AMD and PDL plates (fig. 25C) is the same as that of other

Fig. 24 — Chuchinolepis sulcata n. sp., Xitun Formation, Qujing. Restoration of the trunk-shield in dorsal view, based on the

holotype (V10513.1). (Scale bar 10 mm.)

— 287 —

yunnanolepidoids. The area overlapped by the PMD plate is rather narrow. In dorsal view, the

tergal angle has a fairly anterior position, about 1/51/6 of the mid-line length from the anterior

extremity (fig. 24). Behind the tergal angle, the médian dorsal ridge is less developed. The small

tubercles tum to form the regular ridges, in a pattern similar to that of Hunanolepis (J.-Q. WANG

1991, fig. 13). AU ridges intcrsecl the corresponding plate margins. Between the ridges, the

tubercles are tiny and densely dislributed, and are invisible to the bare eyes, as in C. gracilis.

In viscéral view, the plate exhibits a shallow fossa for the levator muscle in front of the level

of the tergal angle, as in C, dongmoensis (ToNG-DzuY & Janvier 1990, fig. 19). Immediately

behind the fossa, the plate bears a relatively elongated anterior ventral process (prvl, fig. 25C),

which is prolonged posteriorly by a long médian ventral ridge (mvr, fig. 25C).

Three detachcd PMD plates are known. in addition to the PMD plate of the holotype. The

plate has the same shape as that of other Chuchinolepis species, among which the new form is

more suggestive of C. dongmoensis (ToNG-DZUY & JANVIER 1990) than of C. gracilis as to the

length/brcadih index (about 150 in C.. sulcatn and C. dongmoensis, about 110 in C. gracilis).

In external view (Fig. 24), the plate is more or less arched, forming a faint médian dorsal ridge

and a low posterior dorsal angle. Along the médian dorsal ridge and the posterior margin of the

plate, the small tubercles are closely set and do not form the ridges. However, laterally to the

Fig. 25. — Chuchinolepis sulcata n. sp., Xilun Formation, Qujing. A-B, incomplète pectoral fin in latéral (A) and mesial (B)

views, V10513.5; C. AMD plate of the holotype (VI0513.1) in viscéral view (elastomer cast); D, PMD plate in viscéral

view, V105I3.2; E, PMD plate in viscéral view, VI05I3.3. (Scale bar 5 mm.)

— 288

médian dorsal ridge, the smail tubercles form the ridges, which are more or less perpendicular

to the latéral margins. Between the ridges. tiny tubercles could be observed under the microscope.

In viscéral view (Fig. 25D-E), the plate resembles very much that of C. gracilis (G.-R. Zhang

1978). The crisia transversalis interna pasterior (cr.tp, Fig. 25D-E) lies laterally to the well-

developed posterior ventral process and pit (prv2, pt2, Fig. 25D-E), Laterally, the plate overlaps

the ADL plates (cf.PDL, Fig. 25D-E).

The ADL plate is preserved in the holotype, however, its latéral lamina and posterior ex-

tremity are damaged. Externally, the same ridged ornamentation as the AMD and PMD plates,

and the area overlapped by the AMD plate are visible. As in the other Chuchinolepis species,

the ADL plate is long and narrow. The dorsal lamina has a rclatively hroad anterior margin,

ventrally to which cxtcnd the articular fossa and crista transversalis interna anterior.

The PDL plate is relalively long and narrow. As in other yunnanolepidoids, its latéral lamina

is very low, indicating the presence of an independent PL plate. The dorsal corner of the dorsal

margin is conspicuous. On the dorsal lamina, the ridges continue from the AMD and PMD plates

in external view, whereas on the latéral lamina there are three longitudinal ridges parallel to the

ventral margin. The overlapped areas by the PL, ADL, AMD and PMD plates are very clearly

seen in the holotype. In viscéral view, the overlap area for the AMD plate is observed.

Pectoral fin (Fig. 25A; pl, X, 7)

The pectoral fin of C. sulcaia is very suggestive of thaï of C qujingensis (G.-R. Zhang

1984) except for iu ornamentation. As in the trunk-shield, the ornamentation of the pectoral fin

plates is composed of two types; one consists of small tubercles tending to form the regular

ridges on the latéral surface, the other consists of very tiny tubercles between the ridges and

on the mesial and ventral surfaces.

Unnamed antiarch

(Fig. 26A, pl. IV, 12)

IXichonelepis sp. Janvier, 1995: 153, MNHN-CHD02, Fig. 7.

New mathriai., — V10515 from the Xi'tun Formation of Qujing, Yuiman.

Rf.MARKS. — An AVL plate (MNI-IN-CHD02) fromthe Ximii formation (Karly Devonian) of Qiijing. Yunnan

was referred to Xichonalepis with a question mark by Janvier (1995). That specimen bears definitely the same

large ventral fenestra of the trunk-shield as in the .Sinolcpididae (Ritchie et al. 1992) and Vanchienolepis (Tonc-

Dzt;v & Janvier 1990), however. ii retain.s the .simple pecioral fin articulation as the Yutmanolepididae and should

represent a new form of the Antiarcha. A detached PVL plate tV. 10515) which could be referred to Ihis new

form was found by the author from the same locality and horizon, and ils description is given below.

Description

V10515 has a length of 5.4 cm with a very low latéral lamina. Along its dorsal margin is

the area overlapping the PL plate (cf. PL, Fig. 26A). The ventral lamina is rim-like as in the

Sinolcpididae. Its mesial natural margin shows neither the overlapped area nor the overlapping

area, indicating the presence of a large fenestra of the ventral wall. In viscéral view, the crista

transversalis interna posterior (cr.tp, Fig. 26A) lies close to the posterior end of the plate.

— 289 —

Suborder YUNNANOLEPIDOIDEI gen. et sp. indet.

(Fig. 26B, pl. IV, 11)

Matériau. — VI0506 from the Xishancun Formation of Qujing, Yunnan.

Description

This PDL plate (VI0506) is characterized by its oraamenlation, the tubercles forming many

nodes (n, Fig. 26B) on the surface. The nodes are either rounded or elongated, and fairly

developcd on the anterior pan of the plate and along the dor.solateral ridge (dlr, Fig. 26B).. When

compared to the PDL plate of other yunnanolepidoids, this plate looks relatively large, with a

length of 4.2 cm. The dorsal lamina has a posieriorly placcd dorsal corner (d, Fig. 26B), behind

which is tlie area overlapped by the PMD plate (oa.PMD, Fig. 26B). Anteriorly, the plate is

overlapped by the ADL plate. The latéral lamina fomis a more or less blunt angle with the

dorsal lamina. Its ventral lamina is overlapped by the Pl. plate (oa.PL, Fig. 26B).

PECTORAI. fin articulation of THE Chuchinolepididae

Since the pectoral fin articulation of the Chuchinolepididae (= Procondylolepiformes of G.-

R. ZhanCi) had been described and analyzed by G.-R. Zhang (1984), this puzzling problem

was discussed at length by Tong-Dzuy & Janvier (1990), Younü & ZtlANG (1992) and Janvier

(1995). However, no consensus of opinion has been reached. Since knowledge will gain by the

réfutation or corroboration of the hypothèses, another hypothesis on the basis of new observations

is proposed here. The review of the previous hypothèses about the brachial articulation of the

Chuchinolepididae will be helpful to illustrate our hypothesis.

Fig. 26. — A, unnamed antiarch : right PVL plate in viscéral view (elasiomer cast), V10515. Xitun Formation, Qujing; B. Yun-

nanoiepidoidei gen. et sp. indet. : right PDL plate in external view (elasiomer casi), VI0506. Xishancun Formation. (Scale

bar 10 mm.)

— 290 —

Hypothesis of G.-R. Zhang (1984)

This hypothesis suggested that the trilobate, perichondrally ossified scapulocoracoid exposed

in the bottom of the pectoral fenestra was a primitive brachial process which later evolved into

the brachial process of euantiarchs (including the Sinolepididae at that time). With regard to the

dernial brachial articulations between the AVL plate and pectoral fin, three joints were proposed

betwecn the f.ab (Fig. 22C) and ar3v (Fig. 27). between the f.ad (Fig. 22C) and a restored process

of his mml plate (Cdl, Fig. 27), and between the art.v (Fig. 22A, C) and a restored process of

his Cvl plate (Mml. Fig. 27).

The first question raised by this hypothesis is the orientation of the pectoral fin. Since the

available pectoral fins of the Chuchinolepididae are disarticulated, ail reconstructions of the fin

orientation are hypothetical. For the moment, we should décidé which one is the most acceptable,

since it affects directly the interprétation of pectoral fin plates of the Chuchinolepididae. As in

A B

Mm1 Mm2 Mm3 Mm4

Fig. 27. — The left pectoral fin of Chuchinolepis qujingensis restored in latéral (A), mesial (B) and ventral (C) views (modified

after G.-R. ZHANG 1984, fig. 4). (Scale bar 5 mm.)

— 291

the Euantiarcha and Sinolepididac, the pectoral fin of the Chuchinolepididae is covered by four

sériés of dermal plates and is roughly triangular in transverse section. G.-R. ZHANG (1984) re-

ferred the three walls of the fin to the dorsal, latéral and ventral walls. His main argument was

that the tubercles of the dorsal surface were coarser and more regular than tho.sc of the ventral

surface. However, this kind of fin orientation is Just contrary to what happens in euantiarchs,

e.g. Remignlepis. Functionally, the orientation proposed by G.-R. ZHANG (1984) seems unrea-

sonable since the latéral wall would be higher than the mcsial wall, and this would hinder the

movement. Moreover, this hypothesis was correct in supposing that the tubercles of the dorsal

surface were coarser than those of the ventral surface, when compared with the trunk-shield.

But what about the ornamentation of the latéral surface? If we lakc the trunk-shield as a référencé,

in general the omamcnt.s of its latéral wall, arc like those of its dorsal wall, coarser than those

of the ventral wall. The finer ornamentation of the ventral wall was probably due to ils frequent

contacts with the .substrale. The mesial wall of the fin in the Euantiarcha and Sinolepididac has

a fine ornamentation becau.se it faces the latéral wall of the trunk-shield. In the Chuchinolepididae,

if we accepted the explanation of G.-R. Zhang (1984), the ornamentation of the latéral wall

was as fine as that of the ventral wall (G.-R. Zhang 1984, fig. 2). that is contrary to the trunk-

shield. However. if we interpret the fin orientation of the Chuchinolepididae like that of the

Euantiarcha and Sinolepididac. we mect difficulties in the explanation of the dermal brachial

articulation, as wi|| be di.scussed below.

The second question is about the the dermal brachial articulation. As stated by G.-R. Zhang

(1984), there were three dermal joints, among which the joint between the f.ab and ar3v func-

tioned as a fulcrum whereas the others rolated with in a very limited angle. However. among

ail of the pectoral fin specimens. only one kind of dermal articular process has beeii found.

Moreover, the fossa on the parabrachial process, which was regarded by G.-R. Zhang (1984)

as the articular dépréssion recciving the articular process of the fin, seems un.suitablc for the

articulation, even only as a pivot for the rotation of the fin. Our examinations of the parabrachial

fossa (f.ab of this work) show that the fossa lacks ihc siehknnchen texture, typical of the dermal

brachial articulation, and its shape and si/.e makc obstacles to the explanation of its function as

an articulation. The parabrachial fossa is peariform and variable in size, sometimes being fairly

deep. We suggest that the parabrachial fossa is more likely to be a muscle insertion area, as for

the adductor fossa (YouNc; & Zhang 1992).

Hypothesis of YoVNO & ZHANG (1992)

Young & Zhang (1992) suggested that there are only two (not three) proximal dermal

brachial articulations in the Chuchinolepididae, and the brachial articulation of chuchinolepids

is transitional between that of yunnanolepids and that of sinolepids and euantiarchs. They tried

to locate homologies in the structural components of the brachial articulation between the Chu¬

chinolepididae on the one hand and the Sinolepididac -i- Euantiarcha on the other.

The first probleni concerns the axillary foramen. As to the function of this structure, there

had been many divergent opinions. Stünsjô (1931, 1959) and Gross (1933) regarded il as a

passage for the adductor muscle for the pectoral fin. In the samc lime they supposed that the

axillary foramen transmitted aiso nerves and vessels. Watson (1961) noted the small size of

the axillary foramen in Pterichthyodes compared to that of Botliriolepis, and suggested that it

was too small for a muscle to pass, and must hâve been a passage only for the nerves and

— 292 —

Fig. 28. — Restoration nf CUuchinolepis in Icfl laicraJ view. Trunk-shield and skull-roof afler Tono-Dzuy & Janvier (1991,

fig. 18), pectoral fin aller G.-R. Zhang (1984, fïg. 4). (Scale bar 1(1 mm.)

vessels. This idea has been developed by Young & ZHANG (1992), who considered that the

axillary foramen was homologous to the small foramina piercing or encircling the scapulocoracoid

in the Chuchinolepididae and other placoderms. However. considering the différences in size

between the axillary foramen and the foramina of the Chuchinolepididae. it is difficult to un-

derstand this corrélation. If the axillary foramen was a passage only for the nervcs and vessels,

it would be too large in many cuantiarclis. 1 suggcsl here that the axillary foramen of Sinolepid-

idae and Euanliarcha is homologous to the pectoral fenestra of the Yunnanolepididae, Chu¬

chinolepididae and other placoderms. They hâve the same position relatively to the crista

tran'iversalis interna anierior. In addition to a passage for the vessels and nervcs, the axillary

foramen might hâve contained the cartilaginous scapulocoracoid, as in the Yunnanolepididae,

where it forms an endoskeletal joint with the fm. Since the Sinolepididae and Euanliarcha hâve

developed the brachial process and funnel pit (an apomorphy), and the cartilaginous scapulo-

coraeoid in the funnel pit has formed a new endoskeletal joint with the fin, the part of the

scapulocoracoid through the axillary foramen might hâve degenerated in some euantiarchs, e.g.

Pterichthyndes.

Another problem is the insertion area for the abductor muscle of the fin in chuchinolepids.

It is reasonable to assume that the postérolatéral pit (f.ad) on the AVL plate is the insertion area

for the adductor muscle as reinterpreted by YoUNG & Zhang (1992). Since the parabrachial

fossa was considered as the dorsal articular dépréssion. Young & ZHANG (1992) assumed the

région between the f.ab and art.v to be the area for the attachment of the abductor muscle of

fin. However, like G.-R. Zhang (1984), we cannot fmd any definite dépréssion for this attach¬

ment. Moreover, this position seems to be too ventral for the abductor muscle of fin.

Hypothesis in this woric

Since the brachial articulations of the Chuchinolepididae and Sinolepididae + Euanliarcha

are quite different in structure. I suggest that they hâve evolved independently from the primitive

one, which is retained by some yunnanolepidoids. The trilobate, perichondrally ossified scapulo¬

coracoid exposed in the bottom of the pectoral fenestra, as well as the parabrachial process and

— 293 —

fossa, is the synapomorphy of the Chuchinolepididae, whereas the brachial process and funnel

pit are the synapomorphies of lhe Sinolepididae + Euantiarcha. The axillary foramen is ho-

mologous to the pectoral fenestra. In the Chuchinolepididae there is only one proximal dermal

brachial articulation, that is between the art.v of the AVL plate and the ar3v of the pectoral fin.

The fossa on the parabrachial process is the dépréssion for the abductor muscle of fin.

As discussed above, the fin orientation of the Chuchinolepididae in G.-R. Zhang’s (1984)

hypothesis is tlawed with some inconsistencies. For example, the latéral surface of the fin has

the same fine ornamentation as the ventral surface. However, if we accept a fin orientation

similar to that of euantiarchs, i.e. the latéral surface of the fin sensu G.-R. Zhang (1984) be-

coming the mesial (inner) surface of the fin, then we cannot make it compatible with the dermal

brachial joint, since the articular process of the fin is at the mesioventral margin, whereas the

articular dépréssion on the AVL plate would be in a ventrolaleral position. Here we propose

another possibility, i.e. the pectoral fin in the Chuchinolepididae has a laterally compressed shape.

with latéral, mesial and ventral surfaces (Figs 27, 28). In this new model, the ornamentation of

the mesial and ventral surfaces is finer than that of the latéral surface. The only dermal brachial

articulation, more or less ventral in position, is between the art.v and ar3v. Thi.s model is sup-

ported by the two following facts: the preserved dermal plates of the pectoral fin in Fhymolepis

(G.-R. Zhang 1978; M.-M. Zhang 1980; Young & Zhang 1992), which suggesi a laterally

compressed fin in this primitive antiarch, and the dorsoventrally extending scapulocoracoid in

the Chuchinolepididae (G.-R. Zhang 1984; Young & Zhang 1992).

Consequently, the dermal plates of the pectoral fin in the Chuchinolepididae are renamed

(Table 1, Fig. 27). Like the Sinolepididae and Euantiarcha, the Chuchinolepididae hâve four

sériés of dermal fin plates, and the first latéral marginal plate is missing from the proximal plate

ring of the fin.

Tableau 1. — The terminology (in abbreviations) proposed by the varions authors about the dermal plates of the pectoral fin

in the Chuchinolepididae.

This work

Young & Zhang

(1992)

G.-R. Zhang

(1994)

Tong-Dzuy & Janvier

(1990)

Cd1-4

Mm1-4

Cvl

Cdl

Mil

Cv2-4

Ml 2-4

—

Mml-4

Cvl-4

MI2-4

Cd2-4

Cdl-3

—

ar3v

ar3d

ad P

cdpbr

—

ar3v

avp

—

art.v

f.aPv

f.ibr

f.ab

art.d

f.aPd

fpbr

—

f.ab

—

f.ad

s. art

—

gpbr

igr

—

ppbr

prc

—

ppbr

scap

p.pb

—

— 294 —

CLADISTIC ANALYSIS OF THE ANTIARCHA

Historical background

The study of aniiarchs can bc iraccd back tu Eichward (I84Ü) who erected two antiarch

généra (Bafhriolepix and Asterolepix). When the name Placodermi was proposed by M’COY

(1848) for an extinci jawed fish group aniung lhe "ganoids”. three généra of anliarchs (Bothri-

olepis. Asterolepis and Pterichlhys [«tnt’ Ptericlithyodesl) were included in the group; the other

three généra were an arthrodire Coccnsteus, a ptyctodontid Chelyophonis and a heterostracan

agnathan Psainmosieiis. COPK (1885) coincd the namc Antiarcha to distinguish the fishes such

as Bothriolepis and Asteivlepis from lhe other carly vertebrates. The other researchers on antiarchs

in lhe last centiiry incliide Mll.LKR (1841), ACASSIZ (1843, 1844, sce also ANDREWS 1982),

Pander (1856, 1857), Traquair (1888, 1893) and Woodward (1891).

As lo lhe affinity of aniiarehs, COPE (1889) and WOODWARD ( 1891 ) rejecled the homogeneity

of placodernis. and suggcsled thaï antiarchs were closely relaied with ‘osiracoderin’' agnathans,

wherea.s they generally united ptyctodontids lo holoccphalans, and arthrodires to dipnoans. This

opinion was followed by a inajorily of auihor.s uniil Stensio (1931) and Gross (1931), who

retumed antiarchs to placodernis. However, the status of anliarchs among placoderms still re-

mained iinsctllcd (,sec GouiET 1984a), and their classification and interrelationships has under-

gone gréai changes (sec Md.hs 1968).

Stensiü (1931. 1948. 1959) subdivided antiarchs into two main groups. one for Remigolepis

whose pectoral fin wa.s unjoinied, and one for lhe remaining généra which possessed the distal

joint of the pectoral fin. This proposition had strong intluence on the antiarch research and has

been adopled fora fairly long lime (Bero 1940; ObruchEV 1964; SteNSIÔ 1969; G.-R. ZHANG

1984; PANTEEEYHV 1992).

Gross (1965) proposed that, as the whole cxoskeleton is concerned, major subgroups of

anliarchs .shoiild he the asterolepidoids, represented by Axteralepis, and bothriolepidoids, ex-

emplified by Baihriulepis. Remigolepis was regarded as closely related to Asterolepis, and the

absence of the distal joint was assumed to be sccondary. At that lime, the yunnanolepidoids

from lhe Early Devonian of China (Y.rH. Lru 1963; M.-M Chang 1966), which tumed out to be

early représentatives of antiarchs, began to raise interest. and a new suborder of antiarchs, the Yun-

nanolepidoidei, was erected by Miles (1968; sec al.so Gross 1965) to distinguish Ywmanolepis

from the Bothriolepidoidei and Asterolepidoidei. However. sincc lhe detailed Works about Early

Devonian antiarchs of China had not been publishcd. this rank of yunnanolepidoids was not

adopled by DENtsON (1978), who referred Yuimanolepis to the Bothriolcpidae (= Bothriolepidoidei

Miles 1968). In contrast. Df.mson (1978) had acknowledged the Sinolepididae (Liu & P'an 1958),

which was originally u.s.signed to asterolepidoids, as the third major antiarch group. lhe other two

groups being the Bothriolcpidae and Asteolepidae (= Asterolepidoidei .Miles 1968). Hemmings (1978)

removed the family Microbrachiidae from lhe Asterolepidoidei to the Bothriolepidoidei.

Important works on the anatomy and taxonomy of Early Devonian antiarchs from South

China were undertaken by G.-R. Zhang (1978). P’AN & WANG (1978) and M.-M. Chang (1980),

and confirmed that the yunnanolepidoids were the most primitive known forms of antiarchs,

which provided links between lhe advanced antiarchs and other placoderms. However, until then.

— 295 —

these inquiries into the antiarch phylogeny were mainly within the framework of the evolutionary

systematics.

The first cladistic analysis of anliarchs was that of Janvier & Pan (1982). This cladogram

suggested e.xplicitly four major groups (Yunnanolcpiformcs (= Yunnanolcpidoidci Milc.s, 1968),

Sinolepis (the only genus of ihc Sinolepididae then known), Astcrolepida (= Astcrolepidoidei)

and Bolhriolepida (= Bolhriolepidoidei)) within antiarchs, although ihc monophyly of the

Yunnanolepiformes was unccrtain. It was first proposed thaï Sinolepis is tlie sisier-group of the

asterolepidoid + bothriolepidoid pair, and the Sinolepididae was rcmoved front lhe

Astcrolepidoidei. l■'ollowing HemmINGS (1978), Micmbrachius, as well us Wudinolepis, was

placed aiiiüiig the Bolhriolepida. Ilowever, ihis cladogram was incomplète since many antiarch

taxa were not included in the analysis.

Long (1983) proposed a modified classification of MILES (1968). In this grouping, the

Microbrachiidae was still placed within the Astcrolepidoidei, and the Sinolepididae were removed

from the Astcrolepidoidei.

Important contributions to antiarch phylogeny, as well as to paleobiogeography, hâve becn

madc by YoifNG (1984c, 1988) who provided us wilh a fairly complété cladogram of antiarchs,

which is the basis of the présent study. In YoUNG's cladogram, four major monophyletic groups

of antiarchs were rccognizcd. as in Janvier & PAN (1982). and yunnanolepidoids were placed

at the root of lhe cladogram. However, .sinolepids were ranked as lhe .sister-group of bothn-

olepidoids which included lhe Microbrachiidae, and lhe pair of sinolepids + bolhriolepidoids

formed the sister-group of asterolepidoids {cf. J.ANVIER & PAN 1982; RITCIIIE et al. 1992). In

his biogeographic analysis. YOUNG (1984c) pointed out that South China was the area of origin

(or part of the area of the origin) of ail major antiarch groups except for asterolepidoids. However,

in my opinion, a fact should be taken into considération, i.e. the Astcrolepidoidei has a relatively

younger distribution than lhe other antiarchs.

Youno (1990) found some features of Nawagiaspis suggestive of the Astcrolepidoidei

whereas others are suggestive of bolhriolepid alTinity, and the placement of this genus within

his cladogram is rather difficult. In fact, when the range of the Bothriolepidoidei is expanded

to include the Microbrachiidae and Dianolepis in addition to lhe Bothriolepididae sensu stricto,

the boundary belween the Astcrolepidoidei and Bothriolepidoidei becomes more and more ob¬

scure. Another example is Jiang.xilepis (ZtlANC & Lie 1991), which was originally referred to

the Bothriolepidoidei. However, il has many characters of the Astcrolepidoidei, e.g., lhe narrow

anterior margin of lhe AMD plate, lhe loss of the anterior ventral process. and lhe mesially

placed tergal angle. The same problem was encountered in Luquatwlepis (ZHANG & YoUNG

1992), an Early Devonian bothriolepidoid from South China. This antiarch has the independent

PL plate as in Nawagiaspis and some asterolepidoids, and the anterior margin of ils AMD plate

is supposed to be naiTow. In my opinion, ail these problems are related with the problem of the

monophyly of lhe Bothriolepidoidei. which should in turn be resolved within the framework of

the entire antiarch phylogeny.

Zhang & Young (1992) made the phylogenetic analysis of the Bothriolepidoidei using the

Hennig86 program. This was the first phylogenetic work for antiarchs based on detailed character

analysis with a parsimony program. However, lhe results are not so successful as désirable. The

— 296 —

source of the problem, in my opinion, is the presumed monophyly of the Bothriolepidoidei,

which will be examined below on the basis of the phylogenetic analysis of 40 antiarch généra.

Materials and methods

Data

By now. there are 44 antiarch généra (154 species) described in the literature and an unnamed

genus (Janvier 1995; this work). Five of them (Yunlongolepis, Chuanbeiolepis, Taeniolepis, Shi-

menalepis and Hyrcanaspis) are not included in the présent analysis becaiise they are too poorly

known.

Taeniolepis (Gross 1932), from the Upper Devonian (Frasnian) of Latvia, is represented

by nuchal and postpineal plates. Us nuchal plate is excluded from the orbital fenc.stra, and bears

no sensory-line groove, thereby resembling more or less sinolepids and some asterolepidoids.

Hyrcanaspis (JANVIER & PAN 1982) is known from the Early Devonian (Emsian. Janvier

pers. comm.) of Iran. The described specimens includc some disarticulated trunk-shicid plates

and a central plate of the pectoral tin. Its AVL plate bears a relatively large axillary foramen,

somewhal similar to that of Microbrachius, and the dennal plates are omamented with very

thin, almosi parallel ridges. Il is distinguished from othcr euantiarchs by its almosl ventral position

of the precondylar ridge of the brachial process. Hynicanaspis, as well as Microbravhius, was

provisionally regarded as a bothriolcpidoid by Janvier & Pan (1982).

Both Yunlongolepis and Chuanbeiolepis are found from the Early Devonian of Longmenshan,

Sichuan. China (,S.-T. Wanc in HotJ et al. 1988a), The AMD plate of Yunlongolepis, the only

specimen of ihis genus, looks like that of Yunnanolepis, however it lacks the anterior ventral

process and médian ventral ridge. Chuanbeiolepis, a poorly preserved internai mould of ihe trunk-

shield. is suggestive of CInuhinolepis by its anteriorly placcd anterior ventral process. and the

crisia iransversulis interna posterior lying laterally to the posterior ventral proce.ss.

Shimenoiepis, the earliest représentative of anliarchs (J.-Q. Wang |99|b), is known from

the Early Silurian date Landoverian) of Lixian, Hunan, China. However, the information about

this genus is quite scarce. since only an internai mould of the PVL plate (the type specimen)

was referred to il, This PVE plate lias the cri.sia iratisrersalis interna posterior lying very close

to the posterior extremity of the trunk-shield, as in yunnanolepidoids, and this may be plesio-

morphic for antiarchs. Its latéral lamina has a fairly constant height. as in Zhanjilepis (G.-R.

Zhang 1978), and therefore. it was suggesled to be a Yunnanolepidoidei by 1,-0- WANG (1991b).

Therc is no impression for the ventrolateral fossa of the ntmk-shield, which occurs in Yunnanolepis

and Fhyniolcpis. Notewprlhy is an internai mould of the AMD plate from the same level, which

was referred to as an undelermined form of the Chuchinolepididae by J.-O- Wang (1991b).

However. this specimen could aiso belong to Shimenoiepis, since its size and ornameni are com¬

patible with thosc of the holotype. Like the AMD plate of Chuchinolepis, this plate is fairly

long and narrow, and has the clongated, anteriorly placed anterior ventral proce.ss. The Chuchinolepi-

didac (Chuchinolepis) is rather derived considering its pectoral fin articulation, but whether this

early Silurian antiarch shares this derived feature of the Chuchinolepididae or jusl retains the

simple fin articulation as in the Yunnanolepididac is unknown. It is probable that the anterior

ventral process of the AMD plate, which is anteriorly placed, as in Chuchinolepis and this Silurian

antiarch, is plesiomorphic for antiarchs. More material is needed to clarify this problem.

— 297 —

Ten gcnera aniong the rest of antiarchs hâve a poorly known skull-roof [Vietnamaspis (pre-

median plate), Briagalepis (latéral plate), and Monarolepis (luichal plate)] or completely unknown

skull-roof {Zhanjilepis, unnained antiarch, Vanchienolepis, Liiquanolepis, Grossaspis, Lepadolepis

and Wuningulepis). However, since they hâve been well eslablished by other diagnostic charac-

ters, they are included in the phylogenetic analysis in order to minimize the subjectivity in select-

ing the OTU’.s.

Among the analyzcd antiarch taxa (40 gcnera, including the unnamed antiarch which is

indeed a new genus). 22 of thern are monospecific For the other généra, we suppnsed a priori,

that ail of them are monuphyletic and can be used as tenuinal taxa in the phylogenetic analysis.

Characlers anil character codings

Sixty-six characlers werc sclected for the character analysis. Since the sélection of the

characters is the first levcl of a hypothesis, the character independence is taken into considération

and ail characters are ireated as unweightcd in order to avoid biases as much as possible.

As to the multiple-siaie characters, the character analysis is donc to consiroci the character

cladogram. For the branching transformation sériés, the mixed coding (Wiley et al. 1991) is

adopted and one character may be subdivided into .scveral characters. Thcrcforc, each character

used below is a linear transformation .sériés and is Ireated as additivc.

As to the missing data, two situations may occur : onc is the logical inipossibility (coded

in the data matrix as—), and the other is the unavailable information (coded in the data matrix

as ?) due to either poor préservation or absence of fossils. However, bccausc of the constraints

of fossil préservation, therc is a large nuinber of mi.ssing data in the data matrix, that makes

the results of the phylogenetic analysis ncarly uninformative. Two options hâve been suggested

to overcome this difficulty, One is to exclude the leasi known characters or/and the poorly known

taxa from the data set, thereby making the analysis more informative. However. since rnany taxa

are excluded from considération, this kind of analysis has its limits. In order to overcome this

default in the phylogenetic analyses of Paleozoic actinistians. Cloutier (1991 ) has adopted the

complemeniary analy.scs on the basis of the main analysis to include the poorly known taxa.

However. if the re.sulls of the main analysis are largely infliienced by the subséquent coinplemen-

tary analyses, the interprétation of the cladograms is very ambiguous. The other option, used

by Zhang & YoUNG (1992) in the phylogenetic analysis of bothriolepidoids. is to fill in for

some unknown data to give a much more complété data matrix. This has been adopted in the

présent study.

The filling in for the missing data has been sélective and is made for a particular character

of a particular taxon (coded in the data matrix in italics), using the criterion that, among its

supposied closely related forms (according to the previous studics), the character State is well

defined and uniform. This procedure involves more assumptions, but, il is compensaied for by

the less ambiguous and more informative results obtained from the analysis. Moreover, the phy¬

logenetic inference itself is hypothetic deductive (Nelson & Platnick 1984). and ihese inferred

States for unknown characlers are consistent wilh the philosophical background of Cladistics.

As defended by Zhang & Young (1992), no data set can be completely objective. The data

matrix obtained here may provides us with a framework for fiinher research; ihose inferred

States can be eilher corroborated or refuted, and the gaps can be filled in by new material or

more detailed observations. The results of the phylogenetic analysis are just hypothèses at the

— 298

Table 2. — Data matrix of 66 characiers for 40 anliaxch généra. Kujdanowiaspis (arihrodire) and Romundina (acanihoihoracid) are u.sed

as the outgroups of antiarchs. — = logical impossibility coding; ? = unavailablc information. X = inlraspccific variation.s. Inferred

States for unknown characiers are shown in italics. Description of the characiers and character States are in the texi.

TAXA

Kuidanowiaspis

Romundina

Yunnanoleois

Mizia

Phvmolepis

Zhaniilepis

Heteroyunnanolepis

Minicrania

Chuchtnolepis

Unnamed antxarch

Vanchienotepis

Xichonolepis

riüMryiTiïïsr^i

1 1111111112

1254567890 1234567890

00 - 0 00 -

00 - 1 00 -

11000000-0 13Q0000000

11000000-0 1300000000

llOOOOOQ-0 13Û000000C

11000000-0 13Û0000000

11000000-0 1300000000

shania

Liujianeolepis

Sinolepis

Luauanolepis

Wudinolepis

Hohsienolepis

Microbrachius

Vietnamaspis

Dianolepis

Jiangxilepis

Tenizolepis

Nawaeiaspis

Biiagalepis

Monarolepis

Grossilepis

Bothriolepis

Grossaspis

Lepadolepis

Gerdakpis

Wwungulepis

Sherbonaspis

Stegolepis

Byssacanihus

Kirgisolepis

Pterichlhyodes

Hunanolepis

Asterolepis

Pambulaspis

Remigolepis

iioooOoo -0

11000000-0

11000003-0

1101?0?0?0

1110107000

1110i0?000

1110100000

IIJOIOIOOO

IIIOIOVOOO

11201

1)211’OIVO

112?J?07?0

ii2i:?00r0

lOûlQOOOOÔ

1301000000

J???0?00Û0

1301000100

2222222223

1234567890

0000000000

OOOOOOOÛ-l

OOllUOOQX

OOlOOlOQÛO

0010110000

0001010000

OOOlOOOOOO

0001000030

0001000000

?T?1000?31

???L000001

3333333334

1234567390

OOÛOÛOOOOO

OOÛÛÛOQÛOO

X10070IO00

llÛ0?0111O

110090110Û

4444444445

1234567890

OOÛO---

OÛOIO-OOO-

001IQOOIOI

0011000101

GOllOûOlOl

5555555556

1234567850

- 0-000

666666

123456

0-0100

000000001? 00??0?

OÛOûlOÛOOO 00???0

0000000000 000000

OOOOOOOOOO 00???0

1100701000 00111-2101 OOOOOOOOOO

110000'’??? ??il???J0? ?0i???????0

1100?01011 ûllIOOllOO

1211010000 OlOlOOOlOI

1121010001

1211010000

lÛQlOlOOOO

1021010001

ni3io??ooo

lOOlOOlOOO

1001001000

OlOlOOOlOl

0771C00101

0??100010i

1100701000

1100 ? 0 ????

1100 ? 0 ?'

1100?01111

OOllOQQlOl

■?11???10?

OOOOOOOOOO

OOOOlOOOOO

?00??????0

00 ????

?? iJ ??? J 0? ?00??????0

001101?101 0001110000

OOllOIllOl OOOllOOOOO

OOllOi?101

1100001111 001100?101

0001101000

0001121000

o?7ioooioi iino?oiiii ooiio:?ioo oooiiooooo oo????

1001000010 li00?0?l?? ??ii???10? ?£?0???????

lOOlCOOllO li01?00111 C11100?101 100003070? 00????

lOOlCOOlIO 1102700111 0?1100?101 100003000

1001000110 n02?00111 OlllOOOlOl 1000030001 00????

1121 i ????0 1101001001 1001000110 1100 ? Û ? i ?? ?? 111 -? 10 ? 10000 ?????

1120100??0 1101001000 lOOlCOOllÛ 1100?00111 00111-2101 1000021001 00????

U2110ll?0 1101001110 lOOlCOOllû llloyoïlll 00111-2100 100002100? 00????

112 ? 1 ?? 1?0 1001101000 1001000110 1100 ? 011?1 0 ? 111 - 21D1 100002100 ?

11201????! 1101???000 lOOiCOOOlÛ 1100?01i?? 00111-7100 1000031??! 000010

112?l??î?0

il20il???0

1121113110

1121111110

11201000?!

lÛOllOlÛOO

1001101000

lûOiOlOOOO

100)001000

11000 on 00

lOOlCOOllû

1001000110

looiaoûiio

lOQlQOOllO

lOOlOOîO’O

1100?01ilC Q?ll?-?1(?? ?I?P????1?1

H0070011

11201000?! IlOOOOllOO 1001QO?0?0 1100?1?1

1120100001

112O1C109?!

IlOOOOllOO looiQoiûio iiooniiii

1700001?0O

1120100001 1200001100

1001000000

iOOlOOlOOO

1100700100 10111-2101 1000121101 10011?

1110000100 10111-2101 1000121101 100110

iioo?i?i?? ??i:?-?io

110010 ?!??

llOO'Clll?

11201000?! 1200000110 1001000000 110X701111

1120100011 IlOOOOOllO lOOlÙOXÛOO 1100?01111

11201????1 1100????10 1001001070 1100?OC100

1120100001 IlOOOOllOO lOOlGOlOOD 1100101101

1120100111 1100001100 IÇOlOOlûOD 1100701111

1120100010 130001C100 1001001000 1100101111

1120?û00?û 1300001100 iOOiOOÛÛOO 11Û0701111

11200000-0 1300000100 1001000000 1100701111

01000 ?

110001

1010 001

amaa \

lOplO 001???

’IOI? ??!???

noio ooi??i

îlOlD 001

llOlO 0017?1

llOI? 011???

llOlO 011??1

— 299 —

Fig. 29. — Strict con.sensus tree of anliarchs phylogcny conslructcd from 12 MP trees (L= 155, CI = 48) obtained with mhen-

nig*+bb* analysis (Hennig86) of the character matrix in Table 2.

présent stage, which await falsification. As stated by PopPER (1965 : 36) “[ejvery genuine test

of a theory is an attempt to falsify it, or to réfuté it. Testability is falsifiability.”

In some cases, the interspecific variation for a particular genus is also coded as the un-

available information (X in the data matrix), which can be optimized later. In my opinion, this

treatment is more rational.

Outgroups

Since either arthrodires or acanthothoracids hâve ever been con.sidered as the sister-group

of antiarchs (Denison 1975, 1978, 1983; Miles & Young 1977; Young 1980, 1986; Gardiner

1984; Long 1984; FOREY & Gardiner 1986; GOUJET 1984a), two généra of these two groups

are selected as the outgroups in our analysis. One is the actinolepid Kujdunowiaspis (Stensiô

1945, 1963), the other is the acanthothoracid Romundina (0RVIG 1975).

Tree calculations

In the first run, the data set was analyzed with the algorithms hennig* + bb of Hennig86.

Ail the characters were treated as additive (ordered). This generated 100 most parsimonious

(MP) trees (L= 157, CI = 47, RI = 80). By not specifying the * option for the bb algorithm.

— 300 —

Van

Unn

Chu

Het

Zha

Yun

Miz

Phy

Sin

Gre

Day

Xic

Liu

Gro

Ger

Le P

Fig. 30. — Three sets of unre.solved nodes within yunnanolepidoids (A), sinolepids (B) and a.sterolepidoids (C) (ail combinations

= 32 MP solutions). Chu - Chuchinotepis. Day - Dayaoslumia. Ger - Gerdalepis. Gre - Grenfellaspis, Gro - Grossaspis,

Het - Heteroyunnanotepis. Lep - Lepadolepis. Liu - Uujianftolepis. Miz - Aftziu, Phy - Phymolepis, Sin - Sinolepis, Unn

- unnamed antiareh. Van - Vanchienolepis, Xic — Xii honolepis, Yun - ïunnimolepis, Zha - Zlianjilepis.

only 100 MP trees found at certain stage are saved. Therefore, the author performed a second

run of the data set using the combination hennig* + bb* followed by the nelsen command. In

this way, 12 MP trees (.L = 155,CI = 48, RI = 8LFig. 31) and the strict consensus tree ofFig. 29

are obtained. The same data were reanalyzed with the heuristic algorithms of PAUP (version

3.1), which yielded 18 MP trees with the same length as found with Hennig86. This data set is

too large to use the branch and bound algorithm. The strict consensus tree is the same as in

Fig. 29. In facl, there exists 32 MP trees as inferred from the combination of the trichotomies

(Fig. 30). The MP trees were confirmed with MacClade (version 3.0), and the synapomorphies

at each node were easily found with this software.

Character analysis of Antiarchs

1. Pectoral fin scale-covered (0) or modified into a slender appendage covered with small dermal

plates (I).

The pectoral fin of antiarchs is modified into a slender appendage covered with small dermal

plates, and is very specialized in comparison with the pectoral fin of other fishes which is scale-

covered. The evidence from Phymolepis (G.-R. Zhang 1978; M.-M.. Zhang 1980) confirms the

slender appendage to be présent in the Yunnanolepidoidei although it might be not completely

developed (Janvier 1995). A slender pectoral appendage is assumed to be présent in ail antiarchs,

since it can be inferred from the small pectoral fenestra and the articulation structure of the

AVL plate when the pectoral fin itself is unavailable.

— 301 —

2. Pectoral fenestra encircled by more than two plates (0) or by a single plate (1).

The pectoral fenestra of antiarchs is typically developed in Minicrania (Zhu & Janvier

1996) and the Yunnanolepididae, although it is fairly small in comparison with that of other

placoderms. The pectoral fenestra is assumed here to be modified into the axillary foramen of

the advanced antiarchs. In general, the pectoral fenestra of placoderms is encircled by the anterior

latéral, anterior ventrolateral or even posterior ventrolateral plates. In antiarchs, the pectoral

fenestra or the axillary foramen is restricted within the AVL plate, which should be regarded as

apomorphic.

3. Pectoral fin articulation simple (0), sinolepid type (I) and euantiarch type (2).

The simple pectoral fin articulation without any dermal articulation, exemplified by Yun-

nanolepis (G.-R. ZHANG 1978; M.-M. ZHANG 1980), is generally regarded as plesiomorphic for

antiarchs by outgroup comparison. The pectoral fene.stra of the Yunnanolepididae can be directly

compared to that of other placoderms. The advanced types of pectoral articulation include those

of Chuchinolepis (G.-R. Zhang 1984; Yoüng & Zhang 1992), Vanchienolepis (TONG-DZUY &

Janvier 1990), the Sinolepididae (YOUNG & Zhang 1992; Ritchie et al. 1992) and the Euan-

tiarcha.

Turning to the évolution of the pectoral joint in antiarchs, G.-R. Zhang (1984) proposed

a transformation sériés from the Yunnanolepididae to the Chuchinolepididae, then to the

Asterolepidoidei and Bothriolepidoidei including the Sinolepididae. This proposai was developed

by Young & Zhang (1992), who suggested four hypothetical stages in the évolution of the

pectoral fin articulation of antiarchs. i.e. those of the Yunnanolepididae, Chuchinolepis {Procon-

dylolepis), the Sinolepididae and finally the Euantiarcha. However, there is no cicar homology

of the pectoral articulations in Chuchinolepis and the Sinolepididae + Euantiarcha. It is probable

that the pectoral fin articulation of Chuchinolepis is derivcd independently from thaï of the

Sinolepididae + Euantiarcha. The same applies to Vanchienolepis. Therefore, as to the polarity

of pectoral fin articulation of antiarchs, a branching transformation sériés is proposed here. From

the simple pectoral fin articulation derivcd independently three patterns : those of Chuchinolepis,

Vanchienolepis, and the Sinolepididae leading to that of the Euantiarcha. Sincc eilher the artic¬

ulation of Chuchinolepis or that of Vanchienolepis is only represented by one genus among an¬

tiarchs, these two types of pectoral fin articulation is useless for the phylogenetic analysis of

antiarchs at the generic level. Only the third evolutionary path remains here for the analysis.

With the mixed coding method, the articulation of the Yunnanolepididae, Chuchinolepis and

Vanchienolepis is coded as "0”, that of the Sinolepididae exemplified by Grenfellaspis (RiTCHiE

et al. 1992) as '’r', and that of the Euantiarcha as “2”. The fin articulations of Dayaoshunia

(Ritchie et al. 1992) and Liujiangolepis (S.-T. Wang 1987) cannot be observed in the available

material, and that of Sinolepis was only represented by an incomplète brachial process. However,

their pectoral fins are relatively advanced, and their articulations are inferred to be the same as

that of Grenfellaspis.

4. Axillary foramen small (0) or large (1).

The enlarged axillary foramen had been considered as one of the synapomorphies of the

Bothriolepididae by Young (1984c, 1988). Long et al. (1990) quantified the size of the axillary

foramen tn terms of the ratio (height of the foramen to height of the latéral lamina of the AVL

— 302 —

plate) and it was suggested that the foramen exceeding 30% is “large” (ZHANG & YouNG 1992),

a criterion which is followed in this work. As to its phylogenetic significance, ZHANG & YoUNG

(1992) found that it was difficult to be explained because of its scattered distribution and sug¬

gested that it may be correlated with the length of the proximal fin segment. The alternative

explanation is that. if the axillary foramen gives a passage to the adductor muscle of the pectoral

fin, as suggested by Stensiô (1959, 1969), enlarged foramen suggests a stroug muscle and im-

proved movement capacity of fin, Since the axillary foramen is homologous to the small pectoral

fenestraof primitive antiarchs represented by the Yunnanolepididae, the enlarged axillary foramen

is apomorphic whatever its fonction. A large axillary foramen is recorded in Vanchienolepis

(Tong-Dzuy & Janvier 1990), Wudiiwlepis, Microhrachius, Vietnamaspis, Jiangxilepis, Bothrio-

lepis, Gmssilepis (ZHANG & YouNG 1992) and Hynanaspis (Janvier & PAN 1982). The axillary

foramen of Pamhiilaspis (YouNG 1983) is assumed to be small, as in other asterolepidoids.

5. Pectoral fin unjointed (0) or jointed (1).

No complété pectoral fin has yetbeen found in the Yunnanolepididae and Chuchinolepididae.

However, the available material indicates thaï the pectoral fin of the Yunnanolepididae (G.-R.

Zhang 1978; M.-M. Zhang 1980) and Chuchinolepididae (Procondylolcpidae, G.-R. Zhang

1984) is probably unjointed as in Remigolepis (Stensio 1931; PAN et al. 1987), Thus, the Chu¬

chinolepididae and those primitive antiarchs retaining a simple pectoral fin articulation are in-

ferred to hâve unjointed pectoral fins. In contrast. those with more advanced fin articulation are

presurned to hâve the jointed pectoral fin when fossil evidence is unavailable, except for Pam-

bulaspix (YOUNG 1983). Young (1983) suggested a doser relation between Pambulaxpis and

Remigolepiü than either has with Asieivlepis. and whether or not a distal joint is présent in

Pambukispis is important for the analysis. Another genus which is scored as missing data is

Vanchienolepis, which has a more complicated fin articulation structure than the Yunnanolepid¬

idae.

6. Cdl and Cd2 plates in contact (0) or separated (1).

The réduction of the Cd2 plate, which is separated from the Cdl plate by the M12 and MM2

plates, has a limited distribution in antiarchs (Zhang & Young 1992), and occurs only in some

bothriolepidoids. The plesiomorphic State is that the Cd2 plate is in contact with the Cdl plate

(Young 1983, 1984a; Zhang & Young 1992), and is common in other antiarchs, including the

Yunnanolepididae and Chuchinolepididae. Sinolepids were proved to retain the primitive condi¬

tion (RitchîR et al. 1992).

7. Pectoral fin short (0) or elongated (1).

Tbe pectoral fin exiending backwards beyond the trunk-shield is defined here as the elon¬

gated fin, and is assessed as apomorphic. The primitive antiarchs such as the Yunnanolepididae

probably hâve the short, unjointed fins, as in Remigolepis. In antiarchs, the elongated fin is

présent in .Hangxilepis. Bothriolepis (StensiO 1963. 1969), Gro.ssilepis and Liujiangolepis (S. -T.

Wang 1987). The fin length has no clear corrélation with the size of the axillary foramen. The

antiarchs with a large axillary foramen, such as Microbrachius, may hâve the short pectoral fin.

Pan et al. (1987) described a fairly elongated pectoral fin of Sinolepis associated with part of •

the AVL and PVL plates. However, RlTCHlE et al. (1992) suggested that tlie specimen may belong

to a bothriolepid aotiarch. New material is needed to clarify this problem because the Sinolepid-

— 303 —

idae may possess a elongated pectoral fin, e.g. Liujiangülepis (S.-T. Wang 1987). Therefore,

the coding of Sinolepis a.s ‘"unknown” i.s préférable.

8. M12 plate relative to the trunk-shield short (0) or elongated (1).

The length of the M12 plate of the jointed pectoral fin loughiy correspond.s to the length

of the proximal .segment and its posterior end represents the extremity of the proximal .segment.

In Remigalepis and the Yunnaiiolepidoidei whose pectoral fin probably possesses only one seg¬

ment, the M12 plate is generally short and broad; which is, the primitive condition for antiarchs.

In some bolhriolcpidoids, such as Buthriolepis and GrossUepm, the M12 plate is much elongated

with its posterior end .situated behind the level of the posterior end of the MV plate, which

should be inteipreted as apomorphic. However, how to as.ses.s these Iwo States remains a problcm.

Zhang & Young (1992) defined a similar character, i.e. the proximal segment of pectoral fin

shorter or longer than the level of the posterior end of the MV plate. However. it was casily

biased by the relative .size of the MV plate or the space corresponding to the MV plate, especially

when used in the Sinolepididac which have no MV plate. We suggest that the centre of the MV

plate is more constant and thus more siiitable for the référencé. Moreover, the proximal segment

is replaccd herc by tlie MI2 plate, which can be used for the unjointed fins as well.

9. Three (0) or two ( 1) Ml plates of the distal segment.

The plate arrangement of the pectoral fin in the Yunnanolepididac and Chuchinolepididae

suggests that 3 latéral marginal plates in the distal segment, as in Dayaoshania, Liujiangolepis,

Sherhoiwspis, Génial épis and Pterichthyudes, should be plesiomorphic. This character is un-

available in those antiarchs which are assiimed to have unjointed fins.

10. l .ow and elongated (0) or high and short (I) trunk-shield.

JANVtF.R & Pan (1982) suggested that the very large and elongated trunk-shield was one

of the asterolepid {Remigalepis and Asterolepis) synapomorphies. Alternatively, YoUNG (1984c,

1990) proposed that the high and short trunk-shield was apomorphic amongst asterolepidoids,

as judged front the condition in the Yunnanolepidoidei.

11. One (0) or two (1) médian dors.al plates.

Two médian dorsal plates are présent exclusively in antiarchs.

12. Index between width of anterior margin and maximum width of the AMD plate, > 55 (0),

35-55 (1). 15-35 (2), or < 15 (3).

The width of the anterior margin of the AMD plate is a very significant character in antiarchs.

But how to define ils relative narrowness is ambiguous. ZHANG & YOHNG (1992) compared the

anterior margin width with the posterior margin width and proposed that the anterior margin

broader than the posterior one was apomorphic. In our opinion, it is more appropriate to use

the index between the anterior margin width and maximum width of the AMD plate. In general,

a narrow anterior margin may corre.spond to a narrow posterior margin. as exemplified by Yun-

nanolepis, Chuchinalepis, Remigalepis and Asterolepis. The index proposed here, however, could

reflect the relative narrowness of the anterior margin of the AMD plate. Then, the problem is

to décidé the polarity of this character.

Zhang & Young (1992) regarded the anterior pointed AMD plate of Early Devonian yun-

nanolepids and most asterolepidoids as primitive. The broad anterior margin of the AMD plate

— 304 —

was considercd derived in comparison to that in Yunncmolepis. However, this polarity is contra-

diclcd by outgroup comparison. The médian dorsal piale of other placoderms, if it is homologous

to the AMD plate of anliarchs, has a broad anterior margin. [If we accept the proposition of

Y.-H. Liu (1991), i.e. the AMD plate of antiarchs is homologous to the extrascapular plate of

other placoderms, the conclusion will be same, by outgroup comparison.] Growth .sériés in an¬

tiarchs provide furthcr évidence. In BotUriolepb vanadensis (Werdelin & LONO 1986), the index

between the anterior margin width and maximum width of the AMD plate of the juvénile in-

dividuals is much larger than that of the adults. The juvénile Asterolepis orniiui (UPENtECE &

UpeNIEKS 1992) has a relalively broad anterior margin of the AMD plate v^'hereas the adult

Asterolepis has an anlcriorly pointed AMD plate. Therefore, it is considered here that the narrow

anterior margin of the AMD plate is apomorphic. In order to code the character, the subdivision

of the morphological continuum is needed. In this work, we code the index above 55 as primitive

(0). 35-55 as l, 15-35 as 2, and below 15 as 3.

13. Index between anterior and posterior divisions of the AMD plate, <300 (0), 300-500 (1),

> 500 (2).

The AMD plate could be subdivided inlo iwo portions by the latéral corner, which corres¬

ponds to the suture between the ADL and PDL (or MxL) plates. The index indicates the relative

position of the PDL plate with reference to the AMD plate. When the index bccomes larger,

the PDL (or MxL) plate has a more posterior position. The growth sériés in Sinolepis (Liu &

P’ AN 1958; 38) shows that the anterior division of the AMD plate in the juvénile individual is

propoitionally shorler than that in the adult, thus suggesting a polarity. As for the character 12,

we code the index below 300 as plesiomorphic (0), 300-500 as 1, above 500 as 2.

14. Tergal angle of AMD plate cenirally (0) or anteriorly (I) placed.

The position of the tergal angle of the AMD plate generally corresponds to that of the

anterior ventral process whenever the process i.s présent. The outgroup comparison is direclly

related with the homologue of ihe AMD plate in other placoderms, which is more or less con-

troversial. If the médian dorsal plate i.s homologous to the AMD plate of antiarchs, then the

outgroup comparison suggests the posieriorly or cenirally placed tergal angle is plesiomorphic.

However, there is another possibility that the extrascapular plate is homologous to the AMD

plate. If this be tme. then we cannot détermine the polarity by means of outgroup comparison.

Amongsl anliarchs, both States are found in the Yunnanolepidoidei. The anteriorly placed tergal

angle is présent in hothriolepidoids, sinolepids, chuchinolepids and Minicrania (Zhu & JANVIER

1996).

15. AMD plate completely (0) or partiy (1) overlapping the ADL plate.

That the AMD plate overlaps the ADL plate along the entire contact margin in the Yun¬

nanolepidoidei indicates that the very .short overlap margin of the AMD plate for the ADL plate

in Briagalepis, Tenizolepis (Malinovskaya 1992) and Monarolepis is a derived feature (Long

et al. 1990).

16. AMD plate underlapping or partly (0) or completely (1) overlapping the PDL (or MxL)

plate.

— 305 —

There are three types of the overlap relationship belweeri the AMD and PDL (or MxL)

plates in antiarchs. The firsl one is that the AMD plate overlaps the PDL or MxL plate over

the anterior half of their common suture, whereas it is the reverse for the posterior half, as in

Yunnanoh’pis and Remigolepis. The second one is that the .4MD plate overlaps the PDL or MxL

plate over the entire length of their common suture, as in Sinolepis, Liujiaiignlepix (S.-T. Wang

1987), Xicitonulepis, and Asterolepis (KARATAJUTE-Talimaa 1963). The third one is that the

AMD plate is overlapped by the PDL or MxL plate over mosi of their common suture, as in

Bothriolepis, Jkmgxilepis (Zhang & LlU 1991) and Hunnnolepis (J.-Q. WANG 1991a). Since no

outgroup comptirison could be used to déterminé the polarity, the State in the Yunnanniepidoidei

is regardcd hcrc as primitive. The second and the third States reflect the two different evolutionary

directions. As the mixed coding (WiLliY et al. 1991) is used here for the branching Iran.sformalion

sériés, the évolution regarding the overlap relationship between the AMD and PDL (or MxL)

plate is treated as two characters (characters 16 and 17)

The AMD plate of Vanchienuleph (TONG-DZUY & JANVIER 1990) is poorly preserved in

its posterior part, however, the type of ils overlap relationship can be inferred from the infor¬

mation of the PDL plate, which is same as that of Yunnanolepis.

17. AMD plate partly or completely overlapping (0) or underlapping the PDL (or MxL) plate (1)

(see character 16 for comments).

18. Anterior ventral process and pit on the AMD plate présent (0) or absent (1).

The presence of the anterior ventral process and pit in primitive antiarchs, exemplified by

Yunnanolepis and Mintcninui (Zltu & JANVIER 1996). indicates that the loss of the anterior ventral

pit and process in Vanchienolepi.'i. Jiangxilepi.\ and asterolepidoids is a derived featurc for an¬

tiarchs. The anterior ventral process in Micwbrachim was confimicd by Pan (1984; cf.

Hemmings 1978).

19. AMD plate without (0) or with (1) the dorsal spine.

The dorsal spine of the AMD plate is présent in Byssacanthus, Kirgisolepis, Stegolepis and

Jiangxilepis.

20. Latéral process of the PMD plate conspicuous (0) or reduced (1).

A latéral process of the PMD plate is conspicuous in most antiarchs. A reduced latéral

process is found in Vietnamaspis (LONG et al. 1990), Sinolepis and Grenfellaspis (RiTCHiE et

al. 1992), and makes the PMD plate somewhat rcclangular in shape.

21. Crista transversalis interna po.tteiior lying laterally to (O) or behind ( I ) the posterior ventral

pit and process of the PMD plate.

In antiarchs, the position of the posterior ventral pit and process relatively to the cri.sta

transversalis interna posterior is a diagnostic character. The crista transversalis interna posterior

of antiarchs should be regarded as the homologue to the posterior annular crest of other pla-

coderms, such as arthrodires (Goü.iET 1984b), where it is latéral to the médian ventral process

of the médian dorsal plate. This is another evidence for Y.-H. Liu (I99l)’s proposition, i.e. the

PMD plate of antiarchs is homologous to the médian dorsal plate of other placoderms. Therefore,

by outgroup comparison, it is reasonable to assume that the crista transversalis interna posterior

lying laterally to the posterior ventral pit and process is plesiomor[ihic. as exemplified by

— 306 —

Zhcmjik’pis, Minicrania and Chuchinolepis. Aparl from this primitive State, three derived States

are found in antiarchs. One is the posterior ventral pit and proccss lying in front of the crista

traiisversalis interna posterior as in the Euantiarcha. The othcrs are the posterior ventral pit and

proccss lying behind the crista traiisversalis interna posterior. which is represented by two quite

different c<niditio-ns. In the Sinolepididae such as Grcnfeltaspis and Xichonolepis, the crista irons-

versalis interna posterior cxtcnds dorsally with the samc direction as that in the Euantiarcha,

Minicrania (Zhu & JanviRR 1996) and Chuchinolepis, Jt is the posterior migration of the post¬

erior ventral process, which lies ncar the posterior end of the plate, that results in the process

behind the crista. In Yunnanolepis Phyinolepis and Mizia. the position of the ptisterior ventral process

and pit behind the crista transvcrsalis intenta po.sterior is definitcly due to the anteriorly bending

of the crista on the viscéral surface of the PMD plate. These three states represent three different

evolutionaiy trends and form a branching transformation sériés together with the plesiomorphy, which

is subdivided into three characters by the mixed coiling method (characters 21-23).

The cri,sta traiisversalis interna posterior lying behind the posterior ventral process and pit

is widely seen in the Euantiarcha. It is supposed to be aiso présent in those euantiarchs where

this character is unavailable.

22. Posterior ventral pit and process on the crista traiisversalis interna posterior (0) or posteriorly

migrated behind it (1).

The posteriorly migrated posterior ventral proce.ss and pit is seen in Xiclionolepis (G.-R.

Zhang 1980; Ritchie et al. 1992). The condition of Grenfellaspis is slightiy different since its

posterior ventral process and pit arc .separated. Its posterior ventral pit lies on the posterior part

of the crista, and immediately in front of the po.sterior ventral proccss (RlTCHlE et al. 1992).

This may be caused by the robustness of the crista in Grenfellaspis. The posterior ventral process

and pil are poorly preserved in the olher présumable sinolepids, i.e. Dayaosliania, Sinolepis and

Liujiatigolepis. But ihey are clearly different from the State occuring in Yunnanolepis.

23. Crista tran.sver.salis interna posterior lying laterally to lO) or turning anteriorly and in front

of (I) the po.sterior ventral process and pit.

In Yunnanolepis. Phyinolepis and Mizia. the development of the crista traiisversalis intenta

posterior is rather spcctalized. On the viscéral surface of the PMD plate, it becomes fairly strong

and turns anteriorly, close to the anterior end of the plate. As stated above, this State should be

regardcd as apomorphic.

24. Presence (0) or loss ( I ) of the AL plate.

The anterior latéral plate is présent in Phyinolepis and Mizia. However, since the anterior

latéral plate of Phyinolepis and Mizia is extremely small, it is uncertain whether it is homologous

to the anterior latéral plate of other placoderms.

25. Absence (0) or presence ( I ) of the Chang’s apparatus.

The Chanü's apparatus is exclusively found in Yunnanolepis and Phyinolepis. Whatever its

function is. by outgroup comparison it is an apomorphy in antiarchs.

26. Absence (0) or présence ( I ) of the vcntrolateral fossa of the trunk-shield.

The vcntrolateral fossa of the trunk-shield is observed in some yunnanolepidoids, which is

asstimed to be derived becausc of its limited distribution in antiarchs.

— 307 —

27. PDL and PL plates independent (0’) or fused to form a MxL plate (!).

Janvier & Pan (1982) commented upon tlie plate previousiy referred to as the mixilateral

plate in bothriolepidoids and some aslerolepidoids, and concluded lhat the plate in some

asterolepidoids was a true mixilateral plate whereas that in bothriolepidoids was not a mixilateral

plate, but a PDL plate as in Ymnanolepis. In the former case, the latéral lamina of PVL plate

is lower than the latéral lamina of AVL plate. The mixilateral plate was found in Pterichthyodes,

Hunanolepis (J.-Q. Wang 1991a), A.stendepis (Traquair 1914; Stensiô 1931), Gerdalepis

(Gross 1941), and is variable in Byxxacanthus (KarATAJUTE-TaI-IMAA 1960).

28. PVL and PL plates independent (0) or fused to form (or replaced by) a single plate (I ).

It was suggested that in bothriolepidoids the PL plate had been fused to the PVL plate

(Janvier & Pan 1982). In this ca.se. the latéral lamina of the PVL plate is higher than, or at

least equal to, the latéral lamina of the AVL plate. Janvier & Pan's assumplioii lias been cor-

roborated by the condition in sinolepids. In GrenJeUaspis and Daycioslumiei (RlTCUlE et al. 1992),

the main lateral-linc groove traverses the PDL plate almost next to the suture between the PDL

and PVL plates, which rules oui the possibility that the PDL plate of the Sinolepididae incor-

porated the PL plate. The PVL plate of sinolepids may include the élément of the PL plate, as

in bothriolepidoids (Janvier & Pan 1982). The alternative explanation is that the PL plate is

lost in sinolepids and most bothriolepidoids. Nawti};iaspi.i (YOUNG 1990) and Liuptaaolepis

(Zhang & YOUNG 1992) were considered to be bothriolepidoids with an independent PL plate.

29. Semilunar plate paired (0) or unpaired (1).

The .semilunar plate of antiarchs is likcly to be homologous to the interolaleral plate of

other placoderms, which is lypically a rod-like plate possessing a postbranchial lamina. In an-

tiarchs, since the postbranchial lamina becomes a rather internai structure and is less dcveloped,

the postbranchial lamina of the semilunar plate, if the latter is homologous to the interolaleral

plate, is largely or entirely lost. The poslbranchial lamina of Pbymnlepis gtiondi extends onio

the semilunar plate, thereby supporting the homology. The position of the scapulocoracoid rela-

tively to the poslbranchial lamina of the interolaleral plate could not be used as a rcforencc,

since the poslbranchial lamina of antiarchs is rather evenly attached to the ventral wall of the

trunk-shield, and a derived cristu tnmsversulis interna anterior is formed to conceal the scapulo-

coracoid. The outgroup comparison and the condition in the Yunnanolepidoidei suggests that the

paired semilunar plate is plesiomorphic for antiarchs.

30. Absence (0) or pre.sence (1) of a large rectangular aperture on the ventral wall of trunk-shield.

This character is aiso stated as “the médian ventral plate absent”, and has been considered

as one of the most important synapomorphies defining the Sinolepididae (Ritchie et al. 1992).

In the outgroups of antiarchs, both the presence and absence of the médian ventral plate could

be found.

31. Presence (0) or absence (1) of the spinal plate.

The spinal plate is commonly absent in antiarchs, and is only présent in one species of

Yunnanolepis (K porifera). Since the absence of the spinal plate is recorded in the other species

of Yunnanolepis, this genus is coded as X in the data matrix.

— 308 —

32. Poslbranchial lamina external and upright (0) or internai and horizontal (1).

The postbranchial lamina of antiarchs is modified into a rod-shaped structure lying hori-

zontally on the viscéral surface of the ventral wall of lhe trunk-shield and having a fairly internai

position. This apomorphy is partly due to the antcrior extension of the subcephalic portion of

the trunk-shield,

33. Adult ornamentation tubercular (0) or reticular (1).

The outgroup comparison indicates that the tubercular ornamentation in antiarchs is primi¬

tive. The two derived States are as follows : I ) reticular in Bothriolepis (Young 1988) and

Jian^xilepis (ZHANG & Liu 1991); 2) ridgedîn Wudinolepis (K.-J. Chang 1965), Hohsienolepis

(P' AN & Wang 1978), Microbrachius (Hemmings 1978; Pan 1984). Both the tubercular and

ridged omaments are présent in Stegolepis (MalINOVSKAYA 1973), This is a branching trans¬

formation sériés and is coded in two characters (characters 33 and 34).

34. Adult ornamentation tubercular (0), ridged (1) or .subparallel ridges on the dorsal wall of

trunk-shield (2).

Among the antiarchs which bear the ridged ornamentation, the ridges on the dorsal wall

of the trunk-shield are subparallel in Microhrachius (PAN 1984) and Hohsienolepis (P’AN &

Wang 1978). and should be regarded as more derived.

35. Absence (0) or presence (1) of ridged scales.

The outgroup comparison indicates that the scales omamented with tubercles is plesiomor-

phic. To date, ridged scales hâve been found in Asterolepis (Lyarskaya 1977), Wurungulepis

(Young 1990) and Pterichthyodes (HEMMINGS 1978).

36. Absence (0) or presence (1) of a dorsal spongy layer in dermal bone of trunk-shield.

A dorsal spongy layer in the dermal bone of trunk-shield is restricted to Lepadolepis, Gros-

saspis and Gerdalepis and has been used to distinguish them from other asterolepidoids by Gross

(1965) and YoUNG (1984c).

37. Pre.sence (0) or absence (1) of the central sen.sory-linc groove.

Tlie central .sensory-line groove is ai.so lermed as the posterior oblique cephalic line groove

in antiarchs. The outgroup comparison indicates that the absence of the central sensory-line is

a derived feaiurc. The presence of the central sen.sory-line groove has long been considered as

one of the synapomorphics defining bothriolepidoids by YOUNG (1984c, character 24). Since it

is absent in Jiangxilepis (ZtiANG & Liu 1991), Nawagiaspis (YoUNG 1990) and Brtagalepis

(Long & Werdelin 1986), Zhang & Young (1992) regarded the development of this groove

as an apomorphy among the bothriolepidoids. We suggesl that the development of this groove

in some bothriolepidoids might be due to reversion as indicated by outgroup comparison.

As to the distribution of the central sensory-line groove among antiarchs, the homology of

the X-shaped pit lines on the nuchal plate in Ymnanolepis and Chuchinolepis (TONG-DZUY &

Janvier 1990) should be incorporated. G.-R. Zhang (1978) suggested that in Yunnanolepis the

X-shaped grooves were just the posterior pit lines and that there was no central sensory-line

groove. But the case is that the X-shaped grooves consist of two pairs of grooves. M.-M. Zhang

(1980) assigned the anterior pair to lhe central sensory-line groove. and lhe posterior pair to

— 309 —

the posterior pit-line groove. ToNG-DzuY & Janvier (1990) proposée! anolher inlerprctation ;

the anterior one as the centrai sensory-line groove and the posterior onc as the supraoccipilal

commissure. However, the fact that entire X-shaped pit iines are iying in front of the openings

of endolymphatic ducts, siiggests that the posterior pair cannot be assigned to the supraoccipilal

commissure. In placoderms. the supraoccipital commissure is the canal or groove behind the

openings of endolymphatic ducts (DENISON 1978). The posterior pair should be the posterior pit

line. In the outgroups of antiarchs, such as arthrodires, petalichthyids and acanthothoracids. there

are two pairs of pit Iines (middle and posterior pit Iines) on the central plates, which in general

lie on the endocranial ridges caused by semicircular canals. In antiarchs, there is no central

plate. However, the endocranial ridges caused by semicircular canals could be inferred in Yuii-

nanolepis and Chuchinolepis, just below the X-shaped pit Iines of the skull-roof as in the out¬

groups. It is .suggested herc that in antiarchs, since the semicircular ridges of both sides migrated

next to the midiine, the pairs of pit Unes of both sides converged loward each other to form

the X-shaped pit Iines. The same X-shaped pit Iines are also seen in Pterichthyodes milleri and

had been assigned to the middle and posterior pit Imcs respcctively by Hemmings (1978). In

P. milleri, sometimes the anterior pair of X-shaped pit line grooves is in connection with the

central sensory-line groove. which extends from the infraorbital groove (Hemmings 1978,

Fig. IB). As analyzed above, we conclude that in Yuntianolepis and Clmcliinolepis there is no

central sensory-line groove and that the X-shaped pit-line grooves were the middle and posterior

pit-line grooves respectively.

38. Presence (0) or absence (1) of the supraorbital groove.

In antiarchs, the supraorbital groove has been defined as the groove on the premedian plate

between the infraorbital grooves of both sides (Miles, 1968; G.-R. Zhang 1978; Hemmings

1978; Young & Gorter 1981; LONG & Werdelin 1986; YoUNG 1988; Tong-DzüY & Janvier

1990; Ritchie et al. 1992). However, it is difficult to understand the homology between ihis

groove and the supraorbital canal in other placoderms, cspecially thaï of acantholhoracids. In

Yuntianolepis (Tonc-Dzuy & JANVIER 1990), Heteroyunnanolepis (Z.-S. WANG 1994) and

Minicrania (Zhu & JANVIER 1996), an additional pair of V-shaped sensory-line grooves are found

on the postpineal plate and can be compared to the supraorbital groove of Romundina (0rvig

1975), which is also situated behind the nasal openings and pineal plate. Therefore, ihe supraor¬

bital groove in antiarchs is redefined here as the V-shaped grooves on the postpineal plate,

whereas that on the premedian plate, which was defined before as a supraorbital groove, is

better referred to SxENSlÔ’s (1969) original terminology, that is, (he commissure between the

infraorbital sensory Iines. The absence of the supraorbital groove is coded as “1” by outgroup

comparison.

39. Presence (0) or absence (1) of X-shaped pit-line grooves.

X-shaped pit-line grooves are composed of the middle and posterior pit-line grooves, and

are typically developed in Yunnanolepi.s, Chuchinolepis and Pterichthyodes. In Bothriolepis and

Grossilepis in which central sensory-line grooves are developed, the middle pit-line groove is

continuons to the central sensory-line groove. The outgroup comparison suggests that the presence

of X-shaped pit-line grooves is plesiomorphic.

— 310 —

40. Présence (0) or absence (1) of the branch of lhe infraorbital groove diverging on latéral

plate.

This sensory-line groove is équivalent to the postorbital branch of the infraorbital sensory

line in arthrodires and petalichthyids. The absence of this branch of the infraorbital sensory-line

groove in antiarchs is apomorphic by outgroup comparison.

41. Absence (0) or presence (I) of the semicircular pit-line groove.

This pit-line groove is a derived structure in antiarchs. No homologous pit line could be

found in outgroups.

42. Middle pit-line groove issued front lhe infraorbital groove absent or short (0), or long and

extending onto the nuchal plate (1).

By outgroup comparison. the long middle pit-line groove extending from the infraorbital

groove onto the nuchal plate is apomoiphic in antiarchs. In Sherbonaspis. the nuchal plate held

a transverse groove which was assigned as the “occipital cross-commissural pit-line groove” by

YoUNG & GorTKR (1981). In comparison with other asterolepidoids, this groove should be the

transverse middle pit-line groove.

43. Absence (0) or présence (1) of the latéral plate.

The latéral plate, a large plate of the skull-roof taking the place of preorbital, postorbital,

marginal and central plates, is unique to antiarchs.

44. Absence (0) or presence (I) of the premedian plate.

The premedian plate is a skull-roof plate covering the anterior part of the head in front of

the rosirai plate, and is présent in antiarchs, rhenanids and acanthothoracids (Denison 1978;

Goujet 1984a).

45. Preorbital dépréssion présent (0) or absent (1).

The homology problem between the preorbital dépréssion and preorbital recess in antiarchs

had bcen discus.sed by Janvier ik Pan (1982), LONG (1983), YouNG (1984b, c, 1988), Long

& Werdelin (1986). ZHt: & Janvier (1996) hâve made the dctailed analysis on this question

on the basis of new information on Minicrania. It was concluded that lhe preorbital dépréssion

and preorbital recess are not homologous. The preorbital dépréssion is not the suitable place for

the rhinocapsular cartilage. Because the preorbital dépréssion of antiarchs could be traced back

to lhe dépréssion on the “médian prerostral plate” in Romundina (0RVIG 1975), the preorbital

dépréssion is regarded as plesiomorphic for antiarchs.

46. Preorbital dépression extending laterally onto the latéral plates (0) or restricted to the pre¬

median plate (I).

Since the primitive antiarchs - the Yunnanolepidoidei and Minicrania (ZHU & Janvier

1996) - possess the dépréssion Crossing lhe premedian and latéral plates, lhe dépréssion restricted

to the premedian plate, as in lhe Sinolepididae (RlTCHlE et al. 1992) and Wudinolepis (K.-J.

Chang 1965), is assumed to be a derived character.

— 311 —

47. Preorbital recess absent (0), restricted lo the premcdian plaie (I), or extending laterally to

the latéral plates (2).

The preorbital recess is an internai space which bouses the rhinocapsular cartilage and is

roofed by dermal piales. Since no similar structure was found in outgroups, the preorbital recess

is unique to antiarchs. In anliarchs, there exist Iwo types of preorbital recess. In bothriolepidoids,

the preorbital recess lies in the prciTiedian and latéral plates, whercas in aslerolepidoids. sinolepids

and Minienwia, it is reslricled to the poslerior part ot' the premedian plate. Since the preorbital

recess in the yunnanolcpidoid-like aniiarch Miiucrani/i (Znu & Janviur 1996) is small and re¬

stricted to the premedian plate, we propose thaï the recess in aslerolepidoids was less derived

than that in bothriolepidoids.

48. Orbital opening open (0) or enclosed by dermal skull-roof plates (1).

The orbital openings of antiarchs are dorsally placed and enclosed by the dermal skull-roof

plates, i.e., the premedian, latéral, rostral, pineal, postpineal plates. In some cases, the nuchal

plate takes part in the orbital margin.

49. Nasal opening at the antérolatéral corners of the rostral plate (0) or at the anterior margin

of the rostral plate ( 1 ).

In antiarchs. the position of the nasal openings display two States. In most cases, the nasal

openings are at the antérolatéral corners of the rostral plate, whereas in Asterolepis and Re-

migolepis. the nasal openings are at the anterior margin of the broad rostral plate. By outgroup

comparison, one may assess the former stale as plesiomorphic.

50. Narrow (0) or broad (I) latéral plate.

Janvier & P.AN (1982) suggested that during the évolution of aslerolepidoids the latéral

plate was becoming narrower and narrower. Sregolcpis jugani was regarded as the least derived

asterolepidoid, since it retained a relaüvely broad latéral plate (Janvier & Pan 1982: 3S3; Pan

et al. 1987). Their argument was that the broad latéral plate is présent in yunnanolepidoids,

bothriolepidoids and Sinolepis. The shape of the latéral plate has a direct bearing on the shape

of the skull roof. In aslerolepidoids. the skull roof is more or less narrow and elongated, whereas

in bothriolepidoids, yunnanolepidoids and Sinolepis which posses.s the broad latéral plate, the

skull roof is relatively broad and short.

In fact. the shape of the skull-roof and latéral plate in yunnanolepidoids, which was pre-

viously regarded as plesiomorphic because they were the most primitive antiarchs known at that

time, is most likely to be apomoiphic (Zhu & Janvier 1996). Wben we compare antiarchs with

its outgroups. it is quitc clcar that the long and narrow skull-roof should be regarded as plesio-

morphy, which was corroborated by Minicrania (Zhu & Janvier 1996). Minicrania has a rcla-

tively long, narrow skull-roof and narrow latéral plate, however it possesses a simple pectoral

fin articulation, like the Yunnanolcpididae. Therefore, in our character data matrix, the narrow

latéral plate is coded as plesiomorphic.

51. Premedian plate short and broad (0) or long and narrow (1).

The growlh sériés in antiarchs (Werdelin & LONG 1986; Zhu & JANVIER 1996) indicales

that the short and broad premedian plate is a primitive character. In the growth .sériés of Bothri-

olepis canadensis and Minicrania (Znu & Janvier 1996), the premedian plate in the juvénile

— 312 —

individuals is proportionally shorter than in the adult, suggesting that the short premedian plate

is a plesiomorphic character for antiarchs.

52. Anterior margin of premedian plate convex (0) or slightly concave.

The slightly concave anterior margin of the premedian plate is présent in Remigolepis, Pam-

bulaspis, Asteralepis (YOUNG 1984c), Stegolepis (MalinoVSKAYA 1977) and Sherbonaspis

(YOUNG & GORTER 1981).

53. Absence (0) or pre.sence (!) of an unomamented shelf and rostrocaudal groove on the pre¬

median plate.

In antiarchs, only Asteralepis, Pambiilaspis and Remigolepis (YouNG 1984c) bear the un¬

omamented shelf and rostrocaudal groove on the premedian plate, which should be regarded as

apomorphic since it is présent neither in primitive antiarchs nor in outgroups.

54. Rosirai widlh/orbilal width index of premedian plate snialler (0) or larger (1) than 200.

Il has been shown that in the grovvih séries the rosirai margin of the premedian plate grows

more rapidiy than ils orbital margin (WERDELlN & LONG 1986; Zhu & Janvier 1996). In the

Sinolepididae, e.g. Sitiolepis and Greiijellaspis, the rostral margin of the premedian plate extends

much laterally. which is considered as an apomorphy. In the data matrix, the rostral width/orbital

width index superior to 200 is coded as I.

55. Orbital fenestra large (0) or small (I).

The orbital fenestra incorporâtes orbital openings, sclerotic plates, nasal openings, rostral

and pineal plates. The size of the orbital fenestra is mainly affected by the size of orbital openings

and the breadth of the rostral and pineal plates. Compared with other placoderms, it is suggested

here that large-sized orbital fenestra is plesiomorphic, as supported by the condition in one of

the primitive antiarchs Minicrania (ZHU & Janvier 1996).

56. Relative position of the orbital fenestra anterior (0), slightly anterior (1), slightly posterior

(2) or posterior (3).

The relative position of the orbital fenestra could be indicaled by the index belween the

length of the postpineal and nuchal plates and the length of the premedian plate. But this index

has ils défaillis. The large-sized orbital fenestra as in Asteralepis and Minicrania (Zhu & Janvier

1996) will bias the index. In order to overcome this bias, we subdivide the skull-roof length

into two portions by the centre of the orbital fenestra and adopt the index between the length

of the posterior portion and the length of the anterior one. The outgroup comparison indicates

that the posteriorly placed orbital fenestra is apomorphic. For convenience, we code the index

larger than 200 (anterior) as 0, between 200 and 140 (slightly anterior) as 1, between 140 and

90 (slightly posterior) as 2, smaller than 90 (posterior) as 3.

57. Postpineal and nuchal plate long and narrow (0) or short and broad (1).

By outgroup comparison, we assume thaï the long and narrow postpineal and nuchal plates

are plesiomorphic. In order to quantify this character, the length/width index of the postpineal

and nuchal piale is used here. It is proposed that the index higher than 90 is coded as 0 whereas

that below 90 is coded as 1.

— 313 —

58. Nuchal plate without (0) or with (1) orbital facets.

This character could be expressed in another way: the postpineal plate excluded from the

latéral plate or not. The out-group comparison indicates that the nuchal plate with orbital facets

is apomorphic. The orbital facet of the nuchal plate in Briagalepis could be inferred from the

shape of the latéral plate (LONG & WerdelIN 1986).

59. Long (0) or short ( 1 ) obstantic margin.

In asterolepidoids, the poslmurginal plate extends posterolaterally and forms a short obstantic

margin facing posteriorly. The out-group comparison suggests that the relatively long obstantic

margin is plesiomorphic.

60. Ab.sence (0) or presence (1) of the submarginal articulation.

In placoderms. the dermal articulation between the submarginal plate and skull-roof was

first recorded in Boihriolepis by YOUNG (1984b). Later, this articular process was described in

Nawagiaspis (YOUNG 1990). It is suggesied to be présent in Gmssilepis, Dianolepis and Bria-

galepis, as inferred from lhe .structure of the latéral plate (Zhang & YoUNG 1992). This articular

process is absent in the Yunnanolepidoidei (Y.-H. Litj 1963; G.-R. Zhang 1978), Minicrania

(Zhu & .lANVlER 1996), the Sinolepididae (PAN ei al. 1987; RlTCHlE et al. 1992), and

Asterolepidoidei (STENSlô 1969; HemmincS 1978; YOUNC 1984b). Its presence in Microbrachius,

as assumed by Zhang & YOLTMG (1992), is admitted in this work.

61. Endocranial postorbital process short (0) or extending in front of the orbital notch (1).

The long endocranial postorbital process extending in front of the orbital notch on the latéral

plate is présent only in Bolhriolepis and Grossilepix (ZHANG & Young 1992).

62. Absence (0) or presence ( 1 ) of the pronounced postpineal thickening.

It is known only in Remigolepi.s and Pamhiila.^pis (YouNG 1984c).

63. Presence (0) or loss (1) of the prelaleral plate.

The prelateral plate of antiarchs is homologous to the postsuborbital plate of other pla¬

coderms (YOUNG 1984b). By outgroup comparison, the loss of the plate is a secondary feature.

In antiarchs, the prelateral plate has only been described in Bolhriolepis, Grossilepis and

Nawagiaspis. In Mizia (Yunnanolepis) parvus (V4424.3), M.-M. ZHANG (1980: pis 1-2, III-I)

identified a suture in the dermal bones of the cheek which was referred to as the suture between

the suborbilal and postsuborhilal plates, and at the same time she considered thaï lhe pf)Ntsuborbi-

tal and submarginal piales had fused to some extern and the suture between them was fairly

vague. Sincc the dermal check bone complex of A/, parvus fits in the notch on the lareral margin

of the skull-roof and behind the infraorbital groove, this suture is most likely to be between the

postsuborbital and submarginal plates, by comparison with lhe cheek plates in oihcr placoderms.

The suborbital plate occupies a more anterior position and bears the infraorbital groove. This

is confirmed by lhe detached suborbital plate of Yunnanolepis (ToNG-DZUY & Janvier 1990)

which bears lhe sensory-line grooves very simitar to those foiind in other placoderms. In

asterolepidoids, the entire submarginal plate occupies the notch in the latéral margin of the skull-

roof, and the postsuborbital plate is lost secondarily, as in Asterolepis, Remigolepis, Pterichthy-

odes (Hemmings 1978) and liunanolepis (J.-Q. Wang 1991a). However, the condition in other

antiarchs is unknown and coded as missing data.

— 314 —

64. Frelaterai plate with a long anterior process (0) or équilatéral, triangular in shape (I).

The prelateral plate of Bathriolepis and Grossilepis is somewhat équilatéral, triangular in

shape. The long anterior process in Nawagiaspis (YOUNG 1990) is assumed to be primitive, as

suggested by the State in Mizia parvus.

65. Prelateral plate behind (0) or above (1) the mental plate.

In Bathriolepis, Grossilepis and Nawagiaspis (YouNG 1990), the prelateral (= postsuborbital)

plate is between the skull-roof and the mental {= suborbital) plate and the infraorbital groove

traverses through the prelateral plate. By outgroup comparison, this is shown to be a derived

character for antiarchs.

66. Mental plates of both sides separated (0) or meeting in the mid line (1).

Since the mental plate is homologous to the suborbital plate of other placoderms (Young

1984b). the separated mental plates of both sides, as in Bathriolepis (YoUNG 1984b) and Nawagi¬

aspis (Young 1990), are plesiomorphic. The same applies to the Yunnanolepidoidci and Min-

icraniu (Zhu & JANVibR 1996), as inferred from the viscéral surface of the skull-roof. In

Asterolepis, Remigolepis (Nii,.ssON 1941; Lyarskaya 1981) and Pterichthyodes (HEMMINGS

1978), the mental plates of both sides contact each other and form a complété upper biting

margin of the jaw.

Results

The tree calculating algorithm of Hennig 86 (h*, bb*) gives 12 equally most parsimonious

(MP) trees with length of 155, consistency index of 48 and rétention index of 81. A strict con¬

sensus tree with the nelsen option (Fig. 29) shows ihrce trichotomies The same data set has

been calculated using the heuristic algorithm of PAÜP (version 3.1), and 18 minimum-length

trees with a length of 155 are found (consistency index = 0.477, homopla.sy index = 0.523, ré¬

tention index = 0.810, rescaled consistency index = 0.387). The strict consensus tree is same as

that found with Hennig86. AU of the equally shortest trees obtained from Hennig86 are not

included among those from PAUP. In fact, three sets of unresolved nodes (Fig. 30), hâve 32

combinations, indicating 32 equally parsimonious trees, that are confirmed by means of MacClade

(version 3.0) (ail having the length of 155). One of the most parsimonious cladograms (Fig. 31)

is selected for the analysis.

The antiarchs form a clade (Node 1), supporled by nine synapomorphies;

- the pectoral fin modified into a slender appendage covered by small dermal plates

(character I);

- the pectoral fenestra encircled by a single plate (character 2), two médian dorsal plates

(character 11 );

- the loss of the anterior latéral plate (character 24);

- the absence of the spinal plate (character 31);

- the postbranchial lamina internai and horizontal in position (character 32);

- the absence of the central scnsory-line groove (character 37);

- the presence of the latéral plate (character 43);

— 315 —

Pjg 31 — One of the equally parsimonious solutions showing one possible set of character distributions with character change

in brackets. Whcre character changes are not shown, the transformation is 0 to I. Black squares represent a uniquely derived

feature, squares with dots represent reversais, and squares represent parallelisms.

— 316

- orbital openings enclosed by the dermal skull-roof plates (character 48);

- the anterior latéral plate reappears at Node 8 {Mizia and Phymolepis).

The presence of the spinal plate in Yunnanolepis porifera is also a reversion as indicated

by the cladogram. The central sensory-line groove shows three independent reversions in anti-

archs. The presence of the premedian plate (character 44) indicates the doser affinity between

Romundina and the Antiarcha than either with Kujdanowiaspi.i (GOUJBT 1984a), and it is retained

in the analysis, allhough it does not provide any additional phylogenetic information in this

Work.

Five major monophyletic groups are dearly shown in the cladogram. i.e. the Yun-

nanolepidoidei (Node 2). Sinolepididae fNode 11), Microbrachiidae (Node 16), Bothriolepididae

(Node 19) and Asterolepidoidei (Node 28). The Bothriolepidoidei as previously defined (Miles

1968; Long 1983; Denison 1978; Janvier & Pan 1982; Young 1984c, 1988, 1990; Zhang

& Young 1992) turns ont to be paraphyletic.

The Yunnanolepidoidei is redefined to include Vanchienolepis, the Chuchinolepididae, the

Yunnanolepididae, Zhanjilepis, Hclemyunnanolepis and an tinnamed antiarch, and its monophyly

is supported by the anteriorly pointed AMD plate (character 12 |3)) and broad latéral plate

(character 50). The broad latéral plate corresponds to ihc broad and short skull-roof, and has a

parallelism at Node 10, which reverses later (Node 24). Among the Yunnanolepidoidei. three

groups are found (Nodes 4, 5, Chuchinolcpix) and iheir interrelationships cannoi be resolved

with the available information (none of the synapomorphies to support Node 3). Vanchienolepis

and the unnamed antiarch are groupcd logether (Node 4) by the large rectangular aperture on

the ventral wall of the trunk-shield (character 30). This State of the character 30 has once been

considered as the synapomorphy of the Sinolepididae (Young 1984c; RtTCHtE et al. 1992). Ac-

cording to our cladogram, this large ventral trunk-shield aperture must be develcrped inde-

pendently three times (Nodes 4, 11 and an undefennined species of Yunnanolepis).

Node 5 is supported by the centrally placed tergal angle of the AMD plate (character 14).

The cladogram assumes that the anteriorly placed tergal angle, as in Chuchinolepis,

Vanchienolepis, Minicrania (Zhu & JANVIER 1996) and Bothriolepis, is plesiomorphic for anti-

archs, and the centrally placed tergal angle develops indepcndently at Node 28.

Node 6 corresponds to the branching between Zhanjilepis and the Yunnanolepididae. It is

characterized by the ventrolaleral fossa of the trunk-shield (character 26).

Node 7 represents the Yunnanolepididae. which is defined by the crista tramversalis interna

posterior turning anteriorly and in front of the posterior ventral process and pit (character 23)

and ChanG's apparatus (character 25, a reversai in Mizia).

Among the Yunnanolepididae, Mizia and Phymolepis are closely related at Node 8 by the

lack of the supraorbital sensory-line groove (character 38) and a reversai (appearance of the

anterior latéral plate, character 24(01).

Node 9 corresponds to the branching between Minicrania (Zhu & Janvier 1996) and other

advanced antiarchs possessing the brachial process and fiinnel pit. It is characterized by the

preorbital recess (character 47), the loss of X-shaped pit-line grooves (character 39) and absence

of the branch of the infraorbital groove diverging on the latéral plate (character 40). The

— 317 —

preorbital recess is independently derived in Heteroyunnanolepi.i and secondarily losl in the Mi-

crobrachiidae (Node 16), and its presence in the Sinolepididae is confirmed by Grenfellaspis

(Ritchie et al. 1992). Character 39 is a highiy homoplastic character, and shows a parallelism

in Phymolepis and Ihree reversais in Kirgisolepis, Pterichthyodea and at Node 39. Character 40

shows two reversais in Kirgisolepis and at Node 20.

Node 10 represents the branching between the Sinolepididae and Euantiarcha, supported by

six synapomorphies. Le. the sinolepid type pectoral fin articulation (character 3[1]), the jointed

pectoral fin (character 5, a reversai in Remigolepis), the PVL and PL plate fused to form a

single plate (character 28), the absence of the supraorbital groove (character 38), the broad latéral

plate (character 50, two reversais respectively at Node 24 and in Sittnlepis) and the slightly

anterior position of the orbital fenestra (character 56| 11). The distribution of character 28 in the

cladogram suggests that the independent PVL and PL plates in the Euantiarcha arc due to sec-

ondary séparation. In combination with character 27, the mixilateral plate of some asterolepidoids

is formed by the PDL and PL plates after this secondary séparation.

Node 11 defines the Sinolepididae and corresponds to the branching between Liujiangotepis

and the other sinolepids. This node is characterized by the AMD plate completely overlapping

the PDL plate (character 16), the posterior ventral pit and process posteriorly migrated behind

the crista transversaux interna posterior (character 22. unknown in Liujiangolepis), the presence

of the large rectangular aperture in the ventral wall of the trunk-shield (character 30), the rostral

width/orbital width index of the premedian plate larger than 200 (character 54) and the small

orbital fenestra (character 55). Character 16 lias two parallelisms in Grossitepis and Asterolepis,

and character 55 shows three parallelism’s in Chuchinolepis. Yunnanolepis and the Bothriolepis

+ Grossilepis pair.

Node 12 represents an unsolved irichotomy in the Sinolepididae, Dayaoshania and Xi-

chonolepis being either monophyletic as shown in Fig. 31 (also Fig. 30B2) or paraphyletic as

shown in Fig. 30BL This node is supported by two uniquely shared derived characters, lhe

index between anterior and posterior divisions of the AMD plate higher than 300 (character

13[1]) and the preorbital dépréssion restricted to the premedian plate (character 46), and a ho¬

moplastic character, the index between the width of the anterior margin and the maximum width

of the AMD plate smaller than 55 (character 12(1], with a reversai in Sinolepis).

Node 13 is lhe branching point between Grenfellaspis and Sinolepis, supported by the index

between anterior and posterior divisions of the AMD plate higher than 500 (character 13[2]),

the reduced latéral process of the PMD plate (character 20), and a reversai (anterior position of

the orbital fenestra. character 56(0],

The Dayaoshania-Xichonolepis lelationship (Node 14) is supported by the index between the

width of the anterior margin and the maximum width of the AMD plate below 35 (character 12(2]).

Node 15 corresponds to the branching between the Microbrachiidae and the other euanti-

archs. This node is characterized by six synapomorphies;

- the euantiarch type pectoral fin articulation (character 3(2]);

- the AMD plate underlapping the PDL or MxL plate (character 17);

— 318 —

- the crista transversalis interna posterior situated behind the posterior ventral process and

pit (character 21);

- the unpaired semilunar plate (character 29);

- the long and narrow premedian plate (character 51);

- lhe slightly posterior orbital fenestra (character 56[2]), and the submarginal articulation

(character 60).

Character 17 is a homoplastic character and reverses three times in Grossilepis, Byssacanthus

and at Node 37 (reversai once again in Pambulaspis). Character 29 has a reversai at Node 28

and again in Gerdalepis. Character 51 has a reversai at Node 32 and again in Stegolepis. The

loss of the submarginal articulation (a reversai) occurs at Node 28 or 29, Character 56 is a

highiy homoplastic character and shows two parallclisms in the lineage of the Sinolepididae

(Xichonolepis and Lhijiangolepis). In the lineage of lhe Euaniiarcha, the orbital fenestra has a

more posterior position (character 56(3]) respectively at Nodes 16 and 27 (or 26, since it is

unknown in Luquanolepis). Above Node 29, a reversai pattern of character 56 is found in the

cladogram.

Node 16 represents the branching between Wudinolepis and Microbrachius + Hohsienolepis,

and defines the Microbrachiidae. This lineage is united by one uniquely derived character, the

ridged adult ornamentation (character 34[1]), and five homoplastic characters:

- the cniarged axillary foramen (character 4);

- the presence of the central scnsory-linc groove (character 37[0]);

- the middie pit-line groove issued from the infraorbital groove long and extending onto

the nuchal plate (character 42. a parallelism al Node 28);

- the loss of the preorbital rcccss (character 47|0J);

- the posterior position of the orbital fenestra (character 56L3J).

According to the cladogram the axillary foramen is enlarged four times, in:

1) Vanchienatepis;

2) Bothriolepis and Grossilepis (Node 22);

3) Wudinolepis and Microbrachius',

4) Vietnamaspis and Jiangxilepis (Node 25).

The enlarged foramen is predicted in Hohsienolepis.

Node 17 indicates the sister-group relationship between Microbrachius and Hohsienolepis,

supported by the subparallel ridges on the dorsal wall of the trunk-shield (character 34[2]).

Node 18 corresponds to the branching between the Bothriolepididae + Tenizolepis and ail

other euantiarch généra excluding the Microbrachiidae. This node is supported by;

- the M12 plate elongated relatively to the trunk-shield (character 8, a reversai at Node 29);

- two Ml plates in the distal segment (character 9, unknown in the Microbrachiidae and a

reversai at Node 31 );

- the absence of the preorbital dépression (character 45);

- the preorbital recess extending laterally onto the latéral plates (character 47(2]);

- the postpineal and nuchal plates short and broad (character 57);

- the prelateral plate dorsal to the mental plate (character 65, unknown in the Sinolepididae,

Microbrachiidae and Minicrania).

— 319 —

Character 8 shows aiso a parallelism in Wiidinolepis. The alternative explanation is the

derived State of character 8 at Node 15, and later two reversais, respectively at Nodes 17 and

29). Characters 45 and 47 hâve parallelisms in Heteroyiinnanolepis (Z.-S. Wang 1994).

Node 19 represents the branching betwecn the Bothriolepididae and Tenizolepis, and is

characterized by the AMD plate overlapping the ADL plate antcriorly and being overlapped by

the ADL plate posteriorly (character 15). It has a reversai at Node 22.

Node 20 defines the Bothriolepididae, and corre.sponds to the branching between Briagalepis

and ail olher bothriolepids. It is defîned by one uniquely derived character: the nuchal plate

with orbital facets (character 58);

and one reversai: the presence of the branch of the infraorbital groove diverging on the

latéral plate (character 40101).

Node 21 is the branching between Monarolepis and Grossilepis + Bothriolepis, and is sup-

ported by the separated Cdl and Cd2 plates (character 6, unknown in Briagalepis and Tenizolepis)

and a reversai, the presence of the central sensory-line groove (character 37[0]).

Node 22 characterizes the sister-group relationship between Bothriolepis and Grossilepis.

The eight synapomorphies are:

- the large axillary foramen (character 4);

- the long pectoral fin (character 7, unknown in other bothriolepidoids and Tenizolepis);

- the AMD plate completely overlapping ADL plate (character 15[0]);

- the presence of the X-shaped sensory-line grooves (character 39[0J);

- the presence of the semicircular pit-line groove (character 41, unknown in Monarolepis

and a parallelism at Node 30):

- small orbital fenestra (character 55, unknown in Monarolepis and Briagalepis:);

- the endocranial po.storbital process extending in front of the orbital notch (character 61);

- the prelateral plate équilatéral triangular in shape (character 64).

Among them, characters 15 and 39 are reversais, and character 64 is quite ambiguous be-

cause of the large number of missing data.

Node 23 represents anolher larger diversification among the Euantiarcha. and Dianolepis

is at the base of this lineage. It is supported by the narrowed anterior margin of the AMD plate

(character 12111).

Node 24 corresponds to the branching between Jiang.xilepis + Vietnamaspis and

Luquanolepis + Nawagiaspis + Asterolepidoidei. It is supported by a reversai, the narrow latéral

plate (character 50[0J).

Node 25 suggests that Jiangxilepis is the sister-group of Vietnamaspis, with an enlarged

axillary foramen (character 4).

Node 26 is the branching between Luquanolepis and Nawagiaspis + Asterolepidoidei. It is

characterized by a reversai, the PVL and PL plates being independent (character 28[0]).

Node 27 indicates that Nawagiaspis is the closest sister-group to the Asterolepidoidei. This

node is supported by the high and short trunk-shield (character 10, with a reversai later at Node

38) and the posterior orbital fenestra (character 56(3], unknown in Luquanolepis).

— 320 —

Node 28 defines the Aslerolepidoidei and corresponds lo the branching between Hunanolepis

and ail other asterolepidoids. It is characterized by nine synapomorphies;

- the tergal angle of the AMD plate mesially placed (character 14[01);

- the anterior ventral process and pit lost (character 18);

- the PDL and PL plates fused to form a mixdaterai plate (character 27);

- the semilunar plates paired (character 29(01; unknown in Nawagiaspis)-,

- the middle pit-line groove issued from the infraorbital groove long and extending onto

the nuchal plate (character 42);

- the preorbital recess restricted to the premedian plate (character 47[1]);

- the short obstantic margin (character 59);

- the loss of the submarginal articulation (character 60);

- the loss of the prelateral plate (character 63).

The cladogram suggests that the centrally set tergal tuigle is apomorphic for antiarchs, and it

shows a parallelism at Node 5. Tlie .short obstantic margin is possibly présent in Nawagiaspis ac-

cording to YOUNC (1990: 47), if so, character 59 is an uniquely derived character at Node 27.

Node 29 represents the branching between Kirgisolepis + Byssacanthus and the other

asterolepidoids excluding Hunanolepis. Jt is supported by two reversais: the M12 plate short

relatively to the trunk-shield (character SfO]); and the slightly posterior orbital fenestra (character

56[2]).

Node 30 indicates a sister-group relationship between Kirgisolepis and Byssacanthus, and

is defined by two slightly hornoplastic characters: the dorsal spine of the AMD plate (character

19); and the presence of the semicircular pit-line groove (character 41).

Node 31 corresponds to the branching of Pterichthyodes and the other asterolepidoids ex¬

cluding Hunanolepis, Kirgisolepis and Byssacanthus. It is characterized by a reversai: three Ml

plates of the distal segment (character 9[01), a tess informative character because of many missing

data (the ridged scale, character 35). and a uniquely derived character, the mental plates of both

sides meeting in the mid-line (character 66, unknown in Hunanolepis, Byssacanthus and Kirgi¬

solepis).

Node 32 is the branching between the Gerdalepididae and Sherbonaspis -t- Wurungulepis +

Stegolepis -t Asterolepididae. It is characterized by two reversais:

the short and broad premedian plate (character 5110]);

the slightly anterior orbital fenestra (character 56] 1]).

Node 33 defmes the Gerdalepididae and is characterized by the similar dorsal spongy layer

in the dermal bone of tlie trunk-shield (character 36). It represents one of the unresolved tri¬

chotomies in this cladogram since no derived character is l'ound at Node 34.

Node 35 represents the branching between Sherbonaspis and Wurungulepis + Stegolepis +

Asterolepididae. It is united by the fairly narrow anterior margin of the .\MD plate, its index

between the width of the anterior margin and the maximum width below 35 (character 12[2]),

and the anterior margin of the premedian plate slightly concave (character 52).

Node 36 shows Wurungulepis as the sister-group of Stegolepis + Asterolepididae. It is charac¬

terized by the independent PDL and PL plates (character 27[0]).

— 321 —

However, since (he independenl PDL and PL plates of Wurungulepis is an inferred character,

its position at this node is provisional. If the mixilateral plate of Wurungulepis is evidenced. it

will resuit in an additional trichotomy at Node 35. This cladogram indicates that, among the

advanced antiarchs which possess the brachial process. the fusion and séparation of the PDL,

PL and PVL plates occurred twice. The flrst time is at Node 10, where the PVL and PL plates

are fused to form a composite plate, which is separated as in Luquanolepis and Nawagiaspis,

then a second fusion took place between the PDL and PL plates to form the mixilateral plate.

The independenl PL and PDL plates at Node 36 are due to the secondary séparation.

Node 37 corresponds to the branching between Stegolepis and the Aslerolepididae, and is

supported by one reversai, the AMD plate partiy overlapping the PDL (or MxL) plate (character

17[0]).

Node 38 defines the Asterolepididae, and is the branching between Asierolepis and Re-

migolepis + Pambulaspis. It is characterized by four synapomorphies:

- the low and elongated trunk-shield (character 10[0]);

- the index between the width of the anterior margin and the maximum width of the AMD

plate below 15 (character 1213]);

- the nasal openings at anterior margin of the rostral plate (character 49);

- the unomamented shelf and rostrocaudal groove on the premedian plate (character 53).

Character 10 is a reversai. Node 39 suggests a sister-group relationship between Remigolepis

and Pambulaspis with an uniquely derived character: the pronounced postpineal thickening

(character 60).

COMMENTS ABOUT ANTIARCH PHYLOGENY

1) The monophyly of the Yunnanolepidoidei is corroborated in the cladogram, although,

with an emended définition. The simple pectoral fin articulation and large triangular or oval

preorbital dépréssion are plesioniorphic for antiarchs, and cannot be used to define the Yun¬

nanolepidoidei. This onler is characterized by the anteriorly pointed AMD plate and broad latéral

plate, and is expanded to include the Chuchinolcpididae, Vanchienolepis and an unnamed antiarch.

However, Minicrania (Zhu & JANVHER 1996). which retains the simple pectoral fin articulation

and large preorbital dépression, is excluded from the Yunnanolepidoidei. The Yunnanolepidoidei

are found exclusively in the Early Silunan-Early Devonian deposits of South China block, in-

cluding North Vietnam.

2) The Chuchinolcpididae hâve doser affinity with the Yunnanolepididae than either has

with the Sinolepididac + Euantiarcha. They represent an unsuccessfui attempt at developing a

dermal pectoral fin articulation. The same applies to Vanchienolepis.

3) The large rectangular aperture on the ventral wall of the trunk-shield is apomorphic for

antiarchs. and occurs independently in the Sinolepididac and a lineage of the Yunnanolepidoidei.

4) The brachial process and related structures are the successful attempt of antiarchs as to

the pectoral fin articulation. The antiarchs with these advanced structures dominated during the

Middle and Late Devonian.

— 322 —

5) A yunnanolepidoid-like antiarch Minicranki (Zhu & Janvier 1996) is the sister-group

of the advanced antiarchs. sharing with the latter a preorbital recess.

6) The monophyly of the Sinolepididae is confirmcd. Vanchiennlepis and the unnamed an¬

tiarch cannot be referred to sinolepids. As proposed by Young & ZHANG (1992). the brachial

process of the Sinolepididae is rather primitive compared to that of the Euantiarcha. The

Sinolepididae is the sister-group of the Euantiarcha. as suggested by JANVIER & PAN (1982),

and RitchIE et al. (1992).

7) The Microbrachiidae is at the base of the Euantiarcha. It cannot be referred to the

Asterolepidoidei as proposed by Miles (1968). Long (1983) and J.-Q. Wang (1991a).

8) The Bolhriolepidoidei. as previously defined, turns out to be paraphyletic. In contrast.

the Bothriolepididae are still valid as a monophyletic group defined by lhe orbital facet of the

nuchal plate.

9) Nawagiaspi.s. which possesses both asterolepidoid and bothriolepidoid featurcs and was

referred to the Bothriolepidoidei (YoUNG 1990). cornes out as the closest sister-group of the

Asterolepidoidei.

lU) The monophyly of the Asterolepidoidei is well supported in lhe cladogram. Hunanolepis

front lhe Middic Devonian of South China is placcd at lhe base of the Asterolepidoidei.

11) The Pierichthyodidae as defined earlier (Stensio 1948; Hemmings 1978; Young 1984c,

1988; J.-Q. Wang 1991a) fail to be monophyletic. JANVIER & PAN (1982) regarded Stegolepis.

Byssuciintlius. and Pterichthyodes as a paraphyletic group. and this is partly corroborated here.

In our cladogram, the Pterichlhyodidae turns out to bc polyphylelic.

12) Stegulepis is the sister-group of the Astcrolepididac, however, this is only supported

by a reversai. The Aslerolepididae is well supported in the cladogram, as in Young (1984c,

1988), and the unjointed fin o( Remigidepis is a .sccondary condition (Young 1984c). G.-R. Zhang

(1984) considered that the unjointed fin of Remigolepi.s is a primitive antiarch condition, from

which derived lhe Jointed fin of the other euaniiarchs. This is the mosi parsimoiiious treatment

as regard to the fin evoluiion, however it is fairly less parsimonious as to other characters. G.-R.

Zhang's view is partly followed by PAN in PAN et ai (1987). who thought that Reinigolepis

was lhe only représentative among cuantiarehs retaining lhe iinjointed fin. However. if his pre-

ferred cladogram (Pan et al. 1987, Fig. 51) is taken into considération, the jointed fin originated

independently al least four limes, which is very uniikely.

Acknowledgcments

The reseurch was carried out linder the supervision of Dr Ph. Janvier during iny stay Ut the Laboratoire

de Palétmtologie. Muséum national d’Hisloire naturelle, URA 12 CNRS. Paris in 1093-94. which i.s thanked

for the provision of working faciliiies. The stay was financed by grants of lhe Chinese Academy of Sciences.

I express niy .spécial thanks to Dr Ph. .lANVlKR for his valuable comments and suggestions throughout this

Project and his help in improving the manuscripl of tins work. Eor the sponsorsbip of this project, 1 am

very much indehted to Pr. M.-M Chanc tIVPP, Bcijing), who is also deepiy thanked for her gifts of many

valuable specimens and continuons encouragement. 1 am mo.st grateful to Drs Ph. Janvier, D. Goujkt,

H. Lelièvre (Paris) and Pr ToNa-Dzov Thanh (Hanoi University, Vietnam) for the fruitfui discussions

and their generosity in providing the specimens in their care. I am much indehted to Dr D. Goujet for

his help and permission in running lhe Hennig86. PAUP and MacCladc programs in his care. I also thank

— 323

Dr L. Marcus (American Muséum of Nalural History) who introduced me to use the Hennig86 program in

1989. I wish to express my gratitude to Dr Gavin YOUNG (Bureau of Minerai Resources, Camberra) for reading

and commenting on the manu.script and improving it stylistically. Thanks are aiso due to Pr Ph. Taquet (Paris)

for the invitation to his laboratory, and to Mr D. Serrettf. (Paris), who took the photographs.

Manuscript submitted for publication on 17 June 1994; accepted on 22 September 1995.

REFERENCES

AGASSIZ J. L. R. 1843. — Recherches sur les poissons fossiles. 5 vols. Imprimerie Petitpierre, Neuchâtel.

— 1844. — Monographie des poissons fossiles des Viewc Grès Rouges ou Système Dévonien (Old Red Sandstone)

des Iles britanniques et de Russie, .lent & Gassmann, Neuchâtel.

Andrews S. M. 1982. — The discovery of fo.ssil flshes in Scoiland up lo 1845. Royal Scottish Muséum, Edin-

burgh.

ANONYMOUS 1985. — International Code of Zoological Nomenclature. Third édition. London. International Trust

for zoological Nomenclature, University of California Press. Berkeley & Los Angeles; 1-338.

Berg L. S. 1940. — Classification of fishes, both recent and fossil. Trudy zool. In.st., Leningrad. 5: 85-517 |in

Russian and English],

Chang K.-J. 1965. — New antiarchs from the Middie Devonian of Yunnan. Venebrata PalAsiatica 9: 114 [in

Chinese with English summary).

— 1978. — Early Devonian antiarchs from Chuifengshan. Yunnan. In : Symposium on the Devonian System

of South China 1974. Geological Press, Beijing; 292-293 [in Chinesel.

Chang M.-M. 1966. — Notes on some veriebrates from the Early Devonian of Yunnan, China. Preprint : Colloque

international Centre national de la Recherche scientifique 163.

Chi Y.-S. 1940. — On the discovery of Bnthriolepis in the Devonian of central Hunan. Bulletin of the Geological

Society of China 20; 57-72.

— 1942. — Upper Devonian Bothriolepis beds of Yunnan. Scienlific Record, of the Academia sinica 1; 1-2.

Cloutier R. 1991. — Interrelatioilships of Palaeozoic actinistians ; patterns and trends. In M.-M. Chang, Y.-H.

LlU & G.rR. Zhang (eds). Early Veriebrates and rclated prohlems of evolutionary biology. Science Press,

Beijing; 379-428.

COPE E. D. 1885. — The position of Pterichlhys in the System. American Naluralist 19; 289-291.

DENISON R. H. 1975. — Evolution and classification of placoderm fishes. Breviora 432: 1-24.

■— 1978. — Placodermi. In H.-P. SCHULTZE (ed.). Handbook of Paleoichihvologv. Ousiav Fischer Verlag. Stutt¬

gart 2: 1-128.

— 1983. — Further considération of placoderm évolution. Journal of Vertebrute Paleontology 3; 69-83.

Eichwald F. 1. VON 1840 — Die Thier-und Pflanzenreste des allen rolhen Sandsleins und Bergkalks im Nov-

gorodschen Gouvernement. Bulletin de l'Académie des Sciences de Saint-Pétersbourg 7; 78-91.

Fang R.-S., JlANG N.-R., Fan J.-C., Cao R.-G.. Li D. Y. 1985. — The Middie Silurian and Early Devonian

Stratigraphy and Pnlaeontology in Qujing District, Yunnan. The People’s Publishing House of Yunnan,

Kunming, China: 1-171 [in Chinese with English abstract].

FARRIS J. S. 1988. — llennig 86. Version 1.5. Port Jefferson Station, New York.

FORRY P. L, & Gardiner b G- 1986. — Observation on Ctenurella (Piyctodontida) and the classification of

placoderm fishes. Zoological Journal of the Linnean Societv 86: 43-74.

Gardiner B. G. 1984. — The relalionships of plaeoderms. Journal of Vertebrute Paleontology 4; 379-395.

GOUJET D. 1984a — Placoderm interrelatioilships ; a new interprétation, with a short review of placoderm classi¬

fications. Proceedings of the Linnean Society of New South Wales 107: 211-243.

— 1984b. — Les Poissons Placodermes du Spitsberg. Arthrodires Dolichothoraci de la Formation de Wood

Bay (Dévonien Inférieur). Cahiers de Paléontologie Fd. CNRS Paris ; 1-284.

— 324 —

Gross W. 1931. — Asterolepis ornata Eichw. und das AntiarchiProblem. Palaeontographica A 75: 162.

— 1932. — Fossilium Catalogues I : Animalia. Pars 37: Antiarchi. IV. Junk. Berlin: 1-40.

— 1933. — Die Fische des baltischen Devons. Palaeontographicu A 79; 1-74.

— 1941. — Die Bothriolepis-Arien der cellulosa-Mergel Lettlunds. Kungliga svenska Vetenskapsakademiens

Hiiiidlingtir ser. 3 19; 1-79.

— 1963. — Uber die Plaeodemien-Gatlungen Asterolepis und Tiaraspis aus dem Devon Belgiens und einen

fraglichen Tiaraspis -Resi au.s dem Devon Spitzbergens. Bulletin de l’Institut Royal des Sciences naturelles

de Belgique 41. 119.

HemmiNGS s.K. 1978. — The Old Red Sandstone antiarchs of Scotland; Pterichthyodes and Microhrachius.

Palaeonifjgraphical Society Monographs, London 131: 164.

Hkn.NIG W. 1966. — Phylogenetir Systematics. Universily of Illinois Press. Urbana.

Ho(.i H-F. 1988a. — The Devonhw Strurigruphy und Sedimenutry Faciès oj Longinenshan, Sichuan Province.

Geologreal Publishing House, Beijing (in Chinesel.

— 1988b. — The Devonian System of China, Stratigraphy of China, hT 7. Geological Publishing House. Beijing

jin Chinese).

JaNVIKK Ph. 1995. —The brachial articulalion and pectoral fin in Antiarchs (Placodermi). In: M. .^RSENAULT,

H. Lelievri. & Ph. Janvier (eds). .Studies on Early Vertebrates (Vllth International Symposium. 1991,

Miguasha Parc, Qucbec). Bulletin du Muséum national d'Histoire naturelle, Paris 4'' série, section C 17(1-

4): 143-162.

Janvier P, & Pan J. 1982. — Hyn unaspis bliecki n. g., n. sp.. a new primitive euaniiarch (Antiarcha, Placodermi)

from thc Eifelian of northeastern Iran, wiih a discussion on antiarch phylogeny. Neues Jahrhuch fiir Géologie

und Paldoniologie Ahhandlung 64; 364-392.

KARATAJUTE TALIMA.A V. N. 1960. — Byssacunihus dilatalus (Eichw.) from the Middle Devonian of the USSR.

Collection of the Acta Oeologica Lithuanica, Vilnius: 293-305.

— 1963, —Genus Asterolepis liom the Devonian of the Russian Platform. In A Grigelis & V.N. Karata-

JOTE-TALIMAA (eds). The dala of geology of Lithuania. Geological and Geographical Institute of the

Academy of .Science of Lithuania SSR, Vilnius: 65-223 (in Ru.ssian wiih English summary].

Lui s.-F. 1974. — Discovery of ïunnanolepis fauna in Kwangsi and ils stratigraphical sigmficance. Vertebrata

PalAsiatica 12: 243-248 (in Chine.sel.

— 1992. — On the Lower Devonian antiarchi.an.s of Cuangtti. China. Vertebrata PalAsiatica 30: 210-220 [in

Chinese with English summary].

LU' T.-S. & P'AN K. 1958. — Devonian fishes from the Wulung Sériés near Nanking, China. Palaeontologica

Sinica 141: 1-41 (in Chinese with English summary].

Liu Y.-H. 1963, — On the Amiarchi from Chuising. Vertebrata PalAsiatica 7: 39-46 [in Chinese with English

summary],

— 1973, — On the new forms of Polybranchiaspiformes and Petalichthyida from Devonian of Southwest China.

Vertebrata PalAsiatica 11: 132-143 (in Chinese).

— 1991. — On a new petalichthyid, Eurycarospis incdis gen. et sp. nov. (Placodermi, Pisces) from the Middle

Devonian of Zhanyi, Yunnan. In M.-M. ClIANG, Y.-H. LlU & G.-R. ZHANG (eds). Early Vertebrates and

related problems of evoliitionary hiology. Science Press. Beijing: 139-178.

Lui Y.-H. & Wang J.-Q. 19X1. — On three new arthrodires from Middle Devonian of Yunnan. Vertebrata

PaI.Asialica 19: 295-304 [in Chinese with English summary].

Long J. A. 1983. — New boiliriolepid fish from the Late Devonian of Victoria. Australia. Palaeontology 26:

295-320.

— 1984. — New phyllolepids from Victoria and lhe relalionships of the group. Proceedings of the Linnean

Society of New South Wales 107; 263-308.

Long J. A., Burett C. F. Ncan P. K.. & Janvier Ph. 1990. — A new boihriolepid antiarch (Pisces, Placodermi)

from lhe Devonian of Do Son Peninsula, norlhern Vietnam. Akheringa 14: 181-194.

long 1. A. & Wehoelin L. 1986. — A new Laie Devonian bolhriolepid (Placodermi, Antiarcha) from Victoria,

with description of other species from lhe State. Akheringa 10; 355-399.

Lyarskaya L. A. 1977 — New data on Asterolepis ornata from the early Prasnian deposits of the Baltic région.

In: V. V. MENNER (ed.). Essays on phylogeny and systematics of fossil fishes and agnathan. Nauka, Moscow:

36-45 [in Russian].

— 1981. — Baltic Devonian Placodermi- Asterolepidae, Zinatne. Riga, 152 p.

— 325 —

Malinovskaya s. P. 1973. — Stegolepis (Antiarehi, Placodermi), a new Middie Devonian genus from Central

Kazakhstan. Paleotitologicheskii Zhurnal 7; 189-199 [in Russianj.

— 1977. — Systematic position of aniiarchs of Central Kazakhstan. In V. V. Menner (ed.). Essays on phylogeny

and systemalics offossil fishes and agnathan. Nauka, Mo.scow: 29-35 [in Russianj.

— 1992. — New Middie Devonian antiarehs (Placodermi) of Central Ka7.akhstan. In E. Mark-Kurik (ed.)

Fossil fishes as living animais. Academy of Sciences of Estonia, Tallinn: 177-184.

Mansuy h. 1907. — Résultats paléontologiques. In M. H, L.antenois (ed.) Résuitats de la mission géologique

et minière du Yunnan méridional (Septembre IWJ-Janvier 191)4). Dunod & Pinat, Paris : 150-177.

— 1912. — Étude géologique du Yun-Nan oriental. 2" partie. Paléontologie. Mémoire du Service géologique

d’Indochine 1 : 1-146.

— 1915. — Contribution à l'étude de.s faunes de l'Ordovicien et du Gotlilandien du Tonkin. Mémoire du

Service géologique d'Indochine 4 : 1-7.

M’COY F. 1848. — On some new fossil fish of the Carboniferous Period. Annales Magazine of natural Historv

Ser. 2. 2: 1-10.

MtLES R. 1968. — The Old Red Sandstone antiarehs of Scotland. Family Bothriolepididae. Palaeontographical

Society Monographs 122; 1-130.

Miles R. S. & Yoiinc G. C. 1977. — Placoderm interrelationships reconsidered in the lighl of new ptyctodontids

from Gogo, Western Australia. In S. M. ANDREWS, R. S. MILES & A. D. Walker (eds). Prohlems in

vertebrate évolution. Academie Pre.ss. London; 123-198.

Miller H. 1841. — The Old Red Sandstone. Ist ed,. Adam & Charles, Edinburgh.

Nelson G. & Platnick N. I. 1984. — Systemalics and évolution. In M.-W. Ho & P.T. Saunders (eds). Beyond

Neo-Darwinism. Academie Press, New York; 143-158.

NILSSON T. 1941, — The Downtoniun and Devonian vertébrales of Spitsbergen. 7. Order Antiarehi. Skrifter

Svalbard Ishavet 82: 1-54

Oruchev D. V. 1964. — Class Placodermi. In ORLOV (ed.). Fundumentals of Paleontology 11, Agnatha, Pisces.

Nauka, Mo.scow: 118-171 (in Russian].

0RVIG T. 1975. — Description, with spécial reference to the dermal skeleton, of a new radontinid arthrodire

from the Gedinnian of Arctic Canada. In J.P. Lehman (ed.). Problèmes Actuels de Paléontologie - Évolution

des Vertébrés CNRS Colloques Internationaux 218: 41-71.

P’AN K. 1964. — Some Devonian and Carboniferous fishes from South China. Acta palaeontologica sinica 12 :

139-168.

— 1973. — New material of Devonian fish fossils from Central South China. In The mllected papers of

Stratigraphy and Paleontology. Symposium on the atlas of Paleontology of Central South China: 35-44 [in

Chinese).

— 1981. — Devonian antiarch bio.stratigraphy of China. Ceological Magazine 118: 6975.

— 1984. — A new species of Microbrachius from Middie Devonian of Yunnan. Venebrala PalAsiatica 22:

8-13.

Pan J. & DlNELEY D. L. 1988. — A review of early (Silurian and Devonian) vertebrate biogeography and

biostraligraphy of China. Proceedings of the Royal Society of London B 235: 29-61.

P'AN K. & Wang S.-T. 1978. — Devonian Agnatha and Pisces of South China. In Symposium on the Devonian

System of South China 1974. Geological Press, Beijing: 298-333 [in Chinese).

— 1981. — New discovencs of Polybranchiaspida from Yunnan Province. Vertebrata PalAsiatica 19: 113-121,

P'AN K., Wang S.-T., KAO L.-D. & Hou J.-P. 1978. — Devonian continental sediraentary formations of China.

In Symposium on the Devonian System of South China 1974. Geological Press, Beijing: 240-269 [in ChineseJ.

P'AN K., Wang S.-T & Y.-P. LlU 1975. — The Lower Devonian Agnatha and Pisces from South China. Pro-

fessional Paper in Stratigraphy and Palueontology 1: 153-169 [in Chinese],

Pan J., HUO F.-C., CAO J.-X., Gu Q.-C., LlU S.-Y., Wang J.-Q., GaO L.-D. & LlU C. 1987. — Continental

Devonian System of Ningxia and iis Biotas. Geological Publishing House, Beijing (in Chinese with English

summary].

Pander c. h. 1856. — Monographie der fossilen Fische des Silurischen Systems der Russich-Baltischen Gou¬

vernements. Kdnigliche Akademie der Wissenschaften St Petersburg.

— 1857. — Ueber die Placodermen des devonischen Systems. Kdnigliche Akademie der Wissenschaften St.

Petersburg.

— 326 —

PanïF.LHYEV N. 1992. — New remigolepids and liigh armnured aniiurchs of Kirgizia. In E. Mark-Kurik (ed.)

Fossil fishes as livin/; animais. Academy of Science.s of Estonia, Tallinn: 185-191.

PiEN C,-S. 1948. — Noie on ihe occurrence ot Boihritilepis in norihem Kwanglung. Scientific Journal of the

Sun i'al-Sen Universily (N. S.i Universily of Kansa.s, Lawrence I: 1.

POPPF.R K. 1965. - Conjecture and Réfutations: the Grnwth of Scientific Knowledge. Harper & Row, New York.

RiTCliiC A., Wang S.-T.. Young G. C. & Zhang G.-R.. 1992. — The Sinolepidae, a Family of antiarchs (pla-

codenn fi.shes) from the Devonian of South China and Ea.>iiern Au.stralia. Australasian Muséum Records

44: 319-370.

Stf.NSIô e. a. 1931. — Upper Devonian vertebrates from East Greenland, colletted by the Danish Greenland

EZapedition in 1929 and 1930. Meddellesler oin Gronlund 86: 1-212

— 1945- — On the head of eerluin arlhrodires : 11. On the cranium and cervical joint of the Dolichothoraci

t Acanihaspidae). Kungliga svenska Vetenskapsakademiens Handlingur 22 (3): 170.

— 1948. — On lhe Placodertni of the Upper Devonian of East Greenland, 2. Antiarchi: subfamily Bothri-

olepinae, Wilh an attcnipt at a revi.sion of the previousiy describcd species of that Family. Meddellesler om

Gronland 139: 1622.

— 1959. — On lhe pectoral fin and shoulder girdie of lhe arlhrodires, Kungliga svenska Vetenskapsakademiens

llandlingar 8 (4): 1-229.

— 1963- — Anatomical studios on the arihrodiran head. Part 1 Préface, geological and geographical distribu¬

tion. the organizaiion of lhe head in the Dolichothoraci. Cocco.sieumorphi and Pachyosieomorphi. Taxonimic

appendi.x. Kungliga .svenska Vetenskapsakademiens Handlingar 9 (4); 1-419.

— 1969, — Placodermaia. la .1. Piveteaij (ed,1. rrailé de Paléonlologie, Mas.son, Paris 4 (2) : 71-692.

TonG-DZUY T. & .lANVIl-R Ph. 1987. — Les Verldbrds Dévoniens du Viêt .Nam. Annales de Paléontologie (Verté¬

brés InvertéhrésI 73 ; 165-194.

— 1990, — Les Vertébrés du Dévonien inférieur du Bac Bo oriental (provinces de Bac Thaï et Lang Son,

Viêl Nam). Bulletin du Muséum national d'Hisloire naturelle, Paris J' série section C 12 : 143-223.

— 1994. — New Early Devonian vertebrates from Trang Xa (Bac Thai, Vietnam), with remarks on the dis¬

tribution of lhe vertebrates in lhe Song Cau Group, loumal oj Svuiheusf Asian Earih Science UE 235-243.

Traqhair R. H. 1888. — On lhe structure and classification of lhe Asterolcpidae. Annales Magazine of natural

History 2 (6): 485-504.

— 1893. — On lhe Brilish species of Aslerolepidae. Proceedings of lhe Roval physical Society of Edinburg

10: 283-286.

— 1914, — A tnonograph of the tishes of lhe Old Red Sandslone of Britain. Pl.2. — The Aslerolepidae.

Pulaeonlographical Society Monographs 1-4: 63-134.

UpeniLCE 1. & UPENlhKS J. 1992, - Young Upper Devonian antiarch (Asterolepis) individuals from the Lode

i|uarry, Latvia. hi E. Mark-Kurik (ed,). Fossil fishes as living animais. Academy of Sciences of Estonia.

Tallinn: 167-176.

Wang H.-C. 1942. — The stiatigraphical position of Devonian fish-bcaring séries of E. Yunnan with a spécial

discussion on the Tiaomachien Formation of central Hunan. Bulletin of the Geological Society of China

12: 217-225,

Wang J.-Q. 1979. — A new Family of Arthrodira from Yunnan, China. Vertebrata PulAsiatica 17: 179-188 [in

Chinese wilh Engli.sh summary).

— 1982. — New materials of Diniehthyidae. Vertebrata PalAsiatica 20: 181-186 [in Chinese wilh English

summary].

— 1991a. — New' material of Hunanolepis from the Middie Devonian of Hunan. In M.-M. Chang, Y.-H. Liu

& Ci.-R. Zhang (eds). Early Vertebrates and related problems of evolulionary biology. Science Press, Beij-

mg; 213-247.

— 1991 h. — The Antiarchi from Early Silurian of Hunan. Vertebrata PalAsiatica 29: 240-244 [in Chinese

wilh Enghsh summary],

— 1991c. — A fossil Arthrodira from Panxi, Yunnan. Vertebrata PalAsiatica 29: 1-7 [in Chinese with English

summary],

Wang J.-Q. & Wang N.-Z. 1983. — A new genus of Coccosteidae. Vertebrata PalAsiatica 21: 1-8 [in Chinese

w'ith English summary],

— 1984. — New material of Arthrodira from the Wuding Région. Vertebrata PalAsiatica 22: 1-7 [in Chinese

with English summary].

— 327 —

Wang J.-Q. & Zhu M. 1995. — On the âge of the Jiucheng Formation of Wuding. Yunnan. Acta Stratigraphica

Siiiica.

Wang S.-T. 1987. — A new Antiarchi from the Karly Devonian of Guangxi. Vertebrata PalAsiatica 25: 81-90

[in Chinese with English abstract].

Wat.SON D- .M. .s 1961. — Some additions to nur knowledge of Antiarchi. Palaenntology 4: 210-220.

WERDIîLIN L. & Long J. A. 1986. — Allomeiry in the placodemi fish Bathriolcpis cunadensis Whiteaves and

its evolulionary significance. l.ethuia 19: 161-169.

WlLHY E. O., SlFGKL-CAL'.Shv D., BltttoKi) 13. K. & FONK V. A. 1991. — The mmpleat cladkt: a primer of

phylogenelic procedures. Spécial publication No. 19, The University of Kansas. Muséum of Naiural Hisiory,

Lawrence.

WOODWARD A. S. 1891. — Catalogue of the fossil fishes tu the Brilish Muséum of NaturuI Histary. Pt. H.

Brilish Muséum (Naiural History). London. 567 p.

YOUNG G. C- 1980. — A new Karly Devonian placoderm from New South Wales, Australia. with a di.scussion

of placoderm pliylogeiiy. Palaeoutugraphica /\ 167: 10-76.

— 1983. — A new antiarchan fish (Placodermi) from the late Devonian of Soulheastern Australia. IIMR Journal

of Australasiau Ceotugy and Ceophysics 8: 71-81

— 1984a. — .An aslerolepidoid antiarch (placoderm lïsh) from the Early Devonian of the Georgina Basin,

central Aiisiralia. Alchennga 8: 6,5-80,

— 1984b. - Reconstruction of the jaws and braincase in the Devonian placoderm fish Bothriolepis. Palae-

ontology 27: 635-661.

— 1984c. — Commcnts on the phylogeny and biogeography of anliarchs (Devonian placoderm fishes) and

the use of fossils in biogeography. Proeeedings of the Linnean Society of S'ew South Wales 107: 443-473.

— 1986. — The relatitiiiships of placoderm fishes. Zoological Journal of the Linnean Society London 88; 1-57.

— 1988. — Anliarchs (Placoderiii fishes) Ironi the Devonian Aztec Siltstone, Southern Victoria Land, Antarctica.

Palaeontographicu A 202: 1-125.

— 1990. — .New anliarchs (Devonian placoderm fishes) from Queensland, with commcnts on placoderm phy¬

logeny and biogeography. Meinoirs of the Queensland Muséum 28: 35-50.

YoUNG G. C. & GORTER .1. D. 1981. — A new fish fauna of Middie Devonian âge from the Taemas/Wee Jasper

région of New South Wales. Bureau of Minerai Re.sources of Australia, Bulletin 209: 83-147.

YoUNG G. C. & ZilA.NG G.-R. 1992. — Structure and function of the pectoral joint and operculum in anliarchs,

Devonian plucodenii fishes. Pa/acontology 35: 143-464.

Zhang G.-R. 1978. — The anliarchs from the Karly Devonian of Yunnan. Vertebrata PalAsiatica [in Chinese

with english abstract] 16: 148-186.

— 1979. — Some opinions on the description of antiarchs in '‘Devonian Agnatha and Pisces of South China”.

Vertebrata PalAsiatiea 17. 85-89 [in Chine.se].

— 1980, — New matenal of Xiclwiiole/ns qujingensis and discussion on some of il.s morphological charac-

leristics, Venebrahi PalAsialkn 18: 272-2li() (in Chinese with English abstract).

— 1984, — New forni of the Antiarchi with primitive brachial process from Early Devonian of Yunnan. Verte¬

brata PalAsiatica 22: 82-91 [in Chinese with English abstract).

Zhang G.-R. & Lin A5 0. 1991. — A new anliarch from the Upper Devonian of Jiangxi, China. In M.-M.

Chang. Y.-H- Lin & G.-R. ZH.ANC (eds). Early Verrehrates and related problems of evolutionary hiology.

Science Press. Beijing 195-212

Zhang G.-R. & YOUNG G. C. 1992. - A new antiarch (placoderm fish) from the Early Devonian of South

China. Alcheringa 16: 219-240.

Zhang M.-M. 1980, — Preliminary note on a Lower Devontan anliarch from Yunnan. China. Vertebrata PalAsi¬

atica 28: 179-190 [in Chinese with English abstract].

ZHU M. & Janvier Ph. 1994. — Un Onychodontide (Sarcopterygii) du Dévonien inférieur de Chine. Comptes

Rendus de P Académie des Sciences Paris. Série II 319; 951-956.

— 1996. — A small antiarch, Minicrania lirouyii ii. g., n. sp., from lhe Early Devonian of Qujing. Yunnan

(China), with remarks on antiarch phylogeny. Journal of Venebrate Paleontology 16 (1): 1-15.

Zhu m. & Wang J.-Q. in press. — On the Early-Middie Devonian boundary in Qujing, Yunnan. Acta Strati-

graphica Sinica.

Zhu M., J.-Q. Wang & Fan J.-H. 1994. — Early Devonian fishes from Guijiatun and Xujiachong Formations

of Qujing. Yunnan, and related biostratigraphic problems. Vertebrata PalAsiatica 32: 1-20 [in Chinese with

English summary).

— 328 —

Plate I

Antiarcha, Xilun Formation, Early Devonian, Qujing, Yunnan, China.

Yunnanolepis porifera n. sp.

1, 2 - Trunk-shield in dorsal (1) and ventral (2) views, V10507.3, x 2.

3, 4 - Trunk-shield in dorsal (3) and latéral (4) views, VI0507.1, x 2.

5 - Trunk-shield in dorsal view, VI 0507.2, x 2.

6, 7 - Left AVL plate in latéral (6, x 3) and postérolatéral (7, x 8, showing the scapulocoracoid in the pectoral fenestra)

views, V10507.4.

Phymolepis cuifengshanensis K.-J. Chang

8, 9 - PMD plate in dorsal (8) and viscéral (9) views, VI0508.3, x2.

10 - PMD plate in dorsal view, V10508.2, x 2.

— 330 —

Plate II

Antiarcha, Xishancun, Xitun and Xujiachong Formations, Early Devonian, Qujing, Yunnan, China.

Yimnanolepis porifera n. sp.

1 - AMD plate in external view (elastomer cast), VI0499.16, Xishancun Formation, x3.

2 - PMD plate in viscéral view (elastomer cast), V 10499.23, Xishancun Formation, x 3.

3 - Left AVL and Sp plates (elastomer cast), V10499.37, Xishancun Formation, x 3.

4 - External mould of left AVL and Sp plates, V10499.36, Xishancun Formation, x3.

5 - Incomplète AMD and ADL plates in viscéral view, V 10507.7, Xitun Formation, x 3.

6 - Incomplète right ADL and AVL plates in dorsal view, V10507.5, Xitun Formation, x 3.

7 - Incomplète right AVL plate in dorsal view, VI 0507.6, Xitun Formation, x 3.

Mizia longhuaensis n. g., n. sp.

8, 9, 10 - Trunk-shield (holotype) in dorsal (8), ventral (9) and latéral (10) views, V10515, Xujiachong Formation, x 3.

Yunnanolepis sp.

11 - Incomplète trunk-shield and skull-roof in dorsal view, V105I4, Xitun Formation, x 3.

— 332 —

Plate III

Antiarcha, Xishancun Formation, Early Devonian, Qujing, Yunnan, China.

Yunnanolepis porifera n. sp.

Trunk-shield in ventral view (internai mould of the ventral and latéral walls and external mould of the dorsal wall),

X 3.

Trunk-shield in dorsal view (internai mould of the dorsal and latéral walls, and external mould of the ventral wall),

X 4.

Internai mould of an AMD plate, V10499.9, x 3.

Two ADL plates in external view (elastomer casts), V10499.24-25, x 3.

PDL plate in external view (elastomer cast), V 10499.36, x 3.

External mould of a trunk-shield (ventral and latéral walls), V10499.2, x 3.

Trunk-shield in ventral view (internai mould of ventral and latéral walls), V 10499.2, x 3.

AMD plate in viscéral view (elastomer cast), V10499.ll, x 3.

Zhanjilepis aspratilis G.-R. Zhang

9 - Internai mould of a right PDL plate, V1050I.7, x 3.

10,11 - AMD plate in dorsal (10) and viscéral (11) views (elastomer casts), V10501.1, x 3.

1 -

V10499.I,

2 -

V 10499.3,

3 -

4 -

5 -

6 -

7 -

8 -

— 334 —

Plate IV

Antiarcha, Xishancun Formation (1-11) and Xitun Formation (12), Early Devonian, Qujing, Yunnan, China.

Phymolepis cuifengshanensis K.-J. Chang

1 - Left ADL plate in dorsolateral view (elastomer cast), VI 0500.2, x 2.

2 - PMD plate in dorsal view (elastomer cast), V 10500.1, x 2.

3 - Internai mould of a PMD plate, V 10500.1, x 2.

4 - PMD plate in viscéral view (elastomer cast), V10500.1, x2.

5 - Left PDL plate in viscéral view (elastomer cast), V 10500.3, x 2.

6 - External mould of a right PL plate, VI 0500.4, x 2.

7 - Internai mould of a left AVL plate, VI 0500.5, X 2.

8 - Left AVL plate in viscéral view (elastomer cast), V 10500.5, x 2.

9 - Right PVL plate in viscéral view (elastomer cast), V. 10500.6, x 2.

Zhanjilepis aspratilis G.-R. Zhang

10 - Right PVL plate in viscéral view (elastomer cast), V10501.5, x 3.

Yunnanolepidoidei gen. et sp. indet.

11 - Left PDL plate in external view (elastomer cast), V10506, x 1.5.

Unnamed antiarch

12 - Internai mould of PVL plate in ventral view, V10515, x 1.5.

336 —

Plate V

Antiarcha, Xitun Formation, Early Devonian, Qujing, Yunnan, China.

Phymolepis guoruii n. sp.

1, 2, 3 - Trunk-shield (holotype, V10509.1) in dorsal (1), ventral (2) and latéral (3) views, x 1.5.

4 - Trunk-shield in dorsal view, V10509.2, x 1.5.

5, 6 - Incomplète AMD plate in dorsal (5) and viscéral (6) views, V10509.3, x 2.

7 - PMD plate in dorsal view, V10509.5, x 2.

8, 9 - PMD plate in dorsal (8) and viscéral (9) views, V 10509.6, x 2.

— 338

Plate VI

Anliarcha, Xishancun Formation, Early Devonian, Qujing, Yunnan, China.

Heteroyunnanolepis qujingensis Z.-S. Wang

1 - Trunk-shield (V10502.i) in dorsal view (elastomer cast), x 1.5.

2 - Dorsal and latéral walls of the trunk-shield (V10502.1) in viscéral view (elastomer cast), x 1.5.

3 - AMD plate in viscéral view (elastomer cast), V 10502.2, x 2.

4 - Internai mould of a left AVL plate, V10502.3, x 2.

— 340 —

Plate VII

Antiarcha, Xishancun Formation, Early Devonian, Qujing, Yunnan, China.

Heteroyunnanolepis qujingensis Z.-S. Wang

1 - Extemal mould of dorsal part of a trunk-shield (V10502.1), x 1.5.

2 - Internai mould of a trunk-shield (VI 0502.1) in ventral view, x 1.5.

3, 4 - Left AVL plate in external (3) and viscéral (4) views (elastomer casts), V10502.3, x2.

— 342 —

Plate VIII

Antiarcha, Xishancun (1-4, 6-12) Formation and Xitun (5) Formation, Early Devonian, Qujing, Yunnan, China.

Chuchinolepis gracilis K.-J. Chang

1 - Internai mould of a juvénile skull-roof and trunk-shield in dorsal view, V10503.1, x 5.

2 - AMD plate in viscéral view (elastomer cast), V10503.5, x 3.

3 - Internai mould of an AMD plate, V10503.4, x 3.

4 - AMD plate in exlernal view (elastomer cast), V10503.3, x 3.

5 - Right AVL plate in dorsal view, V10510.2, x 3.

Chuchinolepis qujingensis (K.-J. Chang)

6 - Internai mould of an AMD plate, V10504.1, x 2.

7, 8 - AMD plate in viscéral (7) and dorsal (8) views (elastomer casts), V10504.1, x 2.

9 - Internai mould of a left ADL plate, VI 0504.2, x 2.

10 - Left ADL plate in viscéral view (elastomer cast), V 10504.2, x2.

11 - Left ADL plate in external view (elastomer cast), VI 0504.3, x 2.

12 - External mould of a left ADL plate, V10504.3, x 2.

— 344 —

Plate IX

Antiarcha, Xitun Formation, Early Devonian, Qujing, Yunnan, China.

Chuchinolepis robusta n. sp.

1, 2, 3 - Left AVL plate in dorsal (1), ventral (2) and latéral (3) views, V10512, x 2.

Chuchinolepis gracilis K.-J. Chang

4, 5, 6 - Right AVL plate in dorsal (4), ventral (5) and postérolatéral (6) views, V10510.I, x 3.

Plate X

Antiarcha, Xitun Formation, Early Devonian, Qujing, Yunnan, China.

Chuchinolepis sulcata n. sp.

1 - Trunk-shield (holotype, VI0513.1) in dorsal view, x 2.

2 - Trunk-shield (holotype, V10513.1) in dorsal view, with the fragment of AMD plate removed, x2.

3 - Incomplète AMD plate in dorsal view, V10513.6, x 2.

4 - PMD plate in dorsal view, V10513.2, x 2.

5, 6 - PMD plate in viscéral (5) and dorsal (6) views, VI05I3.3, x 2.

7 - Incomplète pectoral fin in latéral view, V10513.5, x 2.

Chuchinolepis qujingensis (K.-J. Chang)

8, 9, 10 - Fragment of left AVL plate in dorsal (8), ventral (9) and posterior (10) views, V10511.1, x 2.

11 - Incomplète right AVL plate in posterior view, VI0511.2, x2.

12, 13 - PMD plate in viscéral (12) and dorsal (13) views, V10511.9, x 2.

Bulletin du Muséum national d'Histoire naturelle, Paris, 4* sér., 18 , 1996

Section C, 2-3 : 349-402

Isolated Dinosaur bones from the Middle Cretaceous

of the Tafilalt, Morocco

by Date A. RUSSELL

Abstract. — The "Grès rouges infracénomaniens" of Southern Morocco, possibly of Albian âge, contain

evidence of one of the most diversified dinosaur assemblages known from Africa, including a relatively long-

necked species of Spinosaurm and abundant but isolated bones of a peculiar theropod ("Spinosaurus B” of

Stromer 1934). Also preserved are the oldest records of abelisaurids and among the oldest records of titanosaurids

in Africa. Theropods are most abundanlly represenied, followed by sauropods; ornithischian remains were not

identified. Bones of infantile dino.saurs are présent, one of which was derived from ■■m individual weighing less

than 4 kg. The assemblage resembles ihat of the Bahariya Formation more than that of Gadoufaoua. possibly

because of a Irophic dependence upon large, freshwater fishes. U was more closely linked zoogeographically to

South America than to Nonh America.

Key-words. — Dinosaurs, Middle Cretaceous, Morocco. biogeography.

Os isolés de Dinosaures du Crétacé moyen du Tafilalt, Maroc

Résumé. — Les Grès rouges infracénomaniens du Maroc méridional, supposés d’üge Albien, ont fourni

l’un des ensembles de dinosaures les plus diversifiés d’Afrique, avec en particulier une espèce de Spmosaurus

à cou relativement long, et des os abondants mais isolés d'un Ihéropode singulier {«Spinosaurus B » de Stromer

1934). Sont également conservés les plus vieux restes connus d'abelisaurides et parmi les plus vieux de tino-

saurides en Afrique. Les théropodes .sont les formes les plus abondamment représentées, suivis par les sauropodes;

aucun reste d’omithischien n'a été identifié. Des os de dinosaures juvéniles sont aussi présents, l’un d'eux pro¬

venant d'un individu d’un poids inférieur à quatre kilogrammes. L’ensemble évoque davantage celui de la For¬

mation Bahanya que celui de Gadoufaoua, peui-ëire en raison d'une dépendance trophique envers de grands

poissons d’eau douce. D’un point de vue zoogéographique, cet assemblage .se relie de plus près, à l'Amérique

du Sud qu’à l'Amérique du Nord.

Mots-clés. — Dinosaures, Crétacé moyen, Maroc, Biogéographie.

D. A. Russell, Nonh Carolina State Muséum of Saturai Sciences, and Nonh Carolina State University, Raleigh, USA: formerly

of the Canadian Muséum of Nature. Ottawa, Canada.

INTRODUCTION

Fed by melting snows in the High Atlas, the Rheriss and Ziz rivers descend to an alluvial

plain (the Tafilalt) surrounding the oases of Erfoud and Taouz in the Moroccan Presahara. The

plain lies within a terrain dominated by ranges of folded Paleozoic strata (the Anti-Atlas). At

the Southern edge of the Tafilalt, the two stream courses merge to fomi the Oued Daoura, which

continues through a broad tableland (hamada) called the Kern Kern into Algeria. Continental

— 350 —

strata of middie Cretaceous âge are expo.sed along both the plateau bordering the Tafilalt to the

north and the base of the escarpment of the Kcm Kern (Figs 1, 2).

Fossil vertebrate remains were collected front these strata by René Lavocat, in the course

of four winter expéditions (often in the company of Fernand Joly and Georges Choubert; Lavo¬

cat 1954a) to lhe Tafilalt bctween 1947 and 1952. These included sawfish and rare lungfish

teeth. cranial parts of a gianl coelacanth, a fragment of a skull possibly referrable to Lihycnsuchus

(H. D. Sues, pers. comm. 1994), and teeth and an anterior portion of lhe skull of a very large

crocodile. Several varieties of theropod teeth were also recovcred, as well as i.solated bones

compared to skeletal parts of Elaphrosuurus, Carchamdontosauriis and Spinosaiirus (see Table 1).

Lavocat's outstanding discovery was the incomplète but partially articulated sauropod skeleton

constituting the type of Rebbachisaurus garasbae (Lavocat 1952, 1954b), which he excavated

during the winters of 1949-50 and 1951-52 with the assistance of the Service géologique du

Maroc.

In 1971, collections were made near Taouz by a party from the Institut und Muséum fur

Géologie und Palaontologie of the Georg-August-Universitât in Gôttingen, led by H. Alberti.

Fig. 1. — Cenomanian-Turonian transgression in Morocco (after CHOUBERT 1952).

— 351 —

Table 1. — Fossil vertebrate taxa previously recorded in the “Grès rouges infracénomaniens” of the Tafilalt.

Hybodontidae

Hybodus sp. (Tabaste 1963)

Sclerorhynchidae

Onchopristis numidus, isolated teeth very common (Lavocat 1948, 1954a, 1955b: 20;

Tabaste 1963; Cappetta 1980: 155-6; Wenz 1980)

Semionotidae

Lepidotes sp. (Lavocat 1954a: 102; Tabaste 1963)

Gigantodontidae

Stromerichihys aethiopicusl (Tabaste 1963)

Eotrigonodontidae

Eoirigonodon tabroumiti (Tabaste 1963)

Ceratodontidae

Ceraîodus africanus, isolated tooth plates rare (Lavocat 1948, 1954; Choubert 1952;

Tabaste 1963; Wenz 1980)

Ceratodus hume/(T abaste 1963)

Ceratodus sp. (Lavocat 1948, 1954a, 102; Tabaste 1963)

Coelacanthidae

Mawsonia lavocati (Tabaste 1963; Wenz 1980, 1981)

Podocnemididae

Gen. indet. (de Broin 1988)

Araripemydidae

Araripemys sp. (de Broin 1988)

Libycosuchidae

Libycosuchus sp. (Buffetaut 1976, 1994)

cf. Trematochampsidae

Hamadasuchus rebouli (Buffetaut 1994)

Crocodilia, fam. indet.

“Thoracosaurus" cherifiensis (Lavocat 1955b); large, long-snouted form (Lavocat 1954a:

102; not Sarcosuchus, de Broin in Taquet 1976: 50; Buffetaut 1989a; may be a

pholidosaurid, Buffetaut in Benton 1993: 698).

Theropoda, fam. indet.

cf. Elaphrosaurus sp. (Lavocat 1954a: 102; 1954b; 1955b: 56)

Spinosauridae

Spinosaurus sp. (Buffetaut 1989a, 1989b)

Theropoda, fam. indet.

Carcharodontosaurus sp., isolated teeth numerous (Lavocat 1954b; Taquet 1976: 39;

Buffetaut 1989a)

Diplodocidae

Rebbachisaurus garasbae (Lavocat 1951, 1954a, 1954b, 1955c: 20, 56)

cf. Ornithischia

Gen. indet. (Lavocat 1954b, 1955c: 20)

— 352 —

Additional sawfish, lungfish and giant coelacanth remains were recovered (Wenz 1980), as well

as those of a trematochampsid and a large, presumably long-snouted crocodile (Buffetaut

1989a). Among the dinosaurian remains were teeth of Carcharodontosaurus and a maxilla referred

to Spinosauriis (BUFFETAUT 1989a, 1989b).

More recently, the inhabitants of the Tafilalt hâve been encouraged by a growing inter¬

national market for extinct vertebrates to search adjacent escarpments for fossil teeth and bones.

Most of the materials now in private ownership probably consist of crocodile and Carcharodon-

tosaunis teeth. Many specimens hâve been obtained from local sources in the Tafilalt by Brian

Fig. 2. — Area of outcrop (in black) of the “Grès rouges infracénomaniens” near the Tafilalt (after Lavocat 1954a). Dotted

Unes represent drainage (“O.” - oued), “F” indicates a fossil occurrence (after Lavocat 1954a, Eberharde, pers. comm.

1994).

— 353 —

Eberharde (of Moussa Direct. Cambridge, U.K.), some of which were in tum deposited in the

Natural History Muséum in London, and a larger proportion of which was acquired by the

Canadian Muséum of Nature. Of the materials deposited in the latter institution, those of fishes

will be described by Stephen CUMBAA of the Canadian Muséum of Nature, of chelonians by

France DE Broin of the Muséum national d’Histoire naturelle, Paris, and of crocodiles by Hans-

Dieter SUES of the Royal Ontario Muséum. Dinosaurian materials are described below.

Because the teeth and bones were collected by non-specialists, only general locality infor¬

mation could be provided by Mr. Eberharde. The new materials do provide morphological

Fig. 3. — Représentative section of the “Trilogie mésocréiacée” in the région of ihe Tafilalt (after JOLY 1962). A, “Grès rouges

infracénomaniens"; B, “Marnes versicolores à gypse”; C, “Calcaire cenomano-turonien”.

— 354 —

information which would otherwise be lost were the specimens to be excluded from paleonto-

logical considération. Hopefully, inadequacies regarding the provenance of specimens will partiy

be offset by a fuller appréciation of the character of vertebrate assemblage inhabiting the Tafilalt

during Middle Cretaceous time.

STRATIGRAPHY AND CORRELATION

The “Trilogie mésocrétacée” which borders the Anti-Atlas région to the east and south has

been separaled into three lithologie entities (Choubert 1952). The sequence (Fig. 3) was sum-

marily described by Joi.Y (1962: 60-62):

"1. Des grès inférieurs, ruuges, déiriliques, alluviaux et continentaux. Leur épaisseur

moyenne est de l 'ordre de 100 à 150 in : mais elle peut varier, surtout lorsque les grès recouvrent

une surface accidentée dont ils ermoient les irrégularités ; en outre, elle s’amincit sur le bord

des bassins de sédimentation. Le faciès présente aussi de légères variations selon les lieux et

surtout selon les niveaux.

À la base, et particulièrement dans les creux des reliefs ennoyés, ils .sont grossiers et hété¬

rogènes. Ils recouvrent .souvent le Primaire par riniermédiaire d'un lit discontinu de quelques

dm, tantôt de caiUoutis anguleux qui évoquent le revêtement des glacis d'érosion, tantôt de

galets roulés comme ceux des fonds d'oueds, fis se composent de passées de graviers ou de

sables en forte proportion émoussés et luisants, intercalés dans des grès ruuges mal cimentés,

à gros grains souvent mats ou au moins picotés, avec parfois des traces de calcaire ou de gypse,

ou, plus communément, des bancs de conglomérats à éléments de Primaire.

Plus haut, les grès sont plus fins, plus uniformes aussi dans leur composition minéralogique

essentiellement quartzeuse, et mieicx calibrés. Ils sont de teinte plus claire, voire même franche¬

ment jaunes : les grains luisants abondent : les stratifications entrecroisées sont plus fréquentes

et plus nettes ; les dragées de quartz sont plus nombreuses. On y rencontre .souvent des passées

argileuses, des traces de plantes et des ossements plus ou moins roulés d'animaux de lagunes

ou de marigots (pois.soiis, crocodihens, dinosauriens).

Enfin, vers le sommet, s'affirment et se multiplient les entrelits marneux et argileux qui

annoncent des conditions tuntvelles.

2. Des couches de passage marneuses, rouges, blanches ou bigarrées, parfois gypseuses,

passant à des marno-calcaires ou des calcaires gréseux. Elles peuvent atteindre plusieurs dizaines

de mètres d’épaisseur. Elles annoncent le retour de la mer et le début de la grande transgression

cénomanienne. Cette formation évoque un régime de lagunes où s’accumulaient encore des dépôts

détritiques, en bordure d'une mer transgressive.

3. Un puissant complexe calcaire, la dalle, formation franchement marine, épaisse de 40

à 75 m, qui marque le maximum d'e.xtension de la mer.”

The three divisions are more generally known, respectively, as the “Grès rouges infra-

cénomaniens” (locally exceeding 200 m in thickness). “Marnes versicolores à gypse” (100-

200 m), and “Calcaire cénomano-luronicn” (100-150 m; CHOirBERT 1952; Basse & CHOUBERT

1959). The ammonite Neolobites vihrayeanus occurs at the base of the “Calcaire cénomano-

— 355 —

turonien” in the Anti-Atlas région (Choubert 1952: 145-146), and more particularly in the west¬

ern Kern Kem (BASSE & CHOUBERT 1959: 72). The Vibrayeanus Zone is considered by Meister

et al. (1991) to represent the lower part of the upper Cenomanian, which accordingly represents

the younger limit for the âge of the Tafilalt fossil vertebrate assemblage.

The older limit for the âge of the assemblage is very poorly constrained. The mineralogy

of the “Grès infracénomaniens” corresponds to that of “séquence B” of the “Couches rouges”

in the High Atlas, which apparently was deposited during Upper Jurassic and Lower Cretaceous

time (Monbaron et al. 1990). Near Anoual, some 150 km north-east of the Tafilalt, “séquence B”

contains a mammal-bearing horizon which may be of Berriasian-Valanginian âge (SioOGNEAU-

Russell et al. 1990; DuFFiN & Sigogneau-Russell 1993). A Barremian detritic épisode is

widely developed across north-we.stem Africa (BUSSON & CORNÉE 1989, 1991). Whether or not

this detritic épisode is related to pre- or intra-Barremian tectonic movements in the Middle Atlas

(CharrièRE 1992), a “faible discordance” above the mammal-bearing horizon in the section

near Anoual (Sigogneau-Russëll et al. 1990: 474) and the Paleozoic erosional surface buried

beneath the “Grès infracénomaniens" in the Tafilalt (JOLY 1962) is an unresolved question.

Possibly in part due to their successional continuity with the overlying Cenomanian carbonates,

the “Grès infracénomaniens” are usually considered to be of Albian to Cenomanian âge {cf. de

Broin 1988; BuffetaUT 1994; CaPPETTA 1980; Wenz 1980).

Exposures of the “Grès infracénomaniens" extend for 250 km, curving east and south from

Erfoud, and then Southwest along the northem escarpment of the Kem Kem (Fig. 2). Fossil

occurrences are rare and usually consist of isolated teeth or a few fragments of abraded bone

(Lavocat 1949, 1954a). Occasionally, large concentrations of water-wom fossils occur, including

Onchopristis and crocodile teeth, and ganoid scales, Teeth of Carcharodontosaurus are about

half as abundant as those of crocodiles, and sauropod teeth are less than half as abondant as

those of Carcharodontosaurus. Shark teeth are rare and apparently limited to outerops near Erfoud

(Brian Eberharde, pers. comm. 1994). Occurrences of associated skeletal parts of large dinosaurs

include the type of Rebbachisaurus garasbae, cited above, and unconfirmed reports of two spéci¬

mens, which were allegedly discovered by local people recently and partly destroyed.

SYSTEMATIC PALEONTOLOGY

Super-order SAURISCHIA Seeley, 1888

Order THEROPODA Marsh, 1881

Family Spinosauridae Stromer, 1915

Spinosaurus maroccanus n. $p.

Type. — Médian cervical vertebra NMC 50791 (Fig. 4a-c, NMC is the acronym for a specimen in the

collections of the Canadian Muséum of Nature), acquired by the Muséum through the courtesy of Mr. Raymond

Meyer.

Referred SPECIMENS. — Dentary fragments NMC 50832 (Fig. 5), 50833 (Fig. 6); médian cervical vertebrae

NMC 41768 (Fig. 4d). 50790 (Fig. 7), dorsal neural arch NMC 50813 (Fig. 8).

— 356

Diagnosis. — Ratio between length of centrum (excluding anterior articular condyle) and height of posterior

articular facet of centrum approximately 1.5 in mid-cervical vertebrae.

Etymology. — The species is named for the “royaume du Maroc” (kingdom of Morocco).

Fig. 4. — Spinosaums marocannus sp. nov., médian cervical vertebrae of NMC 50791 (type) in (a) left latéral, (b) anterior and

(c) po.sterior aspect; (d) NMC 41768 in left latéral aspect. Unless otherwise indicated, scale bar in this and following figures

represents 5 cm; cross-hatching indicates broken surfaces.

Description and discussion

In S. aegyptiacus the ratio belween length of centrum and height of posterior articular facet

approaches I. I, suggesting that the cervical centra were more siender and the neck was longer

in the Moroccan species (see Fig. 9, Table 2A). The pedicles of the neural arch are also elongated,

but the vertebrae are otherwise very similar in the two forms. The cervical vertebrae strikingly

differ from those of “Spinosaurus B” in the greater height of the vertébral spines, and in that

the neck was bowed dorsally, in the manner usual for large theropods.

Two dentary fragments (NMC 50832, 50832: Figs 5, 6) are essentially indistinguishable

from the dentary in the type of S. aegyptiacus (cf. Stromer 1915, pl. I, figs 6, 12a-b), if only

357

Table 2A. — Spinosaurus maroccanus n. sp., measurements of verlebrae: Hc and Wc mcusurcd from thc posterior articular facei.

L represents total length. Le docs noi include thc anierior hcmisphcrical surface. @ nieans "approxiinately”. \X represents

total lengih belween anierior rim of anierior zygapophysis and posterior rim of posterior zygapophysis, LNA represents min¬

imum length of neural arch pediclcs, HV represents height of vertebra from posteroventral edge of cemrum to top of neural

spine. Measurements of the type of S. aef^yptiacus arc from Stromer (1915: 24; thosc preceded by an asterisk were laken

from pl. 2).

LNA

NMC 41768

183

143

103

242

126

NMC 50790

—

@ 152

233

120

289

NMC 50791

195

146

100

328

115

—

Posterior

Type of S. aegyptiacus

vert a

vert b

7190

185

*140

*133

*195

‘100

*108

slightly more than three quarters the ünear dimensions of the Egyptian specimen. They were

derived from opposite dentaries of similar size, and the alveoli are here arranged into an anter-

oposterior sériés (Table 2B) in the assumption that NMC 50832 contains the anterior two pairs

of small, mid-mandibular alveoli présent in 5. aegyplktcus (STROMER 1915, pl. 1, fig. 12b), and

NMC 50833 contains the posterior two pairs (for a total of four pairs). There are no indications

of interdental plates. An unerupted tooth was présent in the second alveolus from the front of

the dentary in NMC 50832. It consists of an enamel shell measuring 45 mm in height. and

18 mm anteroposteriorly and 17 mm transversely at the base. The carinae are smooth, and

bordered by a smooth longitudinal sulcus approximately 2 mm wide. The remainder of the surface

of the Crown is covered with small, 1 mm wide longitudinal ridges which become finer and

converge toward the apex of the tooth.

Table 2B. — Spinosaurus maroccanus n. sp., measurements of length of aivcolar openings in dentary, from front to back.

NMC 50832

NMC 50833

third

pair

first

pair

fourth

pair

second pair

incomplète

incomplet

— 359 —

Of the cervical vertebrae, NMC 41768 is larger than NMC 50791 bul its morphology retlects

a more anterior position in the neck. The postzygapophyseal epipophyses are relatively larger

in NMC 41768, the diapophysis are Icss robust and there is no elevated area for insertion of

flexor musculature on lhe posteroventral edge of the centnini, as there is in NMC 50791. The

epipophyses arc approximately intermediate in development in NMC 50790. Found in association

with the lutter vertebra were two long and slender cervical ribs, the more complété of which

(NMC 507906) measures 493 mm in Icnglh and 10 to 12 mm in mid-shaft diameter. The base

of the tuberculum is 36 mm broad anteroposteriorly and of the correct proportions to articulate

smoothly with a cervical diapophysis; the abraded capitulum vvas apparently subcircular in out-

line. Although the distal end of the rib is missing, the shaft exceeded the length of the centrum

by a factor of at least 3.2. By comparison, in Caniotaurus the length of the fourth cervical rib

exceeds that of the centrum by a factor of about 5 (Bonaparte et al. 1990). Isolated fragments

of cervical ribs could easily be mistaken for omithopod ossified tendons.

A neural arch (NMC 50813) is referred to Spinosaurus because of an exceedingly robust

and buttressed neural spine base (Fig. 8, see also Stromer 1915, pl. I, fig. 19). It was derived

from the anterior portion of the dorsal column, as indicated by the capitular facet being shared

between the centrum and the neural arch. The spine base is much more massive than in a vertebra

similar to those of "Spino.tauru.'i B” (NMC 41850) from the same région of the vertébral column.

The anterior pedicles of the neural arches broadly diverge anteriorly.

— 361 —

SiGILMASSASAURIDAE, fam. nov.

SIGILMASSASAURUS, n. g.

Diacnosis. — Apomorphies in lhe cervical vertebruc for lhe genus and fainily include: spines short, wider

transversely lhan long; inlercentral arliculalions wider tliati high. exceeding length of ccntrum in width; Iransverse

planes ihrough ameriiir and posterior inierceniral articulations converge dorsally. indieating cervical sériés “U”-

shaped in latéral aspect; parapophyses projeel vcnirolaterally beyond nin of intercentral articulations, hy-

papophyses more powerfully devcloped in iniddle of cervical séries lhan in région of cervico-dorsal transition.

Etymology. — The generic name is derived from the ancieiu city of Sigilmassa, former capital of the

Tafilalt and a former centre of commerce in the western Sahara.

StromER (1934) designated theropod .skcletal clement.s collected from a .single site in the

Bahariya Oasis of Egypt as "Spinnsuiints B,” noting thaï two individuals were represented. He

doubted that the limb éléments, which belonged to an animal weighing on the order of 400 kg

(c/ Ander.son et al. 1985). could bc referred to Spinusaurm. However, the vcrtcbrac, which

were derived from an animal the size of AUosaurus or Carnntaurus (about 1.400 kg. cf. ANDER¬

SON et ai 1985; Bonaparte et al. 1990: 30), apart from the shortness of their spines. were

considered to resemble those in the type of Spinosaiiriis aegyptiaciis. BUFI-ETAUT (1989a, 1989b)

also noted that the short spines in the vertebrae of "Spiiiosaurus B” distingui.shed them from

those of 5. aegyptiat us, Indeed, the c-ervical vertebrae of Sigilma.'isasauru.'i differ so greatly from

those of S. aegyptiaciis in lhe relative length and brcadtli of the centra, the angle between the

plane of the intercentral articulations and the longitudinal axis of the centrum, and lhe height

of the spines (see above) thaï lhe two forms are here held to be distinct on a family-group level.

Vertebrae are here identified as Sigilwas.sasaiint.s because of their fundamental similarity to verte¬

brae from lhe cervical, dorsal and caudal régions of the larger specimen described and figured

by STROMER (19.34: 8-11, pl. figs 2-6).

These vertebrae also differ from those of other Bahariya theropod taxa. A mid-cervical

(“Halswirbel b”) in the major specimen STROMER (1931; II) referred to Carcharodonto.saurus

differs in that the width of the anterior central lacet is less than the length of the centrum and

the médian ventral keel ihickens posleriorly, lhe rever.se of conditions in “Spinosaurus B.” Pro¬

ximal caudals referred by Stromer (1931, pl. I fig. 10; 1934, pl. 2 fig. 5) to Carcharodon-

tosaurus and Baharia.'iaurLis bear pleurocoels, unlike in “Spinosaurus B.” The chevrons are not

bridged in “Spinosaurus B,” as they are in Carcharodontosaurus (Stromer 1934: 12).

Sigilmassasaurus brevicollis n. sp.

Type. — Cervical verlebra NMC 41857 (Fig. 10).

Diagnosis. — As for genus; l'or species-level characters see under “Description” and “Discussion”, Sigil¬

massasaurus sp., below.

Etymology. — The species name alludes to lhe shortness of lhe cervical centra.

Referred spectme.ns. — (Listed in u morphologicully anteroposterior sequence. although none are known

to bave been associated): cervical vertebrae NMC 41790, 41774. 41856, 41857 (type); dorsal vertebrae NMC

41858 (Fig. lla-b). 41850 (Fig. Ilc-d), .50428. 50402. 41776, 5(W07. .508«), 41772, 41851; caudal vertebrae

NMC 41854 (Fig. 12a), 41775. 4185.5 (Fig. I2b-d), 41855 (Fig. I2e-g).

— 362 —

Description

The différence in the proportions of the smallest (NMC 41774) and largest (NMC 41856)

cervical centriim in the collection (see Table 3) suggests an order of magnitude différence in

weight betwecn the animais which produced them. Allhough it has not been possible to observe

sequential variation within an articulated sériés of anterior presacral vertebrae, the vertebrae can

Table 3, — Sisilmassosaurus hrevicoilis, mcasurcnieiits of verlehrae: Hc and Wp measured from ihe posierior articular facei,

Wa mea-surcd fmm iht* anïcrior articular facti, L rcpresenls tolal Icnglh. (ff nieans ‘‘approximalely*’. Le cxcludes the anterior

heinispherical surface. Dw measured acinss (hc ends of the diapophyses. The vertebrae are arianged morphologically in an

anlcrior-poslerior séquence. Ali excepl NMC -11629 (S. sp.) hâve been assigned lo 5. brevicoUis.

L Le Hc Wa Wp Dw

CERVICALS

NMC 41790

@ 127

—

115

—

NMC 41774

130

NMC 41856

146

113

145

150

—

NMC 41857

121

102

147

149

@ 336

NMC 41629

150

113

—

@ 115

124

— {S. sp

Posierior

DORSALS

NMC 41858

—

103

101

@ 141

142

NMC 41850

152

122

110

115

110

NMC 50428

—

NMC 50402

195

168

119

—

120

NMC 41776

@ 150

134

116

—

NMC 50407

@ 63

@ 69

NMC 50800

@ 56

@ 59

NMC 41772

162

138

109

@ 81

@ 93

NMC 41851

157

134

101

@ 88

—

Posierior

CAUDALS

NMC 41854

110

—

NMC 41775

NMC 41853

110

NMC 41855

NMC 41862

Posierior

— 364 —

easily be assembled into a morphological sériés. In theropods, the posteriormost cervical vertebra

may be separated from the anteriormost dorsal by the change in the associated rib from a slender

élément lying subparallel to the vertébral column to an angled structure with its long axis nearly

perpendicular to the vertébral column. This change occurs between the 9th and lOth postcranial

segments in Alkisaiims and Monolophosaurus, and the lOth and llth postcranial segments in

Carnotaurus and Tyrannosaurus. In each of these généra, the transverse process becomes hori¬

zontal, or nearly so, on the anteriormost dorsal vertebra (OsBORN 1917, pl. 27; Madsen 1976;

Bonaparte et al. 1990; Zhao & Currie 1993). By virtue of the fact that the transverse process

is horizontal on NMC 41858, this vertebra will be considered as an anterior dorsal.

Fig. 11. — Sigilmassasaurus brevicollis n. g. et n. sp., referred dorsal vertebrae; NMC 41858, postulated anteriormost dorsal in

(a) posterior and (b) right latéral aspect; NMC 41850, anterior dorsal in (c) anterior and (d) right latéral aspect. Scale bar:

5 cm.

— 365

The spines are short, anteroposteriorly narrow and wide transversely through much of the

cervical sequence (NMC 41790, 41774, 41856, 41857), unlike in Spinosaurus where the spines

are elongaled (Stromer 1915). They broaden anteroposteriorly in the anterior dorsal région

(NMC 41858, 41850) where they apparently remain short but slope posterodorsally. The articular

surfaces of the anterior and posterior zygapophyses are, respectively, gently convex and concave.

The superior surfaces of the posterior zygapophyses arc smooth in the cervical région, with no

indications of the epipophyseal tuberosities which occur in Spinosaurus (Stromer 1915). The

diapophyses are short, straight and project ventrolaterally in an anterior cervical (NMC 41790),

become recurved and elevated in médian ccrvicals (NMC 41774, 41856, 41857), are straight,

and posterolaterally directed in an anterior dorsal (NMC 41858), and slightly elevated and an-

terolaterally directed in an anterior dorsal (NMC 41850) which occupied a morphologie position

posterior to the preceding vertebra. Pneumatic foramina on the anterior and posterior surfaces

Fig. 12. — SigUmassasaurus hrevicollis gen. el sp. nov., referred caudal vertebrae; (a) NMC 41854, mid-caudal in left latéral

aspect; NMC 41853. mid-caudal in (b) left latéral, (c) dorsal and (d) posterior aspect; NMC 41855, posterior caudal in (e)

left latéral, (0 dorsal and (g) po.sterior aspect. Scale bar; 5 cm.

— 366

of the diapophysis in the cervical vertebrae disappear posteriorly within the anterior dorsal région,

and an infradiapophyseal lamina, which appears in the posterior cervical sequence, becomes

divided into anterior and posterior branches in the anterior dorsal région.

The cross-sectional area of the neural canal remains relatively constant in the morphological

sequence including cervicals NMC 41856, 41857 and dorsal NMC 41858, averaging 8.6% of

the area of the intercentral articulations. However, in the anterior dorsal (NMC 41850) located

morphologically behind NMC 41858 it i.s only 4.6% of that of the intercentral articulations.

These circumstances suggest that the forelimbs werc rudimentary in thi.s foim, for the neural

canal is proportionately larger in those with large forelimbs (ClFFtN 1990). A very shallow emargi-

nation for the neural canal is présent within the dorsal margin of the posterior central articular

surface in several cervicals (NMC 41790, 41774 and 41856). The emargination is deeper in a

posterior cervical (NMC 41857) and an anterior dorsal (NMC 41858, in which the posterior end

of the right neural arch is necrotic). The emargination is no longer présent in a more posteriorly

situated anterior dorsal (NMC 41850).

Relative to those of olher theropods, the cervical centra are e.xceedingly broad, and the

width of the posterior articular surface exceeds its height by a factor of about 1.5. In AUosaurus

(MadSEN 1976), Camotaurus (Bonaparte et al. 1990) and Spinosauru.’i (sec above) the ratio

is about I 25, and in Deinonychus (O.STROM 1969) and tyrannosaurs (RUSSELL 1970a) it is about

1.00. Behind the anterior end of the dorsal column (NMC 41858), the intercentra) articulations

narrow and bcgin to dccpcn (NMC 41850), becoming vertically oval (NMC 41776, 41772,

41851), The width/length ratio increases (see Table 3) from the anterior (NMC 41774, 41856)

to the posterior cervical région (NMC 41857), after which it diminishes through the anterior

dorsal sequence. Ail preserved dorsal vertebrae are strongly opisthocoelous.

The parapophysis is low in anterior cervicals (NMC 41790, 41856, 41774) but occurs at

mid-centnm height in a posterior cervical (NMC 41857). It straddles the centrum-neural arch

suture in two anterior dorsals (NMC 41858, 41850). In AUosaurus and Tyrannosaurus the par¬

apophysis is confined to the centtvm on ail of the anterior presacral vertebrae back to and in¬

cluding the second dorsal, and it straddles the centntm-neütal arch suture on several more

segments posteriorly (Madsen 1976; OSBORN 1917, pl. 27). A small, posteriorly sloping neural

spine would place the first Moroccan dorsal loward the front of the anterior dorsal sériés, and

the broad neural spine would place the second several segments behind. A pleurocoel situated

above the parapophysfs in cervical centra diminishes posteriorly in the sequence as it is apparently

displaced to a position above the suture for the neural arch (NMC 41858); there are no pleurocoels

on available dorsal centra (the centrum is incompletely pre.served in NMC 41858).

The angle bctwecn the plane of the intcrcentral articulations and the longitudinal axis of

the cervical centra indicates that the neck was strongly arched into a ^’U”-shaped curve (Fig. 13),

a condition apparently unknown in any other theropod. Measurements of available vertebrae

suggest that the surface area of the intercentral articulations may not hâve increased greatly

posteriorly in the cervical sériés, alihough it typically doubles in other large carnivorous theropods

(cf. measurements in Allosauru.s Gilmore, 1920; Monotophosaurus Zhao & Currie, 1993; Car-

notaurus Bonaparte et al., 1990; and Albertosaurus and Da.splelosaurus Russell, 1970a). In ha-

drosaurs, where the neck is also “U"-shaped in latéral profile, the intercentral articulations do

not increase greatly in area from front to back (pers. obs.). Perhaps the neck in Sigilmassasaurus

— 367 —

was composée! of a relatively large number of vertébral segments, as in hadrosaurs (Weishampel

& Horner 1990). It is probable that it was more flexible and capable of a much greater variety

of movements than the back.

Fig. 13. — Sigilmassasaunis brevicoUis n. g. el n. sp., rcconslruclcd anterior presacral vertébral column; vertebrae, from left to

right, are reconstructed and scaled afier NMC 41790. 41774, 41856, 41857 (dotled, interpolated), 41858 (dotted, interpolated)

and 41850.

Although a longitudinal keel is présent, a hypapophyseal tuberosity is not developed on an

anterior cervical from a juvénile (NMC 41774). It is small in the mid-cervical région (NMC

41856). enormously developed on a posterior cervical (NMC 41857) and diminishes in size within

the anterior dorsal région (NMC 41850, 41776). In Deinonychus (OSTROM 1969) and Sinor-

nithoides (RUSSELL & DONC 1993) the hypapophyses are most strongly developed behind the

posterior cervical région, within the cervico-dorsal transition.

In a dorsal in the specimen originally designated as "'Spinosaiims B” the parapophysis is

situated enlirely on the neural arch, and the vertebra musi hâve occupied a morphological position

posterior to NMC 41850. As in the latter vertebra. however, the transverse process is anter-

olaterally directed. The Egyptian vertebra evidently lacks a hypapophysis, but does bear a longi¬

tudinal ridge ventrally (Stromer 1934: 8, pl. I fig. 4). Of threc médian dorsal centra from

Morocco, one (NMC 41776) bears a small longitudinal keel anteroventrally, the second

(NMC 41772) bears a slightly elevated longitudinal rugosity, the third (NMC 41851) bears a

— 368 —

diffuse swelling in this région luarked by longitudinal striae. The dorsal vertcbrae arc arranged

in a hypotbetical anteroposterior sériés in Table 3 on the basis of the narrowing inlercentral

facets which more closely approximate an orientation at right angles to the long axis of the

cenTrum. and diminishing hypophyseal keef The dorsal cinilm are very similar to those of Spi-

nosaunix, and some may here mislakenly be assigned to Sigilniassa.iaunis. Two dorsal centra

from similar anatomical positions represent extrêmes of a possible growih serie.s; one (NMC

50402) is about 5.3 times the lincar dimensions of the other (NMC 50428), which cubed yields

a volume (weight) différence on the order of 150 times. The neural arch suture is open in both

vertebrae.

Scveral caudal vertcbrae aiso appear lo form a morphological serie.s (NMC 41854, 41775,

41853, 41855) consistent with caudals belonging lo ‘'Spinaxaurus B." In the rectangular shape

of the centra^ lhe vertebrae aIso generally rcsemble médian and distal caudals of Onrannsaiirus

(Taquet 1976). This re.semblance is enhanced by lall, poslerodorsally projecting spines. However,

StromeR ( 1934; 10), because of the dividcd nature of lhe spinc in a distal caudal, inteipreted

the structure as representing the anterior zygapophyses. Nonc of the Moroccan caudals described

here possess a divided spine, but NMC 41854 and 41855 otherwise closely resemhle those figured

by Stromer (1934, pl. I figs 4, 6). The "anterior zygapophyse.s” arliculate with nolhing in the

Egypiian vertebra, and would represent a z.ygapophyseal adaptation unique among the distal

caudals of dinosaurs. Il would be more conservative lo interpret the structure as either a divided

or a pathologie .spine; the latter alternative is provisionally adopted here.

In ail of lhe available caudals, the spine rises from the posterior half of lhe neural arches,

and is wider than long. Anteriorly, the smooth lamina of bone roofing the neural canal is very

thin. and broken away between the bases of lhe anterior zyagapophyses. The latter are broken

off, but probably lacked articular surfaces in lhe more distal caudals (NMC 41853, 41855). Broken

but well-developed transverse processes are présent m NMC 41854, and a low, longitudinal ridge

is présent in this région of the centrum in NMC 41775.

Discussion

The cervical vertebrae superficially resemble those of Ouranosauriis (Taquet 1976) in lack-

ing a well-developed neural spine, but differ greatly in possessing pneumatic foramina on the

centrum and neural arches, very broad central articular lacets, powerfully devclopcd diapophyses

and abbreviaied postzygapophyses. The cervical centra akso superficially rescmble those of ti-

tanosaurs in possessing pneumatic foramina, but are much shoriei than in these sauropods {cf.

Bonaparte & PowELL 1980; Powell 1987; Jacobs et al. 1993). In the extrême réduction of

the neural spine, lhe recurving diapophyses and ihc emplacement of the pneumatic foramen on

their anterior surfaces, and the broad intercentral articulations, the cervicals clostely rcsemble an

anterior dorsal from Tendaguru referred by Janensch (1929a, fig. Il) to the .saiiropod '‘Gigan-

tosaurus.” The latter vertebra is reconstructed to measure over I m across the diapophyses,

whereas the largest Moroccan vertebra possessing well-preserved diapophyses probably measured

but 34 cm in this dimension. The position of the parapophysis indicates that the East African

vertebra is from the anterior dorsal région. Yet it lacks a spine. which is well developed in the

anterior dorsals of the Moroccan theropod. It aiso lacks a hypapophyseal peduncle. The resem-

— 369 —

blance is nonetheless striking. The morphological transition from cervical into normal theropod

dorsal vertebrae, however, supports the theropod atTinilies of Sigilmassasaunis.

From the enormous size of the hypapophyses in the posterior cervical région, it could be

inferred that the muscles attached to them would hâve powerfully and rapidly projected the neck

downwards. As noted above. the cervical segments bccomc more robust posteriorly in lheropods

with large skulls. If the skull of Sigihnassa.taiirus was small (relative to conditions in large

theropods), and comparable in size to the skull m rnost hadrosaurs with cervical vertebrae of

approximately equal size and comparable morphology. one is left with the image of a theropod

weighing well over a tonne, with rudimentary forelimbs, a neck adapted for pecking and a skull

of avian proportions relative to the size of the body. Superticially, the tail may hâve resembled

that of an ornithopod more closely than that of a theropod. It is évident that Sigilmassasaurus

may represent a truly unusual group of theropod dinosaurs.

Sigilmassasaurus sp.

Referred SPECIMEN. — Postcrior cervical centrum NMC 41629 (Fig. 14c, f, g), and distal caudal vertebra

NMC 41862.

Description and discussion

The cervical centrum belonged to a large vertebra. but the sutures were insufficiently closed

to prevent the séparation and loss of the neural arch. The colour of préservation is a deep,

reddish brown, and quite distinct from the white to light orange colour of other bonus in the

NMC collection. Its dark colour is consistent with the darker colour of sédiments at the base

of the “Grès rouges,” and suggests thaï the bone was derived from the base of the sequencc. It

could be older than the other specimens by tens of million of years date Barremian through

early Cenomanian tinte spans ncarly 35 million years, Harland et al., 1990).

The parapophysis in NMC 41629 is located at mid-height on the anterior edge of the latéral

wall of the centrum. Il is aiso locaicd at ihis level in a cervical vertebra figured by Stromer

(1934, pl. 1 fig. 2) from Egypt, as well as in NMC 41857, described above. Thus, ail three

vertebrae were from the base of the neck. They can nevertheless be arranged in a regular mor¬

phologie sequence from relalively plesiomorphous to relatively derived:

Sigilma.ssasaurus sp. (NMC 41629, Fig. 14c, f, g):

Centrum narrow, posterior width: length ratio 1.10;

Intercentral articulations indicate neck flexed slightly (Fig. 15);

Latéral wall of centrum planar above pleurocoel;

Pleurococl large;

Parapophyseal articulation faces laterally;

Hypapophyseal peduncle small.

‘’^Spino.saurus B” of STROMER (1934, pl. 1 fig. 2; here Fig. 14b, e, h, stippled):

Centrum intemiediatc, posterior width; length ratio 1.24;

Orientation of intercentral articulations uncertain;

Latéral wall of centrum swollen above pleurocoel;

— 370 —

Pleurocoel intermediate in size;

Parapophyseal articulation faces lateroventrally;

Hypapophyseal peduncle enlarged.

Sigilmassasaums brevicollis (NMC 41857, Fig. 14a, d, i):

Centnim broad, posterior width: length ratio 1.46;

g h

Fig. 14. — Sigilmassasaums sp., posterior cervical vertebrae; NMC 41857 S. brevicollis (type) in (a) anterior, (d) ventral and

(i) left latéral (reversed) aspect; "Spinosaums B” (dotted. afler Stromer 1934) in (b) anterior, (e) ventral and (h) righl

latéral aspect; NMC 41629 S. sp. in (c) anterior, (0 ventral and (g) left latéral (reversed) aspect. Scale bar: 5 cm.

— 371 —

Intercentral articulations indicate strongly flexed neck (Fig. 13);

Latéral wall of centrum greatly swollen above pleurocoel;

Pleurocoel small;

Parapophyseal articulation faces ventrolaterally;

Hypapophyseal peduncle greatly enlarged.

Fig. 15. — Sigilmassasaurus sp.; suggested curvature of cervical sériés, fitting three outlines of NMC 41629 in left latéral aspect.

Scale bar: 5 cm.

The sequence, however, may not represent a single phylogenetic clade; see “Paleoen-

vironmental and Paleobiogeographical Comparisons” below.

Although the centrum of a distal caudal (NMC 41862) is of the same proportions as a

similar element referred to S. brevicoUis (NMC 41855), it is less constricted medially and much

more robust in appearance. The diameler of the neural canal is much greater than in NMC 41855

(19 mm VS 11 mm). It may belong to a distinct, but closely related form, and, as in NMC 41629,

is more darkly coloured.

Family Carcharodontosauridae Stromer, 1931

Carcharodontosaurus sahariens (Depéret & Savomin, 1927)

Referred specimens. — Maxilla fragment NMC 41859; isolated teeth NMC 41908, 41910, 41817, 41818,

41819; cervical vertebra NMC 50792 (Fig. 16).

Description and discussion

Depéret & Savornin (1927) established Megalosaurus sahariens on two teeth collected

from coarse clastics of late Barremian âge near Timimoun in central Algeria, 460 km south-east

of the Tafilalt (DE Lapparent 1960; 15; Lefranc 1983). The species was in turn designated

— 372 —

as the type species of Carcharodontosaurus by Stromer (1931). The denticles of the posterior

carina in teeth here referred to C. saharicus are consistently smaller than in most theropods, as

has been reported in Carcharodontosaurus by Farlow et al. (1991: 175 and équation on 174)

and as in the complété crown in the type of Megalosaurus saharicus figurcd by Depéret &

Savornin (1927, pl. 12 fig. 1. lA). The denticles are approximalely the same size in a well-

preserved replacement tooth in the maxilla of the skeletal specimen described by Stromer.

However, the apex of the crown in this tooth is symmetrical (STROMER 1931, pl. 1 fig. 2), not

posteriorly recurvcd as it is in other teeth referred to this species icf. Depéret & SAVORNtN

1927, pl. 12 figs 1-2; STROMER 1931, pl. 1 fig. 1; teeth referred to the C. saharicus herein).

Measurements of the teeth from the Tafilalt are given in Table 4A.

Table 4A. — Carcharodontosaurus saharicus, measurements of teeth.

Basal W

Basal L

Crown H

Denticles/5 mm

NMC 41908

NMC 41910

NMC 41817

NMC 41818

NMC 41819

A fragment of a left maxilla (NMC 41859) bears impressions of four alveoli (two in an-

teroposterior sequence measure 39 and 35 mm in longitudinal diameter), separated by three septa

measiiring between 7 and 11 mm thick ventrally. These dimensions arc similar to tho.se near the

centre of the left maxilla described by STROMER (1931: 7). The subrectangular outline of the

alveoli differs greatly from circular alveoli in the maxilla from Morocco referred by Buffetaut

(1989a) to Spinosaurus. The latéral surface of the fragment is slightiy concave; the latéral surface

of the nearly complété maxilla described by Stromer is fiat.

A single cervical vertebra (NMC 50792. Fig. 16, Table 4B) is similar to “Halswirbel b” in

the sériés of large vertebrae of lhe specimen referred to Carcharodontosaurus by STROMER (1931:

11-12, pl.. 1 fig. 9). However, the parapophysis is located in a more elevated position than in

the Egyptian élément, near the middle of lhe latéral surface of the centrum. Furthermore, a large,

conical excavation is aiso devcloped between ihc anlcrior zygapophysis and the neural canal,

which was not described by STROMER in the Egyptian specimen. Both characters are typical of

vertebrae occupying a more posterior position within the cervical sériés than the Egyptian vertebra

(cf. Madsen 1976). The absence of apowerfully developed ventral keel in the Moroccan vertebra

may al.so be duc to its position within the cervical column.

Table 4B, — Carcharodontosaurus saharicus, measurements of cervical vertebra (NMC 50792. for abbrevialions. see table 3).

L Le Hc Wa Wp

NMC 50792 148 96 101 107 101

— 374 —

The vertebra differs strikingly from cervicals of Spinosaurus and Sigilmassasaurus in that

the neural spine appears lo hâve been entirely or nearly entirely replaced by a shallow groove

extending along the mid-line of the element, above the neural canal. The zygapophyses are located

dorsolateral to the groove, similar lo their elevated po.sition in cervical vertebrae of Carnotaurus

(in which, however, a .small neural spine is présent, cf. BONAPARTE et al. 1990). The dorsal

mid'Iine of the Egyplian vertebra (Stromer 1931. Il) is noi preserved, so that its morphology

could not be described. The angle (15*^) betwecn the planes of the anterior and posterior inter¬

central articulations of the Moroccan vertebra indicutes that the neck was strongly arched dorsally

adjacent to it, Anteriorly, the neural canal measures 34 mm in height and 25 mm in width;

posteriorly the measurements are 24 and 29 mm, respectively. For other measurements, see

Table 4B. The vertebra is strongly opisthocoelous.

Family Abelisauridae Bonaparte & Novas, 1985

cf. Majungasaurus sp.

Referred SPECIMEN. — Fragment, médian portion of ramus of right dentary NMC 41861 (Fig. 17d).

Description and discussion

The element is badly abraded. As preserved, it measures 35 mm in depth and 134 mm in

length, and was originally probably slightly smaller than the dentary referred to Majungasaurus

crenaiissimus by Lavocat (1955b). It contains portions of nine empty alveoli, the close spacing

of which suggests that there may hâve been as many dentary teeth (17) as in the dentary from

Madagascar. Uniike in Majunga.sauru.s, lhe alveoli increase in length posteriorly (from 6 to

14 mm) but remain similar (rectangular) in dorsal outline. The eroded latéral surface of the bone

appears to hâve been massivcly constructed. Il is peculiar in that it is marked by indistinct

vertical ridges which parallel, but do not coincide with the anterior and posterior limits of the

alveoli. No Meckelian groove is présent on the médial surface of the bone; perhaps the fragment

was separated from a now-missing ventral half before it was buried.

Abelisauridae, gen. et sp. indet.

Referred specimen. — Fragment, médian portion of ramus of right dentary NMC 41859 (Fig. 17a-c).

Description and discussion

The fragment is approximately 20 cm long and contains empty alveoli for four teeth. It

measures 93 mm in depth below the anteriormost, and 98 mm below the posteriormost alveolus.

In dorsal view, the alveolar margin curves anteromedially, as in a Lameta “allosaurid” dentary

(VON HUENE & MatleY 1933, pl. 12 fig. 1), Majungasaurus (LavoCAT 1955a) and Carnotaurus

(Bonaparte et al. 1990). The central two alveoli are essentially complété, measuring 26 mm

in length. 10 mm in width and at least 32 mm in depth. They are slightly constricted medially.

An interdental lamination appears to hâve been présent rather than separate interdental plates

— 375 —

(as is apparently aiso the case in the dentary of Majungasaurus Lavocat, 1955a). Unlike in the

forms ciled above, the external surface of ihe dentary does not appear to be divided into a

wider, convex lower half and a narrower, fiat upper half. There is no evidence preserved of a

posterior emargination in the dentary. Sutures are présent posteroventrally on the médial surface

for contact with a splénia).

Fig. 17. — Abeli.sauridae, gen. et sp. indet., NMC 41859, fragment of right dentary in (a) dorsal, (b) media! and (c) latéral

(inverled) aspect; (d) cf. Majungasaurus sp., NMC 41861. fragment of right dentary in dorsal aspect. Scalc bar: 5 cm.

The elongated alveoli preclude the specimen’s referral to Spinosaurus, and their relatively

small size is incompatible with tooth dimensions in Carcharodontosaurus. The dentary is also

much smaller than in these two taxa.

THEROPODA indet.

The following bones arc sufficicntly charactcristic to suggcst that thcy will be recognized

in materials obtained in the future. Some anatomically similar éléments differ morphologically

to the exlent that they appear to belong to separate species. Ail are separated into informai bone

“taxa” designaled by letlers so thcy can be easily ciled. How thcy could be grouped together

into skeletons of distinct species is purely spéculative at prc.sent. Theropod humeri werc separated

from those of crocodiles through the presence of ihick cortical bone al mid-shafl, presumably

an adaptation to withstand stresses generated by a massive body upon a relatively small forclimb.

In NMC 41865 and 41873 cortical bone is very thick in this région (see also in Allosaiinis

Gilmore 1920, pl. 6, fig. 7; an unusually large proportion of tyrannosaur humeri suslained injury

in life; RUSSEIX 1970a: 18).

— 376

BONE “taxon” a (CRANIAL ROOF)

Two interorbital fragments (NMC 50807 and 50808 (Fig. 18) apparently represent the same

small to medium sized form. They do not resemble the skull roof in any known Southern hémi¬

sphère theropod (cf. von Huene & Matley 1933; Bonaparte & Novas 1985; Bonaparte et

al. 1990). l’he co-ossified fronials are powerfully constructed, as are suturai contacts for adjacent

ossifications. A transverse fiexure occurs anterior to the centre of the orbits, so thaï the anterior

portion of the fronials de.scends al an angle of 25" to the posterior half of the bone. Longitudinal

striations marking the posterior limits of the nasal alac exlend back to the point of fiexure; they

become sironger anterolaierally in the area of contact wiih the prefronlal and lacrimal. In NMC

50808, the frontals extend 50 mrn along the cranial mid-line from the fiexure to their posterior

limit. This measurement is 65 mm in NMC 50807. The parietals. which are preserved in the

latter specimen, form a .saggilal crest rather ihan a fiattened surface along the cranial mid-line.

On the ventral surface of the .skull roof, the horizontal diameter of the orbil is approximately

50 mm in NMC 50808, and a perpendicular druwn through its centre meeis the cranial mid-line

at an angle of 55". Anteriorly, the shallow cavity containing the olfactory bulbs is 18 mm wide;

well-marked twin cavities for the cérébral hemispheres total 27 mm in widlh near the posterior

limit of the frontals.

Fig. 18. — Theropoda, indet. (bone “taxon” A), NMC 50808, co-ossified frontals in (a) left latéral and (b) ventral aspect.

Scale bar: 5 cm.

Bone “taxon” B (cervical vertebrae)

An axis vertebra (NMC 50810, Fig. 19) with closed cen/ruw-neural arch sutures documents

the presence of a small theropod within the assemblage. It differs from a cervical described by

DE LappaRENT (I960: 31, pl. II. fig. 5) from In Tedreft, Niger, in that the cenirum is rounded

rather than .square in ventral cross-section, the pleurocoel is localcd posterior to rather than above

the capitular facct, and in that the two small excavations situated above the posterior exit of

neural canal in the specimen from Niger are not présent. The siender cenirum (45 mm long and

— 377 —

18 mm wide posteriorly) is suggestive of an elongaled neck. A centrum from the mid-cervical

région (NMC 50810) is similar in size but more robustly constructed (34 mm long and 27 mm

wide posteriorly). A small pleurocoel is located above and behind the capitulât facet; another

small opening is présent at the same level, centred within the posterior half of the left side.

Fig. 19. — Theropoda, indet. (bone “taxon" B), NMC 50810. axis vertebra in right latéral aspect. Scale bar: 5 cm.

BONE “taxon” C (POSTERIOR DORSAL VERTEBRA)

NMC 50403 (Fig. 20, Table 4C) is a relatively large vertebra. The posterior central facet

is marked by shallow, radiating suture-like grooves, suggesting proximity to the sacrum. Longi-

tudinally striated luberosities occur between the posterior (but not anterior) zygapophyses, indi-

cating the existence of tendinous links to the vertebra behind. The bone thus probably represents

the last dorsal vertebra. It differs markedly from vertebrae in the anterior dorsal région of Spi-

nosaurus and Sigilmassasaurus in its amphiplatyan nature and in the exceedingly weak, anteriorly

inclined base of the neural spine. Nor is an assignment to Carcharodontoxaurus likely because

Table 4C. — Theropoda indet., measurements (in mm) of posterior dorsal vertebra (NMC 50403).

Length of centrum 158

Width of anterior zygapophyses 35

Width of centrum anteriorly 37

Width of centrum posteriorly 122

Height of centrum posteriorly 120

Height of centrum anteriorly 141

Width of neural canal (between neural arch pedicils) 33

Width of neural canal (within centrum) 9

Depth of neural canal (between neural arch pedicils) 42

Depth of neural canal (within centrum) 13

— 378 —

it lacks pleurocoels, which are présent posteriorly into the caudal sériés of the lalter genus

(Stromer 1931, pl. 1, fig. 10).

Other attributes of the vertebra include an anterior zygapophyseal facet that faces nearly

directly upwards, and a relatively large neural canal. The dorsal margin of the canal is separated

into two halves by a low. longitudinal ridge; ventrally the canal is incised deeply into the dorsal

part of the centrum. The bases of the transverse processes indicate that the structures were only

slightiy elevated distally. Another central fragment (NMC 50404) clo.sely resembles the anterior

half of the centrum in NMC 50403.

Fig. 20. — Thcropoda, indet. (bone "taxon” C), NMC 50403, po.sterior dorsal vertebra in right latéral aspect. Scale bar: 5 cm.

Bone ‘taxon” D (Caudal vertebrae)

A vertebra (NMC 41863. Fig. 21) from the anterior portion of the distal half of the tail

differs from those assigned above to Sigilinassasaiirus in ils more cylindrical form and in the

présence of slrongly developed bases for the anterior zygapophyses. A bony ridge along the

dorsolateral surface of the centrum implies that the élément was originally located immediately

behind vertebrae bearing transverse processes. The anterior border of the much-abraded neural

spine is vertical, suggesting that the spine might also hâve been nearly vertical. In médian

caudals referred by Stromer (1934: 34, pl. 2, fig. 11) to Bahariasaurus the spine is post-

erodorsally inclined. The centrum is less elongated than in médian caudals referred by DE Lap-

PARENT (1960) to EJaphrosmirus. Its length is 86 mm, and the posterior central articulation is

51 mm wide and 50 mm high.

— 379 —

Another caudal (NMC 50797) is smaller and lacks any trace of transverse processes. The

shape of the centrum and vertical anterior base of the neural spine suggests that it was derived

from the same taxon as the previous vertebra. However, although the centrum is smaller (with

a length of 65 mm, and a posterior central articulation measuring 34 mm wide and 30 mm high),

the bases of the anterior zygapophyses are much more robustly developed. This could be a con¬

séquence of a more distal position in the caudal sériés.

Fig. 21. — Theropoda, indet. (bone “taxon” D), NMC 41863, distal caudal vertebra in (a) dorsal, (b) right latéral and (c) anterior

aspect. Scale bar: 5 cm.

Bone “taxon” E (coracoid)

NMC 41806 is a fragment of a coracoid, apparently derived from the left side, which con-

tains the coracoid foramen. There is no suturai surface preserved for contact with the scapula.

The greatest proximal width of the element is 44 mm; it was probably about as large as the

coracoid of Albertosaurus lihratus (NMC 2196). The coracoid foramen is, however, much larger

(30 mm VS 17 mm in the vertical diameter of extemal opening), and is also larger than in the

coracoid questionably referred by Stromer (1934: 33, pl. 3, fig. 14) to Carcharodontosaurus.

There is no channel from the foramen toward the médial surface of the scapula, as there is in

Allosaurus (Madsen 1976, pl. 41).

Bone “taxon” F (humerus)

The distal end of left humerus (NMC 41865, Fig. 22a) measures 156 mm in width. At

mid-shaft, the horizontal and vertical diameters are 69 and 55 mm, respectively, and the cortical

380 —

Wall is approximately 19 mm thick. Because the dislal end of NMC 41865 resembles that of

the humérus in Rarynnyx (Charig & MiLNER 1990, fig. 9.5) in a manncr analogous to similarities

between cervical vertebrae of Spinosaunts maroccanus and Baryonyx (compare NMC 41768,

and Norman & Milner 1989: figure on p. 55), il might pertain to Spinosaurus.

BONE “taxon” g (HUMERUS)

A large lefl humérus (NMC 41852, Fig. 22b-c) can be distinguished from NMC 41865 on

the basis of a much more robust mid-shaft région, and the absence of a large, moderately deep

concavity situated proximal to the radial condyle. Its shaft is very straight, and the long axis

Fig. 22. — Theropoda, indcl.: (a) NMC 41865 fbone “taxon" F), distal end of left humérus in flexor aspect; NMC 41852 (bone

“taxon" G), left humérus in (b) ventral aspect and (c) with the distal end in flexor aspect: (d) NMC 41873 (bone “taxon"

H), distal end of C right humérus in tlexor aspect. Scale bar: 5 cm.

— 381 —

of the articulation for the antebrachium is rotated clockwise (when viewed dislally) about 45°

with respect to the long axis of the proximal articular surface. The latissimus scar on the prox-

imodorsal surface is weak. The élément measures but 150 mm in distal widih, with robust hori¬

zontal and vertical mid-shaft diameters of 89 and 78 mm. respectively (the mid-shaft

circumferencc is 271 mm).

The humérus is imusually long for lhat of a lheropod. measuring 60 cm as preserved, and

probably about 65 cm long when entire. It is thus longer than recorded in Daspletosaunts (35.7 cm;

Russell 1970a), Tyrannosaums (36 cm: Osborm \9{)b), AUoseturus (38.6 cm, Madsen 1976), Tor-

vosaurus (42.4 cm; Galtün & JENSEN 1979). Buryonyx (46.3 cm: CHARIG & MlLNER 1990), Gal-

limimus (53 cm; OSMOLSKA et al. 1972). Segnosaurus (56 cm; PERLE 1979) and Chilanlaisaurus

(58 cm; Hu 1964). The humérus Is known to be longer only in Therizinasaurus (76 cm), where the

shaft is more sigmoid, and the ends more expanded (Barsbold 1976), and in Deinocheirus (93.8 cm),

in which it is much more slender (OSMOLSKA & RONIEWICZ 1970).

BONE “taxon” h (HUMERUS)

NMC 41873 (Fig. 22d) is the distal end of a right humérus. The bone is about the size

and general proportions of the humérus of Deinonychus (OsTROM 1969, figs 55-56). It is unusual

in that the radial condyle is very large and smooth, suggesting a well-developed capacity for

pronation and supination. The élément measures 48 mm in distal width, with horizontal and

vertical mid-shaft diameters of 23 and 21 mm. respectively; at mid-shaft, its circumference is

72 mm, and the cortical wall is approximatcly 8 mm Ihick.

The humérus, as well as the tibia from the Kern Kcm escatpment which was compared to

that of Elaphrasaurus by Lavocat (1954b), were both derived from a theropod approximatcly

equal in size to the type of Elaphrosaums bamhergi (JANENSCH 1925). No spécial affinity to

the Late Jurassic Elaphmsaiiru.s is thereby implied.

Bone “taxon” 1 (manal phalanges)

Three large manal phalanges w'ere probably ail derived from the same species, and are

reminiscent of, but slighlly smaller than a large phalanx from Alrar near the Libyan froniier of

Algeria described by DE Lapparent ( I960, “métatarsien”;. 29, pl. 9, fig. I ). NMC 50794 (Fig. 23)

is probably from the firsi digit, and measures 250 mm in interariicular length, 45 and 50 mm,

respectively. in the height and width of the di.stal articular surface, and 65 mm in the height of

the proximal articular surface. A fragment of a distal articulation (NMC 41800) is essentially

Fig. 23. — Theropoda, indet. (bone “taxon” 1), NMC 50794, manal phalanx in latéral (?) aspect. Scale bar: 5 cm.

— 382 —

indistinguishable from that of NMC 50794. NMC 50805 is probably from the second digit, and

measures 111 mm in interarticular length, 29 and 42 mm, respectively, in the height and width

of the distal articular surface, and 43 mm in the height of the proximal articular surface.

BONE “taxon" J (MANAL UNGUALS)

NMC 41820 (Fig. 24a) generally resembles a manal ungual from the Tinrhert, Algeria,

described and figured by DE Lapparent (1960: 29, pl. 6, fig. II). An extensor tuberosity above

the articular facet is reminiscent of, but broader than in oviraptorosaurs; the flexor tuberosity

is less prominent (cf. CURRIE & RUSSEI.L. 1988, fig. 4). It measures (as presers'ed) 91 mm along

the dorsal curve, 81 mm from the articular facet to apex of the claw, and its articular facet is

28 mm high and 21 mm wide. A cast (NMC 41977, fig. 24b) of a larger, but otherwise similar

claw in the personal collections of Brian EBERHARDE measures 232, 189, 64 and 37 mm in the

same dimensions. This claw is of a suitable size to be supported by the large manal phalanges

described above (bone “taxon" I). When aiticulated with NMC 50794, a digit length of 435 mm

is suggested. This is approximately 90% of tlie length of the first digit in the giant hand of

Deinocheirus, from the Upper Cretaceous of Mongolia (OSMOLSKA & Roniewicz 1970, fig. 2).

The linear différences between the Moroccan unguals suggest a weight différence in excess of

an order of magnitude between the two animais which produced them.

Fig. 24. — Theropoda, indei., manal unguals in

lateraK?) aspect; (a) NMC 41820 (bone “taxon”

J); (b) NMC 41977 (cast, bone “taxon” J); scale

bar: 5 cm; (c) restoralion aftcr NMC 50839A-E.

50842A (bone “taxon” K).

— 383 —

BONE “taxon” K (MANAL UNGUALS)

NMC 50839a-e, 50842a (the form of the unguals is restored in Fig. 24c to the scale of the

largest fragment) are characterized by a proximally squared but distally trenchant ventral edge

of the claw, and by a roundcd dorsal margin which is invadcd by what appear to be vascular

channels .sprcading from the curvcd longitudinal grooves présent on both surfaces of the claw.

A relatively prominenl llexor tuberosity is présent.

The relative extcnl of latéral surface exposed bclow the longitudinal grooves varies between

specimens; whcther this is duc to inicrdigital or taxonomie effects is unknown. The unguals are

small and incomplète, but are estimated to have measured 20 to 50 mm in straight line from

the proximal articulation to the distal point.

BONE “TAXON” L(MANUS UNGUAL)

NMC 50842B lacks the proximal articulation. It resembles the second manus ungual in

Deinonyciius (OSTRüM 1969, fig. 63), but is less recurved. The fragment measures 30 mm from

the broken proximal surface to the distal point.

BONE “taxon” M (FEMORA)

The proximal end of a right fémur (NMC 41869. Fig. 25a-c) differs from the large but

siender fémur referred by Stromer ( 1934: 36, pl. 3, fig. 5) to Bahitriasaunis in the cxceplionally

heavy development of both the fourth trochanter and the scar for the insertion of pu-

bischiofemoralis musculature. Asimilarly powerfully constructed lesser trochanter lies well below

the head of the fémur, uniike in the lutter fémur and that of Carchamdnntosaunis (STROMER

1931, pl. l. fig. 14).

Although both fragments appear to be from femora of similar size, NMC 41869 cannot

directly be contpared to the distal fragment that Stromer ( 1934: 39) compared to the fémur of

Erectopus sauvagei. Near the base of the fourth trochanter, the circumfcrence of the fémur is

approximately 310 mm, suggesling that the weight of the animal from which it was derived

probably djd not exceed onc melric tonne (c/, Antîer.süN et al. 1985).

A small. but ncarly eomplcle right femur (NMC 50382, Fig. 25d) closely resembles NMC

41869 in the morphology of its proximal end and is generally congruent with that of the distal

fémoral fragment referred to Etectopus by Stromer (1934. pl. 3, figs 9a-b).

If thèse bonus hclong to the same taxon, the larger fémur was probably very short and

massive, implying a hind limb which was similarly .short and massive. The élément thus appear.s

to be buill for power rather than speed (<f. robust femora from the Lameta Group of India; VON

Hl'ENE & MATI.EY 1933: 54). and it is tempting to speculate that the hind limb was very short

relative to the si/e of the animal’s body.

The cxtcrnal surface of the smallcr bone is rather smooth and slightly porous. which, together

with the poorly ossified distal end, suggcsls immaturily. The bone is 118 mm long, and is about

13% of the linear dimensions of the Egyptian specimen. A mid-shaft circumfcrence of approxi-

matcly 40 mm suggests a body weight (3.78 kg) which is smaller than that indicated for NMC

41869 by a factor of about 270.

— 385 —

BONE “taxon” N (METATARSAL IV)

A right metatarsal IV (NMC 41770, Fig. 26) is approximately the same length (430 mm)

as an element so identified by Stromer (1934: 56), but its shaft curves markedly distolaterally.

It resembles the metatarsal IV in AUosaurus (Madsen 1976, pis 54-55), although the Moroccan

bone is more robust. Areas of insertion for flexor musculature are particularly prominent, and

the bone is compressed in the mid-shaft région (anteroposterior diameter 67 mm, transverse

diameter 82 mm). In its robustness, the metatarsal resembles the fémur described above. However,

according to the ratios of the mid-shaft diameters of the fémur and metatarsal IV in AUosaurus

{cf. Gilmore 1920: 69, 75), the fémur would hâve been derived from an animal linearly about

half the size of the animal from which the metatarsal was derived.

Fig. 26. — Theropoda, indet. (bone “taxon” N);

NMC 41770, right metatarsal IV in (a) exlen-

sor and (b) flexor aspect. Scale bar: 5 cm.

a " " b

— 386 —

BONE “taxon” O (METATARSAL V)

A left metatarsal V (NMC 50830, Fig. 27) measuring 187 mm in length is much more mas-

sively constructecl and more recurved than in Allosaurus {cf. Madsen 1976, pl. 53). It is inter-

esting that vertebrae of Sigilmassasaunis appear to be relatively common and morphologically

very peculiar, and the robust limb éléments (bone “taxa” M-O) are also relatively common and

morphologically peculiar. It is conceivable, but not demonstrable, that ail of these could be

derived from the same taxon.

Fig. 27. — Theropoda, indet. (bone “taxon” O);

NMC 50830, left metatarsal V in médial

aspect. Scale bar: 5 cm.

Bone “taxon” P (Pedal ungual)

A cast of a pedal ungual fragment (NMC 50987, Fig. 28a), presented through the courtesy

of Mr Kirk Leavesley of New Prague, Minnesota, generally resembles a pedal ungual from In

Abangaril, Niger, describcd and figured by DE Lapparent (1960: 29. pl. 6, fig. 10). The fragment

is characterized by a tlat flexor surface which is inclined at an angle of about 20” to its transverse

axis. It was derived from a claw about the same size as the specimen from Niger. A second,

somcwhat smaller ungual (NMC 50826, Fig. 28b) is morphologically indistinguishable from the

preceding specimens.

Fig. 28. — Theropoda. indet. (bone “taxon” P), pedal unguals in

latéral (?) aspect; (a) NMC 50987 (cast) and (b) NMC 50826.

Scalc bar: 5 cm.

— 387 —

Order SAUROPODA Marsh, 1878

Family Diplodocidae Marsh, 1884

Subfamily Dicraeosaurinae Janensch, 1929

Rehbachisauriis garasbae Lavocat, 1954

Referred SPECIMENS. — Isolated teeth NMC 41808, 41810, 41812; questionably referred teeth: NMC 41809,

41811; cervical spine NMC 41872 (Fig. 29); anterior dorsal vertebra NMC 50844 (Fig. 30).

Description and discussion

Rebbachisaurus garasbae is provisionally referred to the Diplodocidae, allhough new South

American materials warrant a review of its affinities (Bonaparte, pers. comm. 1994). General

diplodocoid relationships are suggested by the reduced suprazygapophyseal laminae and presence

of a vertical ridge on the latéral surface of the dorsal spine in the type specimen (pers. obs.;

see Russell & Zheng 1993: 2094, character 16, state 2).

Fig. 29. — Rebbachisaurus garasbae, NMC 41872, cervical spine in (a) lacerai (?), (b) médial (?), (c) distal and (d) proximal

aspect. Scale bar: 5 cm.

— 388 —

Three pencil-shaped teeth are accordingly referred to R. ganishae. Two of them (NMC

41810, 41812) are slighlly curved, relatively large and resemble each other in colour and pré¬

servation. Both are approximately 9 by 13 mm wide at the base of the enamel crown, and 35 mm

high from the enamel base to the apex of a chisel-shapcd, lingually inclined wear facet. The

enamel surface is slighlly crinkled ventrally and becomes smooth toward the apex; there are no

indications of carinae or vertical striations. Another tooth (NMC 41808) is similar but badly

abraded. Similar teeth aiso occur in the Albian of Tunisia (BouAzrz et al. 1988).

A peculiar bone fragment (NMC 41872. Fig. 29) is tlatlened at one end (where it measures

107 mm in width and 17 mm in thickness) and tapers to the opposite end 270 mm distant (where

it is nearly circular in cross-section and measures 34 mm in diameter). The fragment closely

resembles the longitudinally expanded base of a single ramus of a deepiy divided cervical spine

in Aniargasaurus cazaui (Salgado & BONAPARTE 1991; cast of the type specimen in CMN

collections). In view of resemblances between dorsal vertebrae of Amargasaunni. Dicraeosaurus

and Rehbachi.sauru.i (see below) Ihc elemenl is lentativcly referred to R. garavhae.

The contour of the pleurocoels and absence of zygosphene-zygantrum articulations, as well

as the dimensions of the centrum cicarly suggesl affinilies between an isolated dorsal vertebra

(NMC 50844, Fig. 30, Table 5A) and a posterior dorsal in the type of Rehhachisaiiru.<t garasbae

(LavüCat 1954B; Table 5B, pers. obs.). The greater lenglh of the centrum relative to the height

and width of the intercentral articulations and less steepiy projecting iransverse processes impiy

that the vertebra occupied a more anterior position in the column. Only the basal portion of a

delicalely conslrucled neural spine is preserved, but because the bone is thicker laterally, the

spinc may hâve been divided dorsally into two transver.seiy broadened alae. Among dorsal verte-

brac of diplodocoid sauropods, il would appear thaï those of Dicraensauru.': (JanfnsCH 1929b),

Amargasaurus (SALGADO & BONAPARTE 1991) and Rebbachisaurus resemble each other in the

shape of the posterior dorsal .spines and the dorsolaleral inclination of the transverse processes.

Dorsals of Rebbachitiaimis, however, are easily .separated from those of the other two généra

in the presence of a large plcurocoel, a smoolhly continuons articular surface linking the left

and right zygapophyses and absence of zygo.sphcne-zygantral articulations.

Table 5A. — Rebbachimurus ^arasbaes mcasuremenls of anterior dorsal vertebra (NMC 50844),

Length of centrum 190

Width of anterior zygapophyses 160

Width of centrum anteriorly 171

Width of centrum posteriorly 212

Height of centrum posteriorly 227

Height of centrum anteriorly to base of neural canal 200

Height of centrum posteriorly to base of neural canal 219

Base of centrum anteriorly to anterior zygapophyses on mid-line 450

Diameter of neural canal 45

Right pleurocoel, length 78

height 79

— 390 —

Tarlie 5B. — Rehhachisaurus garasbae, measurements of posterior dorsal vertebra in type specimen (Muséum national d’Histoire

naturelle, Paris).

Length of centrum 195

Width of centrum posteriorly 230

Height of centrum posteriorly 236

Maximum transverse width of spine, as preserved 240

Base of centrum posteriorly to base of posterior zygapophyses on mid-line 456

Diameter of neural canal 34

Base of posterior zygapophyses on mid-line to top of neural spine, as preserved 887

Height of vertebra, as preserved 1 343

cf. Rebbachisaurus sp.

Referred SPECIMEN. — Juvenile anterior dor.sal ceninim NMC 50809 (Fig. 31).

A vertebra! centrum (NMC 50809, Table 5C) f'rom the anterior région of the dorsal column

is smallcr ihan lhat of most previousiy described sauropod specimens (cf. Carpenter &

McIntosh 1994, Table 17.1), but is about 1.3 times longer than a diminutive and relatively

more siender sauropod centrum from the Albian of Sudan (Werner 1994, pl. 5, fig. 2). Both

of these liny African centra differ from the above-described Rehhachisaurus dorsals in that the

pleurocoels arc Icnticular and extend across most of the dorsolateral surface of the centrum. As

pointed out by MclNTOSit (1990: 394), a dorsal from the Laie Cretaceous of northem Patagonia

described by NOPCSA (1902) resembles those of Rehhachisaurus. However, the spine is shorter

and, as in the juvénile centra, the pleurocoel is lenticular and relatively elongated.

In the Patagonian vertebra, the lower position of the anterior zygapophyses relative to that

of the posterior zygapophyses suggests that it is from a position morphologically anterior to that

of the Moroccan vertebra here referred to R. ftarasbae (compare NoPCSA 1902, fig. 2, and

Fig. 31. — cf. Rehhachisaurus sp., NMC 50809, dorsal centrum in left latéral aspect. Scale bar: 5 cm.

— 391 —

fig. 30). Yet its spine is not divided, as was apparently the case in the Moroccan vertebra. The

Patagonian vertebra is not large; the dimensions of its cenlrum apparently exceed those of the

juvénile centrum by factors of two to three. It would thus appear that both a large and a small

rebbachisauroid sauropod were présent in the Tafilalt assemblage.

Table 5C. — cf. Rehbachisourus sp., measurements of juvénile anterior dorsal centrum (NMC 50809).

Length of centrum 28

Length of centrum including anterior cotylus 39

Width of centrum anteriorly 44

Width of centrum posteriorly 48

Height of centrum anteriorly to base of neural canal 36

Height of centrum posteriorly to base of neural canal 38

Right pleurocoel, length 25

height 9

Family Titanosauridae Lydekker, 1885

Subfamily Andesaurinae Calvo and Bonaparte, 1991

gen. et sp. indet.

Referred SPECIMENS. — Isolated teeth NMC 41799, 41801; caudal centra NMC 41773 (Fig. 32), 50793

(Fig. 33); astragalus NMC 41868 (Fig. 34).

Fig. 32. — Andesaurinae, gen. indet., NMC 41773, médian caudal centrum in (a) dorsal, (b) left latéral and (c) posterior aspect.

Scale bar: 5 cm.

— 392 —

Description and discussion

Two small, eroded teeth resemble those referred to Malawisaurus by Jacobs et al. (1993).

They measure about 5 by 8 mm in diameter at the base and approximately 30 mm in height,

as preserved. Smooth carinae are surely présent on NMC 41799, and the abrasion pattern suggests

they were probably also présent on NMC 41801. In the latter tooth, the maximum expansion

of the Crown is doser to the tip than to the base.

NMC 41773 is a gently amphicoclous médian caudal centrum in which the sutures for the

neural arches are located over the anterior two thirds of the element, and the proportions of the

centrum (lenglh 113 mm, height 91 mm. posterior width 80 mm) are similar to those of a médian

caudal referred by JACOBS et al. (1993: 527, fig. 2D) to Malawisaurus dixeyi. These authors

consider the non-procoelou.s nature of médian caudals in Malawisaurus as primitive for tita-

nosaurids, citing conditions in Andesaurus (CALVO & BONAPARTE 1991) from the Albian-

Cenomanian of Argentina. NMC 50793 (Fig. 33) is an eroded distal caudal vertebra in which

the neural arch is also anteriorly placed and the intercentral articulations were probably amphi-

platyan. The length of the centrum is 134 mm.

Fig. 33. — Andesaurinae, gen. indet., NMC 50793, distal caudal vertebra in right latéral aspect. Scale bar: 5 cm.

NMC 41868 (Fig. 34) is a right astragalus which is similar to that of Neuquensaurus

australis (VON HUENE 1929: 43, pl. 17, fig. 1, for taxonomy see POWELL 1992) from the Late

Cretaceous of Argentina, but is less derived in that the element is more oblong in shape and

the dorsal crest linking the articular surfaces is more narrow posteriorly. The bone is 178 mm

wide and measures 127 mm anteroposteriorly and 106 mm vertically.

— 393 —

Fig. 34. — Andesaurinae gen. indel., NMC 41868, right astragalus in (a) anterior, (b) internai, (c) dorsal, (d) external and (e)

poslerior aspect. Scale bar: 5 cm.

SAUROPODA indet.

Referred SPECIMENS. — Isolated teeth NMC 41809, 41811.

Description and discussion

Two sauropod teeth (NMC 4180, 41811) resemble each other in préservation, and differ

both preservationally and morphologically front those referred to Rebbachisauras. One (NMC

41810) is small (9 by 11 mm at the base, 41 mm long); the other (NMC 41809) is large (9 by

14 mm at the base, 48 mm long). The long axis of the teeth is nearly straight, and their surfaces

bear indistinct longitudinal grooves and ridges. The apex is acuminate and lacks an oblique wear

facet, which is présent on the teeth referred to Rebbachisaurus. The teeth differ front those

referred to the Titanosauridae in their larger size and in the absence of carinae.

— 394 —

PALEOENVIRONMENTAL AND PALEOBIOGEOGRAPHICAL CONSIDERATIONS

Within the Tafilalt, the “Grès infracénomaniens” blanketed an erosional surface developed

on metamorphosed Paleozoic strata with as much as 100 m of relief locally (JOl.Y 1962: 157-164).

To the East or North-East was an advancing marine gulf (Ferrandini et ol. 1985). Ail three

units of the “Trilogie mésocrélacée” diminish in thickness loward the West, the “Marnes versi-

colores à gyp.se” progressively becoming more .sandy, and the “Calcaire cénomano-turonien” less

fossiliferous (CnouBURT 1952: 146, pl. 12; Lavocat 1954a; Ferrandini et al. 1985).

Drainage through the Tafilalt was evidently loward the cast, and ihc semi-ariiculated (non-

transported) condition of the type skeleîon of Rehacchisauras suggests that these animais

frequented the surrounding fluvial plain. However, mosl boucs are isolaicd and abraded (Lavocat

1949; Joi-Y 1962: 60). indicating substantial transport. The présence of fresh-water coclacanth

remains derived front fish 3.5 m or more in length (Wt.N2 1980; see Schwimmer et al. 1994

for an Upper Cretaceous marine coclacanth of similar dimensions) implies the existence of sub¬

stantial fresh water bodies upstream, cither within the plain or possibly even beyond it to the

West. The State of préservation of large theropod bones is similar to those of giant crocodiles.

Perhaps carnivorous animais were attracted to the margins of siream courses (skeletal remains

of tyrannosaurs are also very abundant in fluviatile environments of déposition in the Nemegt

Eormation in Mongolia; RUSSELL 1970b; Osmolska 1980), or perhaps they were linked to the

food Chain within large bodies of water. The lutter pos.sibility tends to be supported by spéculation

on the piscivorous habits of baryonychids (Taquet 1984; Buffetaut 1989a). which may hâve

been related to Spinasaurus (ButtTTAUT 1989a. 19S9b). Rare lungfish teeth (Lavocat 1948)

imply only seasonal water abundance in more distal communities.

With the onset of déposition of the “Marnes versicolores à gypse" sédiment transport greatly

slowed, and insufficient fresh water entered the basin to prevent the formation of evaporites.

Lungfish teeth hâve been recovered near the base of the unit (CttoUBF.RT 1952: 145). The Coastal

plain fell below sea level during déposition of the “Calcaire cénomano-turonien,'' which has

yielded well-preserved skeletal material of marine dyrosaurian crocodiles aorth of Erfoud (Eber-

harde, pers. comm. 1994). As noted above, the ammonite Neolobites vibrayeaaits occurs at the

base of the "Calcaire cénomano-turonien" (Choubert 1952; Ferrandini et al. 1985),

A transgres.sion culminating in marine strata with the Vtbrayeanus Zone at their base gradu-

ally brought terrestrial déposition (“Continental Intercalaire") to an end over broad areas of North

Africa (BUSSün & Corné 1989, 1991). In most cases, dinosaurian materials in underlying

Cretaceous strata arc limited to scattered bones the biostratigraphic implications of w'hich are

uncertain (DE Laprarent 1960; Lei'Ranc & Guiraud 1990; 56, Table 8). At only three other

localities hâve associated dinosaurian remains of Cretaceous âge been found in a horizon which

occurs below strata containing the Vibrayeanus Zone.

In one of these, front the Tiourarén Eormation within the lullemmeden Basin in Niger, the

few specimens w hich hâve been collected show no spécial affmity to dinosaurian materials from

the Tafilalt, perhaps due to the likelihood that they were derived from a much older level (MOODY

& SUTCLIFIT 1990, 1991; Sereno et al. 1994). Some 300 km to the east in the same basin,

spectacularly abundant fossil vertébrale remains occur near the middie of the 680 m thick “Série

de Tégama” al Gadoufaoua (Taquet 1976. 1994). This sequence is overlain successively by

— 395 —

deltaic and marine clastics (the latter containing Neolohites vihrayeaniis Meister et al. 1991;

Moody & Sutcliffe 1991). Fish remains in the underlying “Argile de l'irhazcr" arc suggestive

of a late Early Cretaceous âge (MooDY & SUTCLIFFE 1991), and the Gadoufaoua dinosaur as¬

semblage has generally been considered to bc of Aptian âge (Taquet 1976).

A similar transition front fluviatile to ncritic and marine sedimentalion occiirs in the Bahariya

Oasis, in the Western Deserl of Egypt. There. Neoinbiie.s cl. vibrayeatius oecur.s within estuarine

strata immediately overlying dinosaur-bcaring sédiments, and the ammonite speeics oecurs in marine

carbonates which may have been deposited contemporaneousiy in the Southern part of the basm.

The Bahariya assemblage is accordingly considered to tse of Cenomanian (At i.AM 1986) or early

Late Cenomanian âge (DoMiNiK 1985, Werner 1990). Tlius, three assemblages of middle Cretaceous

dinosaurs from the Sahara (Gadoufaoua. Tatilall, Bahariya) were probably situated in general pro-

ximity to each other in tinte. However, fish rcmatns from tliese localities have been assigned to

different spccies. often represent forms of dtffering sizes, and generally support successively younger

âges for the three as.semblages (Martin 1981, 1982; TaquE'I'. pers. conim. 1996).

These ciaumstance.s suggest that the apparently derived Spiiiosoiirus mirvecanus l'rom the Ta-

filalt, and the apparently plesiomorphous S. aegypiiacus front the Bahariya Oasi.s may nttt belong

to a single phylogenetic clade. Similarly, the continuous morphologie sequcnce represented by the

Sigilmassasauru.s sp., "Spinosaurus B" and Sigilma.ssaurus brevicolli.'; could mask an underlying phy¬

logenetic complexily. Because the âge of the three dinosaurian assemblages predates the déposition

of marine strata containing the Vibrayeanus Zone, any vicariance events could nol have been the

resuit of an isolation of terresirial faunas in the we.stem Sahara from their counterparts on the re-

mainder of the continent by a trans-Saharan seaway. Furthcmtorc, the Cenontanian-Turonian scaway

was very shallow (20-30 m, Rryment 1980; 317) and a complété tran.s-Saharan isthnius linking the

North African and lullentnteden basins is postulated to have been of short duration (only a fcw tens

to hundreds of thousands of years, r/i Rryment Hl ütNCii.E 1987; Meister et al. 1991: 94). Effects

on terrestrial vertebrate distributions would have been minimal.

Reptilian abundances in the three assemblages can bc ranked us follows (an ornithischian

occurrence in the Tafilalt by LaVOCAT 1954b. 1955b. is based on an ilium of probable theropod

affinities. Taquet, pers. comm. 1996; for Gadoufaoua, sec Taquet 1976; for Bahariya, see

Stromer 1936; SCHAAL 1984; isolated fragments of turtle armour are rclatively abundant at

Bahariya; Schaal 1984: 31):

Tafilalt (Morocco)

Crocodilia

Theropoda

Sauropoda

Chelonia

(Omithopoda absent)

Gadoufaoua (Niger)

Crocodilia

Omithopoda

Chelonia

Sauropoda

Theropoda

Bahariya (Egypt)

Theropoda

Crocodilia

Sauropoda

Chelonia

(Omithopoda absent)

The assemblage from the Tafilalt closely resembles that of the Bahariya. The similarity is

reinforced by the abundance of giant fresh water fishes in both assemblages (e.g. Hybixbis,

Lepklotes, Ceraiodu.i, Muwsonia\ see Table I; STROMER 1936; 75-76). However. Onchopri.stis

is relatively more abundant in the Tafilalt, as is Ceratodus in the Bahariya (SCHAAt. 1984).

— 396

Another, more continental environment was apparently sampled at Gadoufaoua, which has pro-

duccd lhe only known major Gondwana assemblage where ornithopods outnumber ail other

groLips of dinosaurs (cf. VON HUBNE 1929; VON HUENE & Matley 1933; TAQUET 1976). No

teeth of Oiiclioprisiis or Carcharodontomnrus were recovercd at Gadoufaoua, although they occur

in relative abundancc in the other iwo assemblages. Their absence bas been suggesled to be due

to a grcater âge for the “Série de Tégama” (Taquet 1976), but ecological effects cannot be

excluded (the type teeth of C saharkus were collected from strata inferred by LEERANC 1983

to be of lutc Barremian âge). The dinosaurian assemblage from the underlying “Série de Irhazer”

(Tiourarén Formation) is dominaicd by skeletons of camarasauroid sauropods. although the fish

assemblage is similar to that in higher levels of the Saharan middie Cretaceous (Moody &

Sutcliffe 1990, 1991; Sereno ei ai 1994).

A mammalian microfaunal assemblage, recently discovered in older Early Cretaceous strata

(see “Stratigraphy and Corrélation,’’ above), shows Laurasian as well as Southern Hemispheric

zoogeographic affinilies (SlGOGNEAU-RUSSELl et al. 1990, SlGOGNEAU-RUSSELL 1991). Inter¬

change between Europe and Africa during Early Cretaceous time is also suggested by the di¬

nosaurian record on both continents (RUSSEI.I. 1993. 1995; SERENO et al. 1994). By the time

of déposition of the “Grès rouges infracéiionianien.s’’ in the Tafilalt, terrestrial vertebrates in

West Africa remained linked by a continuons land surface to those of South America (Reyment

& Dingle 1987). This is confirmed by the identification of similar fresh waier fish, lurtle and

crocodilian taxa in what hâve since become separale continents {t’.,ç. BUFlTTAUT & Taquet

1977; WEN7. 1980; DE Broin 1988; Maisey 1991; Bufi etaut & Rage 1993), as well as by a

record of a Rebhachi.'iauriis-Wkç. dinosaur in Argentina (MciNTOSll 1990: 394). The identification

of abclisaurid and titanosaur remains in the Albian of Southern Morocco is also consistent with

Albian-Cenomanian records of lhe two groups in Argentina (BONAPARTE et al. 1990; Calvo &

Bonaparte 1991). In contrast, the absence of diagnostic remains of abelisaurids, diplodocoids

and litanosaurs in Aplian-Albian assemblages in North America (O.STROM 1970) suggests that

faunal relationships between Africa-South America (“Gondwana") and North America were more

remote, in keeping with plate tcctonic evidence. The posl-Ccnomanian record of dinosaurs in

Africa is niuch loo incomplète to serve as a basis for hiogcographic spéculation (RfSSEi.1. 1995).

CONCLUSIONS

The “Grès rouges infracénomaniens” of the Tafilalt hâve yielded abondant, if usually isolated

and often water-worn remains of dinosaurs, possibly of Albian âge. Discrète forms so far iden-

tified include:

Theropoda. small, sp.

Spino.murii!) maroccanus n. sp.

Sigilniassasaiirus brevicollis n. g. et n. sp.

Catvharodontosaitnts .sahariens cf. Majungasaurus sp.

Abelisauridae, sp. indet.

Rebbachisaunis garasbae cf. Rebbachisauriis sp.

Titanosauridae. sp. indet.

— 397 —

To these may possibly be added a stegosaur-like fémur (EberhaRDE, pers. comm. 1994).

In the case of Spinosaurus and Sigilinassasaurus, différences of a species-levei magnitude sep-

arate them from their counterparts in the Bahariya Formation of Egypt. Vertebrae assignable to

Sigilmas.sasaurus dominate dinosaurian vertebrae from the Tafilalt. and include a range of sizes

consistent wiih body weighl différences of a factor of approximately 150. Undetermined femora

indicate the prcsence in the assemblage of infantile individuals of large Iheropods weighing less

than 4 kg. Rcmains of at least ihrec taxa of sauropods hâve been ideiitiRed, but ihose of or-

nithischian dinosaurs, if présent, are not abundant. Variation in the préservation and morphology

of SigilmasMsaunis centra pose the question how much of middie Cretaceous time is represented

by the déposition of lhe "Grès rouges infracénomaniens."

The niimber of different variciies of dinosaurs, represented by isolated éléments found within

a relatively small région, document one of lhe most diverse dinosaurian assemblages known from

Africa. Faunal resemblances are closest to the Bahariya assemblage from deltaic-lagoonal strata of

Cenomanian âge in Egypt, as was originally noted by Lavcx’at ( 19.54a: 102). The abundance of

theropod dinosaurs in both assemblages may be tlic resuit of their linkage to fresh water ccosy.stems

containing gianl fishes. This coastal communily evidently differed greally from an ornithopod-dom-

inated assemblage at Gadoufaoua, which was separated by over lOtK) km from lhe nearest contem-

porary coa.stline Uf- Revment ik Dingi.e I9S7. figs 3, 4; Moody & Surci.lEt'E 1991, llgs I, 2).

The dinosaurian assemblage of the Tafilall exhibits clo.ser zoogeographic affinities to those from the

Early Cretaceous of South America than to those in North America.

Acknowlcdgements

Mr William PiNCH, of Rochester, New York, recognized lhe importance of fossil vertebrate remains

from the Tafilall displayed ai the Tucsoit Gem and Minerai Show Iteiween Id92 and 1994, It is through

his enthu.siasni and initiative thaï some of these fossils were acquired by lhe CMN. Mr Brian Ebrhiiarde,

of Moussu Direct. Cambridge. Bngland. generou.sly supplied information on the occuirence and condition

of preservalion of speeimen.s he consigned to the Muséum. Spceimens were aiso acquired from Mr Raymond

Meyer, of Buffalo. New York, and Mr Horsl Borkaru, of Mineralicn und Fossilicn, Bonn, Germany.

Colleagues from the Mu.scum national d'Histoire naturelle, Paris, generoiisly provided guidance to the

stratigraphy of the Sahara, as well as to Saharan dinosaur collections under their carc: spécial thanks are

due to France RE BROIN, Denise SlGtxtNKAU-RusSHl.l. and Philippe TAQt'ET. Georges BL.SSON graciousiy

reviewed the stratigraphie portions of the mantiscripi in detail, and provided suggestions to clarify the

régional rclationships of the Cretaceous record in lhe Taillait. Valiiable counsel of a geological nature was

provided by André CllARRiERE (Université Paul Sabatier, Toulouse), René GL'IRAGD (Institut des Sciences

de la Terre, de l'Eau et de l'Espace de Montpellier) and Michel Monbakon (Université de Fribourg). José

Bonaparte (Museo Argentino de Ciencias Naiuralcs) and Hans-Dieter .Si.KS (Royal Ontario Muséum) aIso

freely contributed of their paleontological expertise. Il is parlicularly gratifying to acknowledge lhe interest

of His Excelleiicy. Dr Tajeddine BaduoU, Anibassador to Canada from the kiiigdom of Morocco. Philip

CliRRlE (Royal Tyrrell Muséum of Paleoniology) and Philippe Taoi/ET carefully reviewed (hc llnal man-

uscripi and made iiumeroiis suggestions which clarified both incaning and content. The support of this

international group of scholars has been deepiy appreciated.

Manuscript suhmitted for publication on 27 Jiine 1995: accepted on 13 February 1996.

— 398 —

REFERENCES

Allam a. M. 1986. — A régional and paleoenvironmental sludy on lhe Upper Cretaceous deposits of the Bahariya

Oa.si.s, l.ibyan De.sert, Egypt. Journal of African Earth Sciences 5: 407-412,

Anderson J. E.. Hall-M \RTIN .4. iX: ROSSELL D. a. |985. — Long-bone drcumt’erence and weight in manimals,

birds and dinosaurs. Journal of Zoolofiy 207: 5.4-61.

Barsboi.D R. 1976. — New data on lherizinnsanrti.t CI heri/mosaui'iclae, Theropoda). Triuly Sovmestnoi Soviet-

sko-MonnoVskoi Paleonloloaicheskoi tkspeililsii i: 76-92 [in Rus.sian|.

Basse E. & CHOUBERT G, 1959. — Les faunes d'ammonites du “Cdnomano-Turonien" de la partie orientale du

domaine atlasique marocain et rie ses annexes sahariennes. C, R, 20th International Geological Congress,

1959, in L. B. KtiLt.DM (ed.). El Sistema Crelcit ico 2: 59-81,

Benton M. J. (edilor) 1993. - The fossil record 2, Chapman & Hall. London, xvii -i- 845 p.

Bonaparte J. K & Novas K H. 1985. — Abelisaums coinalniensis. n.g.. n.sp., carnosauria del Crelacico tardio

de Palagonia. Ameyhinianu 21: 259-267.

Bonaparte J. i- & POW'ELL J, E. 198(l. — A continental assemblage of telrapods from the Upper Cretaceous

of .\rgcntina (Sauropoda-Coelurosauria-Carnosauria-Avesi. Mémoires de la Sociéle gvologkiue de France

Nouvelle .Série 139: 19-28

Bonaparte: J. F., Novas F. E. & CORtA R. A. 1990. — Camotaurus sasirei Bonaparte, the horned, lightly built

carnosaur from the Middie Cretaceous of Patagonia. Contribution in Science 416:. 1-41.

Bouazi/ -S.. BliEETtAL T E.. UHANMt M.. Jaechr J. J., MARTtN M., Mazin J. M. & LONG H. 1988. — Nouvelles

de'eouverles de vertébrés fossiles dans l’Albien du Sud tunisien. Bulletin de la Société' ^éologic/ue de France

Série 8. 4: 335-339.

BROtN F DE 1988 — Les tortues et le Gondwana. Examen des rapports entre le fractionnement du Gondwana

et la dispersion géographique des tortues pleurodires il partir du Crétacé. Studio Oeohgica Salnianlicensia,

Studio Falaeoi heloniologicii 2: 103-142.

BuFFtiTAOT E- 1976. — F)er Land-Krokodilier Ubycosuciuts Strotner und die Faniilie Libyeosuehidae (Crocodylia,

Mesosuchia) ans der Kreidc Afnkas. Mitteilungen Bayerischen Stoatssainmiung fuer Palciontologie His-

inriscbe Geohigie 16 ! 17-2.8

— 1989a.— New leinains of lhe enigmaiic dinosaur Spinosuunis from the Cretaceous of Morocco and the

affinilies between Spinosaurus and Baryonwx. Nettes Jahrhurcli ftier Géologie und Palciontologie Monatsheffe

1989 2: 79-87.

— 1989b. — New remains of Spinosaurus from lhe Cretaceous of Morocco. Archosatirian Articulations 1 (9):

65-68.

— 1994. — A new' crocodilian from the Cretaceous of Southern Morocco. Comptes Rendus de l'Académie des

Sciences Parts Série 2 319: 1563-1568.

Buffet.al't e. & Rage J.-C. 1993. — Fossil amphibians and reptiles and the Africa-South America connection.

In W. üE'i.iROE & R. L AVOCAT (eds l. The Afrkv-Soulh America coitneclion. Clarendon Press, O.xford: 87-99.

BUFFE-tAEf E & Taquet P. 1977. —The giant crocodilian Sarchostichus in the eaily Cretaceous of Brazil and

Niger. Palaeoiiiology 20: 203-208.

BussoN O. ét Cf.iRNÉF .A. |989. — Données sur les paléodiiuals déduites de la sédimentation continentale du

Mésozoïque saharien. Centre Inlernalional de Formation et d'Eckanges en géologie Publication ixtcasion-

nelle 18, 87 p.

— 1991. — The Sahara from lhe Middle Jurassic to the Middic Cretaceous: data on environments and climates

based on oiilcrops in the Algerian Sahara. Journal of African Sciences 12: 85-105,

Calvo J. O. & Bon.M’ARTE J. F. 1991. — Andesaurits delgadoi gen. et sp. nov. (Saurischia-Sauropoda). dinosaurio

Tilanosauridac de la Formacion Rio Liinay (Albiano-Cenomaniano). .Neuquen. Argenlina. Ameghiniana 28:

303-310.

Cappe'ITA h 1980. Les sélacicas du Crétacé supérieur du Liban II Bato'ide.s. Palaeoniograpbica Abteilung

A 168 : 149-229.

Carpe:nte:r K. & McIntosh J. 1994. — Upper Jurassic sauropod babies from the Morrison Formation. In

K. Carpenter, K. F. HIR.SCH & J. R, HORNER (eds). Dinosaur Eggs and Babies. Cambridge University

Press, Cambridge: 265-278.

— 399 —

Chario a. .1. & Milner a. C. 1990. — The systemalic po.sition of Baryonyx walken, in the light of Gauthier's

classification of the Theropoda. In K. Carpenter & P. J. CURRIE (eds). Dinosaur Sysiematics. Cambridge

Univer.sity Press, Cambridge: 127-140.

Charrière a. 1992. — Di.scontinuités entre les “Couches rouges" du Jurassique moyen et du Crétacé inférieur

dans le Moyen Atlas (Maroc). Comptes Rendus de l’Académie des Sciences Paris Série 2 315 : 1389-1396.

Chol’BERT g. 1952. — Histoire géologique du domaine de l'Anti-Atlas. Notes Mémoires du Sert'ice ftéologiijue

(Maroc) iOO : 7.5-194.

CURRIE P. & Russell D. A. 1988. — Osteology and relalionships of Chirostenotes pergrucilis (Saurischia,

Theropoda) from the Judith River (Oldman) Formation of Alberta, Canada. Canadian Journal of Earth

Sciences 25: 103-115.

DepÉret C, & SavornIN .1. 1927. — La faune de reptiles et de poisson.s albiens de Timimoun (Sahara algérien).

Bulletin de la Société géidoglque de France Série 4 27 : 257-265.

DOMINIK W, 1985. — Stratigraphie und Sedimcntologie (Geochemie, Schwermineralanalyse) der Oberkreide von

Bahariya und ihre Korrelation ziim Diikhla-Becken (Western De.sert, Agypten). Berliner geowissenschaften

Abhandhoigen Reihe A, 62 : 1-173.

DUFFIN C. J. & SlGOG.NEAU- Russell D. 1993. — Fossil shark leeth from the Early Cretaceous of Anoual,

Morocco. Professionnal Paper. Belgium Ceological Survey 264; 175-190.

F.ARLOW J. O., Brinkman d. L., ABI.ER W. L. & CURRIE P. .1, 1991. — Si/.e, shape. and serration density of

theropod dinosaur latéral teelh. Modem Geology 16: 161-198.

Ferrandini M.. Philip J., Babinot J.-F„ Ferrandini J. & Tronchetti O. 1985. — La plate-forme carbonatée

du Cénomano-Turonien de la région d’Erfoud-Errachidia (Sud-Est marocain): stratigraphie et paléo

environnements. Bulletin de la Société géologique de France Série 8 1985 1 : 559-564.

GIFFIN E. 1990. — Gross spinal anatomy and limb use in living and fossil reptiles. Paleohintogy 16: 448-458

Galion P. M. & JENSEN J A. 1979. — A new large theropod dinosaur from the Upper Jurassic of Colorado.

Brigham Yoting Univ., Ceological Studies 26 (2): 1-12.

Gilmork c. w. 1920. — Osleology of the carnivorous dtnosauria in the United States National Muséum, with

spécial référencé tn Anirndemns lAllosaurus) and Ceratosaurus. Bulletin of the U.S. National Muséum 110,

xi + 154 p.

Harland w. b., Armstrong R. L„ Cox A. V„ Craig L. E„ Smith A. G. & Smith d. G. 1990. — A géologie

time scale 1<JS9. Cambridge University Pres.s, Cambridge, p. xv + 263.

Hu S. Y. 1964. — Carnosaurian remains from Alashan, Inner Mongolia. Vertehrata PalAsiatica 8: 43-63 [in

Chine.se. Engli.sh abstract],

HUENE F. VON 1929. — Los saurisquios y ornitisquios de Cretâcio Argentino. Anales del Musen La Plata Ser.

2 3: viii -P 196 p.

Huene F. VON & Matley c. a. 1933. — The Cretaceous Saurischia and Ornithischia of the central provinces

of India. Memoir of Ceological Survey India 21: 1-74.

Jacobs L. L., Winkler D. A., Down.S W. R, & GO.MANI E. M. 1993. — New material of an Early Cretaceous

titanosaund sauropod dinosaur from Malawi. Pakieontology 36: 523-534.

Janensch w. 1925, — Die Coelurosauncr und Theropoden aus der Tendaguru-Schichten Deutsch-Ostafricas.

Palaenntogrophica Suppl. 7 Erste Reihe 1 : 1-99.

— 1929a. — Material und Fomiengehalt der Sauropoden in der Ausbeute der Tendaguru-Expedition. Palaeon-

tographica Suppl. 7 Erste Reihe 2: 1-34.

— 1929b. — Die WirbeI.saule der Gattung Dicraeosaiirus. Palaeontographica Suppl. 7 Erste Reihe 2 : 35-133.

Joly F. 1962. — Eludes sur le relief du sud-est marocain. Travaux de l'Institut des Sciences chérifien Série

Géologie, Géographie Physique 10: 578 p.

Lapparent a. K HE I960. — Les dinosauriens du "Continental intercalaire" du Sahara central. Mémoire de la

Société géologique de France 88A : 1-57.

LAVOCAT R. 1948. — Découverte de Crétacé à vertébrés dans le .soubassement de la hammada du Guir (sud

marocain). Comptes Rendus de l'Académie des Sciences Paris Série 2 226 ; 1291-1292.

— 1949. — Les gisements de vertébrés crétacés du sud Marocain. Comptes Rendus Sommaires de la Société

géologique de France 1949 7 ; 125-126.

— 1951. — Decouverte de re.stes d’un grand dinosaurien sauropode dans le Crétacé du sud marocain. Comptes

Rendus de l'Académie des Sciences Paris Série 2 232 : 169-170.

— 400 —

— 1952.— Les gisements de dinosauriens du Crétacé du sud marocain. Comptes Rendus Sommaires de la

Société géologique de France 1952 2 : 12-13

— 1954a. — Reconnaissance géologique dans les hammadas des confins algéro-marocains du sud. Notes

Mémoires du Sendce géologique (Maroc) 116: 147 p.

— 1954b. — Sur les dinosauriens du Continental Intercalaire des Kem-Kem de la Daoura. Comptes Rendus

l^lh Iniernalional Geological Congress 1952 part IS: 65-68.

— 1955a. — Sur une portion de mandibule de thémpode provenant du Crétacé supérieur de Madagascar. Bulletin

du Muséum national d'Histoire naturelle Paris Série 2 27 ; 256-259.

— 1955b. — Découverte d'un crocodilicn du genre Thoracosuurus dans le Crétacé supérieur d’Afrique. Bulletin

du Muséum naiioiud d'tlisioire naturelle Paris Série 2 27 : 338-340.

— 1955c. — Titres et traeaux scientifique de M- René ■Uivocut. Imprimerie Centrale, Romorantin, France, 95 p.

Lefranc J.-P. 1983. — Corrélation vers le nord et description straligraphique détaillée du Continental intercalaire

(Mésozoïque continental) de la sebkha de Tiniimoun, Gourara. Sahara algérien. Comptes Rendus de l'Aca¬

démie des Scienc es Paris .Série 2 296 : 193-196.

Lefranc J.-P. & Gliraud R. 1990, — The Continental Intercalaire of northwestern Sahara and its équivalents

in the neighbouring régions. Journal of African Earllt Sciences 10: 21-11.

Madsf.N J- h. 1976. —AUosuurus fragilisi a revised osteologv. Bulletin of Utah Geological and Minerai Survey

109 XII + 163 p.

MAI.SEV J. Cl. 1991- — Concluding remarks. In J. G. MAISEY (ed.). Santana fossils: an illustrated atlas. Neptune

City. New Jersey: 422-433.

Martin M. 1981. — Über drci Zahnpiarten von Ceratodus aus der agyptischen Kreide. Mitteilungen der Bay-

erlschen Sraat.i.sammlung Paldoniologie und Hhtorische. Géologie 21 : 73-80.

— 1982.— Nouvelles données sur la phylogénie et la sy.stémaiique des dipneustes postpaléozoiques. Géobios

Mémoire 6 : 53-64.

MclNIOSH J, A. 1990. — Sauropoda. In D. B. WEISHAMPEL. P. DODSON & H. OSMOLSKA (eds). The Dinosauria.

University of California Press, Berkeley; 345-401.

Meister C., ALZOUMA K., Lang J. & MATHEY B. 1991. — Les ammonites du Niger et la transgression trans¬

saharienne au cours du Cénomanien-Turonien. Géobios 25 : 55-100.

MONBARON M., Kubler b. & ZWEIDLER D. 1990. — Detrital rubified sédimentation in the High Atlas trough

in the Mo.sozoïc; attempt at latéral corrélations using correspondenee factor analysis. Journal of African

Earth Sciences 10; 369-384.

Moody R. T. J. et Sutcliffe P. J. C. 1990. — Cretaceous-Tertiary crossroads of migration in the Sahel. Geology

Today 6: 19-23.

— 1991.— The Cretaceous deposits of the lullemmeden Basin of Niger, central West Africa. Cretaceous Re¬

search 12: 137-157.

NOPC-SA F. 1902. — Wirbel eines südamerikanisclien Sauropoden. Sitzungsber. Kaiserl. Akad. Wiss., Math.-Na-

turwiss. Klasse 111 : 108-114.

Norman D. & Milner A. C. 1989. — Eyewitness Books: Dinosaur. Stoddart, Toronto, 64 p.

OSBOR.N H. F. 1906. — Tyrannosaurus. Upper Cretaceous carnivorous dinosaur (second communication). Bulletin

of the American Muséum of Natural Hislory 22: 281-296.

— 1917.— Skclctal adaptations of Ornitholestes, Struihiomimus, Tyrannosaurus. Bulletin of the American

Museian of Natural Hislory 35: 733-771.

Osmolska h. 1980, — The laie Cretaceous vertebrate assemblages of the Gobi Deserl, Mongolia. Mémoires de

la Société géologique de France 139 ; 145-150.

OsMOtJSKA H, & RoNlEWicz E. 1970. — Deinocheiridae, a new family of theropod dinosaurs. Palaeontologica

RoUmica 21: 5-19.

OSMOLSkA H., RONIEWICZ E. & BarSBOLD R. 1972. — A new dinosaur, Gallimimus bullatus n. gen., n. sp.

(Omitlioinimidae) front the Upper Cretuceous of Mongolia. Palaeontologica Polonica 27; 103-143.

OSTROM J. H. 1969. — Osteology of Deinonvehus antirrhopus, an unusual theropod from the Lower Cretaceous

of Montana. Bulletin of the Peahody Muséum of Natural History Yale Univ. 30: 165 p.

— 1970,— Slratigraphy and paleontology of the Cloverly Formation (Lower Cretaceous) of ihe Bighom Basin

area, Wyoniing and Montana. Bulletin of the Peabody Miuseum of Natural History Yale University 35: 234 p.

— 401 —

Perle A. 1979, — Segnosauridae. a new faniily of Iheropods from the Late Crelaccous nf Mongolia. Trudy

Sovmextnoi Snvietsko-Moiij^ol'skoi PaleonlologUheskoi Ekspedilsii 8: 45-55 [in Riissianl.

POWELL J. E. 1987, — Morfülogia del e.squelelo axial de los dinosaurios liianosaiiridüs (Saurischia. Sauropoda)

del esiado de Minas Gérais. Brasil, Anales X Cangressa Brasil Palfanlalngia: 155-171,

— 1992. — Osteologia de Sallasaiirns lorkalus (Sauuipoda-Titanosauridae) de Crelacito Superior dcl Noroeste

argeniino. In J. !.. .SaNZ & A. D. BcSCAi.icisd (edsl Las dinosaurios y su eniorno hiotico, Acias del

Segunda Cursn de Paleanlalagia en Cuenea. Instilulo Juan de Valdes. Ser. Act. Acad, 4: 165-230.

Reyment R. A. 1980. — Biogengraphy üf lhe .Saharaii Cretaceoiis and PaleiKene Iransgressions. Cretaceous

Research I: 299-327.

Reyment R. A. & Dinoi E. R. V. I9K7. — Palaeogeography of ,\friea during lhe Cretaceous Period. Palaeogeo-

grapliy. Palaeoclinmlology. Pnlacoecohgy 59; 93-Il6.

Rus.SELL D. a. 1970a. Tyrannosaurs from lhe Late Cretaceous of western Canada. Canadian National Muséum

Puhiicution in Paléontologie 1, 34 p,

— I970B.— The dinosaurs of Central Asia. Canadian Geographer Journal 81 (6); 208-215.

— 1993. — The rôle of Central A.sia in dinosaurian biogeographv. Canadian Journal of Earth Sciences 30:

2002-2012.

— 1995, — China and the lo.st worlds of the dinosaurian era, Hisrorical Biology 10: 3-12.

Rus.SELL D. A. & DONG Z. M. 1993. — A nearly complété skcleton of a new Iroodontid dinosaur from the

Early Cretaceous of Ihc Ordos Basin, inner Mongolia. People's Republie of China. Canadian Journal of

Earth Scietwes 30: 2 163-2173.

Russell D. A. & Ziienc 7,. 1993. — a large iiiameiKhisaurid from the Juuggar Basin, Xmjiang. People’s Republie

of China. Canadian Journal of Earth Sciences 30: 2082-2095.

Salgadü L- & Bi.tN.yPARTb J. K 1991. — Un nuevo Dictaeosauiidae, Amurgasaurus cazaui gen. et sp. nov.,

de la Forrrutcidn La Amargu. Neocomiano de la Prtvvincia del Neuquén. Argentina, Ameghiniana 28. 333-346.

SCHAAL S. 1984. — Oberkrela/i.sche Osleichlhyes (Knochenlïsche) aus dein Bereich von Bahariya and Kharga,

Âgypten. und ihere Aussagen /ur Paldkologie und Stratigraphie. Berliner geowisseusihaften AhhandUngen

Riche A 53 ; 1-79.

Schwimmer D. R., .Stewart J. D. & Williams G. D. 1994. — Giani fossil cœlacanths of the Late Cretaceous

in the easicm United States. Geology 22: 503-506.

SERENO P. C., WILSON .1. A.. LARSSON il C. e., DUTHEtL D. B. & SUES H. D. 1994. — Early Cretaceous

dinosaurs from lhe Sahara. Science 266: 267-271.

SiGOGNEAI.'-RfS.SELL D. 1991. — Nouveaux mammildrcs thériens du Crétacé inferieur du Maroc. Comptes Rendus

de l’Académie des Sciences Paris Série 2 313; 279-285.

SlGOGNEAU-Ri'SSEt.L D.. .MoSHARO.N M. dt K.AENEL E. DE 1990. — Nouvelles données sur le gi.sement à mam¬

mifères Méso/oîques du llaut-Allas marocain. Geohios 23: 461-483.

Stromer e. 1915. — Ergebmssc der Eorschutigsreisen Prof. E. Stromers in den WUsten Àgypiens U. Wirbelt-

ier-ResIe der Bajarijesliife (unlersles Cenoman). 3. Das Original des Theropoden Spinosaurus aegyptiaeus

nov. gen., nov. spec. Ahhandhingen und Bayerische Akademie der Wissenscluiften Mathematik Naturwis-

sensehaften Kl. 3 : 1-32.

— 1931.— Ergebnissc dei l■orschungsa■iscn Prof. E. Stromers in den Wiisten Agypiens 11. Wirbelticr-Reste

der Bajarijesliife (iintersles Cenoman). 10. Ein Skeletl-Resl von Canhnrodontosiiuriis nov. gen. Abhandlun-

gen and Bayerische Akadetnie der Wissenscliafien Maihernatik-Naliiru issensch({lien RI. 9: 1-23

— 1934.— Ergebni.sse der Forscliungsrcisen Prof. E. Stromers in den Wiisten Àgypiens II. Wirbeltier-Reste

der Bajarijeslufe (untersles Cenoman). 13, Uinosauria, Ahluaullungen and Bayerische Akadetnie der Wis-

senschaften Mathematik Naturwissenschaften Kl. 22 : 1-79.

— 1936. -- Ergebni.s.se der Forschungsreisen Prof E. Stromers in den Wiisten Àgypiens VII. Baharije-Kessel

und - Slufc mil deren Fauna und Flora - Eme ergiSnitendc Zusammenfas.sung. Ahhandhingen und Bayerische

Akademie der Wissenscluiften Mathcinutik-Natiirwissenschaflen Kl. 33: 1-102.

Tabaste N. 1963. — Étude des restes de Poissons du Crétacé saharien. Mémoire de l'Institut Français d'Afrique

noire 68 : 437-485.

Taquet P. 1976. — Géologie et paléontologie du gisement de Gadoufaoua (Aptien du Niger). Cahiers de Paléon¬

tologie C.N.R.S. : 191 p.

— 1984. — Une curieuse spécialisation du crâne de certains dinosaures carnivores du Crétacé: le museau long

et étroit des Spinosauridés. Comptes Rendus de l’Académie des Sciences Paris Série 2 299 : 217-222.

— 1994.— L’Empreinte des dinosaures. Odile JACOB (ed.), Paris. 363 p.

Weishampel D. b. & HORNER J. R. 1990. — Hadrosauridae. In D. B. WEISHAMPEL, P. DODSON & H. OSMOLSKA

(eds). The Dinosauria. University of California Press, Berkeley: 534-561.

Wenz s. 1980. — À propos du genre Mawsonia, coelacanthe géant du Crétacé inférieur d’Afrique et du Brésil.

Mémoires de la Société géologique de France 139 : 187-190.

— 1981. — Un cœlacanthe géant Mawsonia-lavocati Tabste, de P Albien-base du C énom anien du sud marocain.

Annales de Paléontologie, Vertébrés 67 : 1-20.

Werner C. 1990. — Biostratigraphical results of investigations on the Cenomanian elasmobranchian fauna of

Bahariya Oasis, Egypt. Berliner geowissenschaften Abhandlungen Reihe A 120 : 943-956.

— 1994. — Die kontinentale Wirbeltierfauna aus der unteren Oberkreide des Sudan (Wadi Milk Formation).

Berliner geowissenschaften Abhandlungen B Krebs Festschrift: 221-246.

Zhao X. J. & Currie P. J. 1993. — A large crested theropod from the Jurassic of Xinjiang, People’s Republic

of China. Canadian Journal of Earth Sciences 30: 2027-2036.

Bulletin du Muséum national d'Histoire naturelle, Paris, 4® sér., 18, 1996

Section C, 2-3 : 403-415

Conservation de sucres dans les phases organiques

d’os de bovidés fossiles

par Hélène DAVID, Yannicke DAUPHIN, Martin PiCKFORD & Brigitte Senut

Résumé. — Les observations au MEB et les analyses à la microsoncle électronique ont montré que la

microstructure et la composition chimique élémentaire d'un os de bovidé fossile (Plio-pléistocène d'Angola -

plateau d’Humpata) étaient bien conservées. Cependant, la spectroscopie infrarouge (DRIFT) montre une dimi¬

nution des bandes organiques, une augmentation du tau.x de cristallinité et du rapport CO 1 /PO 4 dans l'os fossile

par rapport à l'os actuel. Les spectres obtenus sur les pha.ses organiques extraites confirment la présence de

protéines et de sucres chez le fossile, tout en mettant en évidence leur altération par les processus diagénétiques.

Mots-clés. — Angola, Bovidae, spectroscopie infrarouge, sucres.

Sugar préservation in the organic inatter in fossil bovid bones

Abstract. — SEM observations and électron mieroprobe analyses bave shown thaï tbe microstructure and

Chemical composition of a Plio-pletstocene bovid bone (Humpata Plateau, Angola) arc well preserved. Hovvcver,

infrared spectroscopy spectra exhibii clear différences between Recent and fossil samples. Organic hands weaken

relatively to minerai ones in the fossil, whereas crisialllnily rate and COv/POa ratio increa.se. DRIIT spectra

carried on the extracted organic matrices demonstrate the presence of proteic and sugar contents in the fossil

bone, but the composition of the fossil organic matnces are modified by diagenetic processes.

Key-words. — Angola, Bovidae. infrared spectroscopy. sugar.

H. David, Y. Dauphin, Laboratoire de paléontologie. URA 72J, bât. 504, Université Paris Xl-Orsay.

M. PiCKFORD, chaire de palèuanthropologie et ptèhistoire du Collège de France et Laboratoire de paléontologie du Muséum

national d’Histoire naturelle, 8 nie de Buffon. F-75231 Paris cedex 05.

B. Senut, Laboratoire de paléontologie du Muséum national d'Histoire naturelle, 8 rue de Buffon, F-75231 Paris cedex 05.

INTRODUCTION

L’os est un complexe organo-minéral dont la structure et la composition sont relativement

bien connus. Toutefois, les divers composants de ce tissu ont été inégalement analy.sés et la

majeure partie des données disponibles sur la phase organique concerne les protéines et surtout

le collagène. Collagéniques ou non, les protéines de l’os sont associées à d’autres constituants :

les sucres. Par exemple, le galactose et le glucose permettent d’établir des liens entre les -

molécules de tropocollagène. Les protéoglycannes sont formées par un cœur protéique sur

lequel se fixent des chaînes de glycosaminoglycannes, chaînes principalement composées de

— 404 —

chondroïtine sulfate. Les sialoprotéines et les phosphoprotéines sont des glycoprotéines acides

impliquées notamment dans la fixation du Ca^L l’initiation et la régulation de la calcification.

Le comportement des sucres de l’os au cours de la fossilisation et des phases diagénétiques

postérieures e.st encore mal connu. En outre, la composition des phases organiques fossiles est

rarement corrélée avec l’état de conservation de la phase minérale et celui des autres composants

de la phase organique. Actuellement, aucun critère (sédimentologie, âge...) ne permet une sélec¬

tion rationnelle d'un site favorable à la conservation de la phase organique. Le choix effectué

repose donc sur des indices tels que la bonne conservation de la microstructure osseuse et de

la composition chimique globale. Une étude préliminaire a montré que certains fragments osseux

de bovidés provenant de remplissages plio-pléistocènes du plateau d'Humpata (Angola) répon¬

daient à CCS conditions (David 1994); de plus, la recherche des taxons actuels de comparaison

dont la position systématique était proche ne constituait pas de difficultés. Un protocole per¬

mettant de détecter rapidement la présence et la diagenèse de sucres dans les phases organiques

de l’os fo.ssilc a été mis au point.

TRAVAUX ANTÉRIEURS

L’os frais contient environ 70% de matériel inorganique, 18%' de matériel organique, et

12% d’eau. La pha.se organique est composée d’environ 85% de collagène de type 1 insoluble,

et de 5% d'autres collagènes, des divergences subsistant sur la présence de certains types. À

cela s’ajoutent environ 10% de protéines non collagéniques.

Spectrométrie infrarouge

L’os total

Les données sur les apatites de synthèse sont très abondantes, celles sur les os sont plus

ponctuelles. L’os des bovidés aurait le même comportement que la dahllite (Herman &

Dallemagne 1964). Les spectres DRIFT (Diffuse Réflectance using Infrared Fourier Transform)

de l’os humain présentent des bandc.s d’intensité variable vers 560-600 cm ', 1000 cm ',

1500 cm ' et 3.500 cm'' (SUZUKI 1975). Une bande carbonate a été également détectée vers

850 cm"'. Les bandes des groupes phosphates et de la phase organique (amides I et 11) sont

présentes. Le taux de cristallinité de l’os en cours de croissance serait voisin de 23%. Lorsque

le degré de calcification augmente, les bandes organiques deviennent plus faibles, alors que la

bande à 600 cm ' (PO4) et les bandes carbonates deviennent plus intenses. Les sites de substi¬

tution de CO 1 dans l'apalile de Los ont été localisés par spectrométrie infrarouge (Eu Feki et

al. 1991 ; Rey er a/. 1989, |99!a, 199lb). REV er u/. ( 1990. 1991c) ont également mis en évidence

des changements structuraux dans le domaine v3 PO4 et v4 PO4 lors de la maturation d’apatites

biologiques et minérales actuelles.

Outre les données fournies par les spectres sur la présence d’un composé, diverses méthodes

basées sur les rapports d’intensité des bandes permettent d’accéder à d’autres types d’informa¬

tions.

— 405 —

Termine & Posner (1966) ont ainsi mis au point une méthode de calcul du taux de cris-

tallinité (SF) à partir de la bande v4 PO4 située vers 600 cm ' et établi une corrélation entre le

taux de cristallinité et le rapport Ca/P de l’apatite. Le pourcentage en matière organique peut

également être estimé grâce à divers rapports basés sur les intensités des bandes minérales et

organiques (Dauphin 1993a, b). Les bandes minérales sont situées vers 1 030 cm ' et 1 430 cm '.

Celles associées à la phase organique sont vers 3300 cm ' ou 1 654 cm '. Quatre rapports peuvent

être calculés d’après les intensités de ces bandes. Si les valeurs chiffrées dépendent naturellement

des bandes choisies, sur une série d'échantillons donnés, les courbes obtenues sont parallèles

(Dauphin 1993a, b). L'absence de certaines bandes sur les fossiles ne constitue donc pas un

inconvénient majeur pour l’estimation de la quantité de phase organique selon cette méthode.

La désintégration de l’os de restes humains historiques a été reconnue par Newesely (1989),

la présence de collagène (vers 3 300 cm'') n’apparaissant pas clairement dans ces spécimens.

La phase organique

Les données bibliographiques sur les bandes du collagène, détectées par diverses méthodes,

dans une gamme d'ondes allant de 400 à 2000 cm ' ont été résumées par FUREDI & WALTON

(1968). Généralement, la bande amide A du collagène est légèrement plus élevée que celle des

autres protéines; 3330 cm '. Toutefois, le degré d’humidité fait varier sa fréquence. Il en est

de même pour les bandes amide B, amide I, Il et V (Susi et al. 1971), Fraser & MacRae

(1973) ont signalé en outre une faible bande vers 2940 cm''.

Weiner & Bar-Yosef (1990) ont constaté la conservation des protéines non collagéniques

sur divers os fossiles provenant de seize sites préhistoriques. De plus, les spectres infrarouges

montrent la présence d’argiles et parfois d’une phase organique semblable aux acides humiques.

Caractérisation des sucres

Les données concernant la caractérisation et les quantités des poly- et monosaccharides

associés aux protéines osseuses actuelles sont à la fois nombreuses et limitées. En effet, elles

sont établies sur du matériel fractionné, et aucune donnée globale sur la composition en sucres

élémentaires n’est fournie dans ces travaux spécialisés.

Les fractionnements chromatographiques des résidus de digestion de la phase organique par

des protéases ont montré que les rnucopolysaccharides acides contenaient de la chondroïtine

sulfate, mais au.ssi du galactose, du mannose, du fucose et du xylose (Herring 1972). Les gly¬

coprotéines contiennent une grande variété de sucres, parmi lesquels la glucosamine, la galac¬

tosamine, le galactose, le mannose, le glucose, le fucose et des acides sialiques ont été identifiés.

Deux types principaux d’hétéropolysaccharides seraient associés au collagène. En outre, la mo¬

lécule de collagène elle-même est une glycoprotéine contenant des hexoses (Woodhead-Gal-

LOWAY 1980).

En ce qui concerne l’os fossile, des colorations spécifiques de rnucopolysaccharides ont été

positives sur les coupes d'un dinosaure : Tarbosaurus du Crétacé supérieur du Gobi (Pawlicki

1977), mais la composition élémentaire n’est pas précisée.

— 406 —

MATÉRIEL ET MÉTHODES D’ÉTUDE

Matériel

Le fragment d'astragale de bovidé fossile a été récolté lors d'une campagne de l’Angola

Palaeontology Expédition (PlCKFORD et al. 1992. 1994) dans des remplissages de fissures (plateau

d’Humpata. sud de l'Angola). Il provient de la carrière de Cangalongue 111, sur la rive occidentale

de la vallée de Cangalongue, Le sédiment est composé de brèches grossières, contenant de nom¬

breux fragments de stalagmites recimentées par des travertins. La présence de Serengetilagus,

de Ciigantohyra.K, et surtout de Meiridiochoerus andrewsi permet de dater ces gisements du Plio-

pléistocène.

Le matériel de comparai.son est constitué par un tibia de bœuf actuel provenant d’un élevage ;

il s'agit donc d’un animal assez jeune. Cet os ayant été conservé au congélateur pendant plusieurs

années, il est relativement sec. Une référence supplémentaire a été utilisée ; un os d’un cheval

contenant encore la matrice organique non liée à l’os, incluant les cellules osseuses.

Méthodes d’étude

Contrôle de la conservation de la structure osseuse

L’étude de l’état de la structure os.seu.se a été effectuée au microscope électronique à ba¬

layage. Des cassures brutes ou traitées avec des solutions enzymatiques ont été observées au

microscope électronique à balayage (Philips SEM 505) après une métallisation à l’or palladium.

E.xtraction des phases orgunujues

Après une déconiamination à Thypochlorite de soude pendant 24 heures, les fragments os¬

seux actuels et fossiles ont été rincés à l’eau distillée et déionisée (qualité Milli-Q). puis séchés.

Ils ont ensuité été broyés afin d’obtenir une poudre de granulométrie régulière. Une partie de

ces poudres a été déx:alcifiée à l'acide acétique à pH constant de 4. Les pha.ses solubles et

insolubles ont été séparées par centrifugation. La phase soluble a été dessalée par ultraliltration

sur une membrane Filtron de 3kDa. puis concentrée et lyophilisée. La phase in.soluble est dessalée

par centrifugations successives dans de l’eau (qualité Milli-Q) avant d’être lyophilisée.

La préparation de l’os fossile à l’acide acétique s'est révélée être inefficace (c/. « Résultats »,

« Phase organique insoluble»), Les poudres ont donc été décalcifiées à HCI, à un pH constant

de 2. Les operations suivantes (centrifugations, rinçages, etc.) ont été identiques à celles menées

sur les spécimens actuels.

Spectroniétrie infrarouge

Les poudres et les lyophilisais sont mélangés à du KBr de même granulométrie. Le mélange

contenant environ 5% d’échantillon est séché dans une étuve (35 °C) pendant une nuit.

Le spectrophotomètre infrarouge à transformée de Fourîer (FTIR) Perkin Elmer modèle 1600

est équipé de fenêtres en KBr et d’un accessoire à réflexion diffuse : DRIFT ou DRIFTS (Diffuse

Réflectance using Infrared Fourier Transform) qui permet de travailler sur des poudres. Les mo¬

difications de la phase organique provoquées par les élévations de température et les fortes

— 407

pressions nécessaires à la confection de pastilles pour les acquisitions classiques en mode trans¬

mission sont ainsi évitées.

Le mode d’utilisation préconisé par PERKiN Elmer permet d’obtenir des spectres dont la

qualité atteint celle des spectres établis par transmission. Le temps d’analyse supérieur à 4 mn

correspond à 64 balayages, dans une gamme d’onde de 500 à 4000 cm ', avec une résolution

nominale de 4 cm '. Le système élimine automatiquement les bandes dues à la présence de CO 2

et H 2 O. Afin de rendre les spectres satisfaisant à ta loi de Beer-Lambert. il est necessaire de

procéder à la conversion de Kubelka-Munk, proposée en routine par le système.

RÉSULTATS

Microstructures et composition chimique élémentaire (Fig. 1)

La première étape dans le contrôle de l’état de conservation repose sur l’analyse micro¬

structurale et la composition chimique élémentaire des spécimens.

L’os actuel montre la structure classique de l’os lamellaire, du système haversien et des

zones à structure de type «lamellaire croisé» ou en «contreplaqué» (Fig. la-c). La microstructure

de l’astragale fos.sile de Cangalongue est assez bien conservée : la disposition en lamelles (Fig. Id,

e) et des zones de résorption différentielle sont identifiables. Les cavités osseuses naturelles sont

généralement dépourvues de sédiment, et aucune trace pouvant être attribuée à des micro-orga¬

nismes n’a été observée (Fig. If). Les risques de contamination par du matériel allogène semblent

réduits. La composition chimique élémentaire (microsonde électronique EDS) montre également

des modifications modérées par rapport à Tos actuel : augmentation du Ca, P et Fe, légère di¬

minution du Mg (David 1994).

Os TOTAL (Tableau 1, Fig. 2)

Les spectres infrarouges des poudres d’os total actuel montrent les bandes minérales et

organiques signalées dans la littérature. Les bandes minérales attribuées à PO4 et à CO4 sont

nettement visibles (Tableau I). Certaines bandes organiques sont présentes, mais elles ne sont

pas spécifiques des protéines ou des sucres ; amide A. amide I (deux bandes), amide 11

(trois bandes), et amide lll. Sur les poudres d'os total, certaines bandes se chevauchent. Par

e.Kemple, la présence éventuelle de sucres (1050-1 150 cm ') est ma.squéc par la bande minérale

v3 de PO4 qui esl toujours beaucoup plus intense.

Les spectres réalisés sur le fossile ont le même profil général que l’actuel (Fig. 2). Néan¬

moins, la phase minérale y semble plus abondante et plus cristalline; ceci est particulièrement

net pour les bandes COj. La matrice organique est également présente : les amides A. I et 11

sont conservés, mais avec des pics plus faibles que dans les os actuels; la bande de protéines

située à 1 474 cm ' disparaît. D’après les critères de Stutman et ai. (1965), aucun enrichissement

en F n’est décelable dans l’os fossile.

Une déconvolution du domaine v3 PO4 fait apparaître différentes bandes chez le bœuf actuel.

D’après Rey et al. (1991a), les bandes vers 1 125, 1 110 et 1020 cm ' caractérisent plutôt les

apatites nouvellement formées; la bande 1 145 cm ', quant à elle, indiquerait la présence de

— 408 —

Fig. I. — Comparaison des microslruclures de l'os de bteuf actuel (a-c) el du bovidé fos.silc du Plio-plêislocène d'Angola (d-f);

a, cassure oblique monirant une siruclurc lamellaire. La phase organique a élé partiellemenl éliminée par une .solution de

trypsine, dans un tampon H.iO + Hepes à pli 7,l.‘i, pendant S heures à 37 "C. x .302: b, détail de la surface de l'endéoste.

Tryp.sine, dan.s un tampon H:0 + Hepes à pH 7,15, pendant 4 heures à 37 "C. X 2400 ; e, cassure oblique montrant les

changemenl.s d'orientation des fibres dans l'os lamellaire, Même préparation que b. x 2 112 : d. cassure montrant l'organisation

lamellaire, avec de nombreuses cavités. Comparer avec a. Préparation non traitée, x 27.3 ; e, cassure oblique montant la

conservation de la disposition lamellaire de l’os, et les microporcs sur la surface. Préparation non traitée, x I i)60: f. surface

inleme d'un vai.sseau sanguin, lai disparition partielle de la phase organique pendant la fossilisation rend distincts les mi¬

croporcs. Préparation non traitée, x 1 2h0.

— 409 —

Tableau 1. — Bandes infrarouges présentes dans l’os frais (d’après FuRCDl & Walton 1968), dans le collagène, et les os

actuels et fossiles de bovidés. OS : os non décalcifié. MOI ; matrice organique insoluble. MOS : matrice organique soluble.

ACTUEL

FOSSILE

ATTRIBUTION

LITTERA-

COLLAGENE

MOI

MOS

MOI

TURE

TYPE 1

amide A

3330

3290

3300

3290

3306

3347

3304

amide B

3060-3085

3070

3076

COOH

1743-1734

1742

1684

■■

amide 1

■SB

1655

1654

mmm

1648

1647

■gi

1636

1559

1550

1560

1558

amide II

1547

1540

1541

1540

1532

1515

1507

1505

protéines

1480

1470

CH2, CHS déformations

1447-1464

1443

1457

C03

USD

1446

1457

1410

1424

COO

1400-1410

1402-1411

1400

1418

CH2 déformation

1378-1388

1374

protéines

1344

1343

1350

1339

orotéines

1276

1281

amide III /S04

1238

1241

1242

1240

1233

1225

orotéines

1204

P04

1250

910

1047

1043

1128

sucres

1120

1083

1061

1082

1028

1033

1035

P04

960

963

C03

878

873

874

871

P04

600

604

602

560

567

575

groupements HPOJ". Chez le fossile, les bandes 1 145, I 125 et 1 110 cm * n’apparaissent plus;

or d'après Rey ei al. (1991a), ces bandes sont les plus sensibles aux changements structuraux.

Toutefois la bande I 100 cm ' également susceptible de disparaître aisément, est conservée. Selon

ces auteurs, les bandes I 125 et I 110 cm ' diminuent jusqu’à disparaître en cas d’augmentation

de la cristallinité de l’apatite, ou de dissolution.

— 410 —

FlG. 2. — Spectres infrarouges des os de bovidés actuel et fossile non décalcifiés.

L’examen des spectres DRIFT sur poudre totale indique donc l'existence de modifications

modérées dans la structure et la composition du spécimen fossile.

Les taux de cristallinitc calculés d’après la méthode de Termine & POSNER (1966) sur l’os

de bœuf actuel varient, bien que les prélèvements n’aient pas été localisés dans des zones par¬

ticulières. Toutefois, la moyenne est égale à 0,10, ce qui est similaire à la moyenne obtenue sur

les os de rongeurs actuels (Dauphin 1993a, b). Le taux de cristallinité moyen du bovidé fossile

est supérieur à 0.14.

Le rapport A/COj (phase organique : amide A. bande vers 3 300 cm '/phase minérale, bande

vers 1450 cm ') dans l’os actuel ; 0,82 est similaire à la moyenne établie sur divers rongeurs.

Par contre, le rapport A/PO 4 (amide A vers 3 300 cm''/phase minérale vers I 030 cm ') est plus

faible que chez les rongeurs, puisqu’ il ne dépasse pas 0.33. Les rapports basés sur l’intensité

de la bande de l'amide I (vers 1654 cm"') sont, comme chez les rongeurs actuels, plus élevés

que ceux basés sur l’amide A. Ces rapports phase organique/phase minérale chez le bovidé fossile

indiquent tous une diminution relative importante de la quantité de matière organique, sans que

l’on puisse préciser s'il s'agit de la phase soluble ou insoluble (Fig, 3).

L’analyse DRIFT confirme les résultats des observations microstructurales et de la compo¬

sition chimique élémentaire : les modifications induites par la diagenèse augmentent l’importance

relative de la phase minérale par rapport à la phase organique. Toutefois, si la phase organique

est présente, elle semble modifiée.

Phase organique soluble (Figs 4. 5; Tableau 1)

La phase soluble de l’os actuel montre de nombreuses bandes ; amide A vers 3305 cm ',

amide B vers 3076 cm ', amide 1 vers 1655 cm ', amide II vers 1550 cm ', amide III vers

1240 cm '. Sont également présentes les bandes à 1457 cm ', 1400 cm ' (COO‘ ), 1339 cm ',

1083 cm ' (sucres), 1033 cm ', fréquement signalées dans la littérature pour les protéines. Les

différences observées dans les deux spectres du bœuf et du cheval actuels, notamment la plus

forte intensité des bandes de la phase organique dans le cheval, sont probablement dues à la

différence de fraîcheur du matériel. Rappelons en effet que le cheval contenait la presque totalité

— 411 —

Fig. 3. — Estimation de la quantité de matière

organique contenue dans les os actuel et fos¬

sile d'après les intensités des bandes or¬

ganiques (A : amide A, lA : ainide I) et

minérales (CO3 et PO4) des spectres in-

farouges.

1,2

0,8

0,6

0.4

0.2

A/C03 A/P04 IA/C03 1A/P04

S actuel E] fossile

de la phase organique non liée à l'os. Les faibles quantités de matrice soluble extraites du bœuf

fossile n’ont pas pennis l’analyse par spectrométrie infrarouge.

Phase organique insoluble (Figs 4, 5 ; Tableau 1 )

La phase insoluble, majoritairement collagénique, est abondante dans les os de bœuf et de

cheval actuels (Tableau 1, Fig. 4). Les différences entre le cheval et le bœuf sont moins impor¬

tantes dans lu phase in.soluble que dans la phase .soluble. La bande de Tamide A apparaît à

3290 cm'*. La bande amide B, située à 3071 cm * sur le collagène de type 1. n’est pas identifiée

sur ces spectres. Deux bandes représentent Tamide 1, tandis que Tamide II en montre trois. La

bande à 1457 cm ' duc aux déformations de CHj et CHj est généralement reconnue dans le

collagène. L’amide III montre une bande vers 1225 cm'', déjà visible dans les spectres d'os

total. Dans la zone habituellement considérée comme indiquant la présence de sucres, trois bandes

sont clairement apparentes: I 128 cm ', 1061 cm 1055 cm '. Toutefois, l'attribution exacte

de la bande à 1 128 cm ' reste incertaine, car elle est parfois interprétée comme caractéristique

des liaisons SO4 (Suzuki 1975). L’une des molécules contenant de telles liaisons, abondante

dans les tissus os.seux, est la chondroitine sulfate, mais celle-ci devrait apparaître dans la phase

soluble de Tos. Ces bandes n’ont pas été mises en évidence dans le collagène de type I, seule

une bande vers 1 084 cm'* étant susceptible de représenter les sucres.

Le spectre de la «phase organique insoluble» obtenue par décalcification à Tacidc acétique

du bœuf fossile est pratiquement identique à celui des poudres totales. Les bandes de la phase

minérale sont toutes présentes et très intenses. Par contre, les spectres de la matrice insoluble

fossile résultant de la seconde décalcification (HCl) offrent la même allure générale que ceux

— 412 —

Fig. 4. — Spectres infrarouges des phases organiques solubles et insolubles des os actuel et fossiles.

des spécimens actuels (Fig. 4); ils se caractérisent néanmoins par une bande amide I (vers

1 651 cm ') et une bande amide III (vers I 280 cm ') plus fortes. Une très légère fraction minérale

persiste cependant dans les domaines v4 PO 4 et donc peut-être au niveau de la zone d’apparition

des sucres (v4 PO 4 ).

— 413

m i/A ED i/ii

Fig. 5. — Estimation de la composition des phases organiques des os actuels et fossiles d’après les intensités des bandes organiques.

OS : os non décalcifié ; MOI : matrice organique insoluble ; MOS : matrice organique soluble ; I : amide I ; A : amide A ; Il ;

amide II.

CONCLUSION

D’un point de vue méthodologique, l’utilisation de l’analyse DRIFT permet une estimation

rapide et assez complète de l'état de conservation d'un os fossile. Effectuée sur une simple

poudre, sans autre préparation qu’une décontamination préalable, cette méthode permet d’appré¬

cier les modifications de composition aussi bien de la phase minérale que de la phase organique.

Elle nécessite très peu de matériel, et celui-ci peut être réutilisé par élimination du KBr, soluble

dans l’eau. Il est donc possible, sur des os de grande taille, d’effectuer des prélèvements ponctuels

localisés afin d’établir une cartographie de la diagenèse et/ou un choix des zones à étudier de

façon plus détaillée.

Du point de vue des résultats, les spectres DRIFT ont tout d'abord permis de mettre en

évidence que la méthode de décalcification utilisée pour l’os actuel ne convenait pas aux spé¬

cimens fossiles, malgré les indices de bonne consen^ation fournis par l'analyse microstructurale

et la composition chimique élémentaire : la phase «organique insoluble» du fossile est

pratiquement identique à la poudre non décalcifiée. Dès cette phase de l’analyse, des altérations

sont ainsi mises en évidence. Des résultats similaires ont été obtenus sur d’autres os fossiles

provenant de sites d’âge et de sédimentologie très variés.

Dans le bovidé fossile, la présence de protéines indiquée par les bandes amides détectées

sur les spectres a été confirmée par la fluorescence UV et les analyses en acides aminés (David

1994). Par ailleurs, les modifications de la composition de la phase organique décelée par les

— 414 —

rapports amide I/amide A cl amide I/amidc II dans les spectres DRIFT des phases organiques

isolées (Figs 4, 5) peuvent être mises en rapport avec les altérations des teneurs en acides aminés

par rapport aux phases organiques du bœuf actuel. En ce qui concerne plus particulièrement les

sucres, leur présence est généralement masquée par les larges bandes intenses du PO.^ entre 910

et 1 250 cm ’. 11 est donc nécessaire, pour délecter leur présence, de décalcifier les poudres. Là

encore, les spectres DRIFT ont mis en évidence la prc.sence de sucres chez le fossile, tout en

indiquant des modifications par rapport au bœuf actuel.

En ce qui concerne le fossile, les phases organiques sont nettement modifiées par rapport

à l’actuel. Ceci est en accord avec les résultats obtenus sur la composition en acides qminés

des mêmes spécimens, montrant que les modifications des phases solubles et insolubles n’étaient

pas identiques (David 1994).

Remerciements

Les auteurs tiennent à remercier les nombreuses personnes qui ont permis la réali.salion de la campagne

de terrain en Angola en 1990. En France : Prof. Y. Coppens et Prof. P. Taquet; en Angola : Dr S. Aço

(co-directeiir de l’expédition) et J. Ferreira (directeur du Museu Régional da Huila), ainsi que A. M.

OlIVKIRA, M. BATALHA, V. COKl-Hü, E. FRKIRE, P. BONDO, R. Gomes, R. DA SOUZA, J. Edolardo, A.

Sakatengo, s. Tchihonga, J. Katciiatcha et T M. Gomes, et Son Excellence l'ambassadeur de France

à Luanda. M. S. Filliol et M. E. Roland, conseiller à la Coopération, ainsi que toute son équipe; Son

Excellence l’ambassadeur de la République populaire d’Angola à Paris et son équipe; M. J. Blackshear,

directeur de CONOCO ANGOLA et son équipe.

Manuscrit soumis pour publication le 21 mars 1995 ; accepté le du 29 novembre 1995.

RÉFÉRENCES

Dauphin y. 1993a. — Spectrométrie infrarouge (DRIFT) des os de rongeurs fossiles de Tighenif (Pléistocène,

Algérie). Palàonlologische Zeitschrift 67 (3-4) : 37’7-39-‘i.

— !993b. — Potenlial of the Diffuse Réflectance Infrared Fourier Transform (DRIFT) method in paleonlological

sludies of bones. Applied Spectroscopy 47 (I); 32-5.5.

David H. 1994. — Etat diagénétique des os de quelques mammifères fossiles du plaleau d’Humpata (Plio-pléis-

tocène, Angola). Mémoire de D E.A-, université de Montpellier 2, 59 p. inédit.

El Feki H„ Ri;y C. & Vionoles M 1991, — Carbonate ions in apatites : infrared investigations in the va CO 3

domain. Calcifled Tissue International 49; 269-274.

Fraser R. D. B. & MacR.ae T. P. 1973, — Conformation in fibrous proteins and related synthetic polypeptides.

In B. HORECKER, N. O. KAPLAN, .1. MARMUR & H. A. SCHERAGA (eds). Molecular Biology, Academie

Prc.ss, London, 628 p.

Furedi h. & Walton a. Ci. 1968. — Transmission and attenuated total reflection (ATR) infrared .spectra of

bone and eollagcn. Applied Spectroscopy 22: 23-26.

Herman H. & Dallemagne M. J. 1964. — Les carbonaio-hydroxylapatites et le carbonate des os et des dents

étudiés piu la spectropholométrie infrarouge. Bulletin de la Société de chimie biologique 46 (2-3) : 371-383.

Herring G. M 1972. — The organic matrix of bone. In C. H. BOURNE (ed). The Biochemistry and Physiology

of Bone, Academie Press, New York 1: 127-189.

NewesELY h. 1989. Fossil bone apatite. Applied Geochemistry 4; 233-245.

— 415

Pawlicki R. 1977. — Histochemical reactions for mucopolysaccharides in the Dinosaur bone. Studies on epon-

and methacrylate-embedded semithin sections as well as on isolated ostéocytes and ground sections of

bones. Acta histochemica 58: 75-78.

PICKFORD M., MEIN P. & Senut B. 1992. — Primate bearing Plio-Pleistocene cave deposits of Humpata, Southern

Angola. Hiiinan Evolution 7 (1): 17-33.

— 1994.— Fossiliferous Neogene karst fillings in Angola, Botswana and Namibia. Southern African Journal

of Sciences 90: 227-230.

Rey C., Collins B., GOEHL T., Dickson I. R. & Glimcher M. J. 1989. — The carbonate environment in

bone minerai : A resolution-enhanced Fourier Transform infrared spectroscopy study. Calcified Tissue In¬

ternational 45; 157-164,

Rey C., Shimizu M., Collins b. & Glimcher m. J. 1990. — Resolution-enhanced Fourier Transform infrared

spectroscopy siudy of the environment of phosphate ions in the early deposits of a solid phase of calcium-

phosphate in bone and enamel, and their évolution with âge. I. Investigations in the va PO 4 domain. Calcified

Tissue International 46: 384-394.

Rey c., Renugopalakrishnan V., Shimizu M.. Collins B. & Glimcher M. J. 1991 a. — A re.solution-enhanced

Fourier Transform infrared spectroscopic study of the environment of the CO.^ ion in the minerai phase

of enamel during its formation and maturation. Calcified Tissue International 49: 259-268.

— 1991b. — Fourier Transform infrared spectroscopic study of the carbonate ions in bone minerai during aging.

Calcified Tissue International 49; 251-258.

Rey c., Shimizu M., Collins B. & Glimcher M. J. I991c. — Resolution-enhanced Fourier Transform infrared

spectroscopic study of the environment of the phosphate ion in the early deposits of a solid phase of

calcium phosphate in bone and enamel and their évolution with âge : 2. Investigations in the V 3 PO 4 domain.

Calcified Tissue /nternational 49: 383-388.

Stutman J. M.. Termine J. D. & Posner a. S. 1965. — Vibrational spectra and siructure of the phosphate

ion in some calcium phosphates. Transaction of the Nesv York Academy of Science, ser. II, 27 (6); 669-675.

SUSI H., Ard J. S. & CARROLL R. J. 1971. — The infrared spectrum and water binding of collagcn as a function

of relative humidity. Biopolymers 10: 1597-1604.

Suzuki m, 1975. — Studies on the physicochemical nature of hard tis.sues. Infrared, N.M.R., X-ray diffraction

investigation of hydroxyl-radical, crystalline water and carbonate substitution in biological apatites. In Phy¬

sico-chimie et cristallographie des apatites d'intérêt biologique. Colloques Intem. C.N.R.S., n“ 230, Paris :

77-83.

Termine J. D, & Posner A. S. 1966, — Infrared analysis of rat bone : âge dependency of amorphous and

crystalline minerai fractions. Science 153: 1523-1525.

WEINER s. & Bar-YosEF O. 1990. — States of préservation of bones from prehistoric sites in the Near East :

a survey. Journal of Arcliaeological Science 17: 187-196.

WOODHEAD-GaLLOWAY J. 1980. — Collagen: the anatomy of a protein. Institute of Biology’s Studies in Biology

117: 60 p.

Bulletin du Muséum national d'Histoire naturelle, Paris, 4*^ sér., 18, 1996

Section C, 2-3 : 417-450

Rooting ungulates within placental mammals :

Late Cretaceous/Paleocene fossil record and upper molar

morphological trends

by Lcandro O. SALLES

Abstract. — Basal ungulate interrelationships (Cretaceous/Paleocene condylarths) are seen here in a broader

placental mammal phylogenetic context and hased on a cladistic scrutiny of the upper molar morphology. After

running the ie* opiion ol the Hennig86 .software and applying the Successive weighting algorithth lor a saniple

of 19 taxa and 23 characiers, a fully resolved branching diagrain came out as resuli with 18 componcnis. Utidcr

parentheticul notation die relatioiiships obiained are the following ; H'appotherium (Prokeniwh'sit's (llnholesws

(KenmilexU'.s {Cimoli-sle.f (Prowiomus (Miads (Proviverru {Zulumhdah'-ilex {Lcplidix (Scenopcinus UPoruwys and

Enxigale) (Purgalorius and Pulaechthon)) (O.rypriimis (Dyxni)etodon (Prolungulrtlum and MirrratiiM)))))))»))))»-

Although it is questionahle whether some very basal condylarths are indeed exclusive stem gmups of ungulates,

the results of thi.s study support Ihc monophyly of ungulates. Unguluta is' viewed as a ihember of a large clade,

named Magnorder Herbotheria, and comprising Anagalida densii Sticky & McKenna 1993 and here represented

by rodents and anagalids). Archonta (represented by plesiadapiformes), and Ungulata (exclusively represented by

four basal condylarths).

Key-words, — Cretaceous/Paleocene, placentals. basal ungulates. upper molar morphology and parsimony

analysis.

■ /enracinement de.s ongulé.s dans les mammifères placentaires : les données fossiles

du Crétacé supérieur et du Paléocéne et les tendances morphologiques des molaires supérieures

Résumé. — Une analyse cladi.stique Fondée sur la morphologie des molaires supérieures est proposée. Il

s'agit des relations basales des ongulés (condylanhres du Crétacé .Supérieur/Paléocène) vues dans le contexte

phylogénétique plus large des placentaires. Après .avoir utilise l'option iC* du logiciel HennigSb et appliqué la

pondération successive pour un échantillon de 19 ta,xons el 23 caractères, un arbre complètement ré.solu avec

18 composants émerge comme résultat. Sous annotation parenthéiiques les relations phylogénéiique.s obtenues

sont les suivantes: {PappnihfHum (Pmkennah’stes (Bcbolexiex [Ketiiialestes (Cimakwtex iPrormomax iMUicis

(Proviverro (Zaliimhdalcslex [Li'ptktix iSceiwpagux iiPcirutnyx and Eosigale) (Pitrgatorius and Piilaedithon))

(O.xyprimus (Dysnuytodm (Protungiiluium and Mimttiulo))))))))))))))}. Bien qu’il soit discutable que certains

condylarthies ba.saia apparaissent comme groupes souches propres aux ongulés, les résultats de cette étude ap¬

puient la monophylie des ongulés. Le ta.xon Llngulata est compris comme membre d'un grand clade, ici nommé

Mégaordre Herbotheria qui comprend les Anagalida i.seri.m Stucky & McKenna 1993 et représenté par les rongeurs

et anagalidés), les Archonta (représentés pai les plesiadapiformes) et les Ungulata (représentés par quatre condy-

larthres basaux). Une série d'autres analy.ses de parcimonie sont également présentée.s, mais aucun des résultats

obtenus n'est en contradiction avec celui leprésenté dans l'annotation pareni'nétique

Mots-clés. — Crétaeé/Paleocène. placentaires, ongulés et nuirphologie des molaires supérieures, analy.se

de parcimonie.

L. O. Salles, Muséum national d'Histoire naturelle. Laboratoire de paléontologie, S rue de Buffon, F-75231 Paris cedex 05.

— 418 —

SCOPE

The monophyly of ungulates is one of the ongoing debates regarding high interrelationships

of placental mammals. Available evidence supporting the monophyly of ungulates was charac-

terized by Novacek ei al. (1988) as “embarrassingly poor”, and since thcn not much has been

achieved. Nevertheless, after the First attcmpt to examine the mammalian phylogenetic relation-

ships linder a cladislic framework. (McKenna 1975), approximately lhe same large ungulate

groups hâve been luraped together in a monophyletic group. The latest two broad phylogenetic

analysis on ungulate relationships are proposed by Prothero et al. (1988) and Archibald

(1994).

In this essay, basal ungulate interrelationships are approached taking into account late

Cretaceous/Paleocene condylarths as well as a set of basal placental mammals as a mean of

Fig. 1. — Branching diagram derived from the evolutionary schemes of Van Valen (1978) for ungulale interrelationships. Five

major groups are recognized and ail of them are directly rooted at the basal node numbered 0 together with the Late Cretaceous

genus Protungulatum. This is the major overall analysis of ungulates centered on condylarlhs preceding that of Prothero

et al. (1988).

placing ungulates in a broader phylogenetic conlext. The identity of condylarlhs has for décades

been a controversial subject, and it is nowadays widely cited as an example of a mammalian

“waste basket”. Condylarlhs as conceived by VAN Valen (1978) comprise an assemblage of

highly diverse taxa distributed within the boundaries of the Ungulata (see Fig. 1 for a cladistic

— 420 —

synthesis of the evolutionary schemes proposed by Van Valen). (n this synthesis five major

clades are assigned and ail of them are rooted at the basal node together with the genus Pro-

tunguhitum. It is to be noted that Pantodonta is placed wilhin the large clade (5) where Oxyprimus

is assumed to be the basal taxon. Here, we indeed support the traditional view of condylarths

as a paraphyletic eomplex placed as stem group to a number of high rank ungulate taxa. This

point of view agréés with Prothero et al. (1988) and Archibald (1994). A detailed discussion

of condylarth interrelationships is the suhject of a monograph in préparation.

This paper is focused on the upper molar morphology of the condylarths considered to be

basal with respect lo the entire clade Ungulata. The justification for selecting the dental mor¬

phology is simpiy because a great majority of the late Cretaceous/Paleocene fossil records are

fundamentally known from fragments of maxilla and lower jaw. Thereforc, the idca here is to

put forward a preliminary phylogenelic hypothesis of these basal condylarths in a broader placen-

tal mammal context, avoiding problems of missing data. Moreover, this note is conceived as a

cladistic test to whether molar morphology (internai congruence) can yield somewhat well re-

solved trees with significant consistcncy and rétention indices.

By considering just the upper molars, this phylogenetic analysis is aiso free from assump-

tions of morpho-functional independence bctween the upper an lower teeth. Generalizations re-

garding the upper molar morphology are based on the invariable mophology (the basal

topographical pattern) among the three upper molars and focus on the second upper molar tooth.

The tcrminology applicd follows Brown & Kraus (1979), Crompton & KtELAN-

Jaworowska (1978). HERSHKOVtTZ (1971). and MINCZHEN et al. (1975) (Fig. 2). The notion

of upper and lower levels are inverted in order to facilitate the normal tendency with dental

fragments: the base of the maxilla is read as the lower level and the crown shearing surfaces

as the upper level. In this research for molar homologies within placentals. no morpho-functional

analysis is performed and therefore shearing surfaces are not considered.

Fig. 2. — Placenta! upper molar dental nomenclature partially based on

Bown & Kraus (1979). The abbreviaiions u.sed here are applied

throughout thi.s article, and lhey are the following: cenc = centrocrista;

pasi = parastylc; mest « mcsosiyle; paco = paraconc: meco = meiacone;

pacl = paraconuie: meci = metaconule; peci = precingulum; poci = post-

cingulum; prcc = preproiocrisia; pose = postprotocrista; pela = pre-

talon; pola = post-talon; péri = pericone; hypo = hypocone;

prol = pnjtocüne.

421 —

An assemblage of basal placentals is selected in order to allow different hierarchical levels

of comparison within Placentalia. The classification of Stucky & McKenna (1993, Fig. 3) is

then utilized for sampling the following sélection of placental mammals, taken as représentative

of the major placenta! clades and considered as terminal taxa: Pappoiherium (the only non-

placental, a lower Cretaceous metatherian); four leptictids - Prnkennalestes, Boholestes, Ken-

nalestes, and Leptictis\ four basal feraeans Cimolestes, Prototomux, Pmviverra, and Miacis\ one

lipotyphlan - Scenopagus', threc anagalids - Zalambclulestes, Ensigale, and Paramyx\ and two

basal archontans (plesiadapiformes) - Purgatoriiis and PaUiechthoii. Pappotherium considered

to be reprcsenling the Metatheria and therefore assumed to be the sister group of the assemblage

of placental mammals listed. The character polarizations performed here are often supported by

comparisons with the genus Sulestes, assumed to be rooted al the same basal trichotomy from

which Pappotherium and placentals are branching off. Therefore. Sulestes serves in this analysis

as an anolher outgroup.

In the data matrix analyzed, only Pappotherium is included, because the notion here applied

of a basal metatherian are mostly driven from detailed examinations of the Pappotherium molar

morphology. The reason for including a metatherian in the analysis of parsimony here performed

is simpiy to hâve means to appreciate the bearing of the upper molar morphology on the matters

concerning the monophyly of placental mammals. The character polarity is based on the study

Fig. 3. — Branching diagram derived from the classification of Stlicky & McKenna ( 1993). The que.stion mark over Ihc Ungulata

clade empha.sizes problems regarding ungulate monophyly and interrelalionships. In facl, Stucky & McKenna's classification

does indeed question the frequently accepted notion of Ungulata. But. the polytomy for the component Preptotheria does

reflect the broad controversy that has taken place in the last ten years regarding high placental interrelationships. In the rcsults

section the.se matters are addressed.

— 422 —

of non-therian mammals, Papputherium being only minimally different from the postulated mor-

phology of the ancestral morphotype of therian mammals.

Based on preliminary evidence concerning ratios of parallelisms and convergences of the

dental morphology bctween ungulates and other placentals, it is decided here to limit the number

of ungulates to bc sludicd lo jusi a few généra lhat seems really to represent basal ungulate

forms. Thus at Ihis preliminary stage of our research programme on ungulate basal interrela-

tionships, just four basal condylarihs are examined: Pwtungulauim Sloan & Van Valen, 1965;

Mimatiüa Van Valen, 1978; Oxyprimus Van Valen. 1978; and Dysnoêtndtm Zhang, 1980 (the

taxonomie status of this gcnu.s is addrc.ssed below). Hence, the rooting context of ungulates is

here addressed in the light of these four généra.

There follows a discussion of ail upper molar attributes considered lo be phylogenetically

informative at the basal placental mammal level; and the assignments regarding their taxonomie

ranges arc of course limited to the very sample of taxa described above.

CHARACTER ANALYSIS

The morphological complex scrutinized in this cladistic analysis is the upper molar mor¬

phology and mainly focused on patterns of change preceding and related to the emergence of

the ungulates. The character State distributions of the terminal taxa are shown in the data matrix

(Table I ). and arc only partially referred in this section. The character polarizations are based

on the companson between tribosphenic and non-lribosphenic mammals {sensu ClFELU 1993).

The “ancestral morphotype’’ of tribosphenic mammals is doubled in data matrix, this is due to

the fuel thaï the progrum HennigHb roots networks with outgroups on the initial hypothesis that

the basal node is a trichotomy. Referring of non-placental mammals, the character polarizations

are more accurately iletermined when using Pappotherium rather than Sulestes as outgroup, be-

cause the former has been studied in more detail. In a few cases, it was not possible to establish

the homologous condition in the outgroups, This is generally due to the fact that the outgroups

lack or hâve the .structure(s) evolved in the particular character formulation in a rudimentary

stage of development. For a few features described. it is uncertain whether or not they are aspects

of the same character. But, as far as their variation is concerned, it is argued that although some

clear funclional corrélations can be observed they seem to vary as independent phylogenetic

units.

Given the preliminary scope of this article and the complex literalure regarding the dental

patterns of placental mammals, the characters listed below are not formally referred to previous

discussions. Thus, for a number of features, a sériés of previous interprétations hâve been pro-

posed. The originality of some of those attributes relies on the way they hâve been formalized,

or on the manner the character State partition has been formulated. Nevertheless, the pertinent

literature is folloyved as much as possible. A few of the characters listed below seem to represent

new topographical relationships, for instance characters 9 and 21. It is important to bear in mind

that some of the qualitative multistate characters proposed below might be better formulated in

— 423 —

the context of a set of spécifie measures. However, only small samples of each terminal taxa

hâve been available, making it spurious to attempt a detailed quantitative study.

Table 1. — Data matrix for 23 of the 26 eharacters formulated for placenta! mammals. Note lherefore that the three autapomorphies

referont to eharacters 16, 20, and 25 are excluded from this matrix. The numbered eharacters are described in the text.

Data Matrix

eharacters : 00000000011111111122222

12345678901234578912346

Hyp. anc. 1

Hyp. anc. 2

Pappotheriam

Prokennalestes

Bobolestes

Kennalestes

Leptictis

Scenopagus

Cimolestes

Prototomus

Proviverra

Miacis

Purga tonus

Palaechthon

Sa 1ambdalestes

Boslgale

Paramys

Protungulatum

Dysnoetodon

Oxyprimus

Mimatuta

00000000000000000000000

OOOOlOOOOOOOOOOOOOOOOOO

11002100000000010100000

21001100100000011200000

21003210210000011100000

32123210320100111200000

21224221320100111200000

21013210200000011301000

21013210210100111201200

22123211210101011301100

22111111220201111200100

22224222320210211200011

33224211200000001200001

42124222422220207300002

42333223422220212200003

32224222320200113110001

22223222320201213110001

32224222320200221210002

32224221321210113210001

1. Stylar shelf, labial extension (Fig. 4).

The stylar shelf is characterized by the crown expansion labially to the paracone and meta-

cone. The contraction of this shelf is widely considered to be related to the early evolutionary

history of placental mammals. The réduction of the stylar shelf is thus hypothesized as a derived

unique feature for the Placentalia. Here, this réduction trend is interpreted as comprising a trans¬

formation sériés of five character States.

They are the following: (0) — well-developed. considerably labially expanded; ( 1 ) - partially

contracted; (2) - reduced, ail along its sagittal extension; (3) - residual or very narrow, dis-

playing like a thin border labially to the two cônes; (4) - nearly absent, the labial walls of the

paracone and metacone forming the very labial portion of the crown. The extremes of this trend

are clearly illustrated by the configuration found in Pappotherium, representing the primitive

State 0, and by Paramys for the extremely derived one.

2. Paracone and metacone, height.

It is considered that a trend towards réduction in height of the paracone and metacone

takes place within placentals. Therefore, this feature is formulated to stand for the paracone and

metacone patterns of height above the basal plane of the crown.

— 424 —

right

left

leff

Fig. 4. — Stylar shelf, development (characler I): siale 0, Pappotherium (anicrior loward righl): siale 3. Protungulatum (anierior

loward left): stale 4, Paramys (anierior loward lefl); sish = siylar shelf latéral dimension. This dimension permits a latéral

perception of ihe labial projection of the stylar shelf. Labial view.

Three character States are hypothesized; (0) - paracone and metacone very high; (1) - high;

(2) - low, or simply not characteristically high; (3) - considerably reduced in height.

The condition found in Pappotherium (Fig. 6) is a good example of that interpreted as State

0 and Protungulatum (Fig. 6) of that viewed as State 2.

3. Paracone and metacone, relative positions (Fig. 5).

The character State distributions observed here seem lo covary wiih Ihose of character 2.

The degree of contact between the internai walls of these two cups ranges from a full contact

where the two parts merge with one another to a condition where they are completely apart;

the latter is présent in Paramys.

Hence, four States are formulated: (0) - the two internai walls in full contact, with their

base portions somewhat emersed in one another; (1) - clearly in contact to one another; (2) -

closely placed, maintaining some contact only at the extremities of their basal portions; (3) -

completely apart from one another.

4. Centrocrista, development.

This crista is located in between the paracone and metacone, and connects the upper surfaces

of these two cusps. The development of this crista includes its width and height.

Four character States are formalized; (0) — well-developed; (1) — developed; (2) - not con-

spicuously developed or simply poorly developed; (3) - absent. The State 3 is uniquely displayed

by the considered rodent. Paramys.

In principle, the fact that in Paramys these two cusps are widely apart does not prevent

the existence of a centrocrista. There seems to exist a lopographical dependence between this

feature and the one numbered 3, which turns the judgment of the degree of development of the

centrocrista quite complex sometimes.

5. Protocone, overall size (Fig. 6).

In order to hâve a better understanding of the spectrum of variation regarding the évolution

of the hypocone, the polarization of this character makes référencé to non-tribosphenic mammals

— 425 —

Fig. 5. — This .schcme illustrâtes trends regarding the par*

acone metacone relative positions (charauter 3):

State 0, the distance a reflects a condition wliere lhe

two cônes are very close lo onc anothcr having full

contact bciween their internai walls; State 3, repre-

sents a condition wherc ihc iwo cône» arc away a

part wiih ihc distance b rcflcciing il. The scheme

under siale 3 is derived from the genus Paramys.

(this is the only case where the character polarity is determined referring to basal mammals

other than Pappotherium and Sulestes).

Therefore the primitive character state (0) is considered to be the absence of the protocone,

as represented by non-tribosphenic mammals. The degrees of development of the protocone are

Fig. 6. — Protocone, overall development (character 5): state 0, poorly developed, Pappolherium (anlerior toward right); state 4,

well-developed, considerably larger than lhe paracone and metacone, Protungulatum (anlerior toward left). Linguo-occlusal

view.

— 426 —

formulated in reference to the size of the paracone and metacone, and the condition found in

Pappniherium, where the protocone is very reduced in relation to either one of the two cusps

represents the derived characler State, coded as (1). There, the protocone displays more like an

accessory cusp rather than a major cusp encompassing a significant part of the crown area. State

(2) is an intermediate condition where the protocone emerges as a main cusp but not as developed

as the paracone or the metacone. The other States demarcate a shift in these cusp size rclation-

ships, where the protocone is larger (3) or considerably larger (4) than the paracone and meta¬

cone. State 4 is widely distributed within ungulates.

6. Protocone, transverse expansion.

This is a feature broadly discussed in the literalure (e.g. CiFELLi 1993), here divided into

three States; (0) - protocone poorly expanded transverseJy, more like an accessory than a major

lingual cusp of the crown. and so relatively poorly developed in relation to the paracone and

metacone; (1) - protocone transversely exptinded; and (2) - protocone with a well-developed trans¬

verse e.xpansion. The formalization of this character is not .simple, given the myriad of possible

intermediate degrees of expansions and réductions due to the well-known plasticity of this dental

région in placentals. Nevertheless, it seems clear lhat a major transverse expansion is related to the

évolution of basal placentals. However, until better phylogenetic hypotlieses becorae available re-

garding the sister-group rclalionship of placentals, it is doubtful whether or not the transverse ex¬

pansion of the protocone is homologous with that found in some metatherians. Marshall &

Kielan-Jaworowska (1992) and CtFELU (1993) discussed metatherian-eutherian interrelationships

and arrived at contradictory results. Brielly, ClFELLi presented the metatherians as paraphyletic

whereas Marshall & Kielan-Jaworowska viewed metatherians as monophyletic.

7. The mesiodistal dimension of the crown, where the protocone labial face meets the crown

margin (Fig. 7).

In order to better visualize this morpho-change, it is important to understand that the an-

tero-posterior expansion of the mesiodistal dimension might be the resuit of two independent

7. — Schematic drawings for the mesiodistal dimension of the

crown (character 7): sialc 0. poorly expanded; siale 1, expanded.

The two drawings are derived respectively from Kennateates (dis¬

tance a) and LeptU tis (distance b), and b is to be read larger than

(I. Lingo-occlusal view.

events. Here, the first one is represented by the character 7 and the other by the next one num-

bered 8, which accounls for the antero-postcrior enlargement of the lingual portion of the pro-

tocone. These events are normally read as composing one single character as formulated by

ClFELLI (1993).

At this stage, it seems that three States are distinguishable: (0) - poorly expanded, condition

found in Pappotherium and Sulestes\ (I) - expanded, condition widespread among placentals;

(2) - greatly expanded.

8. Protocone, antero-posterior expansion of the very lingual portion relative to the mesiodistal

dimension of the crown (Fig. 8).

As previously discussed in character 7, the antero-posterior enlargement includes two

independent expansion events. Here, the transformation sériés is composed of four States: 0

- extremely poorly expanded; (1) - slightly expanded; (2) - well-expanded; (3) - greatly

expanded. The development of the hypocone and pericone are not placed as part of this

protocone expansion. Figure 9 exemplifies the problems of interprétation that may resuit

when the development of these cônes are added to account for the protocone expansion. The

lingual portion of the crown changes when these cônes are présent, however, they should be

examined independently.

9. Protocone, labial wall and the trigon base (Fig. 10).

This feature accounts for the development of the protocone labial portion, and the paracone

and metacone are used as topological reference points.

Fig. 8. — Protocone, antero-po.sterior expansion of the

very lingual portion (character 8): the distance a rep-

resents the derived State 2 and is illusirated by the

condition found in genus Protungulatum (anterior

toward Icft).

Fig. 9. — This is the case found in ihe genus Leptictis (anterior

toward right) showing that the hypocone development can

mislcad an attempt lo measure ihe protocone unterxvpost-

erior expansion of the very lingual portion (abbreviated by

prlex). Here lhe overall expansion ofihe very labial portion

of the crown (abbreviated by ovlex) is lhe sum of the pro-

locone and hypocone expansions.

— 428 —

The character State (0) is represented by the cylindrical-like shape of the labial wall of the

protocone ending in a closed angle at the base of the crown, and away l’rom the lingual portion

of the paracone and metaconc, The protocone displays a cone-like shape with a poorly expanded

basal diameter lingually. In State (l), the labial portion displays in a more open angle regarding

that one of State 0. and forms togeiher with lhe niesiodistal portion of lhe Crown a surface that

could be viewed as part of the trigon base, but this surface ends labially aiso away from the

paracone and metaconc (lingual surface). In State (2), a trigon base is aIso formed but il is more

labially expanded reaching the very portion below the basal level of the paracone and metaconc.

In State (3), il ends in an even more open angle, fomiing a ‘’iriie'’ trigon base and reaching the

basal portion of the paracone and metacone; in State (4,) this surface (trigon base) merges with

Fig. 10. — Schematic drawings representing two extremities of the development of the protocone labial portion (character 9):

State 0, the protocone labial wall ends on lhe base crown surface in close angle (a) and away from lhe paracone and metacone;

in the derived siale 3, il is well-devcloped ending on lhe crown surface in a more open angle (P) reaching bolh paracone

and metacone connecting the paraconuie and metaconule, hence forming a rccognizable trigon base.

the crown in a very open angle, nearly in parallel lo the basal plane and reaching the paracone

and metacone (a lypical bunodont pattern of low cusp relief), When the conules are présent,

they seem to be independent of these character States: the conules apparently do not demarcate

the labial extremity of the trigon base.

10. Post-talon basin, development (Fig. 11 ).

The post-talon basin is displayed as a sort of a Hat upper surface or more like a crista or

even as a fossa al lhe upper portion of lhe post-cingulum. which frequently connects the hypocone

when it occurs. In cases of “crista shape" - a narrow fossa is aiso présent between the posterior

wall of the protocone and lhe internai one of lhe crista. The shapes of the post-talon basin are

not considered here to be phylogenetically informative. However, lhe overall development of

the talon seems to be phylogenetically relevant at the basal placental branching level.

This character is interpreted as having three States: lhe primitive State coded as (0) is the

absence of the talon, or the absence of the post-cingulum; the derived State (I) - presence of

— 429 —

the talon in a very rudimentary stage of development; and State (2) referring to a clearly developed

or well-developed post-talon basin.

Fig. U, — Development of the post-talon, or of the upper portion of the posterior cingulum (character 10); State 0, absence or

vestigial stage of development, illustrated by Cimolestes; State 1, developed, illustrated by Oxyprimus. Both anterior toward

right.

11. Position of the labial portion of the post-talon basin (Fig. 12).

The labial portion of the post-talon basin is aiso named pre-talon basin by some other

authors. Here, we named pre-talon basin the anterior portion of the crown where this basin is

placed at the upper portion of the pre-cingulum. But, in this context this attribute refers to the

extent that the post-cingulum grows labially over the upper surface of the crown, interpreted as

the trigon basal Icvel.

Three character States are then proposed to approach this topographical pattern: (0) — the

labial extremity of the post-talon clearly away from the upper crown surface; (1) - near the

level of the crown surface, but not there; (2) - reaching the crown upper surface, composing

this way a condition where the labial growth of the post-cingulum go in the meta-cingulum

bypassing the post-metaconule crista.

Fiü. 12. — Position of the labial portion of the post-talon basin

(character 11); Pnramys (anterior toward ieft) illustrâtes con¬

dition 2 where the mcta-cingulum (meci) mcets the post-

talon (pota). the point of connection is representcd by the

abbrcviation concc.

tneci .

I /Pota

conec

— 430 —

12. Pre-talon basin, development.

Here, the character States are formalized pretty much the same way as indicated for

character 10. The development pre-talon basin is examined in order to interpret the extent of

its emerge over the anterior side of the protocone.

When it is présent in a sort of rudimentary stage, it is coded as State (1); and State (2)

goes for the developed or well-developed stage of the talon; the primitive State (0) refers to the

absence of the pre-cingulum and this way to the absence pre-talon.

13. Position of the labial portion of the pre-talon basin.

This is a very similar morphological situation with that of character 11, where three character

States are formulated to account to the position of the labial extremity of the pre-talon basin

relative to the upper crown surface; (0) - away from the upper crown surface; (I) - near the

level of the upper surface; (2) - reaching the upper surface and connecting the para-cingulum;

characterizing this way a case where the labial growth of the pre-cingulum go in the meta-cin-

gulum bypassing the pre-metaconule.

14. Post-cingulum and pre-cingulum (Fig. 13).

In some taxa the post-cingulum and pre-cingulum are developed in such a way that they

meet lingually, encircling the crown, State (1). The primitive State (0) refers just to the absence

of this condition.

Fig. 13. — Miacis (anterior toward left) illustrating a derived condition 1 where the

pre-cingulum and post-cingulum encircle the lingual portion of the crown.

15. Hypocone, development (Fig. 14),

The hypocone is frequently placed at the very lingual edge, composing therefore the lingual

Wall of the crown with the protocone. This feature refers to the degree of the development of

the hypocone, and, to avoid problems due to the complex variation of this character, just four

States are proposed; (0) - absent; (1) - rudimentary or very small; (2) - developed and well-

developed; (3) - extremely enlarged.

These size coding problems are stated in référencé to the overall size of the crown taking

into account basically the height and width of the hypocone. The reason for not using the pro-

— 431 —

Fig. 14. — Hypocone, development (character 15): State 0. absent {Cimolestes)\ State 1, dcveloped (Scenopagus). Both anterior

toward leh.

tocone size as a référencé is simply to avoid problems of possible independent réductions of

the protocone which would necessarily change tbe hypocone protocone size relationships. There

seems to be a certain ambiguity related to the overall size of development of the hypocone,

given that there are cases where e.,ç. a distinct enlargement of the hypocone diameter is not

followed by a proportional growlh in height. This is the case of Scanapagus, which is questionably

coded as I, and is illuslrated in figure 14. Bolh Pappotherium and Sulesies lack the hypocone;

however. il is relevant to note that sotne other inetatherians do hâve the hypocone, or hypo-

cone-like cusps.

16. Pericone, development.

The pericone is placed at the opposite side to the hypocone on the antero-lingual wall of

the protocone. The development of the pericone is formulated as a binary character: State (0)

absence, and (1) presence.

A point to bear in mind is that one may easily misidentify of similarly placed accessory

cusps by that named pericone.

17. Conules, development (Fig. 15).

This feature stands for the different degrees of development of the conules. Three character

States are formulated; (0) - very poorly developed or nearly absent; (1) - developed; (2) - a

further derived condition where the conules are considerably well-developed, a condition well

illustrated in Paramys.

Fig. 15. — Development of the conules (character 17): .State 0, poor develop¬

ment or vestigial; the genus Zolambdalestes (anterior toward left) seems

to represent a case of this primitive condition.

432 —

18. Conules, shape (Fig. 16).

Four types of conules are recognized but of unclear topographical relationships. They are

the following: (0) - pointed cone-like shape; (1) - wing-like .shape; (2) - fiat cylindrical-like;

(3) - rounded, bunodont-like.

Given that no logical topographical sequence of changes is justifiable for these conule

shapes, this character is considered unordered.

mecl

Fig. 16. — Conules, shape (character 18): State 3, cylindrical-like .shape, il pictures

the condition observed in Paramys (anierior toward right).

19. Conules, position relative to the paracone and metacone.

This character accounts for the extent that the conules are approximated to the paracone

and metacone. hasically the distance between the conules and lho.se two cusps.

Thrce dcrived conditions are proposed, and when lhe conules are away from the two cited

cônes, il is coded as (0). State ( I) represents the case where the conules are close to the cônes

but hâve no contact. In the other two derived States the conules are placed in contact with the

paracone and metacone: lhe siale (2) is gond for a partial contact, and State (3) for full contact

between those cusps.

20. Conules, placement in relation to the anterior and posterior crest of the protocone.

This is a binary character, where State (0) stands for the placement of the conules on both

protocone crests. The derived State ( 1 ) is a case of displacement of the conules with respect to

the crests of the protocone. This feature is indeed an autapomorphy for Paramys.

21. Conules, lingual portion (Fig. 17).

This feature describes the relationships of the lingual portion of the conules with the pro¬

tocone. Two conditions are assigned, one where the lingual wall of the conules are free from

contact with the protocone State (0), of course not referring to its basal portion. The derived

State (1) is characterized by a sort of torsion towards the protocone, resulting therefore in a

considérable contact of the lingual wall of the conules with that of the protocone. This derived

condition seems to be widespread among basal ungulates.

— 433 —

Fig. 17. — Conulcs projcciion lowards ihe prolocone having a large portion of cach conulc lingual wall lying over the labial

surface of the protocone (chai'acter 21): State 0, absence of such a trend (Prolotoinus)-, State 1, présence of this derived pro¬

jection, herc illuslratcd by the condition found in the genus Protungulatum. Poslero-occlusal view.

22. Metacone and metastyle, shearing surface (Fig. 18).

This attribute characterizes a case of carnassial évolution where the metastyle is placed

labially to the metacone forming a carnassial shearing surface, condition coded as State (1). This

condition is reported for Cimolestes as well as other feraeans examined. The primitive State is

the absence of this condition (0).

Fig. 18. — Metacone and metastyle. .shearing surface (= sc; character 22): State 1, the derived

condition where such shearing surface is présent, here illustrated by the condition found

in the genus Cimolestes (anierior loward lefi).

23. Paracone and parastyle, shearing surface.

The paracone and parastyle also form a carnassial shearing surface (1), similar to that of

character 22, but its distribution among the taxa examined is slightly different. The primitive

State (0) is the absence of this carnassial surface.

— 434 —

24. Relative position of the labial portion of the crown (Fig. 19).

This région ot' the crown is viewed as fundamentally composed by the paracone and meta-

cone, and englobing the stylar shelf. In the two plesiadapiformes examined, a posteriorly orien-

tated torsion of the labial portion of the crown is observed. Viewing from antero-lateral angle,

this torsion .seems to permit the paraconule wing to reach the pre-paracrista without having contact

with the pre-cingulum.

The présence of this condition is considered derived (1), and the absence primitive (0).

Fig. 19. — Position of ihc paracont'-meUtconc ponion on lhe crown (characler 24): stale l, posteriorly oriented torsion displacing

the paracone and metacone î'rom a central position on the crown iPalaechthon). In this scheme the arrows show lhe estimated

point.s of torsion, and x is a projection of lhe médial line of the prolocone demarcated hcre by ils tip and serving lhen as

reference for lhe displacement of paracone-melacone région. Anlerior toward right.

25. Paracone and metacone, lingual expansion.

This character stands for the lingual expansion of the paracone and metacone over the par¬

aconule and metaconule respectively.

The présence of this condition is considered derived (I) and the absence primitive (0). This

is another character that cornes oui as an auiapomorphy for Paramys.

26. Prolocone. shapes.

This characler regards the shapes of the prolocone upper portion exposed on the crown

surface. It is nol viewed as a general trend for the three main cusps because the shapes of the

paracone and the metacone seem to covary only to a certain extern wiih that of the prolocone:

for example, there .scems to be no cases where a very high pointed paracone and metacone

coexisi wilh a characierislically shallow prolocone, but no further morphological cohésion among

thèse cusps can be formulatcd as a ruie of variation.

Four general patterns are proposed to account for the variability observed: (0) — cone-like

shapc wilh somewhal high pointed cusps; (1) - slightly rounded. but not representing a typical

bunodoni pattern; (2) — dclïnitely rounded. displaying a typical bunodont like cusp; (3) - some-

what flattened, composing a very shallow lingual crown upper surface. This character State re-

lationships are unclear, in the sense that no consistent argument can be made in order lo advocate

that the transfomiation between one State into another requires more than one step. Hence, this

— 435 —

character is treated as unordered. Notice that this feature is not applicable to the taxa displaying

the protocone in a rudimentary stage of development.

Among the 26 characters formulated, characters 9, 17, 18, and 26 are considered to be

unordered. Table 1 summarizes ail the character information discussed above and gives the dis¬

tribution assigned to each. It is clear that in order to achieve more accuracy in the character

partition for the type of qualitative characters discussed above, a biometricai study followed by

a gap coding system is nceded. But, whulever the improvements might be, while limited to the

upper molar cusp topographical interrelationships, they will be confined to the tribosphenic

placental “ground-plan”: a mosaic of three basal éléments composed by the paracone, metacone

and protocone varying in size, shape and their relative positions along with a sériés of accessory

cusps and crests that also vary in size and shape, greatly modifying the basal tribosphenic System.

ANALYSIS OF PARSIMONY

In this attempt to contribute to a better understanding of the phylogenetic information content

in the placental dental morphology 26 characters are proposed, the majorily being multistate

characters with 4 of them read as unordered. Three characters (16, 20, and 25) are autapomorphies

and are then disregarded in the parsimony analysis performed here to prevent the consistcncy

index from geiiing intlated (CarpEnthR 1988).

After ninning the ie* exact algorithm that generales trees by implicit énumération of the

Hennig86 software (Farris 1988), and applying two hypothetical ancestors in order to bring

about the rnonophyly of the ingroup, 30 most parsimonious trees came out as resuit with a length

of 93 steps, consistency index (ci) of 59 and rétention index (ri) of 80. The outcomes of exact

algorithms are considered to be certain minimal length trees (Darlu & Tassy 1993). The strict

consensus of lhese 30 trees (appendix I) has 17 components (excluding the basal node 39, where

the two hypothetical ancestor are rooted).

Successive approximations weighting (Farris 1969, Carpenter 1988) was then applied,

and it is slabilized atter the third weighting with one tree of length 395, ci of 72 and ri of 88

(Fig. 20). Characters 14 and 22 became of no value. As discussed by SaLI.E.S (1992: 41). Farris

stated that the point made by Carpenter (1988) about the necessity of recuding multistate

characters to binary, when applying the successive weighting option, is not applicable to the

Hennig86 softw-are. Nevertheless, an article that is in press in Cladisiics by Farris, Kluge &

Carpenter may présent different thoughts on this matter. But, at least, the four characters con-

sidered to be unordered are not cases of totally inapplicable "additivity", and on Lipscomb's

manual of the Hennig86 software there is no reference for the necd of having multistate characters

change inio binary ones.

The final tree (Fig. 20) obtained after weighting is later compared with the one obtained

after running the ie* option with ail characters unordered, which first resulted in 205 most par¬

simonious trees with length of 88, ci of 62 and ri of 75, and then the same weighting procedure

is applied - ending up with 12 trees with length of 412, ci of 74 and ri 84 (see appendix 2 for

the strict consensus of these 12 trees as well as directiy from the 205 trees obtained before

weighting).

— 436 —

It is imporlant to undersland the nature of the diagram of Figure 20, in the sense that this

branching pattern is an expression of the congruences among attributes of a single morphological

complex. Therefore, the relationships advocated by this diagram might be better read as patterns

of upper molar évolution rather than a phylogenetic hypothesis among the 18 taxa considered.

These relationships should be simply viewed as statements of overall congruence among the 22

dental traits, and conceming then morphological complex mostly studied by paleo- and neo-

mammalogists in order to interprété the evolutionary history of mammals.

39.

E O Hypothetical 1

1 Hypothetical 2

I fp2 Pappatherium 1 METATHERI A

pap,

Prokennalestes

4 Bobolestes

36.il |î=S KennaiestesJ

LEPTICTIDA

-L,.r

Cimolestes

|p=9 Prototoaus

PLACENTALIA

33 | Hiacis | EERAE

I!=32| f=10 Proviverra_i

1L=3i| lj=lA Zalambdalestes

1!=3o| |j=6 Leptictis~~

1!=29| |-=7 scenopagus

LIPOTYPHLA

16 Paramys |

234=15 Eosigale _1

ANAGALIDA

C 23=15 sosagaj.e

32C3^/irecrtiodARCH0NTAV

C 19 oicypriaue l

iLs2iJ 1=:20 Mlmatuta__

FiG. 20. — Branching diagram for placenta! mammals obtained after applying the le* option of Ihc Hennig86 software and making

use of the successive weighting algorithm. The interprétation of the component 22 as Archonta is very weak in the sense

thaï it may be simply delineating Plesiadapiformes. However. if these are assumed to be basal primates, and primates are

aceepled as members of the Archonta, the interprétation thaï these iwo gencTa represcnis Archonla would be plausible. Lipoty-

phla may be either represented by Scenopagus alone or it may include the genus Leptictis. The position of ihe genus Zalamb¬

dalestes is qucsUonable; SlUCKY & McKüNNA (1993) hâve classified il as a basal lagomorphan. The stated placental

intenelationships are discu.s.sed in the text.

Character optimizations of fealures 3, 5, 9, 10, 14, 15, 19, 22 and 26 hâve assigned more

than one character State for one or more nodes, representing therefore a set of quite complex

alternative paths of evolutionary change. To build a table of nodes and corresponding characters,

somewhat arbitrary decisions hâve been made, in order to simplify the optimization of their

character State distributions. The rationale applied is laler discussed under the transformation

sériés analyzed. But, briefly, it is to choose the possible basal position of the further derived

character State - a procedure that would be more like the ACCTRAN optimization (SWOFFORD

& Maddison 1987). In some cases this rule could not be applied. A simplified version of the

character State distributions on the 17 components in Figure 20 are summarized in Table 2.

— 437 —

Table 2. — Character distributions on the branching diagram of figure 20, and the * symbol identifies homoplasies, character

States placed at more than one node.

Node

38;

5(1)*.

Node 29:

9(3)*, 10(2).

Node

37:

1(1), 2(1), 5(2)*, 6(1),

17(1), and 19(1)*.

Node 28;

7(2).

Node

36;

1(2)*, 9(2)*, 18(1),

and 19(2)*.

Node 27:

8(2), 12(2),

and 15(2)*.

Node

35:

5(3)*, 6(2), 7(1), 9(2),

and 10(1)*.

Node 26;

21(1).

Node

34;

4(1) and 22(1)*.

Node 25:

13(1).

Node

33:

12(1), 15(1)*,

and 23(1)*.

Node 24:

18(3).

Node

32:

2(1), 3(1)*, 8(1), 11(2),

and 14(1)*.

Node 23:

1(4), 9(4), 13(2).

Node

31;

4(2).

Node 22:

24(1)

Node

30;

1(3)*, 3(2)*, 5(4)*.

Node 21:

15(1)*.

INTERRELATIONSHIPS OF PLACENTAL MAMMALS

(UPPER MOLAR MORPHOLOGY, INTERNAL CONGRUENCE)

In the latest matrix révisée! by NOVACEK (1992) regarding extant and l'ossil mammals, the

Crown morphology of upper molars was stated to be informative at Ihree branching levels: the

placental level (narrow stylar shelves of the upper molars); the ungulate basal node (general

bunodont transfomiation including the considérable development in size of the hypocone) and

the third level expressed by the bilophodonty of tethytherians. These dental informations are

here readdressed as discussed in the “Character Analysis” section, but generally the partitions

of the molar mosaic are refined and the character States expanded, frequently resulting in mul-

tistate characters. The following discussion refers to character distributions on the branching

pattern of Figure 20.

The first somewhal surprising resuit involves the overall phylogeny of placental mammals,

here examined from a late Cretaceous/Paleoccne fossil record perspective, which emerges fully

resolved. As expected, Pappotherium is placed at the base, and is followed by a paraphyletic

sequence of lepticlids Prokennalestes, Bobolesles and Kennalesles. Cimalesie.i branches off after

this last leptictid and it is succeeded by the other three feraeans. Protoiomus. Miacis and Pro-

viverra. Hence, the monophyly of Fcrae is nol supported However, If Ferae is assumed to become

monophyletic at some point in this branching area, basally dcmarcatcd by Ciinolestex, feraeans

would be rooted al a rclalively basal position within placcntals which is an inlriguing possibility.

This would support an uncommon interprétation of the phylogenetie status of Ferae.

Zalambdalestes, Leplictis and Scenupagus fonned anolher paraphyletic sequence where the

last is sister group of clade 27. Zalambdalestes is a laie Cretaceous placental frequently inler-

preted as closely related to anagalids, and recently proposed as a basal lagomorphan (STUCKY

& MCKENNA 1993). Assuming thaï rodenis (here represented by Paramys) and lagomorphans

are indeed sister groups, the topology of the placental branching diagram (Fig. 20) does not

support the placement of Zalambdalestes as a lagomorphan. Leptk tis is here placed aparl from

— 438 —

the “closely related” basal paraphyletic assemblage of leplictids, and Scenopagus represents the

“true insectivorans”, the Lipotyphla. Hence, leplictids arc indicated to be paraphyletic and lipoty-

phlans lo be sister group of clade 27.

The monophyly of clade 27 is a subject apart but it is striking that another set of different

fealures seems to indicale a similar scheme of reiationship (now under study by a team ol paleon-

tologists); the main différence is that basal ungulates had not been considered.

Ungulata is here represented by four basal condylarths and it is hypothesized to be a member

of ihis clade 27. Therefore, ungulates are suggested to be closely related to the plesiadapiformes

(clade 22) and anagalids plus rodents (clade 23). The four condylarths came ont rooted at

node 26, with Oxyprimiis at the base and the others three forming a clade numbeied 24, in

which Dysnoetodon branches off from the base and Prottingiilaium and Mimatuta are sister

groups, clade 21. The genus Dysnaeitxkm is here identified as a condylarth-like placental, but

Zh,\ng {1980) suggested that it could be a tillodont. We can assert that at least il is nol a typical

tillodont. But. if one assumes that it might indeed represent a basal form of tillodont, our results

would thcn corroborute ihc old hypothcsis that Tillodontia is derived from a condylarih basal

form (see e.g. GiNCERiCH & GUNNEI.L 1979).

Those relationships discussed above are not really incongruent with any of the other results

obtained after applying different algorilhm combinations for that same matrix illustrated in

Table 2, excepi for a single component comprising two terminal taxa, First, the relationships

summarizcd in Figure 20 are partially supported by the Nelson consensus trcc (appcndix 1)

derived direcily from the 30 mosi parsimonious irces from ihc ic* algorithm. The components

thaï are identical are the following (the numbers on the Icft correspond to the components of

the branching diagram of Figure 20 and the one.s on the righl to ihc components of appendix

1): 37 = 28; 37 =27; 36 = 26; 35 = 25; 32 = 24; 27 = 23; 23 = 22; 22 = 21. None of the com¬

ponents of the branching diagram of appendix I are incompatible with the ones observed in

Figure 20: it is simply a less resolved branching diagram. We want to point out that component

27 émerges without .successive weighting.

Comparing the placental tree in Figure 20 with thaï of appendix 2, in which ail characters

are read as unordered and obtained after applying the ie* and successive weighting options, a

lisi of the identical components between those two diagrams follows (components of Fig. 20 on

the left): 38 = 33; 37 = 32; 36 = 31; 35 = 30; 32 = 29; 28 = 28; 27 = 26; 26 = 25; 24 = 22; 23 = 24;

22 = 23; 21 =21. In tact there is one component front this unordered tree that is not found in

that one in Figure 20; namely component 27. which places Pruviverra and Miacis as sister groups.

Finally, the option of having the characters unordered and not applying the succe.ssive weighting

results in a poorly resolved branching diagram. The relevant point is that it displays one com-

poncni which is identical lo the component 32 of Figure 20, to the component 24 of the tree

in appendix I. and to the component 29 of the tree in appendix 2. Whai is intriguing is that

these components hâve Proviverra and Miacis as members, having then the two other feraeans

out of their assemblages. Hence, these results definitely question the monophyletic status of

Ferae. Nevertheless, at this point we slill understand these results as very preliminary and possibly

simply reflecting the absence of enough character information to bring about the monophyly of

Ferae.

— 439 —

Fig. 21. — Simpiified branching diagram derived from lhat of

Figure 20; the a.s.sumptions made to produce Ihis synihe.si.s

are the folluwing; Ihc ihrec basal iepliclids are assumcd to

be monophylclic; the samc gocs for ihc four feraeans; the

two plesiadapiforme.s could well represent Archonta. Her-

botheria \s the naine given to component 27 of ihc ircc il-

lustrated in Figure 20

Now, comparing the assemblage of placenta! iiiteiTelationships here advocated (Figs 20, 21

for a comparable high leve! synlhesis) with the ones supported by McKenna (1992: 351; here

illustrated in Figs 22A, 22a for a comparable synthesis), and based on the amino acid sequences

for the alpha crystallin A chain, the number of nearly identical components between Ihese two

trees is striking. First, component 37 of the dcntal-based branching diagram (Figs 20, 21) is

identical lo numbered 1 of the alpha crystallin one (Fig. 22). This is to .say thaï Mctatheria is

read as the sister group of Placentalia in both branching diagrams, To our knowledge al| assigned

basal edentates hâve their dental pattern exiremcly differentiated, hence they are not considered

in this study. Therefore, component 5 of Figure 22A having edentates as basal placental cannot

be a test against the branching pattern of Figure 20. A similar but opposite argument goes for

the phylogenelic position of the First ihree basal placentals of the dental-based tree, which are

only known as fossils from the late Cretaceous. However, it is very curions thaï component 23

which has Carnivora branching off from the base is identical lo componeni 34 if we consider

Cimolestes as the basal Ferae; and it may also be that their paraphylelic distribution between

nodes 33 to 30 is due to absence of information, Alternatively, their dental attribiites may be

insufficiently informative in tins character survey to express the monophyly of the feraeans. or

perhaps a .sériés of dental parallelisms exist between feraeans and their relatives. Now. we could

well read component 29 of the dental-based branching pattern as identical to that numbered 30

in McKenna’s scheme. For this we takc Scennpagus and Leptictis as lipotyphlans and ( \\ Leptictis

is not considered. componeni 28 would be identical to 30) and branching off from the base.

Component 27. which is named Herbotheria (and later discussed). is identical lo 32 of the molecu-

lar-based cladogram formulated by McKf.nna. For this we undersland Plesiadapîformes as basal

archontans. component 23 (Fig. 20) standing for Anagalida [sensu STUCKY & McKenna 1993),

and of course component 26 with the four basal condylarths representing LJngulata. .Summarizing,

a total of four nearly identical high rank taxa are reported with the comparison of these two

phylogenelic schemes. This way, both morphological and molecular data support a very similar

scheme of placental mammal interrelationships.

— 440 —

Examining the phylogenetic scheme of Macphee & NOVACEK (1993; 25), here illustrated

in Figure 22b, which is derived from that of NoVACEK ( 1989) after a manipulation of its major

basal polytomy in order to achieve the best optimisation for a sériés of entotympanic characters,

it is clear that the only real incongruence between this branching pattern for placental mammals

and that of McKenna’s (Fig. 22a) is thaï Lipoiyphla is placed as the sisier group of Ferae

(component 7) inslead of sistcr group to the clade composed of Anagalida, Archonta and Un-

gulala. As previously discusscd, this component 30 of McKenna’s .scheme (Fig. 22a), where

Lipolyphla branche.s off from the basal node, is identical to our phylogenetic scheme (Fig. 21).

In that sense we also do not support component 7 (Fig. 22b) proposed by Macphee & Novacek.

Another incongruence between our dental-based branching pattern and the one of Macphee &

Novacek (Fig. 22b) is that, for them Anagalida is the sister group to Ungulata whcreas Anagalida

is sister group to Archonta. Given the scope of this article, it is mosl relevant to note that our

component 27 is identical to the 32 of McKenna's scheme (Fig. 22a) and to component 3 of

Macphee & Novacek (Fig. 22b). It follows from these comparisons thaï we might be indeed

accumulating independent evidence supporting the phylogenetic identity of a new high rank

placental clade, named Herbotheria.

Placental mammal phylogeny is quite a complex subject which awaits better fossil sampling

across the late Cretaceous/Paleocene boundary. This complexity becomes évident by examining

the number of contradictory phylogenetic statements that hâve been published in the last eight

— 441 —

Fig. 22. — The upper two branching diagrams refer lo lhe alpha crystallin A chain based cladogram of McKenna (1992): A. is

directly redrawn from McKenna’s cladogram, it adds only letter L (in bold) for demarcaling the lipotyphlans and lhe Iclter

H (also in bold) for demarcaling the Chiroptera; a, is a simplified version of A. The lower two cladograms refer to the

phylogenetic scheme of Macphee & Novacek (1993); b, is a simplified version of the original B. To facilitate the under-

standing of the componeni différences among McKenna’s phylogenetic scheme and the MaCPHEE & Novacek’s one with

that of Figure 20, their respective simplified derived versions (22a, 22b, and 21) may well be used in the comparison of

these three branching diagrams.

— 442 —

Appendix 1. — Nelsen consensus tree derived directly from the 30 trees obtained after applying the ie* option.

Hypothetical 1

jp 1 Hypothetical 2

|j=^2 Pappotheriim

1-28 fS Prokennalestes

|L-25

Bobolestes

5 Kennalestes

8 Cimolestes

9 Prototomus

6 Leptxctis

7 Scenopagus

10 ProTTlverra

11 Miacls

14 Zalambdalestes

h 17 Protungulatum

23a 18 Dysnoetodon

19 OxypriMMS

11—20 Mima tu ta

|(=12 Purgatoxius

I -21-—13 Palaechthon

=22^

-15 Eosigale

16 Paramys

Appendix 2. — Nelsen consensus tree derived from the 12 trees obtained after applying the ie* and successive weighting options

having ail characters unordered.

Hypothetical 1

Hypothetical 2

2 PappotherluM

3 ProksnnalestBS

Bobolestes

L31I

Kennalestes

Cimolestes

Prototomus

6 Leptictis

14 Zalambdalestes

|[ 10 Proiri verra

11 Miacis

7 Scenopagus

|r^l3 Palaechthon

23^-12

■26j| |j=15

24^^16

Purgatorius

Eosigale

Paramys

Oxyprimus

rr=l 8 Dysnoe todon

21^20

Pro tungula tum

Mima tu ta

— 443 —

years (after the publication of NOVACEK & Wyss 1986), e.g. see articles in Benton (1988) and

SzALAY et al. (1993), This progress report advances some hypothèses of dental irends within

placental mammals, and their alternative paths of evolutionary change are discussed below based

on the branching pattern shown in Figure 20.

DENTAL TRENDS AND THEIR PATHS OE EVOLUTIONARY CHANGE

In this section character State distributions on the branching diagram of Figure 20 are pre-

sented with a discussion of their alternative pathways of evolutionary change. Given the scope

and size of this essay, the transformation .sériés are generally briefly discussed but ail attributes

are considered.

1 . (7 steps. ci 57 - ri 75).

It first appears at node 37. with the placement of character State 1. Thercfore. the partial

contraction of the stylar shelf supports, as expected. the monophyly of Placentalia, represcnted

by component 37. .4t next node 36. State 2 composes a set of four synapomorphies that places

Prokennalestes as the basal placental and ail the oiher taxa forming a large monophyletic clade.

The third appearance of the further derivcd State 3 is placed at component 30 having Zahvnb-

dalestes as sister group of clade 29 (anagalids, including the rodent Paramys, ungulate.s, plcsi-

adapiformcs, plus Leptictis, and the true insectivora - Scenopagux). The furthest derived State

4, where the stylar shelf is considered to be nearly ab.sent, is iiuerpreted to be a synapomorphy

for clade 23. possibly representing the broad notion of anagalids of StuckY & McKenna (1993).

At node 22. there is a reversai from State 3 to State 2. and it cornes to support the monophyly

of Plesiadapiformes or of Archonta if one wishes so: Plesiadapiformes may well be read as a

sort of steni-group of the whole clade Archonta (a matter latcly discussed with Malcolm

McKenna). Another reversai takes place from State 3 to 2 and results in autapomorphy for

Scenopagus.

2. (4 steps, ci 75 - ri 90).

The first trend towards the réduction in height of the cu.sp complex formed by the paracone

and the metacone appears at the ba.sal node 37, State I. Therefore, it is here supported as another

trend of the upper molar morphology related to the emergence of placental mammals. The trans¬

formation to State 2 is placed at node 32. The furthest réduction is placed as an autapomorphy

for Zalambdalestes. Scenopagus is victim of another reversai, from State 2 to State I.

3. (5 steps, ci 60 - ri 86),

The derived State 1 is placed at node 32, marking the first trace of the displacement of the

paracone and the metacone from a full contact, where lhey are somewhat laterally emersed in

one another. At node 30, the further derived State 2 becomes a possible synapomorphy (the above

arbitrarily selected placement), however this State becomes unambiguous only at node 28.

Nodes 30 and 29 can be either represented by States 1 or 2. The character State 3 is placed as

an autapomorphy for Paramys.

— 444

4. (3 steps. ci and ri 100).

It is assigned a trace of réduction of the centrocrista (state 1) at clade 34, where Cimolestes

is rooted at the base. At the lower level 31, a transformation from State 1 to State 2 takes place.

Again, in Paramys, the furthest derived condition (3) is read as an autapomorphy, characterized

by the absence of growth of the centrocrista.

5. (10 steps, ci 40 — ri 64).

This transformation sériés concerns the overall development of the protocone. As expected,

character State 5(1) is placed at node 38. and indeed refers to a basal tribosphenic hierarchical

level, where Pappotherium (the objective outgroup referential includcd in the matrix) is directly

rooted at this node. At the nexi nodes 36 and 37. States 1 and 2 are competing hypothèses for

this character évolution, and at node 35 State 3 is unambiguously placed representing the stage

of growth of the protocone where il bccome.s larger than the paracone. At a lowe.sl level. where

Zalambdalestes is rooted at the basal node (30), the furthest derived State 4 is a potential syn-

apomorphy. However, state 4 becomes unambiguous at node 28, and at nodes 29 and 30, States 3

and 4 are equally parsimonious optimizations. A reversai takes place in Miacis (3 —> 1) and

another in Paramys (4 —> 3).

6. (3 steps. ci 66 - ri 85).

Character state 1, évidence of the transverse expansion of the protoeone here supports the

monophyly of Placentalia (clade 37). State 2 emerges at clade 35 characterizing a further

developed stage of this transverse expansion. A reversai from state 2 towards the less expanded

condition 1 occurs in Miacis.

7. (2 steps, ci and ri 1(X)).

The derived State 1 assigned for the antero-posterior expansion of the labial portion of the

protocone is supporting the monophyly of clade 35, and a transformation to state 2 is determined

to hâve occurred at clade 28, where the insectivora Scenopagus is placed as sister group of

clade 27.

8. (5 steps, ci 60 - ri 85).

The first derived condition of the antero-posterior expansion of the lingual portion of the

protocone is placed at node 32, and the further derived state 2 is supporting the monophyly of

clade 27. Within this clade a reversai towards the less derived state 1 is recorded for Mimatuta

and the différentiation for the furthest derived state 3 occurs in Paramys.

9. (4 steps, ci and ri 100).

The various stages of development of the labial portion of the protocone are determined

to be unordered. Despite the complexity of this character it présents Just one ambiguity at node 36,

where .States 0, I, and 2 are equal optimizations. In Table 2, state 1 is selected on the basis that

the previous node 37 maintains the plesiomorphic condition and the next lower level, clade 35,

State 2 is fixed. This way State 1 would be placed as an intermediate condition between 0 and

2. The transformation to State 3 occurs in clade 29, and the anagalid componenl 23 is .supported

by the furthest derived state 4..

— 445 —

10. (5 steps, ci 40 - ri 80).

The basal possible placement of the derived State 1, regarding the development of the post¬

talon basin, is at node 35. However, it is at node 33 that this condition becomes unambiguously

determined. Despite this ambiguity for the 0 ^ 1 transformation, there is no problem with the

optimization for the following change from States 1 to 2, being assigned to take place at the

hypothetical ancestor represented by node 33.

11. (3 steps, ci and ri 66).

State I is not assigned to any hypothetical ancestor position and the stage where the labial

portion of the post-talon basin reaches the crown upper surface State 2, is determined to be a

synapomorphy for clade 23. This transformation counted two steps given that it has transformed

directly from the plesiomorphic State 0. State 1 stands as an autapomorphy for Mimatuta.

12. (4 steps, ci 50 - ri 86).

This character optimization assigns for the 0 —> 1 character change (appearance of the pre-

talon basin) at the hypothetical ancestor 33, and the transformation from State 1 to State 2 sup¬

ports the rnonophyly of clade 27. A reversai to the plesiomorphic condition occurs in

Zalambdalestes.

13. (3 steps. ci 66 - ri 80).

The evolutionary pathways of this character are limited to clade 25, where the derived State 1

regarding the position of the labial portion of the pre-talon basin is indicated to be supporting

the rnonophyly of this whole clade. The further derived transformation State 2 is assigned to be

placed at hypothetical ancestor 23. State I is homoplastic for the appearance in Mimatuta, re-

sulting thcrefore in an autapomorphy for this genus.

14. (3 steps, ci 33 - ri 0).

The distribution of this character results in ambiguities regarding nodes 31 and 32. There

are two equally parsimonious hypothèses for this character optimization: one that places State 1

as a synapomorphy of clade 32 and consequently with a reversai to the plesiomorpf\ic State at

node 30; and the other interpreiing two independeiit appearances of State 1, one in Miaris and

another in Proviverra. A character step is aiso counted for the derived condition (autapomorphy)

of Mimatuta.

15. (8 steps, ci 40 - ri 76),

This feature concems the hypocone development and its distribution. The basal possible

placement for the first derived condition 1 is at node 33, but it is an ambiguous position since

State 0 is an alternative hypothesis. It becomes unequivocal at node 29 and the further derived

State 2 appears as a synapomorphy for clade 27.. Within this clade a reversai to the less derived

State 1 takes place at node 21, the one indicating the sister-group relationship between Pro-

timgulatim and Mimatuta.

16. The development of the pericone stands here as an autapomorphy for Scenopagus, and there-

fore it is excludcd from the data matrix here analyzed.

17. (4 steps, ci and ri 50).

The first stage of development of the conules is considered as a synapomorphy of clade 37,

State I. This is one of the cases where better refined knowledge of the Placentalia sister-group

— 446 —

relationships could change completely the interprétation of this character State optimizations.

The main remaining question i.s whether the dental similarilies bctween metatherians and eutheri-

ans are due to convergence or common inheritance. The derived State 2 cornes out as an au-

tapomorphy for the condylarth Oxyprinuts. Zulambdalestes and Eoxigale seem to hâve

independently lost the conules. but it is questionable wethet they do not hâve the conules in

very rudimcntary stage of development.

18. (3 steps, ci and ri 100).

This transformation sériés concems the évolution of the shape of the conules. The shape

coded as 1 is indicated to be a synapomorphy for clade 36 (from where Bobokstes is rooted at

the base) and the one coded as 3 is placed at clade 24, with a reversai to condition 1 in Oxy-

primus. There are two alternative paths of evolutionary change regarding the character state 2,

one where it would be placed at node 23 and thereforc rcad a.s a synapomorphy for the anagalid

clade, or as an autopomorphy for Paramys. This is because Eoxigale is coded a.s missing data

(?).

19. (8 steps, ci 37 - ri 50).

The position of the conules relative to the paracone and metacone allows a diverse number

of competing hypothèses of character optimizations. Therefore, the analysis of this character is

limited to the following relevant points. First State 1 is unambiguously placed at node 37, there¬

fore having the status of synapomorphy for ail taxa eonstdered. But at the next nodes 36 and

35 States I and 2 are competing hypothesis, and at the node 34 state 2 becomes unequivocal.

Character state I reappears at node 24 as an alternative path of evolutionary change, or as a

potential shared derived character for the monophyly of the clade that places Dysnoatodon, Pro-

tungulaium and Mimatuta. Character state 3 is an homoplastic character appearing as autapo-

morphies for Cimolestes, Proviverra and Eosigale.

20. The displacement of the conules of the post-protocrista and pre-protocrista is interpreted as

a uniquely derived character state for Paramys, and therefore this is another character excluded

from the performed analysis of parsimony.

21. U step, ci and ri lOO).

The full contact of the very lingual portion of the paraconule and metaconule with the

protocone cornes out in this analysis as the only synapomorphy for ungulates (clade 26).

22. (3 steps, ci 33 - ri 0).

This is another case where the monophyly of the feraeans would better accommodate this

character distribution. At nodes .14 to 31, both the derived shearing carnassial surface and the

primitive one are assigned as possible alternatives. With this scénario it is not worth analyzing

ail the equally parsitnonious character State arrangements. It is to if the monophyly of feraeans

is assumed theii the optimization of this character would be quite different, possibly having the

derived State at the basal position within Ferae.

23. (3 steps, ci 66 - ri 50).

This is another trend for the carnassial shearing surface, and it is placed as a synapomorphy

for component 33, with a reversai to the primitive condition at node 30. The further derived

State 2 cornes out as an autapomorphy for Prototomus.

— 447 —

24. (1 step. ci and ri 100).

The derived torsion of the paracone-metacone région, State 1, is a synapomorphy for the

plesiadapiformes (clade 22).

25. The derived character State I is an autapomorphy for Paramys and therefore is not taken

into considération in the analy.sis of parsimony performed.

26. (5 steps, ci 60 - ri 66).

This is ànolher feature considered to be unordered, and its State distributions are complex.

The really interesting information is that ail the derived States range within clade 27. This sug-

gests that a general bunodont trend may be homologous throughout those taxa of clade 27. This

would be the case if States 1 and 2 were considered as one single condition, it would then resuit

in another synapomorphy for clade 27.

ROOTING THE UNGULATES

The Grandorder Ungulata, represented by the four basal condylarths, is as a monophyletic

clade (26). The single upper molar character that is a synapomorphy for ungulates is the torsion

of the very lingual portion of the conules resulting in full contact with the protocone labial wall

(character 21, Fig. 17). However, SALLES (1994) discussing an earlier version of the data matrix

here analyzed, pointed out that Oxyprimus might be a sister-taxon of anagalids (including rodenls)

plus plesiadapiformes, an assemblage équivalent to clade 25. The general similarity of the upper

molar of Oxyprimus and Pur/^acorius, espccially in wcar pattern on the paracone and metacone

is striking. This might indicate that not necessarily ail upper Cretaceou.s/Paleocene condylarths

are distributed within Ungulata. Or even, that basal condylarths might not be strictly related to

the emergence of ungulates but aiso to other large groups of placental mammals. Following this

view it is not surprising thaï Oxyprimus is pluced at the basal node as sister*iaxon of the other

three condylarths (clade 24). The synapomorphy of this clade 24 is stale 3 of the character 18,

which concems the bunodont shape of the conules.

The dental features frequently referred to as characteristic of ungulates, their general buno¬

dont squarish shape, are here indicated to be in fact related to more basal level relationships

within clade 27. Clade 27 supports the broad notion of anagalids (Stucky & McKenna 1993)

plus the ple.siadapiformes composing a monophyletic group with the ungulates. The three

character siales supporting the monophyly of clade 27 18(2), 12(2), and 27(1)) are related to a

general bunodont dental pattern. The reseniblance of the molar moqthology between early rodents

and primates with basal condylarths is an old subject in the mammal literature, extending back

at leasl to the latc nineteenth century (e.g. COPH 1883; Mattiiew 1897) and the early décades

of this century (e.g. OSBORN 1902). More recently. SZALAY (1963) and Cartmhx (1974) hâve

recognized the.se affinities but assumed that tliey were ail due to adaptive convergences, and,

as earlier discussed, McKenna (1992) and MacpheE & NOVACEK (1993) published phylogenetic

schemes placing ungulates together with anagalids {sensu lato) and archontans (primates). As

mentioned above, other ongoing research, also conceming the maslicalory apparatus, hâve as

their preliminary phylogenetic conclusions the plesiadapiformes as sister group of the rodents

— 448 —

plus anagalids. In the light of the general placental dental patterns represented in Figure 20 it

seems an attractive idea to see the plesiadapiformes representing basal archontans rather than

just primates. This would be then corroborating the idea of reading component 27 as an assem¬

blage comprising Archonta, Anagalida and Ungulata. But, the underlying question is whether

the similarities observed are indeed homologies, or due to convergences for a herbivorous diet.

This question is widely answered ba.sed on a common view that places insectivore-like placentals

as basal to nearly ail major placental clades. Therefore, this question is a priori answered as

due to cases of parallel évolution. It .seems that it is frequently taken for granted that clearly

environmentally corrclated characters are generally the results of independent evolutionary his¬

toriés, and so the possibility of these type of characters are représentative cases of “adaptative

synapomorphies” is discarded without proper examination.

Based on the hypothesis of monophyly of clade 27, including the interrelationships there

proposed, and also on the placement of an additional placental mammal, the genus Tribospheno-

mys, Meng et al. 1994, is a sister-group of Paramys (this resuit is obtained after applying the

combined option of branching swapping mhennig’'' bb* of the Hennig86 based on the same data

matrix and adding a new character). We argue that the very basal history of clade 27 is charac-

terized by primitive bunodont (condylarth-like) transformations. In an ontological interprétation,

it is hypothesized lhat some sort of placental “herbivores” may hâve indeed their very early

history unfolding as part of a monophyletic plexus of phylogenetically related organisms, there¬

fore supporting the view that cases of "herbivore-like" parallel évolution do indeed happen but

at more advanced stages of bunodonty in already differentiaicd branches of clade 27 In other

words, the very first steps towards the bunodont transformation characterizing a condylarth-like dental

pattern are considered to be evidence for tire monophyly of an assemblage of placental mammals

here represented by clade 27, and named Magnorder Herbotheria, and that later, more differentiated

bunodont features corne to be often homoplastic among members of this high taxon. The clade

Herbotheria is discussed also in another article in préparation. But. shortly, it could be provisionally

placed in a classification derived from that one of Stucky & McKenna (1993) and based on the

phylogenetic scheme of Figure 20 (and not taking into considération node 24), as follows:

Supercohort

Cohort

Subcohort

Magnorder

Grandorder

PLACENTALIA Owen, 1837

EPITHERIA McKenna. 1975

PBEPTOTHERIA McKenna, 1975 new rank

HERBOTHERIA new taxon

ANAGALIDA Szalay & McKenna, 1971

ARCHONTA Gregory, 1910

UNGULATA Linnaeus, 1766

We are fully aware of the preliminary nature of this classification, specially concerning the

fact that not ail major groups of Archonta are considered and also that only one morphological

complex is examined. However, we hâve reasons to advance that the new Magnorder Herbotheria

seems to be an idea that indeed carries potential phylogenetic information content.

— 449 —

CONCLUDING REMARKS

First, in order to better understand placental interrelationships, it is clearly necessary to

add new recently described fossil taxa finds across lhe Late Cretaceous/Tcrtiary boundary. Re-

garding our basal ungulate research programme, it is aiso évident that for achieving progress it

will be necessary to cover a broader spectrum of the condylarth phase morphology. In that sense,

a more complété detailed cladistic eharacter analysis of the whole masticatory apparatus for

these basal placental marnmals is an enterprise waiting to be accomplished.

Readdressing the earlier question regarding whether or not the molar morphology is a good

source of phylogenetic information, it is clear that the fully resolved hierarchical pattern here

encountercd for a set of considerably diverse placentals speaks for itself, thus giving a full yes

to this question.

To conclude, the phylogenetic relationships of the placental mammal branching diagram

(Fig. 20) put forward are generally supported by the different options of parsimony analysis

other than the one performed. Hence, despite the preliminary nature of these results, it secms

that the phylogenetic patterns proposed might be indeed representing some historical aspects of

the placental mammal évolution, and we understand that the clade Herbotheria is possibly the

most interesting outcome of this study.

Acknowledgements

During my last trip to the American Muséum of Natural History (AMNH), in the winter of 1994,

this study began as part of my Ph.D project regarding basal ungulate interrelationships. Motivated by

discussions with Malcolm McKenna and .Fin Meng, il became clear to me that an attempi to reexamine

phylogenetic matters concerning the molar morphology of placental mammal would be worlh doing. First,

I would like to thank Richard Tedford for allowing me to study the fossil mammal collections of the

AMNH and the two researchers mentioned from this same institution for kindly cxchanging their idcas

with me. As a formai doctoral sludent of the Université de Paris VTI (Formation doctorale A. I>H RlCQLfeS),

I am also very gratefui to Philippe Taquet for having opened to me the opportunity of developing re.search

programs on mammal systematics at lhe Laboratoire de Paléontologie of the Muséum national d'Histoire

naturelle (MNHN). I specially acknowledge the tinte and suggestions of Daniel Goujet and Etonald Rus¬

sell, who logether with Denise Sigogneau-Russell permitled me to hâve access to ihcir fossil mammal

cast collection. This manuscript was greally improved by discussions with my adviser Pascal Tassy, and

critical reviews by Jerry HoOKER to whoni 1 am e.specialy indcbied, Malcolm McKenna, Pascal TAS.SV,

Anwar Janoo, and Mario DE Pinna. Financial support was providcd by the Conselho Nacional de Desen-

volvimento Cietifico e Tecnolôgico (CNPq), Brazilian Fédéral Government, and an auxiliary funding was

received from the Laboratoire de Paléontologie (URA 12, CNRS) of the MNHN and by the Université de

Paris VII.

Manuscript submitted for publication on 27 July 1994: accepted on 22 October 1995.

— 450 —

REFERENCES

Archibald D. in press. — Archaic ungulates ("Condylarthra") (85 p. ms.). In C. JANIS, L. JACOBS & K. SCOTT

(ed.s). Tertiarx Mcimmals qf Norlh America. Cambridge Univcrsity Press.

BentüN M. J. 1988. — The I‘hyh>iit'iiy and Classification of the Tetrapods. Volume 2. MammaLs. Clarendon

Press. Oxford. Eiiglaiid. .529 p.

CarphntI'R J. M. 1988. Choosing arnona mullistale equally parsimonious cladograms. Cladistics 4 (3): 291-

29b.

Ciini-Ll R. 1993. — Theria of Mctalherian-Rutherian Grade and ihe origin of Marsupials. ht F. Szalay, M.

NovaceK & .M. .McKenna (eds.). Mammal Phylngeny. Placenlals. .Springer-Verlag Press, 2: 205-215.

DaRUL' p. & TaSSY p. 1993. — Lu reconstruction phylogénétique : concepts et méthodes. Masson, Paris, 245 p.

FaRRIS .1. S. 1988. — Hentiig86 (version 1.5) software. Distributed by the author. New York: 41 Admirai Street,

Port Jefferson Station 11776.

GlNOERlCIl P. D. & GCNNELL G. F. 1979. — Systematies and évolution of the genus Esthony.x (Mammalia,

Tillodontia) in the early Eocene of North America. University of Michigan Muscum of Paleontology Con-

irihiuions 25: 125-153,

McKenna M. C. 1975. — Toward a phylogenetic classification of the Mammalia. In W. P. LOCKETT & F. S.

SZALAY (eds.l. Phy/ogeny oj the Primates: a Multidisciplinary Approach. Plénum Press, New York.

— 1992.— The alpha crystallin A Chain of lhe eye lens and mammalian phylogeny. Annales Zoologici Fennici

28: 349-360.

MacphEE R. D. E. & Novacek m. J. 1993. — Définition and relatiunships of Lipotyphla. In F. S/.ALAY, M.

NOVACI.K & M. McKenna (eds.). Mammal Phylngeny, Placvntah. Springer-Verlag Press, 2: 13-31.

MaR.S1I.aLL L, g. & Ki|-.I.aN-JaWOROWSKA Z- 1992. — Relationships of the dog-like marsupials, dellatheroidans

and early tribusphcnic mammals. Lethaia 25 (4); 353-460.

Marsii.all L. G. & OE MuiZON C. 1992. — Allas photographique (MEB) de Metatheria et de quelque Eutheria

du Paléixènc intérieur de la Formation Santa I.ucia à Tiupanipa (Bolivie). Bulletin du Muséum national

d'ilisloire nalurelle, Paris. 4"’' série. 14, C 1 ; 63-91.

Novacek M. J. 1992, — Fossils. topologies, missing data, and the higher level phylogeny of eutherian mammals.

Systenuitu Biology 41 11): 58-73.

Novacek M. J.. Wv.ss A. & McKenna. M. C. 1988. — The major groups of eutherian mammals. In M. J.

BENTüN (ed.). The Phylogeny and Classification of the Tetrapods, Mammals. Claredon Press, Oxford, Eng-

land 2: 31-72,

ProTHERO d.. Manning E. & FI.SCHER. M. 1988. — The phylogeny of ungulates. In M. J. BENTON (ed.). The

Phylogeny and Classification of the Tetrapods, Mammals. Claredon Press, Oxford, England 2: 201-234.

Salles L. O. 1992. — Felid phylugenetics: extant taxa and skull morphology (Felidae, Aeluroidea). American

Muséum Novitates 3047, 67 p.

— 1994. — Rooling ungulates within placental mammals: Upper Cretaceous/Paleocene fossil record. Progress

report in the T II .‘Hmposio .sobre o Cretàceo do Brasil". Rio Claro, Brazil: 109-111.

STI'CKV & McKenna. M. C. 1993. — Mammalia. In M. J. BENTON (ed.l. The Fossil Record II. Chapman &

Hall, London: 7.39-771.

SWOH'ORD D. L. & MAODISON W. P. 1987. — Reconstructing anc-e.stral character under Wagner parsimony.

Maihemallcal Hiosclences 87: 199-229.

SZAL,\V F., Novacek M. & McKenna m. 1993. — Mammal Phylogeny, Placenlals. Springer-Verlag Press, 1,

.321 p.

Van Vai.EN L. 1978. The beginning of âge mammals. Evolutionary Theory 4: 45-80.

Zhang Y. 1980. A new tillodonl-like mammal front the Paleocene of Nanxiong Basin, Guangdong. Vertebrata

PalAsiatica 18 (2); 126-130.

Bulletin du Muséum national d’Histoire naturelle, Paris, 4® sér., 18, 1996

Section C, 2-3 : 451-516

Earth Expansion, Plate Tectonics and Gaia’s Puise

by Martin PICKFORD

“Great spirits hâve always encountered violent opposition froin mé¬

diocre minds.”

Albert Einstein

“Plate Tectonics on the globe is essentially the same as on a plane.”

A. Cox & R. Hart 1986

“Because the Earth is a near-spherical body and a map is projected

on a liai sheet of paper, distortion is inévitable.”

H. OWEN 1983

“Uniformilarianism loday: Plate Tectonics is the key to the past.”

B. F. WiNDLKY 1993

“Plate Tectonics (PT) includes many points of contrary evidence that

hâve been bypassed or ignored."

M. L. Keith 1993

Abstract. — There are four possibilities concerning the size of the Earth, which impinge on ail théories

of its global-scale tectonic évolution - a) ils radius has reniained con.stant through géologie time (the constant

“r" hypothesis underpms much past and curreni thought lo the Theory of Plate Tectonics); h) its radius has

increased wilh the passage of géologie lime (the increasing “r" hypothesis is the main proposition of the Expandiiig

Earth Theoiy); its radius has decreased wilh the ptussage of geological time (an early idea put forward lo e.xplain

the existence of the world's fold mountain belis); and d) its radius has lluctualed over geological lime, sometinies

increasing, sometinies decreasing (an idea put forward lo explain the existence of both compressional and lensional

structures in the Earth's enist).

The firsl two part.s of this paper dcal with iwo of the above conipeting théories of global tectonics - a)

the Plate Tectonics Theory in a globe of constant "r", and b) the Expanding Earth Theory in which "r" has

increased over géologie time. Exainination of thèse théories indicates that there arc serions difficulties in recon-

ciling the availabic geological data wilh the Plate Tectonics paradigm in which it is almosi uiiiversally assumed

that the globe ha.s reinained constant in size and the océans constant in volume through geological time. Many

of the inconsisiencies disappear if lhe saine d.ain are viewed within a model of an expanding Eanh, as long as

appropriatdy sized globes are used for plotting up the data.

The third seciion of lhe paper examines lhe periodicity of evenls in the liihosphcrc, hydrosphère, atmosphère

and biosphère and relates iliese to episodic change.s in the orientation of the axis of rotation of Earth which

induce changes in lhe .Sun/F.arth relationship and lhe development of tremendous stresses and strains in ihc fabne

of the globe, as well as other effecLs. The causes of siich changes are ihelniselves related lo alterations of the

Earth’s fabric during Earth Expansion, principally due lo unevenly distribtited generatuin of oceanic cru.st at

spreading ridges, which induccs imbalancc in the global gyro.scope. The third part aiso looks at soine of lhe

secular changes thaï hâve occurred in varions Earth Systems, including the hiosphere. hydro.sphere and atmosphère.

The paper ends with several major conclusions which attempt to explain the evidently close relationship

that existed between events in lhe lithosphère, hydrosphère, atmosphère and biosphère.

Key-words. — Global tectonics. Plate Tectonics, Earth Expansion, Endogenetic energy. Exogenetic energy.

— 452 —

[.'Expansion terrestre, la Tectonique des plaques et les pulsations de Gaïa

Résumé. — Plusieurs géologues ont déclaré s’interroger à propos de la théorie de la Tectonique des plaques

sur une Terre à rayon constant. Ils formulent une série de questions sur les sources d’énergie et les mécanismes

de déplacement des plaques qui leur paraissent rester sans réponses. Les cellules de convexion du manteau terrestre

animées par l'énergie géothermique, mécanisme retenu par les tenants de la théorie de la Tectonique globale, ne

constituent pas. ti leurs yeux, un mécanisme convuincant pour expliquer le mouvement de.s plaques. En effet,

l'énergie géolhenriique est beaucoup trop faible pour accomplir les mouvements envisages. Il existe, de plus,

beaucoup de faits de divers ordres qui réfutent cette théorie. Ainsi, des problèmes de cartographie, des solutions

parochiales dans deux dimensions de l'espace enfin de.s solutions contradictoires émises sur le mouvement des

plaques. D’autres géologues ont envisagé des mécanismes différents (la tectonique de membranes et le flux de

liquides) pour expliquer la Tetfonique globtile, mais leurs propositions .sont faites dan.s le cadre d'une Terre à

rayon constant L'n nombie croissant de géologues considère la possibilité d’une Terre en expansion au cours

des temps géologiques, et envisage qu’elle ail joué un rôle majeur dans la Tectonique globale.

L’E.xpansion de la Terre n a pas eu une expression uniforme à .sa surface comme le montre la largeur et

la répartition des rides d’expansion océanique. Ainsi, seuls 17% de ces rides d’expan.sion sont répartis au nord

du Grand Cercle constitué par les chaînes téihysiennes. tandis que 815% le son! au sein et au sud de ce cercle.

Cela signifie que l’expansion de l'hémisphère sud a été plus forte que celle de l’hémisphère nord, et que si la

gravité avait été nulle, alors la terre aurait dû se gonfler inégalement. Mats Tactkm de la gravité, source d’énergie

terrestre la plus pui.ssante. a maintenu la forme sphérique du globe tene.strc. en conséquence de quoi, elle a

donné naissance aux chaînes omgéniqiies et aux fosses oeéatiiques qui sont alignées plus ou moins selon un

grand cercle. Cela serait dû au Fait que faire d un hémisphère croît pendant que son périmètre décroît, engendrant

ainsi une compres.sion le long de la zone de contact entre les hémisphères (le Grand Cercle). Une E.xpansion

terrestre qui n’est pas uniforme reiidran compte, d'une piirt. des anomalies dont la Tectonique des plaques est

incapable d'expliquer foiigine et. d'autre part, des données géologiques telles que. I) le rapport i.sotopique

d’béliurn dit „ eosmic n des rides médio-otéaniqiies. 2) la bais.se des puléotempéralures des océun.s, 3) la contem¬

poranéité des phases teetogéniques et celle de l'.ieiivité des rifts, les flux de basaltes, les irrégularités du dépla¬

cement des points chauds, l'existence des rides d’hxpansion océanique abandonnées et d'autres faits encore.

Dans le cadre de l’hypothèse de l’Expansion terrestre, la Tectonique globale résulterait de l’effet réciproque de

l’énergie, de l’expansion et de la gravité.

Deux des mécanismes possibles de l’Expansion terrestre sont discutés ; I ) un noyau terrestre aux propriétés

d’un plasma. 2) une Terre hydrique ides protons disséminés dans des lartices mcialllques). Ces deux mécanismes

peus'ent rendre compte de I Expansion terrestre sans qu'il y an eu une augmentation de la masse de la Terre.

Le mécanisme envisagé est le suivant : le plasma (ou la Terre hydrique) s’altère progressivement en libérant des

protons qui se lient avec des élections pour oonstituer des atomes, un processus qui s’.tccompagne d’un fort

accroissement du volume sans ehangemenl de masse globale.

Lu Itoisième partie de notre .uficle liaile du comportement par à-coups de la planète uin.si que des chan¬

gements séculaires. Au sein du premier d’entre eux, on dénombre les phases leetogéniques. les pulsations des

rifts et les éruptions .sporadiques de flux basaltiques ; tous semblent être liés au cours des lemps géologiques et

avoir été engendrés par les mêmes causes. Au même moment se déroulent dans la biosphère, des événements

majeurs comme le* extinctions globales. Parmi les changements séculaires envisagés, ceux de l'obliquité de la

planète et de la position de son axe de rotation ont dû joué chacun un rôle important en modifiant la périodicité

des saisons et le climat mondial, ce qui s’est traduit par des changements majeurs sur la biosphère.

Des conclusions, qui tentent d’expirqiier les relations étroites qui unissent les événements de la lithosphère,

de l'hydrosphère, de l'atmosphère et de la biosphère, sont tirées.

Mois-cics. — Expansion terrestre. Tectonique des plaques, phases leetogéniques, sources d’énergies endo¬

gènes, ceintures orogéniques. Grand Cercle, réorientation de l’axe de rotation lerreslre, paléotempératures océa¬

niques, ceintures écoclimatiques, obliquité de l’axe de rotation terrestre.

M. PicKoiKu. Muséum uiiliouat tl'Histmrr uutun'lle. ÏMhtiruloin’ ilt' jutléiutifiln^ie, à rue lie Bujfon, f-'-752.il Paris,' Collège de

Franee, Il Place Murcelliu Bertheiat, F-JMXIF Paris cedex OF: Cealogical Surver of Namibia. PO. Bo.x 2168. Windhoek.

yooo Namibia,

— 453 —

AIM

The aim of this paper is to examine the history of varions Earth Systems (lithosphère, hy¬

drosphère, atmosphère, biosphère) for evidence concerning the rhythms, modes and trends of

change that they havc undergone since the Crctaccous. An understanding of these Systems may

throw some light on the évolution of the Earth's lithosphère and biosphère during the Cainozoic,

which may thereby Icad to a better understanding of the close historical relationship that existed

between the Earth’s crust and the biosphère.

INTRODUCTION

For the past three décades. Plate Tectonics has been the prédominant theory in global geo-

tectonics. The Theory of Plate Tectonics resl on several basic assumptions. the validity of which

hâve been questioned. One of these assumptions is that the Earth has remained the same size

since the Proterozoïc. This assumption has been called into question on several occasions but

the majority of geotechnicians hâve refused to respond to the challenges. A second is that the

plates are tortionally rigid, an assumption that has heen succesfully challenged to the extend

that many geotechnicians no longer consider it valid. A third assumption is that the volume of

the hydrosphère has remained constant since the Proterozoïc. And there are several more which

are seldom questioned such as the ideas that obliquity has not changed and the pôle of rotation

has been "fixed" throughoui geological time. A furiher assumption is thaï plate.s are driven by

convection cells in the mantle, of which ridgc push, slab pull, deviatoric tcnsional stresses and

other forces arc considered to be manifestations,

This paper examines several of these assumptions in detail. Il also looks ut many of the

contentious aspects of Plate Tectonics. including the apparent lack of sufficient cnergy to power

the plate tectonic “motor". This and many anomalies that hâve been noted by researchers indicate

thaï sometliing is serioitsly wrong with the ba.sic assumptions of Plate Tectonics. It is concluded

that most of these are invalid. In particular. the assumption that the Earth has remained constant

in size is seriousiy challenged by a variety of data which indicaie that the radius of the Earth

has increased during the Phancrozoic. This conclusion has lcd to the proposition of a theory of

global geotectonics called “Expanding Earth Hypothesis". There are two versions of the latler

hypothesis - “fast expansion" and "low expansion". The trajeclory of continental masses as en-

visaged in the Plate Tectonics and Earth Expansion hypothèses are summarized in Figure I.

FRAMES OF REFERENCE

A troubling problem in global geotectonics, and one thaï is not often addressed adequately,

is the définition of frames of référencé which apply over geological time seules. In fact there

do not seem to exist any “absolute” long term frames of référencé.

— 454 —

Fig. I. — Comparison of threc Ihoories of global tcclonics. A. Plate Tectonicn in a globe ot' constant "r”; B. vSlow Earlh Expansion

(inasstc globe XIW' of modem globe; C. fast F.anh Expansion (Cambrian globe 60% of modem globe). In A. ail movements

of continents tarrows betow darfc masses) arc liorir.onlal (small eartoon above globe shows no radial veetor of movenient).

B. movement of continents i.s muiiily radial jway Irom ihe centre of the Earth but lhere is room for sonie horizontal veclors

of movement (arrowt below dartc masses representing connmenLs)(small cjrtoons above globe shows thaï a combinaison of

radial and hoii/onial movemenis lias occured). C, ail movemenis of continents is considered lo be radial tarrows below dark

musses) (eartoon above globe shows no horizontal veetor of movement)

lu Plate Tectonics analyses, a common practice has been to select one or other of the con¬

tinents, often Africa or South America, as a référencé, and to relate other plate movements to

it. But since the break-up of Pangaea, Africa has moved northwards with respect to the présent

day South Pôle, so such solutions are not “accurate” with respect to any global frame of référencé

such as the présent day latitude/longitude graticule. Inspection of any sériés of palaeogeographic

maps, including cartographically accurate ones. such as those of OwEN (1983a), will highlight

the problem.

The magnetic dipole changes position on a daily basis, but is often assumed to average out

over geological time to provide a notion of where “true north" and “true south" were located

at any desired period in the pasl. This is despite the tact that at présent magnetic north is inclined

at some 14" from the rolalional North Pôle.

The hot-spot (and wet-.spot) frame of référencé has been in use for more than a decade,

but the crustal hot-spoi tracks relatcd to this System show kinks and dog-legs in them, indicating

that the outer layers of the Earth seem to hâve been moving indcpendently of the inner layers,

at least during some geological periods. Fuithermore two of the prime examples of hot-spots

-the Cook-Austral chain and the Marquesas chain— show gross violations of the simple age-

dislance relationship which is supposed to characterize hot-spot tracks (KF.iTFt 199.3).

Ail this means that the search for an absolutc long tcrm frame of référencé for the entire

globe and its contents is probably a futile one.

Lambeck (1979) discussed the question of trames of référencé in some detail and showed

that tectonic motions of the continents or plates cannot be separated from any drift of the axial

dipole. The marine magnetic anomalies, which permit the relative motion of the plates to be

— 455

detemiined throughoul the Cainozoic and Mesozoic do not solve the problem, even if one adopts

an additional frame of référencé such as the hot-spot frame, or by minimizing the motions of

plate margins. The reason is that “to fix any one plate requires three parameters and, if there

are N moviog plates and a wandering pôle, there will be a total of (3N-l)+2 unknowns (if one

longitude is arbitrarily fixed) for any one geological epoch. As independent observations we

hâve the relative motions between the plates and the position of the apparent pôle relative to

one plate, or a total of 2+3(N-l). Thu.s the number of unknowns aiways exceeds the number of

observations. A unique solution is impossible.'' (LA.MBECK 1979: 76)

Under these circumstances, it appears to be absurd to claim that. over geological time pe-

riods, it is impossible for the Earth's axis of rotation to expérience reorientation, either within

the Earth or with respect to the Universe. The Earth's axis of rotation is not fixed with respect

to ail the matier comprising the crust. nor is it fixed with respect to ail the matter making up

the mantle, and it would be surprising if it were found to be fixed with respect to the core.

Even more surprising would it be to fmd that it is fixed with respect to an exlemal frame of

référencé such as the plane of the ecliptic or any other celestial reference. if only because the

entire universe has itself been evolving along with ail the matter in it (Lasker et ai 1993,

LaSKER & ROBUTEI. 1993: Vf.RMEERSEN & Vlaar 1993; WII.LIAMS 1993). However, the “belief”

that the axis of rotation of Earth cannot change appears to be remarkably deepiy ingrained into

the conscience of many gcoscientists.

In the final analysis, there is no idéal solution to the question of trames of reference. but

we should not be straighi-jacketed into thinking that any frame of reference, including the axis

of rotation of the Earth. is fixed. The Earth is not a school globe fixed to a set of gymbols. It

has no gymbols at ail, except those imaginary ones that we give it for mathcmatical and geo-

physical purposes, A fixed or permanent axis of rotation is as imaginary as Euler pôles arc.

METHODS

Carey (1983) and OWEN (1983a. b) hâve highlighted some of the pitfalls of working on

fiat sheets of paper when dealing with structures that are disposed over the surface of a globe.

For example, on a globe, pi, the ratio between the radius of a circle and ils circumference, is

not a constant. On a fiat shect of paper it is. On curved and enclosed surfaces, this gives rise

to some peculiar, even counler-intuitive geometry, which has fooled more than one geotectonician

during the past thirty ycars. On fiat surfaces, the angles of a triangle always add up to I8()'\

On a sphère they can add up to any value between 180" and 540". Draw a circle on a globe,

and then try to represent it on a fiat sheet of paper. Depending upon the projection used and

the position of the circle on the projection, a circle on a globe can take on a formidable array

of forms on a sheet of paper ranging from a circle through an enlire family of ellipses and other

closed shapes to a straight line. Distortion of shape, size (or scale) and rclationships between

neighbouring entilies is the usual resuit of depicting on a fiat sheet of paper the forms seen on

a globe. A scale bar which can be used anywhere on a globe is almost worthless on a fiat plan

of the globe. On a globe the north arrow of a compass always points dircctly towards the North

Great Circle

belt 30" wide

\e>’

^ 0 ‘

— 457 —

MOR and Conservative

plate margin earthquakes

Fold belt and Trench

Earthquakes

Isolated Earthquakes

Fig. 2. — Modern globe^' showing cpicentres of earthquakes of magnitude greater than 4.5 in the period l963-1977'^ the major

plates and the two great circle belts (30” wide) in which occur the world’s fold mountain belts, most of the Earth’s subduction

zones, and most of ils comprcssional earthquakes and volcanoes. The Tethyan Great Cercle belt coniains the Atlas, Pyrenees

and Alpine>Himalayan fold mountain belt and the Sumatra-Soloman subduction zone and is characterized by sinistral torsion.

The Cordilleran Great Cercle belt contains the Rockies and Andes fold mountains belt. and most of the circum-Pacific sub¬

duction zones. It has dextral torsion.

— 458 —

magnetic Pôle. On a flal map of lhe world it .seidnm points direclly at the magnetic North Pôle

but at various angles front it. the "direct" line being depicted by a curved line.

There are many other “surprises” for scientists who study global problems by working on

fiat sheets of paper. For example, we take an ordinary pair of compassés and a globe. We fix

one of the points of lhe compass at a point on the surface of lhe globe (for in.stance at the

North Pôle) and start to draw concentric circles on its surface, bcginning with the points of the

compass close together and ending with ils points far apart. Initially lhe circumfercnce of the

circles being drawn and the parts of the globe’s surface enclosed by them both increase in size,

but we will soon reach a stage — called a gréai circle - beyond which the circumference of the

circles being drawn begins to get smaller, even though the part of lhe globe’s surface enclosed

by the circles continues to increase. We will eventually reach a stage of the exercise where the

circumference of the circle being drawn becomes zéro - the drawing point of the compass has

arrived at the South Pôle - but the entire globe's surface area is enclosed by il. Continue to

increase the distance between the fixed point of the compass and its drawing point, and you

will fail to make any mark on the globe.

This “thoughi experiment” might at firsi glance appear lo hâve little praclical use, but con-

sider the Pacific Océan, the surface area of which is still increasing. Its rim already approximates

a gréai circle. Whai will happen if the surface area of the Pacific Océan continues to augment

by addition of crust at .spreading ridges. as it surely will ? On a constant diameter Earth, its

perimeter will decrease in length, and this will hâve many geoicctonic implications.

On a globe, the two areas on either side of a closed perimeter add up to a constant. This

remains Irue no matter what the size and shape of the perimeterare. The Pacific Rim has increased

in length by about a third of a great circle since the days of Pangaea at the onset of the Jurassic.

On a globe of constant dimensions this would only be possible if the Pacific Rim was shorter

than a great circle ai the time of Pangaea. There are two possibilities on such a globe: either

Pangaea occupied an area less than a hemisphere while the proto-Pacific covered an area greater

than a hemisphere. or Pangaea was greater than a hemisphere while the proto-Pacific was smaller

than a hemisphere. Ail reconstructions of Pangaea show lhat it covered slightly more than a

hemisphere of a globe of today's dimensions (CARbY 1988, fig. 29). Thus,. on a globe of constant

dimensions, the Pacific Rim could only get longer if the pioto-Pacific (or Eopacific as it is

sometimes called) increased in size al lhe expense of Pangaea The proto-Pacific has indeed got

larger since the days of Pangaea - the océan floor palaeomagnetic data shows this ampty. But

the proto-Pacific did not get larger at the expense of Pangaea. During lhe same period the Pangaea

side of lhe Earth has also greatly increased in area by lhe iasertion of the Atlantic. Indian,

Arciic and Southern Océans bciwcen ils varions continental fragments. Someihing musi be amiss

with the assumption of constant “r”.

METHODOLOGICAL PROBLEMS

Two further very real methodological problems hâve been the tendency for geotectonicians

to Work 1) on planar maps of the world rather than on globes and 2) on restricted parts of the

world, rather than examining the globe as a whole. One resuit has been that solutions to

— 459 —

geotectonic problems proposed for a particular area often contradict solutions proposed for a

neighbouring area or for areas on the far side of the globe. The “gap artefact” is a ciassic

examplc of this cartographie problem - .see, for e.xample, the paper by Bi.iu-Duval et al. (1976,

fig. 5) for an erroneously wide Tcthys in the vicinity of Europe, If we begin reconstructing

Gondwanaland on a globe of conslani dimensions, starting in the wcst, by Ihc time we reach

the eastern edge of the recon.struction lhere is an enormous gap beiween Asia and Australia (OWEn

1983a, b). There is much gcological and palaeobiological évidence to indicate thaï this gap did

not exist, and that. in realily, during ihc Mesozoic, Australia and Asia were not as far apart as

the “gap artefact” suggests. If we now try a reconstruction of Gondwana but this time starting

in the east. with Asia and Australia near each olher, lhen by the time we reach the wesl. lhere

is an unacccplably wide gap between Africa and Europe. The only realislic way to close the

gap orgore while mainlaining cartographie and geological integriiy, is to decrease the dimensions

of the globe by a suitable amount. No amount of pulling and tugging will gel a small waistcoat

to button ncatiy onto an oversized waist.

As geologists, we aie trained to ihink in three dimensions - four are better- and even

parochial geotectonic siudies arc usually carried oui in three dimensions. However. the vast ma-

jorily of such siudies consisi of the three dimensions of a liny portion of the Earlh’s crusi. So

restricted are some portions sludied. that at the scale of the globe they elfectively represent two

dimensional, unidimensional or even point siudies. depending upon the size and shape of the

areas examined. For example, at the scale of the globe, the Alpine Fold Belt is hke a length of

lape drapcd across a large globe. Ils width - about 800 km if vse are generous - is a mere 2%

of the circumference of lhe Earth. Ils length from Portugal to the Bay of Bengal is about 30%

of the circumference of the Earth, while ils ihickness usually represenied in cross sections is of

the order of 0.5% of the diameter of the Earth. The average width of the Great Rift System of

Africa is a mere 0.1% of lhe circumference of the globe and the relief between its shoulders

and its tlnor is of the order uf O.025%’ of the Earth's circumference. Al the scale of the globe,

the Great Rift is basically a unidimensional feature - a line on a globe.

As OwEN (1983b) and Carey (1988) hâve explained, many of the objections to the Expanding

Earth Tlieory are invalid because lhe objector failed to consider the différences between planar and

spherical geometry. Légion are the planar raaps that hâve been published on plate lecionic thèmes

which are cartographically inexact. As OWEN writes it, many of these maps are nothing more than

“cartoons”,

For example. PoWEl.L ( 1979) reconstructed the history of the Indian plate during its Journey

from the soulhem hcmisphcre in the middic Cretaceous to its “docking” with the Eurasian land-

mass during lhe Oligocène. The successive ouiline maps of lhe Indian plate do not change in

siz.e or in shape during this voyage on the chosen projection, even though they sliould do so

according to accepted cartographie procedures. It looks for ail the world as though Powell eut

out a map of India and redrew lhe same map in various positions on the fiat base map, forgetting

that this is an invalid procedure from any cartographie point of view, excepl on a globe.

A further example of the pitfalls of working on planar maps of the globe concerns the

difficullies of appreciating lhe three dimensional aspect of global tectonics from a planar rendition

of the data, even if the latter are impeccably accuratcly plotted. DoGLlONt (1993, fig. 1) for

example, provides a standard view of the world with lhe centre of the map occupied by the

— 460 —

western part of the Pacific Océan wiih Africa to the lefl of the map and South America to the

right. On this map he distinguishes ■‘West”-dipping, and “East or NE”-dipping subduction zones.

However, a plot of the subduction zones of the Pacific Océan région on a globe rather than on

a map. reveals that ail these zones are dipping in the same way - away from the Pacific hémi¬

sphère. The Pacific hemisphere is in effect being overridden by the opposite hemisphere, and

where the activity is taking place, a gréai circle zone of cordiljeras and trenches lias formed

(Fig. 2). A similar great circle zone of cordillenis and trenches occurs along the alpine fold beit from

Portugal to the Tonga Trench, where the southem hemisphere is dipping under the nortliem one.

Mutter & MiiTTER (1993) use another standard map (which only extends to 70" of latitude)

in which the centre of the map is in the ea.slern Pacific, bui in which South America and Africa

are in the right hand half of the map. On this map. MUTTER & MuTTER hâve depicted the

direction of plate movements by drawing arrows at right angles to spreading ridges and trenches.

The Pacific plate is shown to be moving westwards with respect to the South American plate

and the castem part of Asian plate, yei northwards with respect to the northeastern part of Asia

and the northwestern extremity of North America, at the same time it is undergoing dextral shift

with respect to North America in the région of California. Is the Pacific plate moving westwards,

northwards or is it rotating anticlockwise ? On the planar map it is difficult to reconcile the

conflicls.

On the same map, the depiction of movement vectors of the Antarctic plate indicate that

ail around its periphery, the edges of the Antarctic plate are moving southwards (an impossibility,

even on a globe). In Tact, what has happened is that the edges of the Antarctic plate, as defined

by the spreading ridges which surround Antarctica. hâve accreted northwards with respect to the

continent, which lias been relatively immobile .since the hreak-up of Gondwanaland. Thus, far

from the Antarctic plate having a southwards plate motion as depicted on the map, it is the

spreading ridge System which has increased in length and thereby shifted northwards with respect

to the continent of Antarctic.

In effect, when working on global tectonic problems, there is no alternative but to work

on globes and not on planar maps.

THE PLATE TECTONICS PARADIGM AND SOME PROBLEMS WITH IT

Définitions

A définition of Plate Tectonics given in the Glossan of Geohgy (Bâtes & Jackson 1990)

is as follows:

“A theory of global tectonics in which the lithosphère is divided into a number of plates

whose pattern of horizontal movement is that of torsionally rigid bodies that interact with one

another at their boundaries, causing seismic and tectonic activity along these boundaries.”

Naturally, there are many aspects to the Theory of Plate Tectonics, and each one of which

can be considered a theory in its own right. These include the concepts of subduction and ob-

duction, the notion that plates are driven by convection cells in the upper mantle and several

others. One should add that most supporters of Plate Tectonics assume that the radius of the

— 46] —

Earth has not changed since at least the Mesozoic (ZiEGLER 1993, refers to it as the “finite”

globe) itself a theory of the Earth about which there is not universal agreement.

There are a few researchers who accept the plate-like character of the lithosphère and some

of the aspects of motion - such as latéral displaçement of plates - implied in the Theory of

Plate Tectonics, but who consider that, in addition, the Eailh’s radius has increased by some

20% sincc the Triassic. These scientists comprise the so-called "slow expansion” school of the

Earth Expansion Theory (OwEN 1983). lu this version of Earth Expansion, some of the continents

hâve expericnccd latéral movemcnt with respect to the underlying mantle. but ail of them bave

in addition e.xperienced radial movement outwards from the centre of the Earth.

An alternative group of geoscientists, the “fast expansion” school of the Earth Expansion

Theory, holds that no latéral (or horizontal) movement of lithospheric plates has occurred, but

that the whole séparation of the continents has been achieved by a proces.s by which the radius

of the Earth has increased by 40% since the Triassic (Carky 1978). In this version of Earth

Expansion, the major movement vector of the continents has been radial, away from the centre

of the Earth.

Origin of Plate Tectonic theory

The origin of the concept of Plate Tectonics is described by Cox & Hart (1986). They

point out that while he was writing his papcr to Nature. WiLSON (1965) eut out some pièces of

paper and moved them around the table top. Cox and Hart recommend that students do the

same in order to understand the basic principles involved. Whilst such a procedure is very ped-

agogical, it is incomplète if not followed up by doing similar thing.s on a globe. These fiat

pièces of paper moving about on a fiat table top help to illustrate the concepts of ridges (or

rises), trenches, polarity, transform faults and dextral and sinistral movements.

The concepts as described by Cox & Hart (1986) are essentially two-dimensional and of

a parochial nature. Indeed. in the entire text amounting to 392 pages, there is not a single pré¬

sentation of a whole-globe view of plate tectonic aelivity. Instead we find the statement that

“plate tectonics on a globe is es.sentially the same as on a plane.” The most complicated .system

discussed by these authors is the “three plate problem.” Nowhere is there a warning to budding

plate tectonicians that their solutions to parochial problems should be compatible with plate

movements everywhere else in the globe.

From hypothesis by way of theory to dogma

For three décades the Theory of Plate Tectonics in a globe of constant dimensions has

dominated the thought processes of most of the world’s geologists and many of its palaeontol-

ogists. This theory seeks to propose a rnechanism - Plate Tectonics - which can explain the

distribution of the continents and océan floors and ail the structures observed in them. Yet, since

its inception, observations hâve been made which cast doubt upon ils validity as a universal

explanation of the évolution of the Earih's crusl. Such is to be expected as a normal facet of

scientific theory - what seems abnormal or contentions is the tendency for some geoscientists

to ignore the criticisms as though they are trivial or of no conséquence. In reality, many of the

problems raised for the Theory of Plate Tectonics during ihc past three décades are serions and

— 462 —

should not tnerely be ignored, These difficulties certainly won’t just "go away”. Some of these

observations are of a parochial nature and are usually not considered by many adhérents to the

theory to be problematic for it. Other criticisms, however, undermine the theory by aiming at

its roots.

Since the earty I960s alteniaiive explanatiuns of ihc évolution of the Earlh’s lithosphère

hâve generally becn ignored, even in the rare event that they could be published in accessible

scientilic joumals. Unloriunatcly, (lie leudency on the part of the scientific community to ignore

or to suppress the publication of alternative points of view - by ridicule, by simpiy ignoring

them, or by unscientific or undeontologic refereeing processes for example - has led to a situation

in which Plate Tectonics Theory has for many scientists become a “dogma” rather than a scientific

concept to be exaniined, questioiied, tnodified and accepted or rejected on the merits of its evi-

dential basis and its prédictive and explanatory powers.

At its inception, the Theory of Plate Tectonics was irealed by most scientists as a genuine

scientific theory to be subjected to the usual scientific scruliny. Consequently, it underwent mod¬

ifications as new data and interprétations became available. As time passed however. in the

minds of many people it has crossed the threshold which séparâtes scientific theory from dogma,

and for many researehers it has now become a matter of unquestioning belief In a scientific

domain, the latter is clearly an unacceptable state of uffairs. There are mimerous publications,

for example in the domain of palaeobiogeography, in w'hich Plate Tectonics has been invoked

to explain biogeographic data without the authors having the slightest inkling of what Plate

Tectonics means or how il is supposed to work. Even in the domain of geotcclonics, there are

those who pcrfomi incredible mental gymnastics in order to gei their field observations to fit

into the plate lectonic paradigm. For those who iraplicitly believe in Plate Tectonics as the princi¬

pal niechanism behind global tectonics. alternative explanations of the évolution of the Earth’s

crust appear to be anathema - they are concepts to be rejected oulright, generally even without

their basis having been read and understood.

For the past three décades, many of the geographical aspects of the geological and biological

changes which occurred during the Tertiaiy hâve been explained within a modcl of mobile con¬

tinents whose movements relative to each other on a globe of constant dimensions, hâve been

determined by “plate tectonics" (Cox & Hart 1986; CONDIE 1989). Indeed, it appears from the

way they writc, thaï for many researehers “plate tectonics" has become synonymous with “con¬

tinental shift", which of course it is not. One is a theory of mechanism to explain an observation,

the other is the observation it-self.

PHILÜSOrHY

In the belief ihat science best advances by the overthrow or modification of prcviously

cherished théories, especially those that hâve tended to cross the threshold into dogmalism, I

hâve been laking a look at some of the different ideas that hâve been proposcd to explain past

events in the Earth's lithosphère and biosphère, 1 do this because for the past two décades I

hâve found that many of the palaeobiological and palaeoecological events of the Cainozoic Era

are incompatible with plate movements as reconstructed on a globe of constant dimensions, or

tbat they cannot be salisfactorily explained by it. As a resuit I bave looked at some of the points

— 463 —

of dissalisfaction lhat other earth scientists and palaeontologists hâve expressed about the Theory

- more properly “théories”, since there are several variations on the basic theme - of Plate Tec-

tonics. I hâve aiso looked at some of the alternative proposais to explain the history of the

Earth’s crust and its inhabitants, both animal and plant, and I pass some of these on to the

reader in the hope thaï lhey will raise questions in his or her mind. I cannot prétend that any

of the ideas are originally mine; mosl of thcm hâve been proposed at one time or another by

my predccessors. The motive for writing this Icugthy teview of the matter is to highiighl the

varied aspects of the théories of global tectonics with a view to stitmilating discussion.

Sontc of the ideûs were considered worthy of research long before the Theory of Plate

Tectonics came onto the scene. Most of them were never refuled or disproven in scientific debate,

they were simply ignored or put to rest once Plate Tectonics "arrived”. because it was considered

by many - crroncously as it happens - that the Plate Tectonic Theory rendered ail other geotec-

tonic hypothèses redondant. What unités these alternative hypothèses is the fact that ail of them

are at présent vigorously denied by the majority of supporters of the Theory of Plate Tectonics.

Focus OF THIS PAPER

In this paper, I focus for the most part on African and European Cainozoic geology and

palaeontology. I am fully aware of the dangers of adopting a parochial Old World focus - which

I tend to do in this paper, because lhat is where I hâve experienced the Earth’s Geology and

Palaeontology - but since this région is so complex, there is much to learn from it. It is well

to keep in mind that the area suhjected to close scrutiny is a minor part of the entire lithosphère,

and that any hypothesis about its évolution should accord with the évolution of the globe as a

whole. There is no use in presenting a solution to the Mediterranean problem for example, if

it only raises contradictions elsewhere in the globe. As Carey (1988) and OwEN (1983b) hâve

pointed out, parochial studies hâve long been the bane of geotectonics.

Problems for Plate Tectonics

Ever since its general acceptance by most western geologists and palaeontologists. the

Theory of Plate Tectonics in a globe of constant dimensions has had ns detractors, most of

whom hâve been ignored despite the fact that they hâve posed serious questions (Carey 1988;

Keith 1993; Stei.ner 1977; OWEN 1976) based on clear ob.servaiions. some of them apparently

fatal to certain aspects of the theory (Table I ). There are enormous problems for Plate Tectonics

regarding increases in the surface area of the globe (SteinER 1977). Subduction, obduction,

folding and under-plaiing simply do not résolve these problems. CAREY (1988) discusses several

enigmas in which the horizontal motion of plates proposed by proponents of Plate Tectonics on

a globe of constant “r” does not accord with the field evidence.

Keith (1993) has compilcd an impressive lisi of paradoxes and puzzles which run counler

to current Plate Tectonics (PT) dogma, including radioactivity in the continental crust and the

upper mantle, the heai flow paradox. anomalous “hot-spot” tracks which are marked by gross

violations of the simple age-distance relationship, and many others. He proposes an alternative

mode! of global tectonics which he has called the Viscous Flow (VF) Model, that he believes

is belter supported by the observations. While having a great deal of sympalhy for the VF model

— 464 —

Table 1. — Comparison of Early version of Plate Tectonic Theory and Fasl Expansion Theory.

Plate Tectonics

Fast Expanding Earth

Substrate

Observations and evidence

Layered Planet with rigid crust

Shapes of continents

Sea-floor magnetics

Mid-ocean ridges

Transform faults

Fracture zones

Trenches & island arcs

Mountain Fold belts

Hot-spot tracks

Earthquake distribution

Volcano distribution

Terranes

Layered Fluid Planet

Shapes of continents

Sea-floor magnetics

Mid-ocean ridges

Transform faults

Fracture zones

Séparation of continents

Continental magnetics

Cartographie fits

Earthquake distribution

Volcano distribution

Rotation of microplates

Mechanisms

Convection in mantle

Accretion at ridges

Subduction at trenches

Collision shortening

Vertical tectonics

Accretion at ridges

Rock flow

Resuit

Plate movements

Séparation + convergence

Plate enlargement

Séparation of continents

Motion of plates

Latéral on surface of globe

Radial to surface of globe

Energy source

Geothermal

Mantle drag force

Ridge push force

Slab pull force

Slab drag force

Suction force

Gravity (secular change?)

Kinetic energy of rotation

Mechanical energy of Earth

Reaction forces

Transform fault résistance

Colliding résistance

Rates

+/- constant (2-10 cm/a)

+/- constant

Assomptions

Constant “r"

Constant océan volume

Pre-Jurassic ocean-floor

Completely subducted

Constant atmosphère volume

Iron/nickel solid core

Rigid plates

Rotation axis fixed

Obliquity fixed

No pre-Jurassic océan

Area of continents +!- constant

Implications

Océan volume not constant

Atmosphère volume changes

Moment of inertie not fixed

Rotation axis not fixed

Obliquity not fixed

Great Circle Torsion

60% Earth radius at Jurassic

— 465 —

Difficulties

Arctic Paradox

Biogeography of Tethys

Many cartographie problems

Large palaeomag. error bar

2 Dimensional thought

Pacific Rim Paradox

Tethys Gape Artitact

Insufficient energy

Northwards vector paradox

Palaepole overshool paradox

Pacific Paradox

Indian Enigma

Missing Archaean crust

Africa enigma

Antarctic enigma

Peru-Chile Trench anomaly

Kermadec Trench anomaly

lapetus océan Myth

Zodiac Fan Anomaly

175 km deep continental keel

Distribution of radioactive éléments

10 km depth to brittleductile transi¬

tion

Overlapping spreading centres

No known mechanism of ex¬

pansion

Moment of inertie

Lack of expansion on Moon,

Mars, Mercury and Venus

Palaeozoic Invertebrate bio¬

geography

"G" change less than 1%

since 4.6 Ga

Sea-levels

Palaeomagnetic radius

Comparisons of Modified Plate Tectonic Theory and Slow Expansion Theory

Main différences from early

version of PTT and Fast EET

Plates déformable

Rates of motion variable

Deviatoric tenslonal stresses over

mantle plumes

Continental undertow

Microplates

Ridge jumps

Overlapping spreading centres

Some latéral plate motion

Rates of motion variable

80% Earth radius at Jurassic

Some subduction and obduc-

tion

Pulsed Tectogenesis

proposed by KEITH (1993) -especially when one considers the fact that over géographie .scales

of 10^ km and lime span.s of 10* years, even solid rock behave.s like a fluid - I consider that

it suffers from some of the same problems as the PT niodel, in parlicular lhe assumption of

constant “r” and inadéquate sources of energy to aecount for the viscous flow.

The question of energy suppiy to power the plate tectonic motor has still not been success-

fully addressed by supporters of the theory. despite (he fact thaï these questions hâve been posed

from the very dawn of the era of plate tectonics (ALVAREZ 1990; Wilson 1993). As Alvarez

(1990) points out, the lack of a mechanism for WEGENER's theory of continental displacemcnt

caused it to die. yel a similar lack of a mechanism has not prevented the modem version of

the same theory - continental displacement by plate tectonics - from gaining wide.spread accep¬

tance, even among those that previousiy rejected it on those very grounds. Currently proposed

energy sources, such as geothermal energy (Table 2) appear to be much too feeble to drive the

plate tectonic motor(s) (Marchal 1991). Apart from which, geothermal energy has been sug-

gested as one of the major sources of energy that maintains the Earth’s magnetic field (Table 2)

and would apparently be unable to suppiy enough energy to drive bolh Systems. Other proposed

— 466 —

Table 2. — Energy sources on Earth. For explanalion of the number/p/number convention, see Marchal (1991): (e.e.

2.139p29 = 2.139 x lO’'').

Endogenetic (telluric) energy

a) Kinetic energy of rotation of the Earth

2.139p29 joules

b) Gravitational energy of the Earth

- 2.44p32 joules

c) Mechanical energy of the Earth

p26 joules

d) Geothermal flux

3 to 4p13 watts

Exogenetic (extra-telluric) energy

a) Tidal energy

pi2 watts

b) Solar energy

(39% of solar energy receipt is immediately re-radiated to space)

61% of 1.8p17 watts

Comparison

A large earthquake may dissipate up to pi 9 joules of mechanical energy. (Data from Marchai 1991.)

There are about 10'* earthquakes every year.

Over a period of 1 million years geothermal flux would release about 10^^ joules of energy into the

fabric of the Earth.

The energy required to produce and maintain the Earth’s magnetic field is between 10’° and 10** watts;

it is thought to dérivé from radioactive decay or from latent beat of solidification of the inner core.

Heat loss from continents is about 12 x 10*^ watts (= 55 mW/m^), beat loss from the océans is about

30 X 10*^ watts (= 95 mW/m^); of the latter, the average measured oceanic heat flow is 67 mW/m^,

the différence from 95 mW/m^ being heat losses at océan ridges by hydrothermal circulation. (Data

from CoNDiE 1989.)

energy .sources such as rldge push, slab pull and trench suction (COX & Hart 1986), deviatoric

tensional stresses (ZlF.Gl.ER 1993; Wll.,SON 1993; Doglioni 1993; Pavoni 1993) and continental

undertow (ALVAREZ 1990) hâve overtones of perpétuai motion, and as such are unlikely to solve

the problem of energy supply to make the plate tcctonic motor work.

A further problcni wilh manllc convection as a mechanism for plate motion, is lhat the

convections cclls are supposed to romain in rclatively stable positions relative to ridges and

trenches over geological lime frames (Pavoni 1993), yet at the same time they hâve to change

their volume, and therefore thoir gcomcti'y. as the plates expand or contract in size. The African

plate, for example, is ctmsiderably larger now lhan it was during the Jurassic, and the same

applies to the Australian plate, the Antarctic plate, the South American plate, the Eurasian plate,

as inspection of a global map of plates and spreading ridges will show For much of its history,

even the Indian plate got larger with the passage of time. and it was only during the Late

Crctaceous and Cainozoic lhal its norlhem parts began to deereasc in area. There appears to be

a fundamental contradiction here between the observation that oceanic crust is generated along

a well-defined global System of spreading ridges -implying stabiliiy of the spreading System-

and the observed enlargemeni of most of the plates in the globe - implying, in a globe of constant

dimensions, a changing System of convection cells. The adoption of a two-layered mantle in

— 467 —

which only lhe deep layer is convecting (Alvarez 1990) does not solve the problem of vvhat

energy source is supposed to drive the convection cells.

Furthemiore. calculations presented by STEINER (1977) hâve indicated that whiist individual

spreading centres hâve generated new crust at varying rates, the overall global sea-floor spreading

phenomenon is a co-ordinated global process, where at any given time. high-spreading rates in

one océan basin are coinpcnsated by low rates in another. Il' this is so. then a speed-up in crustal

génération at one spreading ridge would hâve to be balanced out exactiy by a slow-down in the

others. What the overall conlrol of such a mechanism is -especially if spreading is visualized

as being relaied in somc way to convection cells in the manlle - remains to be addressed by

Plate Tectonics adhérents.

In contrasl. in the Expanding Earth paradigm. expansion is expected to occur preferentially

along a global spreading System and to produce the enlargement of plates as a conséquence of

activity along this System. Furthemiore, the localized expression of expansion could vary greatly,

a speed-up in one area being natnrally accompanied by a slow-down elsewhere.

ASSUMPTIONS ARE THEORIES

We need to keep in mind that “Continental Displacemenl” and “Plate Tectonics” are two

quite separate théories. So is the theory that the Earth’s dimensions hâve remained constant

through Phanerozoic lime — the lïnile globe concept of ZiEOLER (1993). The former is based on

observations of the Earth’s crust. both continental and oceanic, and proposes that continents

today occupy positions which are different from Ihosc thaï Ihcy occupied in the pasl. The latter

is a theory to explain how and why the continents corne to occupy different positions diiring

the passage of geological time. Jt .seeks lo présent a mechanism (plate tectonics) to explain the

observation icontinenial displacement). Foundenng or modification of the Plate Tecionic Theory

will not lead to automalic rejection of the Theory of Continental Displacemenl, allhough many

of the details of continental displacemenl history might change if currenl mechanisms of plate

tectonics should prove to be inappropriate models to account for it (Keiih 1993).

Plate Tectonic theory lias “evolved” during the past two décades, to the exlent that plates

are no longer thought of as rigid slabs of rock floating on a liquid magma - in faci, as shown

by Keith (1993) there are many geotectonicians who still regard plates as being rigid. For ex¬

ample, Kearey & ViNE (1990) hâve published a textbook in which the following définition

appears: “Within the basic Theory of Plate Tectonics. plates are considered to be intemally rigid

and to act as extremely efficient stress guides. A stre.ss applied to one margin of a plate is

transmitted to its opposite margin with no deformation of the plate inlerior.” However, such

extreme views are now considered lo be far from mainstream plate tectonic theory.

Nowadays, plates arc gcnerally considered lo be subjecT to deformation of varions sorts

(shear, transcurrent faulting. fragmentation, flow, folding, nappe activity and so on). What has

not changed in Plate Tectonics Theory - indeed, any possibility that il has occurred is vigorously

denied by the majority of geotectonicians - is that the dimensions of lhe Earth are assumed to

hâve remained constant at least since the Jurassic period if not since the PreCambrian.

A subsidiary “belief is thaï lhe volume of the workTs océans has remained constant through

geological time. There are several other “fixist” (or “permanentist”) notions which underlie much

— 468 —

current thought in geoscicnces, inosi if not ail of which will prove to be incorrect once they

are investigated more closely by scienlists with unclutlcred minds. This paper addres.ses several

of these “fixist” ideas with a view to demonstrating their inhérent weaknesses. I include among

these the notion that the Earth's obliquity is fixed. that its axis of rotation is as fiimly defined

as il is in a school globe, that the volume of the atmosphère is more or less constant, that the

Sun's output has been more or less constant over geological time, and that the Earth’s core is

like a solid iron-sulphur cannon-ball and has been so at least since the Jurassic.

SPKCIAI PI.I-ADING

A well-used tactic on the part of geotcctonicïans has been to keep the globe constant in

dimensions but to make huge areas of ocean-tloor disappear down subduction zones or to override

continental criist at obduction zones, and to induce large areas of continental crust to crumple

up into the world's fold bclts or to slidc undcmeath each other to produce “double-thickness”

crust. So much so. that it has been said that this is a well-cstablished fact of geolectonics. By

this means, proponents of an Earth of constant dimensions endeavour to keep the total surface

area of the globe constant while acknowledging that vast areas of oceanic crust hâve been

generated at an exponential rate in the western, eastern, Southern and northern hemispheres since

the Jurassic Penod tSTtflNüR \'-)ll).

In regard to subduction, lhe implication is that il would increase through geological time.

Yet the history of Panthalassa (/.c. the Eo-Pacific on a modem dimensions Earth) is lhe reverse:

fast subduction in the remote past to gct rid of the Panthalassa crust (pre-Mesozoic) to allow

the génération of lhe post-Mesuzoic crust that we sec now (STEINER 1977).

Case historiés

The Mediterranean région

Earth scienlists who work in lhe Mediterranean région expérience perhaps the greatest dif-

ficLilties. A referee of an early version of this paper even attested that this section was an “ab-

solutcly unacccptabic assurnplion on the évolution of the Mediterranean” but hc neglected to

say why hc thought so, or to provide any evidence to lhe conlrary. A review of the lilerature

on lhe région will .show thaï barely iwo authors agrce on the inlcrprclalion of the Mediterranean,

yet some patterns of its history seem to be clear (OWUN 1983a; CareY 1988).

According to many plate tectonicians, the entire région between Africa and Eurasia ought

to be in compression, because accordittg to these scienlists, Africa is moving northw'ards across

a broad front againsl a more or less immobile Europe (BERTHEI.SON & Sengor 1990). Yet many

of the structures in the Mediterranean and in neighbouring land masses am extensional in origin

(Doglioni 1993). Spain and Turkey. for example, are littered with “endorheie” basins which

hâve filled with continental sédiments containing rich fossil assemblages. Many of the islands

and land masses in the Mediterranean région hâve been rotated sinistrally - i.e. lhe northern

hémisphère has sheared clockwise relative to the Southern hemisphere as observed from above

the surface of the Eailh - along its entire length, According to Carry (1988). this bell of sinistral

shear extends right round lhe globe following a great circle known us lhe “Tethyan Shear”.

While there are some aspects of C'ARF.Y's Tethyan Shear that appear to be controversial, there

— 469 ^

is a great measure of truth in what he says, as a plot of the Alpine fold ranges and associated

trench Systems will show - many of these structures lie within lO" of a great circle.

In the Mediterranean région, this shear zone has caused fragmentation of the Earth’s crust

within a belt more than 700 km wide running from Spain in the west to Arabia in the east. It

has resulted in an offset of some ISOO km bctwecn the northem and Southern ''hcmisphcres" in

the longitudes ol Africa and Europe. Instead of Africa moving northwards towards Europe, it

actually appcars to hâve incrca.sed its distance frorn il. Examinalion of the rotated "microconti-

nents" (Moesia. Rhodope, Iberia, Turkcy. etc.) within ihe Mediterranean région indicate that the

distance between Africa and Europe has increased during the Tertiary (CARLY 1988: OwttN

1983b). This contention is bonté oui by lhe tensional nature of the Red Sea Rifl and its extensions

southwards into mainland Africa, and of the extensional nature of scveral of the Mediietranean

deep basins (OWLN 1983b). If Carl-y's proposais concernmg Tethyan Shear are salid. then ail

previous plate lectonic explanations of the évolution of the Mediterranean Basin can be discarded

or will hâve to be greatly modified. Furthermore, if the radius of the planel has changed (OwEN

1983a, b), then ail reconstructions of the history of the Mediterranean ba.sed on lhe assumption

of a constant dimensions Earlh, are false.

Adhérents to Plate Tectonic Theory hâve identified numerous “microplates" in the Medi¬

terranean région, but few can agréé on how many plates there ought to be and in which direction

they are supposed to hâve moved (compare for instance the figures in Dewey et ni. 1973; BeR-

THELSON & Sencor 1990; BiJU-DuvAL et ol. 1970; Carry 1988), In fact. the resulis sometimes

look like ad hoc science, a sort of patchwork quilt in which each investigator ignores or is

unaware of what his neighbour has donc. Il is rare to see a corivincing synthesis of lhe lectonic

history of the entire région, not to see how the évolution of the Euith's crust in the Mediterranean

région relates to that of the resl of lhe lithosphère. Carey (1988) is onc of ihc few vvhn regard

the Mediterranean as part of a circum-global structure, the Tethyan Torsion. The only carto-

graphically précisé reconstruction of the hi.story of lhe Mediterranean région is thaï by OwEN

(1983a, b). AH olher efforts exhibit greater or lesser dislortion of the outlines. sizes and positions

of plates in order to producc a “likely” slory or to illustrale a concept.

Orogenic Belts and Great Circle.v

It is no coincidence in my opinion, that loday’s orogenic belts lie along arcs of great circles

or are not greatly removed from such arcs, as Du Toit (1937) noticed. I hâve plotted out the

Alpine fold belt and ils associated trench Systems, as well as the circum-Pacitlc cordillera and

trench System, and it is évident that these immen.se Systems arc confined to a zone that is less

than 10” either side of a great circle. Even ilunigh Prolerozoic, Palacozoic and Mesozoie orogenic

belts hâve been fragmenled and somewhat distorted, it is clear that they also originally lay along

great circles. Wilhout exception, the pre-Mesoz.oic fold belts were considerably shorter than the

Cainozoic belts (ZlF.ca.ER 1993). which I take to provide good évidence thaï the globe was smaller

in pre-Mesozoic times than it is now. The Old World and New World Tertiary fold belts may

owe their origin to the increase in area of one side of lhe globe al lhe expense of lhe olher.

On a globe, a perimeter and the area enclosed on one side by it can both increase until

the perimeter becomes a great circle - even if il is not precisely circular due to inhomogeneilies

in structure or in its history - aller which any further increase in area can only induce a decrease

in perimeter. Such is lhe Pacific rim (or perimeter). If however, lhe globe is expanding in volume

— 470 —

at an appropriate rate, the length of such a great circle - e.f’. lhe Pacific rim - need not become

shortcr tn accomodate the increase in surface area enclosed by il - the Pacific Océan - depending

upon the rate of increase in volume. The presence of the Cordilleran fold belt indicates that

Earlh Expansion has not been able to compensate entirely for inhomogeneous -i.e, one-sided

(geotumors)- expansion of the globe, and the resuit has been a différence between the rate of

increase in surface area of the Pacific and of the length of its rim, resulling in compression of

the Pacific rim great circle relative to the area (on the Pacific side) that it encloses. Compression

structures thus typify more than half the length of the Pacific rim great circle.

At right angles - a .spherical nght angle - to the Cordilleran fold belt. there is the Alpine

fold belt. This also lies more or less along a great circle. allhough Ihc Indian plate has dislorted

it since the Eocene, producing kinks in the arc. Folding, thrusting. nappe formation and olher

tectonic activity in the Alpine fold belt appears to be due to shortening or contraction of a great

circle perimeter related to lopsided growlh of lhe manlle: more in the southem hemisphere than

the north. This has in effect led to greater production of occanic crust in lhe circum-antarclic

région and lhe Southern Atlantic. Indian and Pacific Océans than in the Northern Océans. This

predominanily one-sided (i.e. single hemisphere) expansion has lcd to the apparent net northwards

displacemeiu ot ail continents (excepl Anlarctica) and older oceanic crust relative to lhe présent

equator. recorded as lhe palaeornagnetic nortlnvard movement effect, a feature thaï has been

observed and commeiited upon since lhe dawn of lhe “Plate Tectonic” Era, if not before. This

net northwards displacement of lhe continents produced an effect which is well described by

Carey (1988) but which slill confuses maiiy “traditional” thinkers. As the northern continents

were apparently displaced northwards, lhey carried lheir palaeornagnetic signatures with them.

Thus the Mesozoic “équatorial" magnetic signature - i.e. the zone where magnetic déclination

was nearly zéro, or horizontal, has been carried norihward.s, which gives ihc impression that

Africa and South America hâve migraied northwards since the Jurassic. The Mesozoic mid-lati-

tude magnetic signature, where the déclination is about 45'-' (positive or négative doesn’t matter),

has also moved northwards, giving lhe impiession thaï Eurasia and North America hâve also

shifted northwards during and since the Mesozoic. Those scienlisis who assume thaï lhe dimen¬

sions of the globe hâve remained constant, are forced to conclude that ail continents excepl

Antarctica hâve been convergmg on the ArIic, lhe northern ones more slowly than the Southern

ones. The maps of SMtrti ei al. (1981) represent a classic example of this trend of thought-

in the maps, the Arctic Océan hecorne.s smaller and smaller in extern as geological time passes,

and as lhe continental palaeornagnetic signature tells the authors that lhe continents hâve shifted

northwards. Yel, studios of the palaeornagnetic signature of the floor of ihc Arctic Océan reveal

that it has increascd in area since the Cretaceous, not dccreased. In olher words, relative to the

présent North Pôle, the continents hâve been shifting southwards since the Cretaceous. Yet,

despile this insurmountable contradiction, most geotectonicians continue to maintain their faith

in constant “r”.

Pulaeontology

Palaeontologists hâve often noted that some types of palaeobiological data seem to contradict

reconstructions of plate movements on a constant dimensions globe, either in the details of move-

meni directions or of timing of events orof past positions of plates (Haleam 1981 ). For example,

according to currently accepted versions of plate tectonic history. India was deep in lhe Indian

— 471 —

Occan throughoiil lhe Cretaceous, Palaeocene and thc l'irst half of the Eocene (PatriaI' &

Achache 1984). Yel throughoul ihis pcriod India appears to hâve been much doser to Asia

lhan the plate reconstructions suggest (Sahni 1984), so much so that it failed to develop an

endemic fauna - cndemism, evcn in land manimals appears to hâve been weakly expressed,

although the terrcstrial fossil record is rathcr poor in thc latter domain. During the Early Oligo¬

cène. India was inhabitcd by a few typically African l'aunal éléments such as Hyracoidea iPlCK-

FORD 1986a) suggesting thaï there may have been an inlerchange of terrcstrial faunas betvveen

Africa and India al .some stage during thc Early Oligocène. At lhe same lime, il aiso contained

some typically European l'aunal éléments, such as Anthnicalheriiim muiiniim. which mdicate

faunal inlerchange with Europe bcforc or during lhe Stampian Period iMiddle Oligocène) iPlCK-

FORD 1987). Ail Ihis supports lhe observation by ITaLLAM (1981) that the palaeobiogeographic

évidence reveals disparities with plate tectonic reconstructions in which India was well separated

from Eurasia. He concluded that the width of the Tethyan barrier may have been overestimated

by many researchers.

Climate

The climatic changes which took place al the onsel of the Middie Miocene are sometimes

explained in ternis of plate tectonics. For example, lhe development of dryshod access across

the Tethys Seaway would automatically have severed lhe oceanic circulation between the Indian

and Atlantic Océans (Hallam 1981) thereby perturbing global climales. Al thc beginning of the

Middie Miocene, Europe's climate became more tropical. But during the Upper Miocene. it be-

came less tropical - the boundary zone between lhe Palacarctic and Elhiopian Realms began to

shift soulhwards (PiCKFORD 1986b)- with the resuit thaï almosl al| lhe Elhiopian lineages even-

tually died out in Europe. On its own. the northwards shift of Africa towards Eurasia can hardiy

account for the onset of tropical conditions in Europe and then tor the subsequeni lindoing of

these conditions. At leasl not withoul a certain amount of spécial pleading. As far as I am aware,

no one has yel proposed that lhe Upper .Miocene climatic changes resulicd from plate leelonic

activity between Africa and Europe, although uplifl of lhe Himalayas has been proposed as a

possible contender for lhe onset of monsoon climatcs during the latest Miocene (OtiAOE et al.

1989), contemporaneous with faunal changes in the Indian subcontinent. Al the same lime, the

dessication of lhe Mediterranean during the Messinian Crises has been correlated (or mis-corre-

lated) with faunal changes in Europe and North America (Alroy 1992).

Energy and Pi.ate Tectonics

Despite the tille of their book {Plaie Tectouies - How h warks). the very pcdagogical book

by Cox & Hart (1986) only gets onto what drives the plates on page 339. In the following

ten pages, the authors champion the idea that some kind of thermal convection is responsible

for plate movemeni. Yet not a word is said about the quantitics of energy available in thc supposed

convection cells, nor of lhe amount of energy rcquired to cause thc observed motions, The last

four pages of the ten which are devoted to what drives thc plates, are given ovet to a discussion

of “Driving Forces". These are listcd as Manlle Drag Force, Continental Drag Force, Ridge Push

Force, Slab Pull Force, Slab Drag Force, Transform Fault Résistance, Colliding Résistance and

Suction Force. Of these, Ridge Push and Slab Pull, if they occur, are gravity driven. while the

472 —

so-called drag forces and résistances would actually tend to impede plate movements. The

amounts of energy required for plate movements are simply not addressed by Cox & Hart

(1986).

Nor are they di.scu.ssed by ZlEGLER (1993) who merely States that “Earth’s mande convection

System... facilitâtes escape of thermal energy from tlie Eaith's interior” or by DOGUONJ (1993),

Pavoni (1993) and Wil soN (1993). In a sériés of companion papers. these authors discuss ihe usual

forces supposed tu move the plates, such as ridge push, trench suction, deviatoric tensional stresses,

slab drag and so on. U is évident from the papers thaï there is no consensus among these or any

othcr authors conccrning ihc mechanisms thaï movc plates. Nonc of them address the question of

an energy source and ils magnitude, excepi for invoking thermal energy in the mande.

The situation is similar to the debate among non-mechanics concerning what moves a car.

For one it is the frictioual force betwecn the rubber lyrc.s and the road surface. For a second it

is the turning of the propellor shaft. while for a third it is the up and down motion of the

pistons in the cylinders of the engine. Without mentioning the source of the motive power, which

is of course the fuel in the tank, the debates are rather futile, evcn though valid within the

parochial view of each debater. As every mechanic knows, ihe fuel is the crucial factor, without

which none of the resl would bc possible.

In the plate teclonic model of évolution of the Earth’s crusl, the energy sources which are

supposed to resuit in continental displacement hâve not yet been satisfactorily identified (MAR¬

CHAT 1991). Gcothermal flux on its own appears to be insufficient to account for such activity,

apart from the fact that the same energy has been said to produce and maintain the Earth’s

magnetic field (CONDIE 1989). In any case, as MARCHAT points out, the distribution of radioactive

éléments in the lithosphère and mantlc would tend to impede the formation of convection cells.

The rotation of the Earth rnay well aiso interfère wilh the formation of convection cells.

Exogenetic. or extra-tclluric energy sources, such as a.sironomic (solar, tidal) supplies, are

far loo wcak to drive the plate tectonic motor (Table 2), even if they were combined with ge-

othemial energy. Endogenelic, or telluric energy sources, such as gravitational energy and kinetic

energy of rotation, are the only known oncs which are potentially powerful enough to supply

the rcquisite amounts of energy for continental displacement. Of thc.se two, the kinetic energy

of rotation of the globe may be important in inducing latéral displacement of the continental

crusls with respect to their underlying mantle and for inducing other cffects such as “Tethyan

Shear".

A third, hypothetical source of energy is encompassed by the idea that the cote of the Earth

is a proton plasma, rather than a solid iron-sulphur “cannon-ball”. Capture of électrons by protons

in the plasma would Icad to the création of enormous expansion energy, as well as of energy

released by the fusion. The volume occupied by atoms is orders ol magnitude greater than the

volumes occupied by either protons or électrons on their own. A plasma core decaying progres-

sively to an atomic stale - the outer core - and the production of heavier éléments to form the

mantle would yield vast quantities of energy which might be sufficient to drive both Earth Ex¬

pansion and some plate movements. Indeed. Owen (198), 1992) has espoused the idea of a

plasma core for the Earth, and considers that the combined energy of a decaying plasma core,

geothermal energy, Earth Expansion and the very slow rotational effect would be sufficient to

produce the movements of continents.

— 473 —

Allègre et al. (1993) hâve reported that rccent developmcnls in rare gas geochemistry

suggest lhat the terrestrial atmosphère was generated by degassing part of ihe Earlh’.s mande,

but that neither Néon nor Hélium fit the general piclure. The ratio of "^’Ne/^*Ne ratios in MORB

(between 10 and 13) and in air (9.8) are very different. The ^*’Nc/'‘Ne ratios of up to 13 in

upper mande derived basalts cannol be explained by any kriown nuclear reaction. ALLÈGRE et

al. (1993) postulate that mo&t of ihc ‘“Ne is of cosmic origin, reaching the mande basalts by

subduction of micrometeorites. However, it is perhaps more likely that the “cosmic" signature

of ^“Ne dérivés from the Earth s own cosmic-like core, which would tend to support the idea that

the Earth’s core is a proton plasma. Most of the ^He in the atmosphère, also said to be of cosmic

origin, could likewise be derived, not from the cosmos, but from ihc Eanh’s plasma core. Similarly,

the hydrogen that is lost to space duc to photo-dissociation in the upper atmosphère may well be

replaced by juvénile hydrogen resulting from decay of protons in a proton plasma core to the atomic

State by addition of électrons. If the source of ^‘^Ne and ^He is the Etulh’s core. then it renders lhe

hypothesis of cosmic origin followed by subduction reworking uniikely.

The mechanical encrgy of rotation could lead to slight differential movemeni between con¬

tinental crust and occanic crust (Carey 1976, 1988) because the continents are on average 4.6 km

further from the axis of rotation than are the océans (mean altitude of continents: 870 métrés

above .sea-lcvel; mean dcpth of océans: 3.740 métrés below sea-level ). The.se deformations would

tend to be elongated in the north-south sense, and would either be compressive (along west

coasts of continents) or exlensional (on east sides of continents). The séparation of Euro-Africa

from the Americas may well be partiy driven by such a mechanism, while the trench and cordillera

Systems of Norih and South America could be partly due to the same phenomenon. However,

such powers probably play a minor rôle in the overall global geotectonic scheme. Nevertheless

they should not be forgotten.

Lack of sufficient energy suppiy to drive its moiors, suggests that plate tecionics driven

by convection cells in the rnantle is not a suiiable model for explaining lhe évolution of the

Earth’s lithosphère, a conclusion already reached by a number of independent studies iCarey

1976, 1983. 1988; MARCHAI. 1991). If convection cells exist in the rnantle, then they probably

play a minor rôle in the évolution of the Earth's crust. Nor can slab pull, ridge push and trench

suction solve the problem of motive power for plate movemenls, mainly because wilhout a further

source of energy powenng some part of the cycle, the machine would grind to a hait very

rapidiy. As described by Cox & Hart (1986) the ridge push/slab pull model closely resembles

a perpétuai motion machine in which one force drives the other which provides power for the

first one. and so on ad infinitum. Much more powerful forces, such as gravity or the kinetic

energy of rotation of lhe Earth scem to be implicalcd.

THE EXPANDING EARTH HYPOTHESES

Introduction and origin of concept

A distinct possibility is that the assumption of a constant dimensions Earth is invalid, and

that the Earth has expanded significantly since the Jurassic. Only in an expanding globe can

the sum of the areas of two surfaces defined by a perimeter on its surface increase in value.

— 474 —

The hypothesis of expansion of the Earth has becn seriously proposed (allhough the idea has

often been vehemently rejected) as a way of resolving virtually ail the problems encountered in

reconstiucting past continental configurations. Among thosc who hâve proposed Earth Expansion,

there are two principal groups; tho.se whose reconstructions of Pangaea at the beginning of the

Jurassic cover the entire globe {i.e. there was no proto-Pucific Occan) - callcd fast expanders-

and those whose reconstructions of Pangaea at the beginning of the Jurassic cover a little more

than a hemisphere, a large proto-Pacific Océan (Panthalassa) covering the other half - the slow

expanders. The fast expansion hypothesis. chiefly advanced by CARlîY (1983, 1988) begins with

a Jurassic globe about 60% of modem dimensions. The slow expansion hypothesis, whose princi¬

pal advocate is Owh'N (1983b) begins with a Jurassic globe about 80% of modem dimensions.

In the slow expansion model, extrapolation backwards into geological time beyond the Jurassic

results in a Pangaea which covers the entire surface of the globe sornetime in the Late PreCam-

brian. Understandably, somc scientists arc reluctant to extrapolatc backwards from the Jurassic,

because the constraints providcd by palacomagnetic patterns of the ocean-floor are not available

for prc-Jurassic periods. Ncvertheless, the fact that orogenic belts appear to form along arcs of

great circles, provides some information and the hope is thaï pre-Jurassic reconstructions of Pan¬

gaea may be fortheoming. along with details of crustal deformation during the Palaeozoic and

Early Mc.sozoic due to Tethyan-type torsion, flow, tolding, faulting and other surficial phenomena.

If such reconstructions become available, varions problems of Palaeozoic and Early Mesozoie

biogeography - there are many for which there seem to be no solutions under the constant “r”

hypothesis - may be resolved.

CONTEMPT. RIDICLIt.E AND BIASED TREATMENT

The advancement of geological and palaeontological sciences was delayed and hampered

for four décades by the rejection of Wegener’s iheory of continental displacement on the grounds

that he had proposed an unsatisfactory mechanism to account for il. In palaeontological circles,

reluciance to accept continental displacement led to the proposai of numerous ad hoc explanations

for palaeobiological observations, including the notorious transoceanic “land bridges,” “floating

islands" and other fancifui faunal dispersion mechanisms and pathways. Du Toit (1937) wrote

an cloquent commentary on varions reactions of the scientific community to the concept of “Con¬

tinental Drift." With minor rewording his comments would appiy cunously closely to recent

réactions to the Hxpanding Earth Theory. Cuirent reluciance on the part of many geoscientists

to accept Earth Expansion as a hypothesis to be tested on its own merits — that is. not to be

rejected out of hand - uncannily re.semhles the résistance that Wf.gENEr's concept of continental

displacemeni expericneed for forty ycar.s.

CoNDin ( 1989) for example, in his textbook on Plate Tectonics and Crustal Evolution, dis-

missed Earth Expansion in less than a page of text in an ill-infomied and radier condescending

way. He began by saying that “evidence for an expanding Earth are (sic) either ambiguous or

are (sic) based on ad hoc assumptioiis." He then proceded to cite as the chief evidence, the

argument of océan volumes put forward long ago by Egyed (19.‘i6) but which has not been

used as support for Earth Expansion by its proponents for the past three décades. After dismissing

this piece of evidence - which is a pyrrhic victory since it hasn’t been taken seriously as support

for Earth E)xpansion by expansionists themselves — he continues by stating that “other évidences

— 475 —

for an expanding Earth are even more tenuous,” again citing examples that are not used by the

majority of proponcnts of the theory, including lhe odd notion that the Earth is expanding ac-

cording to IlUBBLE’s Law and the idea that significant Earth Expansion is related to decreases

in the gravitational constant over time.

Consequently his ilNinformed dismissal of Earth Expansion on the basis of false criteria is

unfortunate, especially so because it gives sludcnis the incorrect impression at lhe outset of their

careers that those in authority hâve thought deepiy about the possibility of Earth Expansion, but

that they consider it so feeble a hypothesis that it merits Icss than a page of discussion in a

288 page hook. Invidious is the impariing to students of lhe idea that Earth Expansion rests

solely on such weak evidence as océan volume, HUBBLE's Law and changes in gravitational

forces. Barely a word is meiitioned about the weaith of other evidence for expansion which has

been published in support of lhe theory. In effect, students hâve been denied access through

their texlboitks to ideas which could shape the way they view lhe world for tlie rcsl of their lives.

The textbook by Cox & Hart (1986) introduccs and dismisses Earth Expansion in two

sentences, referring to it as a pre-Plate Tcctonic Theory offered in lhe laie I95(fs to e.splain the

observation that mid-ocean ridges had rifts along their crests. Anyonc reading Cox & Hart’s

book and not knowing lhe lileralure. including mosi students, would be forgiven for thinking

that nothing has apparcntly been written about expansion since the 195()s. The alternative that

these authors imply is that Plate Tectonics provides ail the answers to global teclonic questions,

which renders such old ideas as Etu-lh Expansion unworthy of further considération. No réfutation

is offered by these authors. they merely discard the idea with the words “this theory raised so

many other questions that it never gained much credence.” We are not informed by the authors

of the nature of these “other questions" that were raised by Earth Expansion, many of which

are discussed herein.

Tethys and Tethyan Shear

For much of the Mesozoic and Cainozoic, the Old World consisted of two broad régions

of continental crust separated from each other by the east-west oriented Tethys Seaway, most

of which was floored by continental crust (OWEN 1983a, b). There was some genuine oceanic

crust north of India, but not the incredible area reqiiired by proponents of the constant dimensions

Earth. For details of a more crédible history of the Tethys and of the exient of Tethyan oceanic

floor, the reader is referred to OWEN's (1983b) maps. This relatively shallow water-filled de-

pression proved to be an effective barricr to many groups of leiTestrial veriebrates and inverte-

brates (Haleam 1981), with the resuit that the evolutionary historiés of northem and .Southern

continental faunas tended to follow different irajectories. Yet from time to time. land-bound

animais and plants managed to cross the Tethys Seaway. During the Palaeogenc, numerous mam-

malian lineages crossed Tethys from north to south, while very few managed to cross in the

opposite direction (PlCKEORD 1990b), suggesting that lhe water flowing through the Tethys .Sea¬

way had a southerly drift component. During the Tertiary, however, the seaway ceased to be a

continuons water body strelching from lhe Atlantic to the Indo-Pacific (Adam.s et al. 1983).

There were points of dry land contact between India and Asia on the one hand, sometime during

the Eocene and subscquenlly (Patriat & Achache 1984), and between Africa and Europe on

the other hand at the beginning of the Middle Miocene and in following periods (Adams et al. 1983).

— 476 —

The development of dryshod access between northern and Southern continents led to greatly

increased interchanges of their terrestrial faunas and floras. Some of these interchanges were

impressive enough to be idcntificd as first order faunal turnover puises.

However, the development of dry land access between continents has not been the only

cause of faunal turnover puises. Since the Middle Miocene, Africa and Eurasia hâve been more

or less in constant contact, yet during this period there hâve been several turnover puises, about

12 to 13 m.y. ago. about 8 to 7 m.y. ago and during the Plio-Pleistocene. Some other factor or

factors other than merely the development of dry-land contact between two land masses hâve

played a rôle in driving these post-contact turnover puises. The same applies to the India/Asia

pair and to the North America/South America couplet, where faunal turnover puises occurred

closely in time with those that took place in Africa (Alrov 1992, Barry et al. 1985, 1990).

Palaeontologists usually talk about climatic forcing, or other environmental prime movers as the

causes of such faunal turnover puises (e.g. Al.ROY 1992) although what led to the climatic or

environmental changes in the first place has seldom been specified.

The rôle that the Tethys Seaway has played in the évolution of Afro-Eurasian Tertiary faunas

is crucial. The only way to understand its biological rôle is to appreciale the history of its lithic

substrate (geotectonics) and that of the seawater that tlllcd it (eustacy) (Hai.LAM 1979). For the

former, three major classes of geotcctonic behaviour need to bc examined, ail of which could

be active concurrently;

a) Tethyan Torsion (or geotectonic movements oriented generally east-west);

b) Vertical tectonics (uplift of mountain chains, downthrow of basin-floors);

c) Continental Displaceraent (or geotectonic movements oriented generally north-south)

(Carey 1988).

As far as the Mediterranean région is concerned, there has been a long list of contradictory

descriptions and a laige vanety of hypothèses proposed to explain its complex geology. For

example. Df.weY et al. (I973i suggested that during the Cainozoic Era Africa had moved several

hundred kilométrés westwards relative to Europe. CarEY (1988) in diametric opposition to

DeweY and colleagues, observed that Africa has been offset several hundred kilométrés eastwards

relative to Europe. A third stance (BerTHELSON & Sengor 1990) is that there was little if any

longitudinal offset between the two continents. Many authors hâve suggested that Africa has

moved norihwards towards Europe during the Tertiary (e.g. Dewey et al. 1973) whereas. Carey

(1988) considered that the opposite happened - the distance between Africa and Europe increased

by 700 km during the Tertiary, a view which is supported hy DOGl.tONl (1993) and OwEN (1983b),

but which is denied by many geotectonicians. including a referee to an early version of this

paper. The lutter merely stated thaï the idca was “false" wiihout élaboration. With such a diversity

of observations and opinions, it is necessary to re-examine ail the availablc data with an open

mind. In iny opinion, lire only rcconstructions of the évolution of the Mediteitancan which corne

near to representing its actual history are those by OWEN (1983b) which are closely constrained

by cartographical, geological and palacomagneiic data. Without exception, ail the other versions

that I hâve examined - perhaps hâve 1 missed some - are based on one or more “distortions”,

either cartographie or geological.

The eustatic history of the Tethyan Seaway is equally, if not more complex than that of

its crust. Not only is there the complicated geotectonic history of the Earth’s lithosphère to

— 477

consider because it changes the geometry of the basins holding the water (Hallam 1979; Haq

et al. 1987), but there is atso the contribution made by changes in the volume of the world’s

océans (Carey 1988), such as for example occur during glacial and interglacial cycles, or during

long term éruptions of neonate waters from the mande at mid-ocean ridges and other volcanic

sources. Furthermore, the comet-like trail of hydrogen left by the Earth as it orbits round the

Sun (OWEN 1992) suggests lhat photo-dissociation of water vapour in the upper atmosphère lias

been going on since the origin of the Earth. an observation that indicates that there are continuai

net losses of water from the planel. Without the more or less continuons éruption of immense

quantities of water from the interior of the Earth. the planetary surface would long ago hâve

gone bone dry. Variations in the rates of photo-dissociative losses of water in the atmosphère

or in rates of éruption of neonate waters from the mantle may well hâve played a rôle in de-

tennining the volume of the océans during the past. Thèse would lead to global scale transgres¬

sions or to régressions, depending upon the photo-dissocialion/nconate water éruption budget.

At firsl glance. il would seem thaï .somc of the geotcctonic aspects of the globe eannot be

explained by Earth Expansion, but they arc in any case probably ultimatcly reluted to it. especially

as expansion was not expressed evenly over the surface of the globe Among these are the Tethyan

Shear and its related structures which encircle the globe, noted carly on by Du ToiT (1937) but

confirmed by CAREY (1988). Carey (1988; 285-286) propo.ses that “the Tethyan Torsion seems

to be due to the interaction of gravity and rotational inerlia." During the lifetime of the Tethys

Sea, much more oceanic crust was generalcd south of Tethys than north of it, al the saine lime

that ail continents (except Antarctica) moved appreciably noilhwards. As a resuit, the moment

of inertia of the northern hemi.spherc increa.sed compured to that of the .Southern hemisphere.

Because of tins différence in inertial moment, a sinistral torsion began to operate along the

Tethys Seaway, "wilh the northern çide lending to lag in rotation wiih respect to the Southern

side.” Carey (1988, lïg. 80) lias calculated thaï in the viciniiy of Africa and Europe, sinistral

torsion lias displaced Africa eastwards by sonie 700 km relative Lo Europe since the Cretaceous.

Whilst Carey's explanation of the Tethyan Shear is conlenlious. the existence of the shear

zone is less so, even though one référée of ihis paper slaied thaï “ihc Tethyan Shear with sinistral

componenl is unacceptable", but without slaling why he thought so. There is little doubt lhat

Africa has sheared eastwards relative to Europe by a di.stanee of about 700 km. In the région

of New Guinea to the Tonga Trench there has also been Ireraendous shear in a sinistral sense

to a similar cxleiil. South America lias sheared eastwards relative to North America, thus fonning

part of the same global shear zone. Under these circumstances, 1 do not see why the concept

of a sinistral Tethyan Shear zone is unacceptable. The idca may nced modifying, and previous

explanations may be incorrect, but declaring it "unacceptable” does not do away with the observed

sinistral offsets between Africa and Europe, between New Guinea and South-East Asia, and be-

Iween South America and North America,

Elsewhere in this paper, I link Tethyan Shear to the process by which the Southern hemi¬

sphere is underriding the northern one with a clockwise component of movement. Similarly,. the

Pacific rim dextral shear zone is underriding the opposite hemisphere with an anticlockwise

component of movement, seen on land as dextral shear in California.

— 478 —

Earth Expansion, the lithosphère and the planetary gyroscope

We should not reject the reality of continental displacement on the grounds that Plate Tec-

tonics is not an entirely suitable or complète modcl lo accounl for it. No one has yet satisfactorily

explained how the Harth's magnelic field originaied or is maintained, but this does not mean

that it doesn't exist. We can't reject the plate tectonic concept or aspects of it withoul having

deterrnincd an alternative explanation for continental displacement. The evidence thaï the distance

between mosi continents has increased during the past 200 m.y. cannot bc lightly denied (Owen

1983a, b), the only exception being provided by India and Asia. But even in the case of India,

from the break-up of Gondwana iintil the Cretaceous, it was distancing itself from its neighbours.

It has only been during the latter third of ils post-break-up history thaï il has closed the gap

between itself and Asia.

As Carey (1988) has poinled oui. after the Jurassic Period the distance between each of

the continents increased. Subduction of continental and/or oceanic crusl on a globe of constant

size does not accounl for this observation, which suggests that there musl bc anolher explanation.

Furthermore. reconsiruciions of past continental positions based on cartographically accurate plot-

ting of océan floor magnctic anomalies on a globe of constant dimensions resulis in gaping

gores or gaps (Owen 1983b). Making reconstructions using the same data on globes that are

smaller than the présent day référencé globe, éliminâtes ail these cartographie anomalies, as

long as globes of suitable sizes are selected (Owen 1983b).

Hallam ( 1981 ) concluded that the problem of excessive widih of the Tclhyan Océan during

the Cretaceous and Palacogcne “disappears if one accepts the novel reconstructions of Owen

because the Tethys (9cean is eliminated, but these are based on the assumption that the Earth

has expanded by some 20% since the Triassic." Hallam reported that he fell unable to promote

the idea of Earth Expansion “because of the serious geophysical difficulties posed by such a

rapidiy expanding planet." He did not speeify what the serious gcophy.sical difficulties were.

There can be liltle doubl that the principal cause of the increasc of the distance between

continents is their movemenl radially outwards with respect to the centre of the globe: Ihc pro¬

ximale mechanism is ocean-floor spreading and the symmetrical génération of oceanic crusl each

side of the spreading axes. Crustal masses can shift laterally about the surface of the globe by

a variety of mechanisms (llow, iransfonu faulls, nappes, folds, rifts, tethyan type torsion, differ-

ential rotation drag and so on) ail of which can rcsult in increased or decreased séparation between

two land masses.

The expansion of the Earth is not analogous to blowing up a balloon on which cutouts of

the continents hâve been attached. In such a niodel, expansion would tend to be homogeneous,

which is patently not the case for planet Earth. Instead, expansion is inhomogeneous: far more

surface area has been added in the Pacific hemisphere llian in the Atlantic/Indian hemisphere,

and the Southern hemisphere has increased its surface area greally with respect to thaï of the

northern hemisphere. The ocean-tloor palaeomagnciic patterns confirm this amply. In such un-

evenly expanding masses, there would be a tendency for the expanding body to swell out of

shape - the geotumor concept of Carey (1988)- but the force of gravity in a planet as large

and as dense as Earth, ensures that the geoid remains quasi-sphcrical. The adju.stments between

the tendency for the planet lo swell out of shape. due to inhomogeneous expansion, and the

— 479 —

tendency for gravity to maintain a spherical form in it, lies at the heart of the global geoteclonic

System, the surface expressions of which are visible world-wide, but principally in the fold moun¬

tain belts and rift valley Systems as well as the palaeomagnetic patterns that stripe the océan

floor.

The discovery of the palaeomagnetic pattern of the ocean-floor was early and correctly

interpreted as providing evidence that the continents hâve indeed been moving around. The mech-

anism by which the magnetic stripes were generated at spreading ridges is not seriously

Note how A increases from 1 to 3 while perimeter of circle

increases from 1 to 2, and then decreases from 2 to 3

Fig. 3. — From left to righi, the upper row shows the effect of drawing circles on a globe, in the way depicted in the second

row. From “1” to **2” area A increases in size, as does the length of ils perimeier: Area B decreases in sizc as ils bounding

perimeter increa.ses in length. Beyond the grcal circle stage shown in “2*’ to “3”. area A continues to increase in size, even

though ils perimeter decreases in length, while area B decreases in size. In a dynamic situation m which Ihere is uneven

expansion of one hemisphere of a sphere of constant dimensions - the sum of areas A and B is con.stani- such as is caused

by uneven génération of cnisl al spreading axes, and a force such as gravity which maintains the spherical shape of the

sphere, compressive forces will bc generated along a gieat circle zone as the perimeter of the expanding hemisphere is forced

to become shoricr. Even in an expanding sphere - in which the sum of areas A and B increases — the combination of uneven

expansion of one hemisphere and the force of gravity will Icad lo compression along a great circle zone. For this reason,

the Earih’s great fold belts and torsion zones occur along or close lo arcs of grcat circics.

— 480 —

contended, although what it means in terms of Plate Tectonics or Earth Expansion has been. In

fact, the spreading ridge evidence could as legitimately be considered the surface expression of

Earth Expansion rather than of Plate Tectonics. It does not uniquely support the Plate Tectonic

Theory as opposed to any othcr. The main problem with Earth Expansion which also applies to

Plate Tectonics. is the question of what happons at depths greatcr than 15 km where we hâve

so far been unable to "observe” or deduce satisfactorily what has been going on. even with high

resolution seismicity. As Coh in & Eldholm (1993) point out “ail models of the internai structure

of the Earth are built on inference."

Subduction of oceanic crust, principally near the Pacific rim, at first glance seenis to provide

evidence thaï réfutés Earth Expansion, in the sense that no crust should be subducted if the

génération of new cru.st is .solely by expansion. However, uneven expansion on a globe, which

is a feature of the Earth's .spreading ridge Systems, automatically leads to the genesis of cordillera

and trench Systems along Great Circle zones, along which limited subduction can take place.

This is becau.se gravity maintains the spherical shape of the globe against the. tendency for spread¬

ing ridges to develop a non-spherical globe.

Expansion has indeed not been uniformly expressed over the surface of the globe (Steiner

1977). Steiner calculated that both sea-floor spreading rates and subduction rates increased with

the passage of time, but that throughout the post-Jurassic period there appears to hâve been an

excess of sea-floor spreading over subduction. His conclusion was that an expanding Earth is

strongly indicalèd.

The génération of new crust has been highiy uneven. with much greater quantities generated

in the Pacific hemisphere than in the opposite une, and much more in the Southern than in the

northern hemisphere. Furthermore, génération of oceanic crust in the Pacific hemisphere appears

to hâve been exponential (Steiner 1977), so that in the constant “r” plate tectonic paradigm,

the subduction burden must also hâve been exponential. Such appears to hâve been the case,

but global subduction rates seem to hâve been lower than global ocean-floor spreading rates

throughout the post-Jurassic penod.

These facts, together with the tendency for gravity to maintain the spherical form of the

Earth are of the greatest importance. In our thought experiment above, we showed how on a

globe of constant dimensions, the perimeter of an expanding area reaches a stage - the Great

Circle - where the perimeter begins to shorten even though the area enclo.sed by it continues to

increase. In an expanding globe, uneven expansion has a similar effect, although the mathematics

is more complicaled. If there were no gravitational force to maintain sphericity, the expanding globe

would simply become distorted. much as a balloon with a weak patch becomes misshapen as it is

blown up. However. the force of gravity - the mosl powerfui .source of eriergy in the globe - maintains

the Earth's almost spherical form. The crust expériences and records the adjustments at so-called

subduction zone.s and fold belts, usually expressed along arcs of Great Circles (see Fig. 4).

Continents such as Antarctica and South America, which are surrounded by mid-ocean ridges

whosc traces appear to be enlarged “caricatures” of the continents they encircle, hâve suffered

littlc if any latéral shift over the surface of the Earth. Apart from its northern margin, Ihc same

can be said of Africa. These continents are said to hâve passive margins. Instead, the increase

in distance between these continents has been principally due to radial movement away from

the centre of the Earth by a mechanism of insertion of oceanic crust symmetrically on either

— 481 —

side of lhe mid-ocean ridge Systems. Since ihe Crctaceous Period, about 5000 km of séparation

of such continents, measured along Great Circle traces, can be accounled for purely by Earth

Expansion in the slow model of Earth Expansion (OwEN 1983b). In lhe fast model of Earth

Expansion espoused by Carey (1988). ail the .séparation of continents would bc due lo Earth

Expansion, and virtually none would be duc lo latéral movcmenls about lhe surface of the globe.

In passive margined continents, the continental crust appears lo be firmiy welded onto its

underlying mantle. Il is difficuli to imagine a System of convection cells with descending walls

beneath Africa, Australia and Antarctica. In Africa, in particular, lhe evidence frnni diamondifer-

ous intrusions into the crust. suggest that the diamond-bearing sub-cratonic rocks hâve noi moved

significantly since lhe Early PreCamhrian. The diamonds them.selves crystalhzed during lhe Pre-

Cambrian, yet ihey erupted lo the surface al varions limes during the Palaeozoic and Mesozoic

Eras (up lo Cretaceous limes). Under a convecting mantle System, such diamonds would probably

long ago hâve been carried away from beneath the continents perhaps to be spewed up in mid-

ocean ridge Systems.

Latéral shift over the surface of lhe globe is Icss important for ail continents excepl India,

which is lhe only large land mass to hâve moved any appréciable distance lalerally across lhe

surface of lhe globe (ihcrc are many so-called microconlinents or terranes that hâve donc so,

principally in Tethyan type torsion zones). But even in the case of India, initial séparation from

its neighbouring continents from the Jurassic until the Turonian (Upper Cretaceous) was due

largely to radial movement outwards from lhe centre of lhe Earth by the mechanism of sym-

metrical growth of oceanic crust at mid-ocean ridges, and il was only during lhe Latc Cretaceous

and Tertiary that India began its latéral movement acro.ss the surface of the globe (OWF.N 1983b)

and in so doing, decreased its distance from Asia. From the Jurassic until lhe Turonian its sép¬

aration from Asia had increased in a similar way and by comparable distances to its séparation

from its other neighbours (Madagascar. Africa, Antarctica. Australia). As a resuit, until the

Turonian, India used to be surrounded on ail sides by a mid-ocean ridge System which looked

like an enlarged caricature of the subcontinent. Today, ail that remains of this “caricature” are

the Carisberg and mid-Indian Océan ridges, the norlhern parts having been incorporaled into the

ophiolite belts of Oman, Afghanistan, Pakistan and India. Thus, India only became an exception

to the usual expression of passive margined continents during the Upper Cretaceous -meaning

that il conformed to the general pattern for more than half the time since the onset of the breakup

of Pangaea.

In other words, from the Jurassic until the Turonian, every continent - including India -

was dislancing itself from ail its neighbours, Only since lhe Turonian has India approached Asia,

thereby producing an anomaly and a conundrum. TTie only way to increase the distance between

ail continents is to increase the size of the globe, and this means that Earth Expansion must

hâve occurred.

Energy for Earth Expansion

The primary factors controlling the geometry and physiology of Earth Expansion and

lithospheric évolution is the interplay between Ihree energy source:

a) gravitational energy, the most powerfui known source of endogenetic energy in the globe

(Carey 1988; Marchal 1991);

— 482 —

b) the kinetic energy of rotation of the globe, the second most powerful source of energy

in the globe;

c) expansion energy.

Two versions of expansion energy bave been proposed (OWEN 1992; Larin 1993), both of

which involved lhe progressive decay of protons in a proton-rich corc inlo an atomic State by

escaping Iroin the core after which they combine with électrons aiready présent in abundance

in the niantle than either protons or électrons. OWEN (1992) has proposed that the Earth has a

plasma core which is progressively decaying to an atomic siale, not only resulting in great ex¬

pansion, but aiso in the génération of fusional heat. which contributes lo the Rarth’s geothermal

budget OwEN (pers. comm.) has suggested thaï changes in gravitaiional forces may affect weak

and strong force.s, and thereby play a rôle in plasma decay,

Larin il993), in conlrast. has suggested thaï lhe primordial Earth was hydrogen enriched,

with most of the hydrogen locked up within metallic crystal lattices. The dissolved hydrogen in

this situation is in lhe proton form, the atomic form heing loo large. With the passage of geological

time, protons cscape from the metallic crystal lattices. capture an électron and become hydrogen

atoms which are considerably larger than proton. The long terni resuit is expansion of the body

of the globe as it adjiisis to lhe increases volume of hydrogen so released. An attractive aspect

of Larin’s hypothesis is thaï metallic hydrides can be produced in the laboratory at pressures

of 2-3 megabars, and their behaviour studied.

In both Owi'n's (1992) and Larin's (1993) concepts. Earth Expansion is accomplished

without an increase in mass of the globe. This is an important point to note, because it has

often been assumed by plate tectonicians thaï expansion would automatically be accompanied

by mass increase. for which there is no evidence Indeed, the study of growih incrément in

corals. rhythms in varvites and other evidence concerning ancient orbital and rotational parameters

of the Earth, indicate that no signifiant increase in ihc Earth's mass has occurred since the Proierozoic.

A second important aspect to émergé from OwEN's and LARIN’s ideas. is thaï the changes

in llic Earth's radius would bave positive polarity. Decrease of ‘r’’ would not be possible unless

hydrogen was decomposed into protons and électrons and the protons once more inserted into

a plasma or a metallic crystal lattice. Al some stage in the early history of the planet, this

probably happened. but since al leasi the Archaen. the process has been one of protons decaying

to lhe atomic State.

Larin discussed many of the implicatons of his concept of the hydridic Earth, one in par-

ticular of which is of pertinence to the présent article. He suggested that the volume of the

hydrosphère would hâve increased with the passage of geological time. as hydrogen released

from atomic lattices reacted with oxygen to form water.

AIso discussed by LARIN are the global tectonic implications of a hydridic Earth, in particular

Earth Expansion, géosynclinal processes, flood basait épisodes, évolution of the océans and many

others. Larin appears to fall into the category of “fast expanders" discussed in lhe présent article,

in which lhe notion of subduction is rejected as a mechanism of global tectonics. For him, the

séparation of the continental masses has been by radial movement away from lhe centre of lhe

Earth, and none of il by latéral motion across or over the mantle. Needless to say, he discounts

the idea of mantle convection as a source of energy to drive plates, and along, with it, the

whole Theory of Plate Tectonics is considered untenable.

— 483 —

Hunt et al. (1992) aiso support the concept of Earth Expansion and consider that Plate

Tectonics is obsolète and should be discarded because it bas rendered geosciences torpid and

stagnant for the pasi three and a half décades. The authors provide several lines of evidence in

support of their contention that Plate Tectonics cannot possibly explain the distribution of the

contincnt.s and the océans. In their buok ExpaïuJiiig Geospheres they enter into some detail about

the Expanding Earth concept, although the book is not primilary conccrned with this theory. In

particular they discuss the importance of hydrides, carbides and silicides in modifying the planet’s

volume and its crustal and superficial features, and they di.scuss the origins of tlood basalts. the

origins of the hydrosphère, the nature of the Earth's core and a variety of other factors that

hâve played important rôles in the évolution of the Earth.

Geothermal energy within the inantle can only play a rniiior rôle in terms of overall global

tectonics, although its localized expre.ssions can appear to be dramatic (COFf'lN & Et.DHOl.M

1993). The iiltimate causes of Earth Expansion - such as could be produced by the capture of

électrons by protons in a plasma core which thereby produce atoms, the volume.s of which aie

considerably greater than the volume of either the original protons or the original électrons-

are a matter for further research, but the reality of Earth Expansion should not be denied because

for the time being we remain unsure of its ultimate cause. If we were to appiy the same logic

to glacial deposits, we would hâve to reject the former existence of glaciations because as yet

we do not fully understand what causes them.

GREAT CIRCLE TECTONIC BELTS

The two great Tertiary fold belts lie approximately along arcs of great circles mutually at

right angles to each other. Although the fold belts do not intersect, the great circle tectonic

zones that they are part of do so. once in the région east of the Himalayas and once just west

of the Isthmus of Panama. The Rockies and Andes lie along an arc of a great circle called the

Counter-Tethyan Torsion by Carey (1988), a Great Circle (rend which continues through Ant-

arctica, the Indian Océan via the Ninety East Ridge, the Eastern Himalayas. China. Kamchatka

to Alaska. The Alpine fold belt runs from Portugal and Morocco in the west to the Himalayas

in the east. deviating from a great circle where India has conlacted Asia and where the Alps

hâve been distorted by the Tethyan Torsion system. This System continues round the globe via

Malaya and Indonesia. across the Pacific to the Isthmus of Panama and thence across the Atlantic

to the Mediterranean,

Reconstructions of ancient fold belts of the Mesozoic and Palaeozoic reveal that these also

tended to develop along arcs of Great Circles (Eyees 1993. fig. 9.1).

Under the Theory of Plate Tectonics, there is no explanation for the observation that the

Tertiary and prior fold belts tend to lie along arcs of great circles. or to deviatc only slightiy

from such arcs, and thaï they are often parallelled by trench Systems, thrust Systems and torsion

zones. The Telliyan .system has sinistral torsion, the Cordilleran (Counter-Tethyan) one has dextral

torsion, despite the hold statement of one of the referees of this paper who declared that the

concept of torsion was “false”, but without saying why he thought so. Where the Tethyan torsion

zone emerge onto land, such as for example in New Guinea and between Africa and Europe,

and between South and North America, the evidence for torsion is difficult to deny. Where the

— 484 —

Pacific rim torsion zone émerges onto land, such as in California and New Zealand, dextral

torsion can be observed with the naked eye.

I consider it likely that the spécial relationship between great circles and fold belts is not

merely coincidental or accidentai, but is related to the way that the latter formed. I hâve already

explained how circles drawn on a globe from a point (or ptrle) can increase in surface area and

in perimeîer length until they become great circles. Beyond this stage, the surface area within

the circles continues to increase, but only if the perimeter of the circles decreases (Fig. 4).

The ihree fullowing paragraphs hâve bcen criticized by a référée of this paper.

In a globe such as the Earth in which new crusl is being generaied prefercntially in certain

sectors of the surface {Le. not unifonnly distributed over the surface of the globe), a stage will

soon be reached in which a great circle perimeter will be foreed to shorten in order that the

surface area in that hemisphere can increase. In a globe in which surface areas of widely separated

sectors are increasing in size. shortening along onc great circle trend is insufficient to accomodate

ail the increase. The resuit will be the lendericy for shortening of a second great circle more or

less at right angles to the Itrst, which will producc a second fold belt and related structures at

right angles to the first. Wherc these belts cross, there w'ill be evidence of great tectonic activity

producing complex fragmentation and mega-shearing of the crust. Shortening along a third great

circle mutually at right angles to the first two might be necessary in some crust expansion con¬

figurations, but seems not to occur on Earth.

In the case of the Earth. major surface area increase has occurred in the Pacific région. Its

related Great Circle fold belt is the Cordilleran System and its continuation round the Counter-

Tethyan torsion zone. The scetmd major area of expansion has occurred ail round Antarctica

and in the Southern parts of the Atlantic. Indian and Pacific Océans, its related Great Circle

fold bel! being the Alpine-Himalayan System and the Tethyan torsion zone.

Fig. 4. — Génération of Gréai Circle fold bells in an

expanding Earth. Frame A. the Triassic globe 80%

of the size of the modem globe undergoe.s uneven

expansion so thaï hemisphere A expands ut a slower

raie lhan hemisphere B- Frame B, repre.senLs the

hypoihdical shapc of such an e.xpanding globe if

therc wcrc no gravitaiional force acting on the

mavv- The globe wimld expand ont of shape. lhe

A hemisphere would end up significantly smaller

than the B hemisphere. even ihough l>olh were

large» lhan they were in a Triassic glohe.. Frame C,

the r<*rce of graviiy mainiuins the %phcrica] shape

of the uiicvcniy expanding globe, catising cnisi in

hemisphere B iihe more rajiitdly expanding hemi-

sphcfc) lo converge lowurds lhat of hemisphere A.

therehy fomiing a Great Circle ev>mprcss»onal front

{note (he struighl anows of frênne B am uow com-

pressed- pmduciug a hnnzonfai componeni of

movemeni loward*; ihe Great Cireje ct>mpression

front). Figuie l explalns why the compression oc-

curs along a Great Circle belt. These Great Circle

compressive forces produce the world fold belts.

subduelitjn zones and island ares (see Figure 4).

Thin arrowH-, torce of giavily; holîow arrows; ex¬

pansion energy.

— 485 —

The CTUSl on thc opposite side ot* the Great Circle to the expanding hemisphere resists the

tendency for shortening of the Great Circle. The resuit is that the edges of the expanding hemi¬

sphere override or underride the edges of the other hemisphere. It is this tendency, drivcn by

hemisphere expansion (unevenly distrihuted Earth Expansion) and gravily. that gives lise to thrust

fault Systems, folding of the crust into fold belts. trench Systems and other phenomena related

with the Great Circle tectonic belts of thc world. To a great extent strength inhomogeneitic.s

within the EartITs crust will déterminé where the shortening will begin. but once having started,

shortening activity will tend to continue along thc same trends. Furthermore. since it is uniikely

that overriding or underriding of crust would occur in a perfectly vertical sense. there will be

a tendency for some latéral displacement to take place, which would give rise to either sinistral

or dextral shear components. Once sinistral or dextral shear was established in a Great Circle

System, it would tend to continue shearing in the same sense. The Tethyan Shear lias been sinistral

throughoul the Cainozoic, while the Cordilleran Shear has been dextral. Thus the southem hemi¬

sphere can bc thought of as “screwing" itself under the northern hemisphere with a clockwise

motion, while the western hemisphere is “screwing” itself under the eastern hemisphere with a

counterclockwise motion.

Changing curvature of continents

One of the objections to Earth Expansion is that the curvature of continental surfaces would

hâve to change as “r” increased. DOOLF.Y ( 1983) for cxample, was unabic to rcconcile the rcquired

readjustment of curs'ature of Australia to the gravity data available for thaï continent. Carey

(1988) has responded to such criticisms, showing that Dooley’s gravity analysis was tlawed,

and in addition he has pointed out that thc numerous joint .Systems that characterize almost ail

continental rocks are probably a resuli of millions of microscopie adjustments due to curvature

change.

There may well be larger scale structures related to change in surface curvature of continents.

For example, the régional tensional re-entrants from the Guif of Guinea and on the opposite

East African coast may represent “sirelch marks” related to such curvature change in the continent

of Africa. The eontinent-wide basin and swell .structure of Africa (Carey 1988) could aiso rep¬

resent alternating compressional and tensional accomodaiions to changing curvature. As such,

a.spects of “membrane" tcctonics may play a rôle in the development of some of the continental

rifts and of their basin and swell structures.

One of the swells which has been studied is that of Namibia. It is represented by a north-east/

south-west oriented domain of higb topography, of w hich the Otavi Mountain Land is the north-

east extremity. On its Southern margin the elevated tenain is bordered by the Waterberg Thrust.

which suggests that the swell is essenlially compressional in origin. The Waterberg Thrust is a

fault over 240 km long with al leasi 4 km of overthrusting of Damara Marbles (PreCambrian)

over Mesozoic sédiments of the Omingonde Basin (Post-Triassic al the top) (LOdtkf 1970).

The outerop of the thrust is oriented north-east/south-west, with the marbles overthrusting younger

rocks lying to the south of the fault. This thrust took place after initial breakup of Gondwanaland.

Since breakup of Gondwana, the western coast of Southern Africa has been in tension and

has experienced uplift of over a thousand métrés. Since the Mesozoic there has been little or

— 486 —

no laterally direcled comprcssional forces in this région: many of the Mesozoic and Palaeozoic

strata of Namibia still lie essenlially unlilled and unfolded on top of PreCambrian rocks. How

then, can the existence of the Waterberg Thrust with its unmislakeable, even if localized. signs

of compressive forces, be explained? One possibility is that it represents a localized crumpling

of the crust due to change in curvature of the African continent as a resuit of Earth Expansion,

the energy source being gravity. A second possibility is thaï it was a response on the part of

the crust to forces set iip during a period of axial réorientation, the energy being drawn from

the kinetic energy of rotation of the Earth. What seems sure is ihat il does not hâve a ready

explanaiion wiihin the Plate Tectonics paradigm, since it occurs near a "passive margin” and is

oriented at right angles to the mid-Atlantic spreading axis.

PHYSICAL GEOLOGY (G Al A’S PULSES AND SECULAR CHANGES)

iNTROnUCnON

In Greek mylhology, the Earth goddess was Gaia. Lovelock (1990) has suggested that the

Earth bchaves like a super-organism which he calls Gaia, and which is, among oiher allributes,

able to fall ill and subsequently to heal itself. While Lovelock’s thesis is highiy conlentious,

and the subject of dcbatc, there is general applause for his attempts to view the biosphère as a

whole. While I do not ascribe to the Gaia cuit which has emerged from I.OVKLOCK's work, I

do maintain that the solution to many geological problems is only possible if one examines the

Earth as an entity, and not as a sériés of disjoinied sub-iolals.

Being a goddess, Gaia did not hâve a puise of human kind, but it is certain that she has

experienced numerous c.xogenic and endogenic pulse-like bchaviours during her long history.

These range in scale from day and night (daily). phases of the moon (mcnsual), winter and

summer (annual). the sunspot cycle (c.ll ycars), and glacial and interglacial cycles (0.5 m.y.

+/-), to tectogeiiic phases (7 tn.y. +/-) and many others at various scalcs of frequencies. Ail of

these and more bave left their mark, eilher in the geological record or in the fossil record of

beings which hâve inhabited the biofriendly planet Earth.

Tectoüenic phases

A second phenomenon that at first glance appears difficult to relate directly to Earth Ex¬

pansion, is the pulsed nature of geotectonic proce.sses (Scherba 1987; SCHWANN 1987; Stille

1924; C.AREY 1983). ScHERBA (1987) and ScHWANN (1987), following the researches of Stille

(1924), reccnlly summarized the lleld evidence for tectonic activity and sédimentation in the

parr of the Alpine fold belt which runs from Portugal to the Pamirs. These authors hâve confirmed

that the inlensity of teetogenesis fluctuaied in space and in time. Figure 5 extracts data concerning

olistostrome formation during the Cainozoic in 27 areas within this belt (Scherba 1987). To

the right of the figure is a hislogram of the number of areas experiencing olistostrome activity

at various limes. According to Scherba's data, there were several periods during which no olis¬

tostrome activity is known lo hâve occurred, and others during which up to 13 areas out of 27

experienced such activity. During the past 45 million years, for which there is sufficient data.

— 487 —

there were seven major periods during which olistostrome activity occurred. Because olislos-

tromes are generally generated under conditions of high energy, usually, but not aiways due to

uplift of source areas, and are often iriggered by earthquakes, lhe pulsed nature of the olistostrome

record suggests that orogenesis, to which most olistostromes owe their origin. was also pulsed.

These puises were called tectogenic “pha.ses” (or “epochs”) by Stille (1924). although it is

clear that they were not very regular in their timing. Pulsed tcctogcnesis suggests in turn that

energy expended during lectogenesis was supplied in puises (Schwann 1987), from which it

seems probable that whatever causes orogenesis does not run at a steady rate.

During the Cainozotc Era. lhere were at least nine tectogenic "phases” (Laramide I,

Laramide 2, lllyrian, Pyrenean. Savic, Styrian, Attic, Rhodanian and Valachian) (Fig. 3, Tables 3-

8) each of which lasted, on average, about 2.4 million ycars, and which were separated by rela-

tively tectonically “quiet" periods ranging in duration from 4 to II million years. Similar pulsed

tectogenic activity characterized the Mesozoic (ScftWANN 1987).

No convincing mechanism has yet been suggested to account for these phenomena, although

short term fluctuation in the gravitational "constanl” G has been proposed, partiy because the

implied rate of such geotectonic processes rcquires immense quantitics of energy which only

gravity can offer - in fact, the kinetic energy of rotation of the Earlh offers enough energy to

accomplish such activity. For the moment, to many geologists pulsed geotectonic activity remains

an empirical observation although some scientists even doubt the validity of the observation ihat

geotectonic processes can be pulsed in nature. Having seen some of the olistostromes in Oman

and Italy sandwiched between undisturbed strata, I consider that the pulsed nature of tectogenic

TECTOGENIC

PHASE

VALACHIAN

RHODANIAN

ATTIC

STYRIAN

SAVIC

PYRENEAN

ILLYRIAN

Fig. 5. — Terliary olistostrome activity in 27 sectors of the Alpine fold belt and a histograrn of the number of areas undergoing

such activity in each period of 1.25 m.y. There were seven major peaks during the past 50 m.y. (Data from Scherba 1987.)

— 488 —

Table 3. — Tectogenic phases of ihe Cainozoic Era. Tcctogenic activity took place during about 33% of the Tertiary, whilst

about 66% of the era was tectogcnically *‘quier’.

Tectogenic phase

Period of activity

m.y.

Peak of activity

m.y.

Stratigraphie

corrélation

(/ = boundary)

VALACHIAN

6,3-2.5

3.7

Laie Pliocène event

RHODANIAN

9-7.5

Late Miocene event

ATTIC

13-9

Mid-Late Miocene

STYRIAN

19-16

Early/Mid Miocene

SAVIC

28-23

24.2

Early/Late Oligocène

PYRENEA

41-34

35.5

Eocene/Oligocene

illyrian

47.5-44

45.2

Mid Eocene event

LARAMIDE 2

53-52

52.5

Palaeo/Eocene

LARAMIDE 1

66-64

CretaceousTTertiary

phases has bcen convincingly demonstrated by Scherba and his colleague.s. The famous Neogene

“nappes" of Daroca (Spain) and Perpignan (France) are of interest. because they moved into

place during tectogenic phases, yet they are relutively far from high relief features. The forces

that drove huge sheets of Cambrian liniestones to override Neogene sédiments (MN 04) at Daroca

during the Styrian or Attic phase were essentially horizontal. This suggests that the energy supply

was the kinetic energy of rotation of the globe, rather than gravitational energy. which has pre-

doniinantly vertical components of action. However, the Daroca nappe has been interpreted to

be probably extensional in origin, relaled to the gravitational collapse of the orogen (DOBLAS

pers. comni.), in which case gravity would have been the main force involvcd in its emplacement.

RIFTING PULSES

My studies of the Gregory and Albertine Rifts of Africa reveal that rifting tended to occur

in puises. Further work has shown that the major rifting épisodes coincide reasonably closely

in time with tectogenic phases of Scherba (Table 5). For example, in Kenya, major tectonic

activity look place about 17 to 16 ni.y., about 12 to 13 m.y., about 8 to 7 m.y. and during the

Plio-Pleislocene. In the Western Rift, Uganda ihere were major movemenis about 12 m.y., about

6 to 7 m.y., about 4 m.y., and during the Pleistocenc. In view of the enormous amounts of

energy required and the pulsed nature of the work donc. I feel that such pulsed aelivities may

be related lo the second most powerfui source of energy in the globe, the kinetic energy of

rotation (Table 2).

Continental (traps) and oceanic flood basalts (plateaux)

Orogenesis and its localized conséquences such as olistostrome formation and nappe em¬

placements, represent only a part of the gcotectonic activity that occurred during these phases.

Thcrc are others. including rifting, mega-fissure éruptions and possibly the Telhyan Shear which

encircles the globe. Major rifting epi.sodes in the African Rift System generally occurred at the

same time as tectogenic phases, as did major Irap volcanism, including the Deccan Traps of

— 489 —

Table 4. — Teniary Trap Eruptions.

Name of Trap Field

Period of activity

m.y.

Corrélation to

tectogenic phases

Ethiopian Basalts (Afar)

1.5-4.5

Valachian

Plateau Phonolites (Kenya)

12-13

Allie

Ethiopian Basalts (Central)

9-13

Atlic

Columbia River Basalts (USA)

15-17

Styrian

Ethiopian Basalts

21-32

Savic

Ethiopian Basalts

36-43

Pyrenean

North Atlantic Flood Basalts

36-43

Laramide 2

Deccan Traps (India)

54-56

65-67

Laramide 1

India at the same time as the Laramide 1 phase, The Ethiopian Traps during the Pyrenean Phase,

the Columbia River Basalts of North America during the Styrian phase and the Plateau Phonoliles

of Kenya coïncident with the Atlic phase, even though the record is less complété (or less well-

documented) for such activitics (MacDougall 1988; Carlson 1992: White, 1991).

CoFFiN & Eldholm (1993) hâve reviewed ihe available information concerning “large ig-

neous provinces" which consisi essentially of vast fields of lava, principally basaltic, which

erupted during rclatively brief geological periods (ranging between 1.5 and 4.5 million years).

The volumes erupted are indeed vast, the Ontong Java Plateau containing some 36 millions km^

of basait which erupted in less ihan 3 millions years. As COFFtN & EldhOLM stress, this rate

of éruption is equal to or greater than the rate of emplacement of new crust over the entire

global spreading ridge System. These aulhors conclude that the éruption of continental and oceanic

flood basalts was due lo brief but powerfui puises of magmatic activity. Furlhermore they dis-

cussed some of the effects thaï such pulsed igneous activity might hâve on the hydrosphère,

lithosphère, atmosphère and biosphère. The formation of the Ontong Java Plateau would hâve

caused a global .sea-level rise of 10 métrés unless there were isostatic adjustment of the crust

below the plateau.

The éruption of basalts is usually accompanied by aquaeous and gaseous phases, meaning

that during flood basait éruptions, immense volumes of water and gases would hâve been released

into the hydrosphère and atmosphère, depending on whether the éruptions took place on land

or under the océan. These injections of water and gases were sufficiently voluminous that they

would hâve altered seawater and atmospheric composition on a global scale. These in turn would

plausibly hâve affected the bio.sphere, and Coffim & Eldhoi.m make several provocative cor-

Table 5 — Main Periods of Rifting Activity in East Africa.

Period of Rifting

m.y.

Rift System

Corrélation to

tectogenic phases

5 to 4 and younger

Gregory and Albertine

Valachian

8 to 7

Gregory, Nyanza and Albertine

Rhodanian

13 to 12

Gregory, Nyanza and Albertine

Attic

17 to 16

Gregory and Nyanza

Styrian

— 490 —

relalions between the formation of large igneous provinces and mass extinctions, including the

extinction of the dinosaurs at the time of the éruption of the Deccan Traps, India, and the Early

Miocene/Middle Miocene faunal turnover puise.

Kinks in hot-spot tracks

ViNK ( 1984) dcscribed numerous hot-spot tracks in various parts of the globe and proposed

that hot-spots could be used as a Frame of référencé for determining relative plate motions through

geological time. The classic cxample, the Hawaiian-Emperor chain has long straight sections

with a simple age-distance relation.ship. Hawaii being at the young end of the chain. The proposai

of a hot-spot frame of référence has heen widely used in plate lectonic siudies. COX & Hart

(1986) for example, illustrate the concept with a drawing of a hall with nails driven into it, the

bail representing the corc and the nails the hot-spot plumes. The purpose of this drawing is to

reveal how .solidly fixed the hot-spots arc relative to the corc. However, as has been pointed

out by Kl-lTH (1993) and olhers. lhere are hot-spot tracks who.se behaviour does nol conform

to the nails in a bail model. including the Marquesas chain and the Cook-Austral chain which

show important violations of the simple age-distance rclationship which would be expected if

the nails were indeed fiimly hammered into a solid core. There are aiso serious peirologic and

geochernical problems with the hot-spot model. The Azores “hot-spot" for exaniple. is more

likely to be a wet spot lelated to a lowered solidus (due to the addition of COi and H^O to

mantle peridotites), rather than a hoi-spot due solely to elevated (emperatures.

An important point about hot-spot tracks is that they are comprised of relatively straight

sections mterrupted by dog-Ieg kinks, called cusps. On a parochial scale. hot-spot kinks could

be related to changes in direction of movement of plates. On a global scale, it could be due to

a shift of the entire crust relative to lhe mantle VtNK (1984) reported that lhere was a dog-Ieg

in the leeland hot-spot at 36 m.y.. which coincides closely in time with the "Grande Coupure”

and lhe Pyrenean T'edogenic Pha.se, A further dog-Ieg occurred al lhe KT boundary, which was

also a period of major faunal changes and lectogenesis, and a major one look place about 110-

120 m.y. ago (Barremian-Apiian). It is perhaps not coincidenial that major changes took place

concurrently in the lithosphère (as recorded by dog-legs in hot-spot tracks) and the biosphère

(as represented by the fossil record). Major tectogenic phases, first order biosphère changes and

bends in hot-spot tracks may well be varied aspects of lhe same overall cause - an episodically

shifting axis of rotation of the globe which would cause the Earth's crust to shift relative to

the core and the mantle. thereby inducing stresses and strains over the entire globe, and aUering

the positions of biogeographic realms.

The Tristan hot-spot has an intriguing history. From ils inception sorne 120 m.y. until 80 m.y.

it was located precisely on the mid-Atlantic ridge System, and thus produced a bilaterally sym-

metrical pair of hot-spot tracks - lhe Walvis ridge on the African side and the Rio Grande Rise

on the South America one. At 80 m.y. it shifted position relative to the mid-Atlantic ridge and

intra-plale hot-spot. located within lhe African plate (MiLNER et al. 1995; O’CONNOR & DUNCAN

1990; O’CoNNOR & LE Roex 1992) However. as far as lhe Walvis Ridge is concemed. ils behaviour

remained the same, whereas it stopped feeding material to the Rio Grande Rise which ends

abruptiy some distance from the mid-Atlantic ridge. This history suggesi that there is no con¬

vection ccll below the African plate, because, if there was such a cell. then the course of the

— 491 —

Tristan hot-spot should hâve bccn dcviaied by lhe convecting mantle through which it passes.

If the convection cell was strong enough to niove the African plate, then it follovvs that it should

hâve moved the Tristan plume. But it did not, which indicates that below the African plate the

convecting mantle model is incorrect.

In a récent publication. Norton (1995) demonstrated that the 43 m.y. cvenl in the Emperor-

Hawaiian scamount Chain, which has often been taken lo prove a change in direction of plate

movements at that time - the “43 m.y. events" - exhibits behaviour that is clearly delinked

from that of lhe varions piales thaï comprise lhe global geoiectonic machine. Reconstructions

of plate positions using the Hawaiian hot-spot as a référence arc thiis refuted.

It should be noted thaï the 43 m.y. event affected not only the lithosphère, but aiso the

biosphère. The most important faunal change of the Palaeogene apart from the “Grande Coupure”

occurred at this time (WOODHURNF. & SWISHER 1995). providing yet anolher example of the

close link between tectogenic events and biospheric changes.

Extinct spreading riooes

Several of the oceanic plates contain extinct spreading ridges (ESR). The most extensive

System occurs on the Nazea plate between South America and lhe active Pacific spreading ridge.

Others are smaller. but nevcrlheless provide evidence of shifling axes of crustal génération. Some

extinct spreading ridges. or portions there of may hâve been subducted. for cxample in the Cocos

Plate.

A plot of the âge of abandonmenl of these ESRs reveals thaï most, if not ail of them

became extinct during periods of tectogenesis in the Alpine fold belt (Table 6). Since teclogenesis

spanned approximalely 33% of Cainozoic time. the corrélations between abandonment of ESRs

Table 6. — Extinct spreading ridges and oiher anomalies in the océan Floor.

Extinct Magnetic Tectogenic Period of Kink in hot- Flood basait

Spreading Anomaly Phase abandonment spot track

Ridge

Sea of Japan

Attic

Cocos

Attic

Nazea

Attic

Iceland

Attic

Cocos

Styrian

Marion

Savic

Phillipines

16-17

Pyrenean

Iceland

lllyrian

Greenland

21-22

Laramide 2

SE Australia

Laramide 2

Baffin

Laramide 2

Coral Sea

Laramide 2

Agulhas

Laramide 1

Biscay

Phoenix

Wallaby

(m.y. +/-)

12.5

Plateau Phonolite

12.5

Plateau Phonolite

12.5

Plateau Phonolite

12.5

Plateau Phonolite

17-20

Columbia River

25-27

Ethiopien 2

37-41

xxxx

Ethiopien 1

45-47

49-53

54-55

North Atlantic

54- 55

55- 56

xxxx

Deccan/Mascarenes

72-76

121

xxxx

Ontong Java

— 492 —

and of tectogenic phases is not likely to be due to chance alone. (l is more probable thaï there

is a causal relationship between lhe two phenomena. It is noted that one of the ESRs correlates

in âge witb the major kink in the Hawaii-Emperor hot-spot track and there seems to be a cor¬

rélation between the âge of ESRs and of flood basait éruptions.

The simultaneity of these events suggest that there was a shift of the crust with respect to

the mantle on a global .scale, which caused increased tectogenesis, flood basait activity, increased

rifting, abandonment of parts of established spreading ridges and the birth of new ones and

changes in the axes of some hot-spots.

Causes of pui.sf.d activity in the lithosphère

Whatever the ultimate cause of tectogenic phases, rifting puises and episodie trap éruptions,

the contemporaneity of activily suggests that they were ail responses to the same overall prime

mover. This ultimate cause led to the génération of compressive forces in some sectors of the

Earth's crust at the same lime that il caused cxtensional ruptures and shears in olher parts. Earth

Expansion is a possible candidate although it is not spccifically known to occur in puises and

it would be unlikely ta produce conlemporaneous pulsed activity in widely scallered parts of

the globe. A more likely cause is the rcoricniaiion of the axis of rotation within the globe,

someihing that would produce stresses and strains throughout the fabric of Ihc globe and which

would naturally tend to be pulsed, due to lhe gyroscopic properties of spinning objects.

In the liieralure on Plate Tectonics, one seldom sees références tu pulse-likc behaviour.

Indeed, a basic assumption of lhe theory u.sed to be thaï rates of spreading al any one mid-ocean

ridge system were constant, although Patriat & Achache (1984) concluded thaï lhe spreading

rate of lhe ridge thaï “drove” the Indian plate into the Asian plate slowed down considerably

once contact between the two continents had been made. Subséquent radioisolopic daling of

DSDP borings has providcd a gcochronologic control on the characteristic palaeomagnetic sig¬

natures, which seems to confirm lhe constancy of rate of spreading at mid-ocean ridges. Such

constancy of rate has been laken to provide support for the proposai that geothermal energy is

the source of power for the plate tectonic motor.

Axial réorientation

It seems counter-intuitive to expect that a constant rate continental shift motor could produce

pulsed mountain building activily in the Alpine and other fold belts of the world. Inhomogeneous

expansion of the Earth, in which considerably more oceanic crust is generated in the Southern

hemisphere than in the north. logether with gravity driven adjustments to restore the geoid to

its quasi-spherical form, is unlikely to act in puises. H is perhaps more likely that the imbalance

induced m the rotating Earth by inhomogeneous expansion, would tend to ihrow the planetary

gyroscope out of equilibrium, which would then hâve a tendency to adopt a new axis of rotation

as the imbalance reached a critical value, Such readjuslments of the axis of rotation would nat¬

urally be epi.sodic rather than conlinuous.

Axial réorientation is not a new idea, its pedigree going back more than 110 years. Gold

(1955) considered that large-scale “polar wander" was inévitable if the long-tenn rheology of

the Earth is inelastic. GoldREICH & Toomre (1969) pointed out that in a quasi-rigid body such

as the Earth, large-.scale displacements of the pôle can be readily explained by small relative

— 493 —

displacements of material in lhe manlle. Furthermore, they suggesled that if mantle convection

causes the inertia axis to shift, so will the rotation axis change. Lambeck (1979) discussed the

matter at some length.

Rotating bodies tend to rotate round a relatively stable spin axis, but if lhere is a change

in the distribution of mass within a rotating body, the moments of inertia will change, which

will lead to a shift in the orientation of lhe axis of rotation (Vermeersen & VI.AAR 1993;

Lasker & Robutel 1993). lu rotating bodies in which the distribution of mass changes, there

is usually a thrcshold of imbalancc to cross before the axis will actually change. The axis of

rotation would tend to shift from one quasi-stable orientation to anolher. in which case the kinetic

energy of rotation of the globe would bc tapped in puises. Since the kinetic cncrgy of rotation

of the Earth is lhe second most pcrwerl'ul source of endogenelic energy in lhe globe (Table 2),

the Work donc during axial reorientation may be considérable. Orogenic bells, the Tethyan and Pacific

rim shear. nappes, continental rifts, abandonment of spreading ridges - giving rise to exlinct .spreading

ridges - and trap type, fissure éruptions may reprcscnl diverse aspects of such work,

Earth Expansion doe.s not occur uniformiy over ils surface. Most new crusi is generaiedat

the mid-ocean ridges which arc not disposed equally over lhe Earlh’s crust. Vertical uplifl of

parts of the lithosphère aiso occurs at various scales up to epcirogcnics and the eniire globe,

but gravity maintains lhe almost sphencal shape of the globe. The resuit of this non-uniform

style of expansion is that the global gyroscope can be rendered oui of balance (partly due to

the génération of geottimors due to unequally distributed insertion of new occanic crust at spread¬

ing axes, crust whose density is grcater than that of continental crust), This concept has been

described in detail by C'AREY (1988). Wilh lhe progressive production of gyrosopic imbalance,

the axis of the rotating Earth will eventually reach a threshold beyond which il will gradually

change ils orientation to a new' position of maximum moment of inertia. This implies that the

pôles of rotation are not fixed, a phenomenon which has been known for nearly a century (Carey

1988; LAVALLARD 1988).

EfFECTS of axial REORIENTATION

Repositioning of the Earth’s axis of rotation will lead to the génération of stresses and

strains in the fabric of lhe globe. Tarling & T.arling (1971), in an argument aimed against

the concept of pôle repositioning, considered that such a repositioning would lead to fragmen¬

tation of lhe globe, or al least that it would induce “major adjustments in the lithosphère", for

which they said they could see no évidence. They werc apparemly implying that polar reposi¬

tioning would release so much energy thaï it would cause fragmentation of the globe. However,

if repositioning occurred over geological lime scales raiher than rapidiy, then much of the stress

and strain would be released non-destruclively. mostly as heat (.see section on océan warming

peaks) and as adjustments to the lithosphère such as nappe movements, fold mounlains, riliing

and shears. We must keep in mind thaï at the scale ot hundreds of kilométrés and of millions

of years. the apparently rock .solid lithosphère hehaves like a fluid (Marchal 1968). In my

opinion, much of lhe geological evidence currently interpreted within the framework of Plate

Tectonics, such as fold mountain belts, nappes and rift valleys, represents the “major adjustments

in the lithosphère” thaï Tarling & Tarling had so much difficulty envisaging while they were

rejecting the concept of axial réorientation.

— 494

Lithosphère at the equalor is rotating at a speed of 1670 km/h while that at the pôles is

rotating at 0 km/h. If the position of the equator were to change, then crust formerly at the

equator would be rotating at a slower speed than previously, whereas crust underlying the re-

positioncd equator would be rotating at a higher speed. Thèse différences would induce accél¬

ération or décélération forces In the F.arih's crust. These forces would act tangentially to the

radius of the Earth. In most places, the crust would ab.sorb the forces so generated by folding,

flowing, faiiliing and compression/relaxation, but occasionally the crust would rupture and sheets

of rock would override nearby rock masses. Where these sheets corne to overlie younger rocks,

they are rccogni/.ed as nappes, such as the well-e.\posed example of Precambrian (Damara) Mar-

bles which overlie .lurassic quart/.ites along the 240 km long Waterberg Thrust in Namibia, and

the Cambrian limcstones lying on top of Middle Miocene terrigenous strata at Daroca, Spain,

which dates from the Middle Miocene. Where they corne to overlie older rocks or rocks of the

same âge. such sheets of rock tend not to be recognized as nappes, but are generally mapped

as concordant rock units.

The orientation of the major transform traces in the .^tlantic suggests at first glance that the

spin axis of the tiarlh has changed little during the pas! 200 millions years However, examination

of the orientation of major transform traces in the other océans indicates thaï these traces are not

causally linked to the axis of rotation of the Earth - they are independent Systems, just a.s the axis

of rotation and the magnetic dipole axis are delinked. At présent, for example, the magnetic axis is

inclined at somc 14° from the rotutional axis, and it is not fixed in geographical position.

Axial repositioning -inducing a "graticule shift” - would hâve other conséquences, includ-

ing climatic and biospheric changes (Creber & Chaloner 1984). One of these is related to

the fact that the rotating Earth has an équatorial "bulge” (polar radius; 6357.774 km; équatorial

radius: 6378.16 km). As the axis of rotation shifts. so would the position of the équatorial bulge,

and this would induce stresses and strains in the Earih's crust, as well as other cffects such as

transgression.s and régressions of water bodies due to différences in the tluidity of the lirho.sphere

and the hydrosphère. This is because the équatorial Gréai Circle has a circumference of

40,075 km, whereas polar Créai Circles hâve circumferences of about 39,945 km, a différence

of 130 km. If the equator changes position, the crust ahoad of il will beconie slighlly “siretched"

while that behind il will bccome slighlly “compressed”. Some of this strclching could be acco-

modaled at spreading ridges, but in wide continents such as Africa, .some localized effects such

as rifling and tilting of joint blocks may be expected. Such membrane tectonics has support

from a number of scientists (Doblas, pers. comm.).

As far as Africa is concerned, lherc has been a southwards propagation of rifting (Early

Oligocène in Red Sea, Late Oligocène in Ethiopia, Lower Miocene in Northern Kenya, Upper

Miocene in Southern Kenya and Western Uganda, Pliocène in Northern Tanzania and Plcisiocene

in Southern Tanzania and Malawi). Easl African volcanic rocks also tend to decrease in âge

from north to south, the onset of volcanism usually postdating the earliest signs of rifting by a

million years or so (basal strata in the African Rift System are usually non-volcanic), also sug-

gesting a southwards propagation through gcological time of the leosional cracks up which magma

flowed to rcach the surface. Could this southerly propagation of rifting and volcanism be related

to the proposed southerly movcmeni of the equalor during the late Cainozoic? Or is it more

likely to be related to Coriolis forces acting over geological lime periods?

— 495 —

POLE ORIENTATION AND ECOCLIMATIC ZONES

The graticule which geographers use as a référencé for locating points on the surface of

the globe (latitude and longitude) moves in phase with axial repositioning, because the graticule

is defined by the pôles, while East is defined by the sense of rotation of the globe. Thus if the

pôles shift, so would the equator, because the pôles define the equator.

Because the pôles of rotation of the Harth are oriented at a high angle to the ecliptic, the

solar radiation arriving at the Earth's surface is not tmiformly distributed (Williams 1993).

With modem obliquity, averaged over the year, solar energy reccipt is maximal at the equator

and minimal at the pôles (COLLtNGiKit'RNE 1976). This non-uniform distribution of solar energy

captured at the Earth's surface is responsiblc for generaling (hc ecoclimatic zones which encircle

the globe (tropical, subtropical, temperate. boréal, taiga and polar zones). Thus a change in orien¬

tation of the axis of rotation of the Eaith with respect to its surface (graticule shift), would

resuit in a comparable shift in the position of its ecoclimatic zones and this would in turn affect

the biosphère (CREfiER & ChalONER 1984). The latitudinally arranged ecoclimatic zones would

track changes in graticule position.

PULSED ENERGY EXPENDITURE

If the concordances highlighted in Figure 6 are indeed manifestations of a single underlying

cause in five different sy.stems (the solar System and four Earth Systems), then wc may conclude

that whatever was driving the changes acted in puises which usually lasted about 2 to 2.5 million

years, and which wcrc separated from each other by about 7 m.y. (ranges belwecn 4 and 11 m.y.).

The available évidence suggests that lhe most active periods of tcctogcnesis depended iipon

pulsed energy suppiy. It is less likcly that the gravitational constant changed in a pulsed manner,

or that Earth Expansion was it.scif pulsed (see Carey 1988, for a discussion about pulsed ex¬

pansion al scveral lime seules). The mosl likely mechanism that présents itsclf is that the axis

of rotation of lhe globe changed periodically from one quasi-stable orientation to another. and

that during the period of change (2 to 2.5 m.y.) the Earth's trust tapped into liie kineiic energy

of rotation, causing worldwide stresses and strains wiihin lhe fabric of the globe, recognized by

geologisis as nappes, rift valleys and trap fields that charactcrizc the lithosphère. Aparl from

gravity and lhe progressive decay of a plasma corc to the atomic state, other endogenic and

exogenetic energy sources are loo fecble to resull in the sort of lithospheric damage observed.

In a rotating, non-solid, expanding body, such as the Earth, the di.stribution of rnass within

the body is changing constantly. The Earth is a rotating fluid body. because, at the scales of

100 km and of géologie tinte, even “solid’’ rock behaves like a fluid (Marchai 1968). In the

Earth, therc arc several categories of redistribution of mass: almosphcric, hydrospheric,

lithospheric and probably aiso of the deeper layers of the Earth. Forces generated by the redis¬

tribution of atmosphcric masses (winds) are aiready known to lead to penurbaiions in lhe orien¬

tation of the axis of rotation of the globe (Lavallard 1988) by distances of .scveral métrés in

several years. Océan circulation patterns may hâve played a rôle in determining pôle position,

especially following fundamental changes in océan circulation patterns as the resuit of the opening

and closing of seaways between neighbouring continental mas.scs.

— 496 —

Redistribution of lithospheric mass, especially the iineven génération of oceanic crust at

mid-ocean ridges (greater quantities of new crust hâve been generated in the western and Southern

hemispheres than in the eastern and northern ones) hâve induced imbalance in the planetary

gyroscope in miich the same way that redistribution of mass in a revolving car wheel (due to

wear and tcar, or lo the addition or loss of extraneous mass lo the rims) can lead to “wheel

wobble”, Such “wheel wobble" results from the rotation of unbalanccd masses on a fixed axis

of rotation, tire “wobble" only becoming évident after an imbalance threshold has been crossed

(in the case of car wheels, the threshold is related cither to the amount of mass imbalance or

to the speed of rotation, but in rotating planets, the speed of rotation docs not change rapidly,

in which case the threshold for “wobble" will relate primarily to mass imbalance).

The car tyre, spinning on its fixed axis is prevented by its axlc fittings from adopting a

more balanced spin axis, with the resuit that il wobbles. Correction of wheel wobble is achieved

by adding suitable masses at appropriate locations on the rim to reslore balance. The Earth’s

axis is not fixed. and it is therefore free to take up any convenient orientation that maximizes

balance, but because of rolational inertia, an unbalanced spinning globe must cross a threshold

of imbalance before it will take up a new axis of rotation. Having adopted an axis of maximal

moment of inertia. the Earth w'ill tend to continue rotating about this axis until imbalance grows

sufficiently large that the imbalance threshold i.s crossed; thus changes will tend to be pul.sed.

If the internai forces generated by imbalance and axial reorientation are dissipated over geological

time. then planetary disintegration need not occur. If the réorientation is loo rapid, disintegration

may take place (Tarung & Tarung 1971). That planetary disintegration has happened in the

Solar System is suggcsicd by the présence of Ihc asleriod beli (a fragmented hypothelical planet

called Aster) between the orbils of Mars and Jupiter.

It is therefore proposed thaï the orientation of the spin axis of the Earth (and of other

planets and moons) has changed during geological history, and that stresses and strains induced

within the fabric of the globe by such changes hâve resulted in tectogenesis, rifting and volcanic

activity. as well as major changes in world climate (both by shift in the positions of ecoclimatic

belts relative to the surface of the Earth and by changes in the degree of seasonality due to

obliquity changes) and in the biosphère. Spin axis reorientation has occurred infrequently (9 times

during the Cainozoic) and took place over significant spans of geological time (2 lo 2.5 m.y.).

At a rate of 1 melre per year (measured at the Earth's surface) the spin axis could migrate up

to 2.3" of latitude in 2.5 m.y. if the movement was unidirectional. Spin axes would he quasi-stable,

which means thaï réorientations would be episodic, leading to pulsed changes in Earth Systems

sensitive to such changes (lithosphère, biosphère, atmosphère, hydrosphère). They would also

induce adjustments ta the cxtra-terreslrial relationsliips of the Earth, due to gyroscopic effects

(precession, lilting, nutation), which would aller the Earth/Sun relationship (graticule shift and

obliquity shift), which would resuit in climatic changes on Earth.

THE HYDROSPHERE

Océan Palaeotemperatures

Throughoul the Cainozoic there has been an overall cooling irend in the world’s océans,

interspersed by small warming peaks or by température “plateaux”. Various explanations hâve

been proposed to account for this global cooling trend (e.g. Barron 1985). Most attempts are

— A91 —

based on changing the amount of Sun’s energy reflected into space from the Earth by changing

some parameler on the surface of the Earth or in its atmosphère. I consider it more likely that

the trend towards global cooling is probably due to an increase in the total volume of seawater

that accompanied Earth Expansion (Carby 1988). As the volume of seawater increased over

geological lime, the solar energy lhat supplies most of the océan s beat energy - over 99% of

it, most of the rest being endogenic in origin - beeame dispersed through greater and grcaler

volumes of seawater. The increase in surface area of the Earth conipcnsaled lo some extcnt

because the total solar energy eaptured by the Earth would increase as its surface area increased,

even if lhe solar energy llux rcmained constant (mean solar llux at Eiarth's surface today is

about 350 watts per square melre). The température of the océans therefore decreased throughout

the Terliary, the cooling curve following approximately a slope of volume (L^) over area (L“).

Endogenelic heal (gcothermal llux) plays a rôle in warming the world’s océans, but its effects

are considerably less (mean geothermal beat flux a( lhe Earth’s surface is about 0.818 watts per

square meire). Howevei. durtng tectogenic phases (.see section on energy supplies) greater quan-

tities of endogcnctic beat energy may hâve been released into the environment, which coiild

hâve conlribuied to shorl-lived (c. 1 -2 m.y.) océan warming épisodes. Altematively, siight "pulse-

like" activily of the Sun could accoiint for these short-lived océan warming peaks and température

plateaux. A further possibilily is thaï temporary slow-downs in the éruption of neonatc waters

from the mantle occurred from time to lime, leading to a dccrca.se in lhe volume of lhe world’s

océans, and Ihus to an increase in its température as the incoming solar energy was absorbed

by smaller quantities of water.

The Miller & Fairbanks (1985) curve of 6'*'0 values obtained from benthic Foraminifera

of the Pacific and Atlantic océans shows that several short-lived warming events inierrupled the

overall tendency of global cooling that characterizes lhe Tertiary. Comparison of this curve with

information about tectogenic phases reveals thaï 6 oui of lhe 13 main warming peaks occurred

closely in time with 6 of the 9 tectogenic phases (Laramide 2, lllyrian, Pyrenean. Savic, Rho-

danian and Valachian Phases). From this we tenlatively conclude that tectogenic phases lended

to release great quantities of heal energy into the environment, including lhe world's océans,

which thereby experienced warming peaks.

Eustiicy

It has been suggested that eustatic régressions (Haq et al. 1987) of lhe sort identified by

+ symbols in Figure 6, were cold (Van Couvering 1988), that transgressions were warm, and

that régressions coincide with major faunal turnovers (Ginsbürg 1984, 1986). Examination of

the data (Fig. 6) reveals that there is not an obvious close corrélation between oceanic température

and changes in sea-level, nor of regre.ssions and faunal turnovers. On the contrary, most of the

major faunal turnover puises appear to hâve occurred during periods of high sea-level. This may

suggest that the .so-called eustatic tran.sgressions and régressions may not hâve been worldwide

events, but were more localized phenomena. !n any case, lhe Vail Curve shows so miich activity

that one could propose alinost any corrélation and hâve a good chance of finding a match. The

lack of corrélation between lhe Vail Curve and the ô'**0 curve is therefore probably significant.

The two phenomena of global cooling and of eustacy are not related to the same overall cause.

They appear to be varying completely independenlly of each other.

— 498 —

300 ^ 0 3 2 ^18 g 1 0

» I ■ ■ -'-'-1-

m.y.

Fig. 6 . — Coirelation of Oxygcn 18 curve (from Millkr & Fairbanks 19X5), sea-level curve (from Haq ei al. 1987), leetogenic

phases and faunal turnover pul.ses of the Tertiary and Quaiemary Eras.

However, each of the 9 tectogenic phases of the Cainozoic coincides in time with high

sea-levels, which are for the most part not interrupted by major régressions (more than 50 métrés

drop in .sea-level). Ont of 26 such large scale régressions which havc becn documented for the

Cainozoic Era. only two coincide in time with a tectogenic phase as recognized in Figure 3. 24

of the 26 régressions occurred during period.s of relative tectogenic quiescence. If régressions

were randomly scattered through time, wc would expect about 30% of ihem to hâve occurred

during tectogenic phases (tectogenic activity spanned in total about 30% of Cainozoic time:

22 m.y. versus 65 m.y.) This pattern of exclusion is so strongly expressed that it seems likely

that during tectogenic phases there was something that prevented or impeded great and rapid

sea-level régressions, whereas during periods of tectogenic quie.scence, such régressions not only

became possible, but occurred relatively frequently. The post-Valachian régressions are known

— 499 —

to be related to the build-up and subséquent decay of polar ice caps (Quaternary Glaciations),

and it bas been suggested lhat some of the more marked régressions and transgressions during

the Tertiary inay also hâve been related to the waxing and waning of polar ice-caps. It would

thus appear that the formation of polar ice-caps may hâve been intpeded during phases of tccto-

genesis, and that they could only form (and then decay) during leclogcnically quiet periods.

However, there were probably other factors affecting sca-levcl. inctuding the addition of massive

volumes of neonate water to the océans during emplacement of new lithospherie material at

spreading ridges, as well as changes in rates of photo-dissociation of water molécules in the

upper atmosphère, and major changes in basin shapes and depths during continental displacement

activity.

The polar and équatorial radii of the Earth differ by 21,386 km (polar radius; 6357,774 km;

équatorial radius; 6378.160 km). During graticule shifl. changes in polar and équatorial radii of

the lithosphère would lag slightly behind polar reorientation, bccause the lithosphère lakes time

to rebound isostatically (10 Ka for the rebound of Scandinavia after the waning of the lasi glacial

cap). The radii for the hydrosphère would adjust virtually instanlaneously, and lhe différences

between the two would probably lead to short-lived (scvcral Ka) transgressions in erstwhile

polar parts of the globe and régressions in erstwhile équatorial régions, shore-line topography

permitting. Transgressions and régressions would thus be taking place at the same time, albeit

in different parts of the globe (Weüenkr 1962). Because of this lhe assumption of contem-

poraneous same sense (either sea-level drop or sea-level increase) worldwide sea-level changes

during the so-called eustalic transgressions and régressions depicted in the Vail Curve, may not

be valid.

The ATMOSPHERE

During the Cainozoic Era there were several major reorganizations of world climatic .Sys¬

tems. During the Early (Ypresian) and Late Eocene (Priabonian), for cxample, Europe and North

America were tropical, wliercas aller the Grande Coupure, which correlates in time with the

Pyrenean Teclogenic Phase, they becarae temperate. At the end of the Lower Miocene, coincideut

in time with the Styrian Tectogenic Phase, worldwide seasonality increased dramatically in sever-

ily (Morales et al. 1993; Picki-ORD in press) and Europe became more tropical than it had

been before (palm-lrcc.s grew in Poland, Germany, France and Spain; mangroves grew in France),

whilc Central Africa became less tropical (forest in Zairc gave way to dune fields. the Namib

and Sahara Déserts became humid, large rivers flowed northwards from the Sahara towards the

Mediterranean, St Helena became sub-temperate, the Cape Flora occupied a much greater area

than it does now) (PiCKi*ORD 1991, 1992). At the end of the Miocene, at the time of the Rhodanian

Tectogenic Phase. Europe once more became temperate and Central Africa regained its tropical

character (forest returned to Zaire to grow on sand dunes over 200 métrés thick, desert returned

to the Sahara and Namib régions, the Cape tlora was squee/ed inio a liny corner of Southern

Africa). At about the same lime, monsoon climatic régimes were established in the Indian sub¬

continent (Quade et al. 1989). Following the Valachian Tectogenic Phase began the Quaternary

cycle of polar glaciations.

Reorganization of oceanic circulation patterns during the Cainozoic (such as occurred when

the Tethys Seaway connection between the Indian and Atlantic Océans was severed, or when

— 500

the Isthmus of Panama was. closed, ihcreby interrupting the équatorial currenl that used to flow

between the Pacific and the Atlantic), imdoubtedly played an important role in modifying global

climatic Systems, but such changes would be unlikely to lead to fundamental global scale changes

in degree of seasonality, because seasonality is primarily determined by the relationship of the

Earih to the Sun (obliquify).

The nature of some of the global scale climatic changes that occurred during the Cainozoic,

lend support to the concept that the orientation of the pôle of rotation of the globe has changed,

not only with respect to the globe itself (graticule shift), but also with respect to the plane of

the ecliptic (obliquity shift). T\vo pièces of evidence lend support to this idea, and are worth

repeating. The worldwide increase in seasonality thaï occurred at the end of the lower Miocene,

at the time of the Styrian Tectogenic Phase, suggcsts that the relationship between the Earth

and the Sun changed (its obliquity increased). At the same time. shift of the ecoclimatic belts

(Europe becoming more tropical, Central Africa less tropical) indicates thaï the coordinate grat¬

icule shifted, so that the equator was locaied further north in the Old World, and furtlier south

in the New World (PicKPüRD I99()b. 1991).

Earth Expansion implies an increase of the surface area over which the atmosphère is spread

-pre.seni surface area of the globe is 510 x 10* knr; Middle Miocene surface area was about

495 X 10* km^; Oligocène surface area was about 479 X 10* km'. The atmosphère itself probably

increa.sed in volume in pha.se with Earth Expansion, because of outgassing of the manlle during

volcanic activity and génération of occanic crust at spreading ridges, As the surface area of the

globe increased, there would bave been grcater total quantities of solar energy imercepted at

the Earth’s surface, even if the solar flux rcmained constant. If solar flux remained constant at

today's value of 350 watts per square mètre, the total energy captured at the Earth’s surface

during the Early Oligocène would hâve been about 1.32 x 10"'* joule.s/yr. During the Middle

Miocène it would have been about 1..37 x ) 0'’’joules/yr. Today it is about 1.4 x 10^"'joules/yr.

(These figures are calculaied as the amount of Sun's energy impinging on a circular dise of

great circle dimensions for a period of one year.) In a constant volume atmosphère, this might

lead to atmospheric warrning, but as far as we know, world atmospheric température tended to

decrease during the Cainozoic. The latier, if substantiated might indiçate that the volume of the

atmosphère - like thaï of the océans — was not constant during the Cainozoic, but further research

is probably needed to settle this point.

Evidence erom the solar sy.stem

PiCKFORD (1991) recently suggested that the variety of orientations of the axes of rotation

of the planets and moons in the .solar System provides additional evidence in support of the idea

that spin axes of heavcnly bodie.s are not fixed, but that they change over geological time. This

is not a new idea (Carey 1988). Williams (1993) provides basic data for the solar System and

discu.sses obliquity changes in Earth.

Uranus spins on its axis orienied al 98“ to its orbital plane, and four planets, including the

Earth, spin on axes which are markedly oblique to their orbital planes. In fact only two planets

in the solar System (Mercury' and Jupiter) rotate on axes which are nearly perpendicular to their

orbital planes and spin in the same .sense as the Sun. The .spin axis of Venus is also perpendicular

to its orbital plane, but il rotâtes on this axis in a rétrogradé sense, suggesting that it has “tumed

— 501 —

over” since it separaled from the Sun (if indeed the fission hypothesis applies Lo the origin of

the solar System plancls (Marchal 1968)). The satellite Triton orbits round the planet Neptune

in a rétrograde sense. Carey (1988) suggests that early during its histor>' the norlh rolational

pôle of the Earth was below the ecliptic rather than above it as it is today.

The great variety of pianetary orientations and spin directions could be cited as evidence

that the fission hypothesis for the origin of the solar System is incorrect, However. the fact that

the orbital planes of the planets arc almost parallel to onc another suggests a unity of origin.

probably by fission (Marcual 1968), in which case the varied orientations and spin directions

that characterize the solar system's planets andtheir moons would be due to post-fission changes.

The orbital planes of the planets in the solar System are nearly parallel to une another,

Pluto having the most oblique orbital plane at 17.2“ from that of the Earth. If planets originally

had axes of rotation perpendicular to their orbital planes, then pianetary spin parameters niust

subsequently hâve changed over géologie time. Such change.s would engender marked changes

within the planets, including the génération of stresses and strains in the fabric of the planet.

and would aller the Sun/planet relationship which would induce climatic changes on the planet

(Carey 1988; Williams 1993)

PHYSICAL BIOGEOGRAPHY

Biopulses

One of the more obvious kinds of biological pulsations that hâve occurred are “faunal turn¬

over puises”, which are relatively brief periods during which there were widespread (interconti¬

nental to global) changes in fieras and faunas an order of magnitude greater than “normal"

background evolutionary changes. First order faunal turnover puises are known to hâve occurred

episodically since the Vendian (Late PreCambrian), and include classic events at the PreCam-

brian/Cambrian boundary. at the end of the Permian and at the Lower Trias/Upper Trias boundary.

More recent events such as those that occurred al ihc KCT boundary (Cretaceous/Tertiary Bound¬

ary) and the “Grande Coupure" (Eocene/Oligoccnc Boundary) are infrequenl events. During the

Cainozoic Era there hâve been fewer than a dozen which could be called major worldwide events,

although there were many less important, more parochial faunal changes.

The literalure contains many descriptions of thèse faunal turnover puises, as well as ex-

planations, or partial explanations of whal may hâve caiised them (Vrba 1985: ALROY 1992;

McKenna 1985). The variety of mechanisms that hâve been proposed is surprisingly great, sug-

gesting that most of them are ad hoc explanations. Many of them are based on events which

were noi global but which were too parochial in scale. For example, the Messinian event which

supposedly led to the dessication of the Mediterranean at the end of the Miocene might be

expected to be correlated with changes in faunas and floras in neighbouring parts of the Old

World, but would it hâve induced changes in the New World such as was recently proposed by

Alroy (1992) for the Great Plains faunas of North America? Climatic change lias often been

invoked to explain faunal turnover puises, although what caused the climate to change in the

first place has seldom been evinced (Vrba 1985). Changes in physical geography of the Earth

— 502 —

hâve been suggested. these changes usually being related to plate teclonic activity. Volcanism

has been evoked, as hâve impacts of cornets wilh the Earth and there are many other hypothèses

of biological and environmental natures.

I would like to examine some of these suggestions in order to see if they can be ascribed

to one or to a few basic phenomena such as tectogenic phases, eustatic history, pulsed decrease

in global oceanic températures during the Cainozoic and others.

The ACCORDION MODEL OF ECOCLIMATIC BELTS

Other changes which hâve been proposed to hâve occurred to latitudinal ecoclimatic belts,

is that their latitudinal extcnt has varied over geological time, the belts becoming wider or nar-

rower (Pantic 1986. fig. 1). For exaniple. for Europe to bccome more tropical during the Mio¬

cène period, it suffices simply to propose that the tropical belt extended to 35'’-40‘’ latitude

either side of the eqiiator. This would mean that up to 64% of the surface of the globe enjoyed

a tropical climate (the area of the globe covered by the latitudinally arranged ecoclimatic belts

at various times in the pasi can easily be calculated if their latitudinal extenls are known, using

the formula given in Williams (1975. 1993). The area of a low-latitude zone symmetrical about

the equator is equal to the total surface of the Earth multiplied by ihe sine of the. limiting latitude).

Changes in latitudinal width of ecoclimatic belts of the Earth could also be caused by pre-

cession and nutation of the Earth (WILLIAMS 1993) but these would be rather minor compared

with changes thaï are reported to hâve occurred during the Eocene and the Miocene.

Under the accordion model of ecoclimatic belts, the présent day configuration, in which

the tropics cover little more than 10'' of latitude either side of the equator - i.e. the tropics cover

only 17% of the surface of the globe of which only ( 1 % occurs on land - suggests that since

the Miocène there has been a drastic réduction in the areal extern ol the tropics. During the

Pleistocene, wilh ils glacial/inlerglacial cycles, lhe latitudinal widths of the ecoclimatic belts are

often envisaged as having undergone several cycles of expansion and contraction, rather like

the bellows of an accordion while il is being played.

Al présent, about 39% of solar energy reachmg the atmosphère is immediately reflected

back into spacc (Caron ei al. 1992). If this figure were to change, then one could expect the

ecoclimatic belts to become narrower or wider, as the case may be. It is doubtful. however, that

widening of ihc tropics could bc so great thaï Ellesmere Island al 79"N latitude would become

warm-temperate or tropical wiihout a displacement of the position of the belt,. or without a change

in obliquity. as has been postutated recently. It is more likely that the width of the ecoclimatic

belts flucuiated by a few degrees of latitude only (Pantic 1986), and that their position on the

globe changed in phase with axial reorientation, or perhaps with average rotation - spin plus

nutation.

The tropical belt is characlcrized not only by higher annual mean températures than

temperate, boréal or polar régions, but also by a higher précipitable water content in the atmos¬

phère (Peixoto & OORT 1990). In tropical zones the average précipitable water content of the

atmosphère is belween 45 and 50 kg m* yr, while near lhe pôles values drop to as low as

2.5 kg nr yr. For Ellesmere Island to become tropical in terms of température and atmospheric

water content, similar figures as occur today in lhe tropics would hâve to occur there. Under

the accordion model of expanding ecoclimatic belts, huge quantities of water would hâve to be

— 503 —

transferred into the atmosphère worldwide. The volumes involved if such a model is invoked,

are so great that global sea-levels would be affected. If on the other hand. one invokes a shift

in the position of the ecoclimatic belt relative to the Eartli's surface (i.e. a graticule shift), then

the volumes of atmospheric water need not change very greally in order to achieve the same

resuit of tropicalising Ellesmere Island, as long as the graticule shift was of the appropriate

direction and degree. Graticule shift would aiso solve the problems of "seasonality*' and “polar

night" that the current location of Ellesmere Island endures, which renders il uniikely thaï tropical

condition.s suitable for crocodiles and lurtles could ever hâve been produced lhere without a

graticule shift.

The graticule shiet model of ecoclimatic belts

Repositioning of the ecoclimatic belts of the world does appear to hâve happened. For

example, during the Eocene, Europe and North America were both endowed with tropical cli-

mates. Dawson et al. i 1976) report the présence of lurtles, crocodiles and warm»tempcrate mam-

mals in the Eocene Eurêka Sound Formation ai 79"N latitude. If we consider thaï the North

American land mass has moved slightly .southwards since lhe Eocene,. conséquent upon expansion

of the Arctic Océan, then at ihc lime of sédiment déposition. Ellesmere Island would hâve been

slightly further north lhan it is loday.

For Ellesmere Island ai 80‘’N to bc tropical or subtropical during the Eocene withoul axial

reorientation, more than 98.4% of the Earlh's surface would fall within the tropical to sub-tropical

zone. Such a high figure would imply thaï dramatic changes occurred in the Harth/Sun energy

relalionship, however caused. Yet, during the sarne period, the distribution of oceanic organisms

reveals lhe presence of latitudinal stratification patterns not very different front thosc of today’s

ecoclimatic belts, excepl that their orientation and position on lhe globe were different. It is

thus more likely that the équatorial belt merely sbifted its position on the globe so that il in-

corporated Ellesmere Island, without the belt it.self becoming greally widened latiludinally.

Ralher abruptly - over a period of about 1 m.y. - at the Eocene/Oligocene boundary, both

Europe and America became more temperale. Traditionally this change has been atlributed to a

narrowing of the tropical belt - the accordion model - brought about by global oceanic cooling,

for which ô'*’0 data provides the mo.st convincing support (MlLLEH & Fairbanks 1985) The

accordion hypothesis would only work if the amount of solar energy being caplured by the Earth

varied, either because of variation in lhe output of energy by lhe Sun, or if lhere were changes

in the rétention of energy by the Earth. or if the obliquity of the Earth relative to the plane

of the ecliptic changed (Williams 1993) or if lhere were significant precession or nutation

effects.

An apparent problem is that palaeomagnetic data do not seem to indicate a major différence

between the position of the pôles before and after the "Grande Coupure”. However, because

there is today a différence of 14” between the axis of rotation and the magnetic axis, and because

palaeopole positions are calculated by averaging numerous palaeomagnetic measurcments - the

resuit being assumed to coincide with lhe ancient rotational pôle - the problem is more apparent

than real. The magnetic pôle could move ihrough more than 28" of latitude relative to the ro¬

tational pôle before it would show up as a measureable palaeomagnetic signature. Similarly, lhe

rotational pôle could shift through more lhan 28" of latitude before palaeomagnetic maps would

— 504 —

reveal any movcmenl. As Carey (1988) showed, palaeomagnetic methods are coarse tools for

determining the fine points of Earth history.

Further changes in the orientation of the earth’s axis of rotation during the Upper Miocene

eventually led to a southwards shift of the climatic belts in the Old World towards their présent

day positions. Indeed, the process is slill going on - in Africa the biological diversity “equator”

is slighily ttorth of the géographie equator, and is moving southwards. This is seen as a southwards

“advance" of the Sahara and related belts. although this process has been ereatly altered by

human aelivity in the recent past The présent day biodiversity “equator' of South America is

slightiy South of the géographie equator. and is moving northwanJs. From this il appears that

shift in the géographie position of the equator leads the shift in ecoclimatic belts. The current

lag belween the géographie and biodiversity equators in Africa is about 4“ of latitude (about

450 km), and a similar figure characterizes South America. In .South America, the géographie

equator has shifted northwards slnce the Middle Miocene (PiCKf-ORD lyOOb). For example, Platyr-

rhini - the New World monkeys - used lo inhabit Palagonia (55" South latitude) during the Middle

Miocene, a long way south of their présent day distribution limits in Argentina (35"South) (PtCK-

FORt) 1990b). During the Middle Miocene, St Flelena. which is today at a latitude of 16"S and

enjoys a tropical climate, was covered in plants of a subtemperate to austral kind. This suggests

that Si Hclcna lay wiihin a subtemperate lo austral ecoclimatic helt during the Miocene, and

that its présent day Iropicality is due to a subséquent southwards shift of the tropical belt in

the Old World (PiCKFORD 1991). This kind of evidence appears to support the idea that both

obliquity and pôle position of the Earth hâve changed during the Tertiary Period.

Obliquity and seasonality

If polar repositioning occurred, ihc obliquity of the Earth’s axis relative to the plane of the

ecliptic would change, becausc in rotating gyroscopes, a change in one mcchanical parameter

induces a change in anoiher, causing prccession, tilting, nutation and oihcr phenoinena. Sea¬

sonality is a fonction of the rclationship between a planet and its Sun. In planets whose axes

are pcrpendicular to the plane of the ecliptic, seasonality is minimal. In pltmets wilh low obliquity,

seasonality is weakly expres.sed. In planets wilh high obliquity', .seasonality is marked (Trfvisan

& ToNOtORGl 1987). Thus any change in obliquity of a planet will resuit in changes in the

degree of seasonality on thaï planet (Carey 1988; Wit.t.tAMS 1993).

PiCKFORD (in press), 'Van df.r MEUI.F.N Sl Daams (1992) and Morales et al. (1993) show

that the degree of seasonality greatly increased in tropical Africa and in Spain at the end of the

Lower Miocene and the heginning of the Middle Miocene. This change in sea.sonality was ac-

companied by a major latitudinal shift in African végétation and climatic belts (PiCKFORD 1992).

Namibia. which today is an arid to semi-arid counlry. became humid during the Middle Miocene

(Ward & CORBF.TT 1990). At the same time. Zaire, which is currently covered in tropical forest.

became a dune covered desert, and Libya and Tunisia which are now arid counlrics. were in-

habiied by tropical forest plants and were traversed by Ganges-sized rivers (PiCKFORD 1992).

At the onset of the Middle Miocene. Europe’s climate became more seasonal and more tropical.

These changes in the degree of seasonality led to some dramatic, and even startling effects in

the biosphère, such as the development of bony cranial appendages in numerous lineages of

— 505 —

herbivores and perhaps to the development of estivation behaviours among many animais, such

as terrestrial gastropods (PiCKFORD in press).

The BIOSPHERE

The basis for the subdivision of geological time is biological change observed in the fossil

record. While there is a relatively constant presence of background biological change in the

fossil record, due mainly to autochthonous adaptive changes, there were periods in the past

during which major floral and faunal changes occurred more or le.ss at the samc lime in many

parts of the globe and in many eco.systems, both marine and terrestrial. These are recognized

as floral and faunal “lurnovers" or “turnover puises".

There are ihree main ways that faunas and ttoras respond to environmental changes. The

constituent specics can remain where they are and either adapi to the new condition.s or go

extinct, or they can shift their géographie range in order to remain within their preferred habitat

type which is shifiing ii.s géographie position (Picxford l<!l90bF

In examining one of these major faunal turnovers such as the “Grande Coupure", one is

not w'itnessing the graduai spread of species, each one of which is adapting to sirange conditions

in its newiy colonized territory. Instead. one is in the presence of major changes in the géographie

distribution of animais and plants due to major displacements of biogcographic realms. In olher

words, they are related to changes in the climalic and other factors responsible for mainlaining

the positions of biogcographic boundaries, wiih the resuit thaï they arc displaced relative to the

surface of the globe. If geography (absence of water barriers, mountain ranges) permits il, thon

such “pre-adapted" faunas and tloras will quickly spread to the neighbouring continents or seas

wherever suitable conditions exist for them. If there arc none, then the species either adapi to

the changed conditions or they locally go extinct.

In Figure 6. the major palaeobiological turnover puises (mo.st if not ail of which were world¬

wide and which affected marine and continental faunas and floras alike) which occurred during

the Cainozoic are listed, such as those that took place near the Cretaceous/Tertiary boundary,

the Eocene/Oligocene boundary (the “Grande Coupure”) and the Lower Miocene/Middle Miocene

turnover (the gréai Afro-Eurasian faunal and floral inlerchange), among oihers (Pickford 1990b).

It is évident from Figure 6 and Table 8, thaï cach of the major faunal turnover puises coincided

in time with a tcctogcnic pha.se, which indicates that the lithosphère and lhe biosphère were

each responding in appropriaic ways to lhe samc ovcrall cause. It should be noted that the ex¬

tinction of the dinosaurs, ammonites and belemnjtes ai lhe end of the Cretaceous occurred during

a tectogenic phase - lhe Lararnide 1 Phase - and is thus not radically different in pattern or

cause from olher major turnover puises such as the “Grande Coupure" (Eocene/Oligocene bound¬

ary) which occurred al lhe .same lime as the Pyrenean Tectogenic Phase. The proximal cause of

the extinctions may well hâve been sudden change in global température due either to enhaneed

volcanism - such as lhe Deccan Traps which spanned the K/T faunal events - which affected

the Earth’s solar energy budget. The iiltimate cause may be of a geotectonic nature, such as

axial reorientation or tectogenesis.

During the “Grande Coupure", at least sixteen families of mammals invaded Europe, most

of which probably originated in Asia (Steininger et al. 1985). The invasion was part of a major

— 506 —

Table 7. — Major faunal turnover puises of East Africa.

Period of Turnover Puises

Nature of Faunal

change

Corrélation to

Tectogenic Phases

4 and younger

Omo mammal changes

Valachian

8 to 7

Modem lineages of

African mammals replace

arebaie ones

Rhodanian

12 to 11

'‘Hippariori’ event,

Dendromurines

Attic

17 to 16

80% change in mammal

faunas

Styrian

redistribution ol' lloras and faunas of ihe world. Contemporaneous effects in the seas and océans

were cqually dramatic. Europe, which had been tropical during the Eocene, became lemperate

during the Oligocène. Thus there was an influx of prolo-Palaearctic species into Europe, possibly

aided by the establishment of dry land contact between Asia and Europe at about this time. At

the same time, many of the mammals and reptiles adapted ro tropical Europe, such as primates,

crocodiles and turtles, became extinct there when its climate became temperqte These same

groups continued to survive in Africa and Southern Asia which by this time werc tropical in

nature. Similar changes occurred in North America, which contained a rich primate fauna during

the Eocene, but which had disappeared front the continent by the onset of the Oligocène.

The major changes in the distribution of land floras and faunas may have been aided by

significant fluctuations in sea-lcvcl thaï occurred during the Pyrenean Tectogenic Phase, These

sea-level changes werc not cau.sed by ice-cap formation, but may be related to transgression/re-

gression following axial réorientation (WliGHNi^R 1962. fig. 41) as described in detail below. It

is ai.so po.s.sible thaï the balance between photo-dissociation of water in the upper atmosphère

and the éruption (rf neonatc waters front the mantle into the hydrosphère was upset, a positive

budget resulting in global seule transgressions, a négative budget leading to régressions.

At the end of the Lower Miocène, the gréai faunal interchange occurred between Africa

and Eurasia, which is recognized in Europe as the base of the Middie Miocene (boundary of

mantmal zones MN3/MN4) (MCIN 1975). In Ea.st Africa, there was an 80% change in the mammal

faunas al this time, scveral littcagcs inmtigrating front Europe, but many entering East Africa

for the first lime from other parts of the continent. This faunal turnover puise was accompanied

by floral changes (palms, mangroves and tropical hardwoods colonized Europe, while formerly

forested régions of East Africa became semi-arid) which indicate thaï Europe became more tropi¬

cal thait it was during the Oligocène and Lower Mioccnc. In effect, large parts of Eurasia became

suitable for species adapted to prolo-Ethiopian biogeographic conditions (wet season/dry season

cycle instead of winter/summer cycle, for example) (PiCKFORD 1991, fig. 12). Interchange was

probably aided by the establishment of dry land connections between AfricaÆurope and

Africa/Asia.

— 507 —

Wallace envelopes - latitudinal diversity gradients

An interesting point about the mid-Miocene At'ro/Eurasian vertebrate interchange is that as

the diversity of some iineages of african large mammals decreased in Africa (primates and rumi¬

nants for example), (PiCKEORD 1991. figs 7. 8) the same Iineages tended to expérience increases

of diversity in Europe. In North America, large mammal faunas of the Gréai Plains aiso tended

to increase in diversity from Power Miocene to Middie Miocene limes (Alroy 1992) and sub-

sequentiy to decrease in Upper Miocene and Pliocène tirnes. In the Old World, ihis lendency

strongly sugge.sts that the Wallace ‘envelopes’ which describe latitudinal taxonomie diversity

gradients, shifted position, so that lheir peaks lay nearer Europe than ihey do loday - currently

they overlie tropical Africa near the equator. This of course implics that the position of the

equator has changed (gralicule change) (Pickeord 1991. fig. 20) since ihc mid-Miocene.

There can be little doubt thaï two-way faunal inlerchange look place during the Middie

Miocene and subsequently, but the development of dryshod access beiween the continental masses

was nol the only factor involved in the faunal inierchanges. Of gréai imporlance were biological

factors such as the position of zoogeographic realms relative lo each olher. Ai présent. Africa

and Eurasia are in dry land contact, yct alinost no inteahange of tcrrcstrial forms is taking

place for the simple rea.son thaï most of Africa currciilly lies wiihin a dilfercni biogeographic

région from Eurasia. Prévention of interchange is rcinforced by the présence of the Sahara Dc.sert

and the Mediterrancan Sca, but the main factor rcsponsible for ihe récent lack of inicrmingling

of and interchange between Palaearclic and Ethiopian faunas is hiogeographical in origin. That

the Mediterrancan is not an insurpassable banicr over géologie tiinc is shovvn by the tact that

during the Pleislocene, each tinte thaï there was a glacial period, the Maghreb was successfully

colonized by Palaearclic faunal éléments including deer. wild boar and bears. Conversely, some

characteristic mammals of the Ethiopian zoogeographic région such as Hippo/wiamus and

Theropiihecus (MoyA Sola et al. 1989-90) crosscd inlo Spain during intcrglacial periods.

For Europe to hâve heen colonized “en musse" by proto-Ethiopian mammals the conditions

in Europe would hâve had to hâve been more or less similar to those that exisled in Africa. If

not, only a few Iineages would hâve made the Iransfer after adapting to proto-Palaearclic con¬

ditions. In facl, it is widely accepted that al the heginning of the Middie Miocene. Europe’s

climate became more tropical and more seasonal than it had been during the Lower Miocene

period, thereby making conditions in Europe suitable for a wide variety of African mammals.

Thus continental shill only partiy explains the mid-Mioeene Afro-Eurasian faunal interchange,

and the faunal puises that came afterwards. nolably about 12 to 13 m.y. ago, 8 to 7 m.y. ago

and during the Plio-Pleistocene.

In many ways, the Middie Miocene Afro/Eurasia faunal interchange resembles the Great

American faunal interchange that took place during the late Cainozoic and early Quaternary

(MarshaI.L e/a/. 1982). which wa.s aided by the establishment ol the Isthmus of Panama between

North and South America. In the case of the two Americas, however, the results of the interchauge

were rather one-sided, even though approximately the same number ol' Iineages transferred in

each direction. Those Iineages invading North America from the soulh generally failed to undergo

evolutionary radiations and most of them died out after a rclatively short sojourn in the north.

Lineages invading South America from the north, in contrast. usually experienced successful

evolutionary radiations, many of which are slill extant.

— 508 —

Table 8. — Summary of main lectogenic. rifting. volcanic. oceanic and faunal events of the Terliary. Note that the absolute time

scale is lakcn from Scherba (1987) which is ont of date duc to révision of the gcological lime scale. However. the changes

are nol incorporated into this chart. even though the aulhor is aware of the différences, because he wishes to retain the data

as close as possible to the original author's writings.

POST-VALACHIAN

VALACHIAN PHASE: 4+/- m.y.

POST-RHODANIAN/PRE-VALACHIAN

RHODANIAN PHASE: 4+/- m.y.

POST-ATTIC/PRE-RHODANIAN

ATTIC PHASE: 12.5+/- m.y.

POST-STYRIAN/PR E-ATTIC

STYRIAN PHASE: 17.5+/- m.y.

POST-SAVIC/PRE-STYRIAN

SAVIC PHASE: 25+/- m.y.

POST-PYRENEAN/PRE-SAVIC

PYRENEAN PHASE: 37.5+/- m.y.

POST-ILLYRIAN/PRE-PYRENEAN

ILLYRIAN PHASE: 47+/- m.y.

POST-LARAMIDE 2/PRE-ILLYRIAN

LARAMIDE PHASE 2: 53+/- m.y.

POST LARAMIDE 1/PRE-LARAMIDE 2

LARAMIDE PHASE 1: 66.5+/- m.y.

Four major régressions. Cooling plonge in S’®0 curve.

Major orogenesis and olistostrome activity. Increased rifting

in Western and Eastern Rifts. Uplift of Ruwenzori and as-

sociated mountains. Isthmus of Panama closed. Onset ot

Quaternary Giaciations after océan warming peak.

One major régression.

Orogenesis and olistostrome formation less than maximal

but still important. Major accélération in rifting activity in Gre-

gory, Nyanza and Albertine Rifts. Ecodimatic belts shift sou-

thwards in Old World, northwards in New World- Installation

ot “modern" mammalian lineages (including hominids) in tro¬

pical Afriua Wîjiriiliip peak in océans

Two major régressions.

Major orogenesis and olistostrome activity. Plateau Phonolite

éruptions. Mid-Miocene/Upper Miocene faunal changes.

Two major régressions. Warming peak in océans followed

by cooling plunge

Major orogenasis and olistostrome activity. Daroca (Spain)

nappe emplacement. Onset of rifting in Qregory and Nyanza

Rifts. Africa “docks” ta Eurasia. severing océan circulation

between Indian and Atlantic Océans via Ihn Mediterranean

Seaway. Early Miocene/Middie Miocene faunal turnover

puise. Marked increase In seasonatity worldwide. Slight

océan warming. Ecodimatic belts shift northwards jn Old

World, southwards in New World. Circum-antarctic current

established. Columbia River Basalts.

One major régression. Océan température steady.

Widespread orogenesis and olistostrome activity. Onset of

rifting in Red Sea Rift and Ethiopia. Late Oligocène faunal

turnover. Warming peak in océans.

Five major régressions. Cooling plunge in océan waters fol¬

lowed by slight cooling and warming towards end of period.

Major tectogenic activity with widespread olistostrome for¬

mation. Lirai Seaway closed. Eocene/Oligocene faunal turn¬

over puise (“Grande Coupure”). Southwards shift of

ecodimatic belts in Europe and North America. Océan war¬

ming peak. Eruption of Ethiopian Traps. Dog-Ieg in hot-spot

tracks,

One major régression. Océan cooling followed by warming

at end ot period.

Widespread orogenesis and olistostrome activity in Alpine

Fold Belt. Mid-Eocene faunal turnover puise. Océan warming

peak.

Three major régressions. Océan cooling.

Orogenesis and olistostrome activity, Palaeocene/Eocene

faunal turnover puise. Océan warming peak

Five major oceanic régressions. Warming trend in océan wa¬

ters.

Widespread orogenesis and olistostrome activity. Deccan

Traps éruptions. South Atlantic opens. Cretaceous/Tertiary

faunal tufnover puise (extinction of dlnosaurs, ammonites,

belemnites, rudists, etc.) Océan warming. Dog-Ieg in hot-

spot tracks.

— 509

The differential successes of the two American inlerchanging faunal groups is explained

by Wallace’s Ruie. In most groups of organisms on Earth, diversity is highest near the equator

and is lower at higher latitudes (Stehli 1968; PiCKFORD 1991, fig. 2). North American lineages

invading South America would be approaching the equator (its Early Quaternary orientation was

not very far from its présent position, possibly a few degrces of latitude furiher South thaii it

is now), and would ihus he entering the “high diversity” part of their Wallace "cnvelopc". They

would therefore hâve a lendcncy to increasc in diversity (or not to decrease). In contrast, South

American lineages moving northwards would bc entering the low diversity parts of their Wallace

“envelopes” and their diversity would nalurally tend to decrease (or not to increase).

CONCLUSIONS

Several conclusions can be put forward.

1 — Of the four possible théories concerning the size of the Earth - 1) constant “r”, 2)

increasing “r”, 3) decreasing “r” and 4) fluctuating “r” - the available evidence now favours

the view that “r” lias been increasing over geological time. It is less likely that “r” has remained

constant, and even less likely that it has been decreasing or fluctuating.

2 — During the Cainozoic Era there hâve been major coincidences in the timing of world¬

wide changes or events in the lithosphère, biosphère, hydrosphère and atmosphère, as well as

in the Earth/Sun relationships which define seasonality and the location of ecoclimatic zones on

the Earth‘s surface. William OF OCCAM would suggesi that we search for a single underlying

cause, which we call Gaia's Puise after the Greek Earth goddess.

3 — Plate Tectonics on its own in a constant dimensions globe, is unlikely to explain the

pulse-like behaviour m these varied Earth Systems, nor the changes in Earth/Sun relationships.

In any case, the source of power sufficient to drive the plate tectonic motors (one for each

plate?) remains to be identified. gcothermal energy (expressed as convection cells in the mantle

which arc supposed to move the crustal plates) being far to feeble to accomplish the activity

observed. This suggests that Plate Tectonics on its own may well be an inappropriate mechanism

to explain displacement of the continents and the origin of oceanic crust.

4 — There are two known - and possibly a third - sources of energy capable of accom-

plishing the work expended during continental displacement, polar repositioning, Tethyan shear

and other manifestations of geotectoniç activity. The two known sources arc gravity and the

kinetic energy of rotation of the globe. The third possible source of energy is the progressive

decay of a plasma core to the atomic .State which would not only lead to tremendous increase

in volume (the volume of atoms is orders of magnitude greater than the volume of protons and

électrons) but would also generatc vast quanlitics of heat energy.

The pulsed nature of many of the changes indicates that kinetic energy of rotation was

being tapped periodically, while gravity, the most powerfui source of energy in the globe, has

of course been constantly at work iu maintaining the almost spherical shape of the globe against

the tendency of other forces to bend it out of this shape. Volumétrie increases duc to decay of

a plasma core to the atomic State would lead to Earth Expansion. Thus the figure of the Earth

— 510 —

is predominantly shaped by the interplay of at least two, and possibly tbree major sources of

endogenetic energy and several minor ones - boih endogenetic and exogenetic.

5 — There appear to bave been at least five main mechanisms which bave contributed to

the évolution of the Earih’s lithosphère during the Cainozoic:

Earth Expansion which produees an increase in the Earth's radius which leads to a net

radially outw'ard movement of the continental masses and océan floor, causing the distance be-

tween neighbouritig continents to increase. A major surface expression of this expansion is the

insertion of symmeirical slices of mantle material either side of spreading axes. It also leads to

the transferral of vasl quantities of neonate waters from the Earth’s interior to its surface, where

much is subsequently lost by photo-dissociation in the upper atmosphère.

“Lcad" and "Drag" effecfs induced in the Earth - a quasi-sphcrical spinning fluid mass-

due to différences in radial distance of various Earth layers from the axis of rotation. One global

structure possibly due to this eflect is Tethyan Torsion which has produccd a shear zone some

700 to 1000 km wide right round the globe in which numerous “microconlinents” hâve been

rotated sinistrally. Thèse include Spain and other Mediterranean microcontinents. India, Seram,

New Guinea. Mesozoic Mexico, and the Greater Antilles. A second structure at right angles to

the Tethyan Torsion zone is the Circurn-Pacific Tor.sion, in which rotation of microcontinents or

“terranes" has gcncrally been dextral.

Axial réorientation which leads to pulsed changes in the lithosphère, including tectogenic

activity, rifting, trap volcanism and shift of équatorial bulge, and which induces substantial

changes in other Earth Sy.stcins (hydrosphère, atmosphère, biosphère) and celestial pararneters

(graticule and obliquity changes).

Unequal disposition of spreading ridges at the Earth’s surface has led to unequal expansion

of the globe, producing “geotumors”. The Earth's figure remains quasi-spherical despite the un¬

equal growth, duc to the action of gravity, the most powerful energy source on Earth. Due to

the constraints of spherical geometry on a globe, this unequal expansion of océan basins, espe-

cially the Pacific Océan on the one hand and the circum-Antarctic océans on the other soon

reachcd the stage beyond which their Great Circle perimeters had to become smailcr than Great

Circles. The rcsulting relative compression of these Great Circle perimeters has been the for¬

mation of great circle mountain fold belts, the Cordilleran and Alpine orogenies respcctively

and their related thru.st Systems, trench Systems and torsion zones. Sinistral and dextral torsion

associated with these great circle compression zones may be related to the tact that the Earth’s

crust and mantle are not homogeneous, which would make it uniikely that compression would

be perfectiy vertical. A “screw" effect would thus tend to develop spontaneously. and once having

developcd would be maintained in the same sense during further great circle compression.

The latéral (as opposed to radial) movement of continental crust has occasionally occurred

(India is the prime example, but there are many smaller lerrancs whose présent positions are

best explained by latéral movement), but on a global .scalc such movements of continents are

less important than those due to the combination of uneven Earth Expansion and the maintainance

of a quasi-spherical shape by gravity. Rotation ot continental masses and smaller terranes has

also occurred (Australia, Iberia, Italy, etc.) but are in total a relatively minor -even if dramatic -

aspect of global geotectonics.

6 — There hâve been at least 9 major tectogenic phases during the Cainozoic Era, each

of which lasted between 2 and 2.5 m.y. and which were separated from lhe others by between

4 and 11 m.y. (average 7 m.y.). Similarly episodic tectogenic activity characterized the Mesozoic

Era.

7 — The extinction of dinosaiirs, ammonites, belemnites and rudists at the end of the

Cretaceous coïncides in time with a major tectogenic phase (lhe Laramidc I Phase and its as-

sociated volcanic and rifting activity). As such, lhe K/T faimal turnover puise does not appear

to differ radically from other major Cainozoic faunul limiover puises thaï coincided with lecto-

genic phases, such as the “Grande Coupure" (Pyrcnean Tectogenic Phase), the Lovver Mio-

cene/Middle Miocene (Styrian Tectogenic Phase) faunal turnover puise and the End Miocene

faunal turnover puise (Rhodanian Tectogenic Phase) ail of which were characterized by massive

faunal turnover (up to 80% faunal change) including the extinction of nunierous mammalian

lineages.

8 — Several tectogenic phases coincided in time with first order fissure éruptions (Deccan

Traps (65 m.y.), Ethiopian Traps (45-35 m.y.), Columbia River Basalts (16 m.y.). Plateau

Phonolites ( 12.5 m.y.)) and with épisodes of rifting activity in the African Rift System ( 17.5 m.y.,

12.5 m.y., 7.5 m.y., 4 m.y.).

9 — Several tectogenic phases coincided in time with the abandonmeni of spreading ridges

and the fomiaiion of nevv ones some distance away. The contemporaneity between tectogenesis

and abandonment of spreading ridges suggests thaï the Eaith's crust shifted with respect to the

mantle, with the conséquence that some of the upwelling masses at spreading ridges were shifted

laterally relative to the crust, Ihercby forming a new spreading ridge and leaving an extinct

spreading ridge off to onc side.

10 — In general, periods of tectogenesis coincided with periods of high eustatic sea-levels,

whereas major régressions in sea-level tended to occur during periods of prolonged tectogenic

calm. During periods of polar repositioning (graticulc shifl) the "équatorial bulge" would change

position, but because of the différence in “relax limes" (udjustmenl to isostucy) between the

lithosphère (c.lü Ka for Scandinavia after the waning of the last glacial cap) and the hydrosphère

(0 Ka) areas near the “relaxing polar Oattening” would tend to expérience transgression while

areas near the "relaxing équatorial bulge" would tend to expérience short-lived régressions. It

is noted however, that there are many causes for transgressions and régressions which operate

at many seules from the very parochial {e.g. localized tcctonic activity) to the global (e.g. gla¬

cial/interglacial cycles).

11 — The cooling trend thaï has characterized the world’s oceanic waters during the past

55 m.y. is probably due to incrcases in lhe volume of the world's océans, the "new" water

having been released into lhe hydrosphère from lhe mantle at the same lime as the hasaliic

rocks which accreted at mid-ocean ridges and elsewhere, lhe two being parts of the same overall

process, Earth Expansion. The total amount of solar cnergy per square métré çaplured by sea-

waters may hâve remained relaiivcly constant through géologie time. but as time passed the

energy so captured was diffused through greater quantities of water (length“ compared to length^),

hence resulting in a relatively steady cooling trend in the world’s océans throughout the Caino¬

zoic.

— 512 —

12 — During the same period, puises nf energy were expended during tectogenic activity,

some of which (as beat) may bave contributed to océan warming, recorded as “warming” peaks

and température plateaux in tbe palaeotemperature curves of the world's océans. Altematively, but

perhaps less likely, the Sun may bave a “ptilse-like" behaviour on a time scale of some 5 m.y.

13 — Major events in the biosphère and atmosphère during the Cainozoic. such as modi¬

fications in seasonality patterns, are of a type which are compatible with, if nol demanding a

change in the rclalionship belween the Earth and the Sun. Three categories of global climatic

change are noted:

- reorientation of ecoclimalic belts on the surface of the Earth due to graticule shift;

- seasonality changes due to obliquity shift;

- fluctuation of latitudinal width of ecoclimatic belts (see point 16 below).

14 — Major reorgani/alions of world océan circulation patterns due to opening (Graham-

land/Patagonia) or closing of seaways between continental masses (Asia/Europe, Africa/Eurasia,

North and South America) often coincided in lime with tectogenic phases, the two probably

being different aspects of the same overall cause.

15 — The nature of first order faunal turnover puises suggests that their proximate cause

was large scale shift in the positions of biogeographic realms relative to the Earth's surface.

Since the primary pattern of biogeographic realms is latitudinal in expression, this suggests that

first order faunal turnover puises resulted from latitudinal shifts in the boundarics between realms.

This in tum indicales thaï polar reposilioning occuned, which thereby indueed graticule shift.

16 — The latitudinal estent of global ecoclimalic belts has fluctuated during lhe past. At

times, the tropics were latitudinally wider than they are now, .sometimes narrower. At times, the

polar ecoclimatic belts were wider than they are now, at other times narrower. The cause of this

accordion-like behaviour appears to be differential capture of solar energy at the Earth's surface

(including its atmosphère), due either to fluctuations in the Sun's output of energy, or of the

Earth’s efficiency in capturing it, or to both. At présent some 39% of the Sun's energy reaching

the Earth is immediately rcflecied back inlo space. Presumably there is iiothing sacrosanct about

this figure, which could increase, thereby leaving lhe Earth with less retained solar energy, or

it could decrease, in which case the Earth would retain more solar energy. These changes would

probably show up as accordion-like changes in the widths of the world’s ecoclimatic zones.

Acknowledgements

1 am panieularly anxious to ihank Professor Yves Coppbns (Collège de France) for his coiitinued

help and encouragement. My sincere lhanks go to Professor Philippe TaoUPT (Institut de Paléontologie),

Hervé Lbliévre. Hugh Owt.N, Brigitte Spnut, Léonard GiN-SHtiRCj, Chri.stian MARCit.xL and Jorge Morales

for support, di.scussion.s and encouragement. Thanks are aiso exlended lu Prol'c.ssor Samuel Warren Carey

(University of Hobart) and Dr Hugh Owen (Natural History Muséum. London) for solving many things

about Earth History which had been troubling me, and whtch are now clearer in my mind, thanks to the

Expanding Earth Theory. I thank Drs Miguel Doblas and Chri.stian Marchal for reviewing the text and

for making pithy and intere.siing comnienis about an earlier version of this text.

Maniiscripi submitted for publication on 6 Januarv 1995; acrepted on 2 October 1995.

— 513 —

REFERENCES

Adams C., Gentry A. & Whybrow P. 1983. — Dating Ihe terminal Tethyan event. Utrecht Micropaleontology

Bulletin 30: 273-298.

Allègre C. J., SardA P. & STAUDACHER T. 199.3. — Spéculations about the cosmic origin of He and Ne in

the nuerior of tlie F-arih. Eunh and Blanetary Scienttfic Letier 117: 229-233.

Alroy J. 1992. — Conjunction aniong taxonomie distributions and the Miocene mammalian biochronology of

the Great Plains. Paleohiology 18: 326-343.

Alvarez W. 1990. — Géologie evidence for the plate-drivtng mechani.sm; The continental undertow hypothesis

and the Australian-.Antarctic discordance. Tectonics 9: 1213-1233.

Barron E. J. 1985. — Explanaiions of the Tertiary global cooling trend. Palaeogeogmphy Palaeoclimatology

Palaeoecolgy 5t); 45-61.

Barry FLYNN L. & PILBEAM D. 1990, — Faunal diversily and turnover in a Miocene terrestrial sequence.

In R. Ross & W. .Allmon (eds). Causes of Evolution: A Paleontological Perspective. Chicago, Univ.

Chicago Press: 381-421.

Barry J., Johnson N., Raza S. & Jacobs L, 1985. — Ncogene mammalian faunal change in Southern Asia:

corrélations with climatic, tectonic and eustatic events. Geutogy 13: 637-640.

BATES R. L. & Jackson J.A. 1990. — Glo.'^.uiiy of Geology, 3rd édition. Alexandrin, Virginia, American Geo-

logical Institute.

BERTHELSON a, & ShNGOR A. 1990. —Como se formo Europa? In Eiiropn Cientifica. Barcelona, Ed. Labor,

pt 3: 108-123.

BiJU-DllVAL B., DERCüURT J. & LM. PlCHON X. 1976. — La genèse de la Mediterranée. La Recherche 7: 811-822.

Carey s, W. 1976. — The Expanding Eunh. Amsterdam. Elscvier: 1-488

— 1983. — The Expanding Eanh: A Symposium. Brunswick. Impact Printing (Vie.) Inc.

— 1988-— Théories of dit- Eanh and Universe. Stanford, Stanford Univ. Press.

Carlson R. W. 1992. — Meltmg of wet lithosphère. Nuntre 358: 20-21

Caron J. .M., Gauthier A., SCHAAF A,. Ulysse J. k Wozniak J., 1992. — Comprendre et enseigner la

planète Tetre, Parts, Edilion.s Ophrys: 1-271,

Coffin m. F. & Eldholm O. 1993. — Large Igneous Provinces. Scieniijic American 269 (4): 26-33.

COLLINGBOUKNE M. 1976. — Radiation and sun.shine. In T. T. J. Chandler & S. Gregory (eds). The Climate

of the Briiish Isles. London, Longmans 4: 74-95.

CONDIE K, C, 1976. — Plate Tei tonics and Crustal Evolution. New York, Pergamon: 1-288.

— 1989. — Plate Tectonics and Crustal Evolution, 3rd édition. Oxford, Pergamon Press.

Cox, A., and Hart, R. B.. 1986. -- Plate Tectonics: How il Works. Boston, Blackwell: 1-392.

Creber g. T. & Chaloner VV. O. 1984. — Climatic indications front growth rings in fossil woods, fn P.

Brenchley (ed.). Fos.uls and Climate. London, John Wiley and Sons: 49-74.

Dawson M„ West R. & Wa.nn L. 1976.— Palacogene terre.strial vertebrate.s: northernmosl occurrence, Ellesmere

Island. Canada. Science 192: 781-782.

DEWEY J., PITMAN W., RYAN W. & BONNtN J. 1973. — Plate Tectonics and the évolution of the Alpine System.

Bulletin of the Ceological Snciety of .America 84: 3137-3180.

DOGLIONI C, 1993. — Geological evidenee for a global tectonic polarity. Journal of the Geological Society of

London 150: 991-1002.

Dooley J. C. 1983. — A simple phy.sical test of Earth Expansion, In S. W. Carey (ed.) The E.xpandmg Eanh:

A Symposium. Brun.swick, Impact Printing (Vie.) Inc: 323-326.

Du Toit A. L. 1937. — Qnr Wondering Conlinenls. l.ondon, Oliver and Boyd: 1-361.

EBINGER c. J. 1989. — Tectonic development of the western branch of the East African Rift System. Geological

Society of .America Bulletin 101: 885-903.

Egyed L. 1956. — Détermination of ch!inge.s in the dimensions of the Earth from paléographie data. Nature

178: 534.

EyleS N, 1993. — Earth’s glacial record and its tectonic .setting. Eartb-Science Reviews 35; 1-248.

— 514 —

Ginsburg L. 1984. — Théories scientifiques et extinction des Dinosaures. Comptes-Rendus de l'Académie des

Sciences, Paris, 298; 317-320.

— 1986.— Régressions marines ou catastrophes cosmiques: Comment juger les théories sur l'extinction des

Dinosaures. Société Nationale Elf Aquitaine (Production) 10: 433-436.

Gold T. 1955. — Instability of the Earth’s axis of rotation. Nature 175: 526-529.

Goldreich P. & TOOMRE A. 1969. — Some remarks on polar wandering. Journal of Geophysical Research 74:

2555-2567.

Hallam a. 1979. — Relative importance of plate movements, eusiacy and climate in controlling major biogeo-

graphîcal changes since the Early Mesozoic. In G. NELSON & D. ROSEN (eds). Vicariance Biogeography.

New York. Columbia Univ. Press: 303-330.

— 1981.— Biogeographic relations between the northem and southem continents during the Mesozoic and

Cainozoic. Aufsütze 70: 583-595.

Haq b U., Hardenbol J- & VaIL P. R. 1987. — Chronology of fluctuating sea-levels since the Triassic.

5c(ence 234: 1156-1167.

HUNT C. W., Collins L. G. & SKOHELIN E. a. 1992. — Expanding Ceospheres. Energy and Mass Transfers

fwm Earth's Interior. Calgary, Polar Publishing.

Kearey P. & VINE F .1. 1990. — Global Tectonics. Oxford, Blackwell .Scientific: 1-302.

Keith M. L. 1993. — Geodynamics and inantle flow: an alternative earth mode!. Earth-Science Reviews 33:

15.3-337.

LambeCK K. 1979. — The history of the Earth’s rotation. In M. W. McElhinny (ed.). The Earth: Its Origin,

Structure and Evolution. London. Academie Press: 59-81.

Larin V. N. 1993. — Hydridic Earth: The New Geology of Our Primordially Hydrogen-Rich Planet. Calgary,

Polar Publishing.

Laskar J.,, fOUTEL F. & Robutel P. 1993. — Stabilization of the Earth’s obliquity by the Moon. Nature 361:

615-617.

Laskar J. & Robutel P. 1993. — The chaotic obliquity of the planets. Nature 361: 608-612.

Lavallard J.-L. 1988. — Pourquoi la Terre tourne moins vite? Sciences et Avenir 494: .34-39.

LovelocK J. 1990. — The Ages of Gaia: The biography of Our Uving Earth. London, Bantam Books: 1-252.

LÜdtke g. 1970. — The geology and structure of the area between Etjoberg and Walerberg in central South

West Africa/Namibia. Geological Survey of Namlhia Open File Report: 1-8, map.

MacDougall J. D. 1988. — Continental Flood Basalts. Dordrecht, Kluwer Academie Publishers: I-.341.

McKenna m. 1985. — The great American terrestrial interchange and reorganized oceanic circulation in the

latest Tertiary, South Afncan Journal of Science 81: 258.

MarchalC. 1968. — Figures d'équilibre séculairemem stable des masses fluides hétérogènes en rotation. Bulletin

d’Astronomie Série 3 3: 341-360.

— 1991.— Les causes possible de la dérive des continents. Sciences, Publication de l’Association Française

pour l'Avancement des Sciences 91-93: 60-72.

Marshall L.. Webu S.. SepkosKI J., & Raup D. 1982. — Mammalian évolution and the Great American

Interchange. Science 215: 1351-1357.

MarGUUS L. 1991. — Gaia and the colonization of Mars. GSA today 3 (11): 277-280 + 291.

Mein P. 1975. — Résultats du groupe de travail des vertébrés. Reports on R.C.M.N.S. Working Groups, Bratislava:

79-81.

Miller K. & Fairbanks R. 1985. — Cainozoic d’*0 record of climate and sea-level. South African Journal

of Science 81; 248-249.

Morales J., PiCKFORD M. & SORIA D. 1993. — Incrémental pachyostosis in a Lower Miocene giraffid species

from Spain, and its palaeobiological implications. Geohios 26 (2): 207-230.

MOYA Sola S.. Pons Moya J. & Kohler M. 1989-90. — Primates catarrinos (Mammalia) del Neogeno de la

peninsula Iberica. Paleontologia / Evolucio 23; 41-45.

Mutter C. Z. & Mutter J. C. 1993. — Variations in thickness of layer 3 dominate oceanic crustal structure.

Earth and Planet. Scientifîc Letter 117: 295-317.

Norton L O. 1995. — Plate motions in lhe Norlh Pacific: the 43 Ma nonevent. Tectonics 14: 1080-1094.

— 515 —

O’CONNOK J M. & Duncan R. A. 1990. — Evolution of the Walvis Ricige - Rio Grande Rise Hot-Spot System:

Implications for Afiican and South American plate motions over plume. Journal of Geophysical Research

95: 17,475-17,502.

O’CONNOR J. M. & LE ROEX A. P. 1992. — South Atlantic Hot-Spot-Plume .Systems: 1. Distribution of volcanism

in timc and .spacc. Earth and Ploneiary Scienlific leiter 113: 343-364.

OWEN H. G. 1976. — Continental displacement and espansion of the Eiarih during the Mesozoic and Cainozoic.

Philosuphkat Transaclhm af lhe Royal Society of Utnduii A. 281 (1303): 223-291.

— 1981. — Con.stanI dimensions or an e.spanding Earth ? In L. R. M. COCKS (ed.). The Evolving Earrh. London

and Cambridge, Cambridge Unis. Press; 179-192.

— 1983a. — Ocean-floor spreading évidence of global expansion, tu S. W. Carey (ed.). The Expanding Earth:

A Symposium. Brunswick, Impact Ptinttng (Vie.) Inc; 31-58.

— 1983b. — Atlas of Continental Displacement, 200 Million fears io lhe Présent. Cambridge, Cambridge Univ.

Pres.s: 1-159.

— 1990.— Of gaps and globes. Journal of Riogeagraphy 17: 693-695.

— 1992— Has lhe Earth increased in size? In S. CHATTERJEE & N. HUTTON (eds). New Concepts in Global

Tecionks. LubbocK, Texas Tech Univ. Pre.ss: 289-296.

— (in press).— The Mesozoic lO Recent Development of the Pacific Océan: a Review of Sonie Ideas.

Pantic N, K- 1986. — Global Tertiary climatic changes, palaeophytogeography and phyto.stTatigraphy. Lecture

Notes in Earth Sciences: Global Bio-Events. Heidelberg. Springer-Verlag 8: 419-427.

Patriat P. & Achache .1. 1984. — India-Eurasia collision chronology has implicalion.s for crustal shortening

and driving mechanisms of plates. Nature 311: 615-621.

Patterson C. & Owen H. G, 1991. — Indian isolation or contact ? A re,sponse lo Briggs. Svstematic Zoology

40: 96-100.

Pavoni N. 1993. — Pattern of mantle convection and Pangaea break-up, as revealed by the évolution of the

African plate. Journal of the Geologicul Society of London 150: 953-964.

Peixoto J. P. &. OORT A. H. 1990. — Le cycle de l'eau et du climat. La Recherche 221 : 570-579.

PiCKFORD M. 1986a. — Première découverte d'un Hyracoïde paléogène en fiurasie. Comptes-Rendus de l'Aca¬

démie des Sciences, Paris 303: 1251-1254.

— 1986b. — Oeochronology of lhe Hominoidea: a summary. ht J. Else & P. Lee (eds). Primate Evolution.

Cambridge. Cambridge Univ. Press. 123-128.

— 1987.— Révision des Suiformes (Arliodaclyla; Mammalia) de Bugti (Paki.stan). Annales de Paléontologie

73 : 289-350.

— 1988.— The agc(s) of the Bugti fauna(s). ht The Palaeoenvironment of EasI Asia front the Mid-Tertiary.

Hong Kong. University of Hong Kong Press 2: 937-955.

— 1990a.— Uplift of the roof of Africa and its beaniie on lhe évolution of Munkind. Humait Evolution 5:

1-20.

— 1990b. — Dynamics of Old World biogeographic realms during the Neogene: implications for bioslraiigraphy.

European Neogene Mammal Chronology. NATO, ASl Sériés. New York, Plénum: 41.3-442.

— 1991. — Whal caused lhe firsi steps towards the évolution of Walkie-Talkie primates? Cahiers de Paléoan-

thropolngie Editions du CNRS: 275-293.

— 1992.— Evidence for un arid climate in western Uganda during the Middle Miocene. Comptes-Rendus de

l'Académie des Scinttees Paris 315: 1419-1424.

— (in press). — Fossil landsnails of Easi Africa and rheir palaeoecologieal significance. Cahiers de Paléonto¬

logie Editions du CNRS. Paris.

POWELL C. McA, 1979, — A specu|,'njve teclonic history of Pakistan and surroundings; Sonie con.straints from

the Indian Océan, hi Farah & DE JONG (eds). Geodynamics of Pakistan. Quetta. Geological Survey of

Pakistan: 5-24.

QUADE J.. CERLINC T. & Bowman J. 1989. — Development of Asian nionsoon revealed by marked ecological

shifl during the latest Miocene in northern Pakistan. Nature 342: 163-166.

Sahni a. 1984. — Cretaccous-Palaeocenc terrestrial faunas of India: lack of endemism during drifting of the

Indian Plate Science 226: 441-443.

— 516 —

SCHERBA I. G. 1987. — Stages and phases of formation of CainoMîc olisiostromes in the Alpine Fold Belt.

Global corrélation of lectonie movements. Chichester, John Wiley and Sons; 49-82.

SCHWANN W. 1987. — Summary of the timing of Geotectonic events during Cretaceous and Tertiary times in

the Balkan Peninsula. Global corrélation of tectonic movements. Chichester, John Wiley and Sons: 83-95.

Smith a. G., HURLEY A. M. & Briden J. C. 1981. — Phanerozoic Paleoconlinental World Maps. Cambridge

Earth Science Sériés: 1-102.

Stehli F. G. 1968. — Taxonomie diversity gradients in pôle location: the receni inodel. In E.T. DRAKE (ed.).

Evolution and Environment. New Haven and London, Yale Univ. Press; 163-219,

Steiner J. 1977. — An expanding Earth on the basis of sea-floor spreading and subduction rates. Geologv 5:

31.3-318.

Steininger F-i Rabeper g. & RÔGL F. 1985. — Land mammal distribution in the Mediterranean Neogene: a

conséquence of Geokincmatic and climatic events. In D. STANLEY & F. WEZEL (eds). Geological Evolution

of the Mediterranean Basin. New York, Springer Verlag: 559-571.

Stille h. 1924. — Grundfragen der vergleichenden Tektonik. Berlin: 1-443.

Tarling D. h, & TaRLING M. P. 1971. — Continental Drift, London, G. Bell and Sons.

TREVISAN L. & TONGIORGI E. 1987. — Il Nostro Universo: La Terra, Torino, Terza Edizione, UTET.

Van Colvering J. A. 1988. — Sea-level change. In Encyclopedia of Huinan Evolution and Prehistory. New

York and London. Garland; 505-510.

Van der MEULËN A. & Daams R. 1992. — Evolution of Early-Middle Miocene rodent faunas in relation to

long-term palaeoenvironmenta! changes. Palaeogeography Palaeoclimatology Palaeoecology 93: 227-253.

VermeERSEN L. L. a. & Vlaar N. J. 1993. — Changes in the Earth’s rotation by tectonic movements. Geo-

pltysical Research Letters 20 (2): 81-84.

ViNK G. E. 1984. — A hoi spot niodel for Iceland and the Voring Plateau Journal of Geophvsical Research 89:

9949-9959.

Vrba e. 1985. — Environment and évolution; alternative causes of the temporal di.stribution of evolutionary

events. South African Journal of Science 81: 229-236.

Ward J. & CORBETT I. 1990. — In M. K. SEELY (ed.). Namih Ecology: 25 Years of Namib Research. Transvaal

Muséum Monography 7; 17-26.

Wegener A. 1962. — Die Entstehung der Kontinente und Ozeane. Die Wissenschaft, Braunschweig, Friedr.

Vieweg & Sohn 66: 1-231.

White R S. 1991. — Ancient floods of fire. Satura! History 4: 51-60.

Williams G. E. 1975. — Late Precambrian glacial climate and the Earth’s obliquity. Geological Magazine 112:

441-465,

— 1993.— Hi.story of the Earth’s obliquity. Earth Sciences Reviews 34: 1-45,

Wilson J. T. 1965. — A new class of faults and their bearing on continental drift. Nature 207; 343-347.

Wilson M. 1993, — Plate-moving mechanisms: constraints and controversies. Journal of the Geological Society

of London ISO: 923-926.

WiNDLEY B. F. 1993. — Uniformitarianism today: Plate Tectonics is the key to the past. Journal ofthe Geological

Society of London 150: 7-19.

WOODHIIRNE M. & SwiSHER C. 1995. — Land mammal high-resolution geochronology, intercontinental overland

dispersais, sea-level. climate, and vicariance. In Geochronology Time Scales and Global Stratigraphie Cor¬

rélation. SEPM Spécial Publication 54: 335-364.

ZIEGLER P. A. 1993. — Plate-moving mechanisms; their relative importance. Journal of the Geological Society

of London 150: 927-940.

Bulletin du Muséum national d’Histoire naturelle, Paris, 4® sér., 18, 1996

Section C, n”* 2-3 : 517-554

Earth’s polar displacements of large amplitude:

a possible mechanism

by Christian Marchal

Abstract. — Earth rotation and the position of pôles are essential éléments of climate. For centuries the

motion of the pôles with respect to the Earth’s surface has remained very small, but this was not necessarily

the case during geolngical times. The orogeny of a large mountain range, such as Himalaya, modifies the Earth’s

moments of inertia by a few millionths and this is sufficient for a large displacement of the "stable equilibrium

position” of the pôles. The corresponding motion of the pôles remains usually very slow, as slow as the corres-

ponding orogeny, and the Earth’s équatorial bulge doesn’l forbid the transformation. This bulge is only slowly

deformed until its installation in the new position of equilibrium.

Key-word.s. — Earth, rotation, polar motions.

Une cause probable de grands déplacements des pôles terrestres

Résumé. — La rotation de la Terre et la position des pôles sont des éléments essentiels du climat. De nos

jours, les pôles terrestres se déplacent très peu par rapport à la surface du sol, mais il n’en a pas nécessairement

été toujours ainsi. La surrection d’un grand massif montagneux, comme l’Himalaya, modifie de quelques mil¬

lionièmes les moments d’inertie de la Terre, ce qui suffit à déplacer très largement la «position d’équilibre stable»

des pôles. Bien entendu le déplacement correspondant des pôles reste généralement lent, à la mesure de la lenteur

de la surrection du mas.sif montagneux, et l’existence du renflement équatorial de la Terre n’empêche pas la

transformation: ce renflement e.si simplement peu à peu déformé jusqu'à être installé dans la nouvelle position

d’équilibre.

Mots-clés. — Terre, rotation, mouvements des pôles.

C. Marchal, ONERA, F-92320 Chatillon.

INTRODUCTION

Terra australis incognita

The idea of Earth equilibrium and of the necessity of a great Southern antarktos goes back

to the ancient Greeks and was renewed by the voyage.s of Colombus, Vasco da Gama, Amerigo

Vespucci and Magellan. The most famous geographers of the Renaissance, Martin Waldseemiiller,

Gerardus Mercator, Ortelius discussed the Terra australis incognita and drew tentative maps.

This idea remained alive for two centuries until the voyages of Cook. The Terra australis is a

— 518 —

myth but the Earth bas another kind of equilibrium: its pôles are, within one kilometer, at their

position of stable equilibrium.

The question of the motions of the Earth axis is essential for understanding the évolution

of climatcs and life. The long lerm motion of the Earth’s axis with re.specl to the stars and to

the Earth's orbital plane is now wcll-known, at least qualitativcly (LASKAR 1993; Laskar &

Robiitri, 1993; La.SKAR et al. 1993) These studies show thaï the presence of the Moon is

essential. The obliquity of planel Mars has large variations and a long-term erratic évolution

while the Earth's obliquity is stabili/ed by the attraction of the Moon and remains in tlie vicinity

of 23", which is of course a décisive élément for the Earth’s climales. We will here specially

consider the rotation of the Earth and the motions of its pôles with respect to the Earth's crust.

This subject has received considérable attention for more than a century (see for instance:

Love 1911; Poi.ncaré 1952; Le Mouel & Courtillot 1981 ; Legros 1987; Spada et ul. 1992;

Ricard et al. 1993). Since 190ü polar motion has been very small. its amplitude is less than

15 meters. but we will sec ihat it can sometimes hâve a large drift, up to 200 meters per year,

wheri the Earth is in a State of disequilibrium. That latter velocity is rauch larger than the relative

velocity of lithospheric plates.

The huge progress in the knowledge of Earth through geology, seismology, geophysics,

plate tectonics, geodesy and space geodesy, paleogeography, analysis of polar motions, of Earth

rotation, of motions of artificial satellites, etc. leads to a qualitative description of the phases

of Earth di.sequilibrium.

Thcre are aiso very large progress. in paleomagnetism (Ricard et al. 1993). Unfortunately

the motion of the magnetic axis is poorly related to that of the géographie axis and these two

axes can hâve a large angle {e.g. 60” presenily for planel Uranus).

PRELIMINARY NOTICE (NUMERATION SYSTEM)

In this paper for very large and very small numbers we will use the convention of numération

by “figures and sizes”:

Mass of Earth = M = 5.9737 x 10‘^ kg = 5.9737 p24 kg,

figure size

Constant of Newton’s law = Cavendish constant = G = 6.6726 x lO'" m^/s^kg

= 6.6726 nll mVs^kg.

figure .size

The leller p means “positive power of ten'' and the letter n means “négative power of ten”,

while the figure is always between one and ten (we cannot Write 0.59 p25 instead of 5.9 p24).

The size is thus always well defined, it is of course the most important part of very large and

very small numbers, il is even sometimes the only known part.

— 519 —

MAIN ELEMENTS OF EARTH

We will specifically consider the following éléments.

The REFERENCE ELLIPSOID (BURSA 1992)

The international Earth reference System (I.E.R.S.) uses an oblate ellipsoid of révolution

with :

(1) Earth équatorial radius R = 6378 136 m

(2) Earth polar radius Rp = 6356 751.3 m

(3) flattening = (R-Rp)/R = 1/298.257

Mass and moments of inertia (Bursa. 1992)

(4) Mass of Earth = M = {5.9737 p24 ± 3 p20} kg

The product of the mass M by the constant G of Newton’s law is known to 9 digits:

(5) GM = Earth gravitational constant = 3.986 004 41 pl4 m^s'^

The three main moments of inertia A, B, C (with A < B < C) are given by:

(6) A/MR^ =0.329591 ±n6

(7) B/MR^ = 0.329 599 ±n6 n6=10^ (see preliminary notice)

(8) C/MR^ = 0.330 678 ±n6

Their différences are known with much higher accuracy:

(9) (C-A)/MR^ = 1.086 258 n3 ± 2n9

(10) (C-B)/MR^ = 1.078 996 n3 ± 2n9

The moment C practically corresponds to the polar axis, while A corresponds to an équatorial

axis with longitude 14.95'* West. Hence the B axis is équatorial and has the longitude 75.05'* East.

The Earth gravitational potential

The analysis of the motions of artificial Earth satellites has led to a very accurate knowledge

of the Earth’s gravitational potential outside the Earth (Balmino et al 1978; Lerch et al 1979;

Marsh et al 1990; Reigber et al 1993).

This outer potential U has several usual expressions, for instance the following one with

spherical harmonies;

f •» k

GM ’v- r R ^

(11) U = -^ 1+XS 7 Jk„Pkn(sin(p)cosn(L-Lkn)

[ k=2 n=0 V ^

with:

(12) GM and R given in (5) and (1)

520 —

r, (p, L are the ordinary geocentric spherical coordinates:

r = satellite -Earth center distance

(13) j (p = geocentric latitude (positive towards North)

[ L = geocentric longitude from Greenwich meridian (positive towards East)

Pkn(sin tp) is the Legendre polynomial or the associated fonction of order k, n:

, (costp)"

kn(sin(p)- d(sin (p)'''^''»

(sin^ tp - I

Silice -90” < (p < + 90° we hâve cos tp = + V 1 - sin^ tp and:

(15) P 2 ,o = (3 sin^tp- l)/2; P 2 ,i = 3 sin tp cos tp; P 2,2 = 3 cos^tp; etc.

Finally the J^n and Lk„ are the coefficients of the Earth’s potential. They are related to the

distribution of masses inside the Earth.

For n = 0, with the mass dm at the point r, tp, L:

(16) Jko = TTrT j dmi^Pkolsincp)

MR M

and for n < 1:

Jkn cos nLkn = — — " ^ j dm r‘‘ Pkn (sin tp) cos nL

I (k + n) ! MR*' "m

Jkn sin nLkn = —f dm r^ Pkn (sin tp) sin nL

(k + n) ! MR‘‘ •'m ^ ^

For the Earth, with its polar axis as the main axis of inertia, we obtain:

(18) I J20=(A + B-2C)/2MR-; J2i=0; J 22 = (B - A)/4MR^

[ L 22 = — 14.95° = longitude of the A — axis

These expressions and accurate knowledge of the outer potential U explain why the différ¬

ences (9) and (10) are known with greater accuracy than the ratios (6), (7) and (8).

The coefficients (k - n) !/(k + n) ! of (17) introduce artificial very small numbers that reduce

the real importance of the corrcsponding terms.

In order to re-establish the true importance of successive terms it is customary to use the

following Earth coefficients Ckn and Skn:

(19)

Ckn

Skn

Jkn cos nLkn Jkn Sin nLkn

(k + n)

(k-n)!(2k+l)Vn

with n = 0 = > Vn = 1

n > I = > Vn = 2

— 521

For instance with the model Grim4 (Reigber et al 1993):

C 2 o = -4.841 656 236 964 4 n4

(20) C2:=S2i=0

C22 = 2.439 315 544n6

S 22 = - 1.400040 70 n6.

Which gives:

J 20 = C 20 VT= - 1.082 627 246 95 n3

J 21 =0^_

J 22 = V 5(C^2 + 5 ^ 2 )/12 = 1.815 485 964 n6

^ I L 22 = 0.5 Arc tan (S 22 /C 22 ) = - 14.926 8"

(C - A)/MR- = - J 20 + 2 J 22 = 1.086 258 218 88 n3

(C - B)/MR^ = - .I 20 - 2 J 22 = 1.078 996 275 02 n3 .

These results agréé with (9), (10) but their accuracy is illusory: they represent an average

while the tidal variations of Ck„ and S^n correspond to the inaccuracies of (9) and (10) and are

of the order of a few n9 {i.e. a few lO '*, see preliminary notice).

These inaccuracies are well illustrated by the comparison of different models. For instance

the ninth Goddard Earth Model, called GEM 9 (Lerch et al 1979), gives:

Caü^-4.841 655 5 n4

C 2 i=- 2.1 nlO

(22) \ 821 =-4.06 n9

C 22 = 2.4.34 00 n 6

S 22 = - 1.397 86 n 6 .

The différences between the two models are indeed of the order of a few n9.

On the other hand C 2.1 and S 2.1 are not zéro for the GEM 9: there is an angle of about

2.6" between the polar axis of this model and its main axis of inertia.

The.se models give C^n and Stn many values of k and n (for instance for ail values

with n Ê k < 20 for the GEM 9). These coefficients only hâve a slow decrease for large k and

n and the convergence of (II) is not good: at the scale of kilometers the Earth has many ir-

regularities. Fortunately we will not need ail these coefficients.

Beyond those of (20) and (22) only two coefficients are not between + n 6 and - n 6 :

(23)

j C 3 1=2.028 26 n 6

I 833 = 1.411 40 n 6

in the GEM 9.

These coefficients Ci;n and 8 icn allow us to draw pictures of the “geoid”, as in the Figure 1.

The “geoid”, very near to mean sea-levcl, is an equipotenlial surface of the différence

U - 0.5 (OTT^ cos^ tp, the latter term being given by the centrifugal forces with:

(24) (0 = rate of rotation of Earth = 7.292 115 n5 Rd/s.

— 522 —

Fig. !. — Lcvcl curveb of ihc gcoid (in mcicrs) wilh respect !o ihc following référence cilipsoid: équatorial radius = 6378 155 m;

flattening = 1/298.255 (Balmino, Rkicber & MOYNOr 1978)*

This equipotential surface can be comparée! to some reference ellipsoid such as (1), (2),

(3) (the chosen equipotential surface is one with a volume équivalent to that of the ellipsoid).

The level curves are given in meters and range from - 108 meters (near India) to + 82 meters

(New Guinea).

The polar motions at the Earth surface

The polar motions, and the corresponding variations of latitudes, hâve been accurately

measured for a century, These motions hâve two main parts (Bureau des Longitudes 1984).

A short period part with Iwo periodical components (Fig. 2)

A yearly component related to major meteorological phenomena such as the monsoon.

A component with the “Chandler period’* of 435 days and corresponding to the following

phenomenon. If a free body with main moments of inertia A, B, C does not exactly hâve pure

rotation about iu C-axis it will hâve a motion similar to the Euler-Poinsot rotation of rigid

bodies. For these rigid bodies the period P of polar motion is given in terms of the period Pq

of main rotation by:

(25) P = PoV (AB/(C-A) (C - B)} .

In the case of the Earth, Po is 23 h 56 min 4.1 s (/.e. one sidereal day) and (25) gives

P = 303.61 days. However the Earth is not a rigid body and its properties, essentially its elasticity,

enlarge the period of polar motions from P to the Chandler period of 435 days (LOVE 1911).

— 523

The second part of the motion of North Pôle is a very slow drift towards Canada that

remains when the short period components are removed (Fig. 3). This slow drift is certainly

related to plate tectonics and to the postglacial rebound.

Fig. 2. — The motion of the North Pôle during the four years 1972-1975. O is a conventional origin, Ox is in the direction of

Greenwich and Oy in the direction of Canada. At the Earth’s surface 0.1" is only 3.1 m.

CONCERNING THE INNER STRUCTURE OF THE EaRTH

The inner structure of the Earth and the distribution of its inner masses is a difficult question.

A small part of our information has an astronomical and mechanical origin. The motions of the

pôles with respect to the stars (precession. nutation) are related to the main moments and axes

of inertia. while the motions of the Moon and of the artificial satellites give us the total Earth

mass and the coefficients of its outer potential. These coefficients are known fonctions of the

inner distribution of masses (équations (16) to (19)). However they only give very partial in¬

formations. indeed an arbitrary pomt-mass and a spherically symmetrical distribution with the

same total mass and the same center hâve exactiy the same outer potential.

Geology and vulcanology give us much information about the Earth’s crusl while geodesy,

and especially space geodesy. yield accurate informations about the Earth’s dimensions.

Our main source of information about the interior of the Earth is seismology. The largest

earthquakes give rise to powerful waves propagating in ail directions up to the antipodes with

— 524 —

velocities depending on local conditions. The measurements of propagation durations by seis-

mometers reveal the existence of several discontinuities at various levels within the Earth.

We are thus led to the modem Earth models such as the PREM Is given in Table 1 (DziEWONSKl

& Anderson 1981). Notice thaï this model in terms of only one parameter, the depth, is necessarily

an average approximation: the mean thickness of the lithosphère is about 217 km but it varies from

a few kilometers below the oceanic ridges to about 300 km below the continents.

Canada

Fig. 3. — The slow drift of the Noith Pôle toward.s Canada when the short period components appearing in Figure 2 are removed.

THE FOUR EARTH EQUILIBRIA

Because of the pre.sence of outer bodies, the Sun, the Moon, etc., our planet cannot be in

a State of perfect equilibrium. It permanently undergoes tides in its océans and its crust. Further-

more its inner sources of energy draw along large mas.ses at a very slow rate. Fortunately these

phenomena remain very small, at the scale of the Earth, and we can define the four following

equilibria that are more or less near to the true Earth; the hydrostatic equilibrium of a non-rotating

Earth (Fig. 4A); the hydrostatic equilibrium of a rotating Earth (Fig. 4B); the isostatic equilibrium

of a non-rotating Earth (Fig. 4C); the isostatic equilibrium of a rotating Earth (Fig. 4D). The

last equilibrium is very near to the true Earth but notice that the first one is already good; it

is the basis of the PREM l.s model of Table l.

The hydrostatic equilibrium of a non-rotating Earth

This First equilibrium (Fig. 4A) is simple and well-known. It has spherical symmetry. The

different layers are spherical and concentric with density increasing with depth.

— 525 —

The HYDROSTATIC EQUltffiRlUM OF A ROTATING EARTH

This second equilibrium (Fig. 4B) remains simple. It bas spheroidal symmetry; the axis of

rotation is also an axis of symmetry of révolution; the équatorial plane (normal to the axis of

rotation through the center of mass) is a plane of symmetry; on straight Unes either parallel or

perpendicular to the axis of rotation the density increases when we approach the center of mass.

This spheroidal symmetry is very general for hydrostatic equilibria (see appendix 1).

The Earth has a rather slow rotation and ils hydrostatic equilibrium bas level surfaces that

are concentric and coaxial oblate quasi-ellipsoids whose oblateness increases trom the center to

the surface. These quasi-ellipsoids are slightiy depressed al mid-latitudes (the dépréssion is only

about 4.3 m at the Earth’s surface at latitudes ± 45°).

Fig. 4. — The four equilibria of Eaiih. A, hydrostatic equilibrium of a non-rotating Earth. The level .surfaces are concentric

spheres; B. hydrostatic equilibrium of a rotating Earth. The level surfaces are concentric and coaxial oblate quasi-ellipsoids

(slightiy depressed at mid-latitudes); C. isostatic equilibrium of a non-rotating Earth. The sea-level is almost a sphere; D,

isostatic equilibrium of a rotating Earth. The sea-level is almost an oblate ellipsoid. B, D: The Earlh’s angular momentum

H is 5.85989 p3.5 kg m’/s.

— 526 —

Table 1. — The PREM Is (Preliminary Référencé Earth Model Is). Parameters : depth z, density, seismic velocities; Vp: longi¬

tudinal waves; Vs: transversal waves; P: pressure: g; graviiy; r; radius. Notice thaï everywhere: z + r = 6371 km = mean

Earih radius. Also notice ihat V.s = 0 in the outer core. This layer represents 15,6% of the Earth’s volume and 31% of its

mass; it is considered as liquid.

z (km)

density

Vp (km/s)

Vs (km/s)

P (kbar)

g (cm/s2)

r (km)

OCEAN

1,02

1,45

981,56

6371,0

1,02

1,45

0,299

982,22

6368,0

LITHOSPHERE

Upper crust

2,60

5,80

3,2

0,303

982,22

6368,0

2,60

5,80

3,2

3,364

983,31

6356,0

Lower crust

2,90

6,80

3,9

3,370

983,32

6356,0

24,4

2,90

6,80

3,9

6,040

983,94

6346,6

Upper mantle

24,4

3,38

8,11

4,49

6,043

983,94

6346,6

3,37

8,11

4,48

11,239

984,37

6331,0

3,37

8,08

4,47

17,891

984,93

6311,0

3,37

8,07

4,46

24,539

985,53

6291,0

3,37

8,07

4,46

24,546

985,53

6291,0

115

3,37

8,05

4,45

36,183

986,64

6256,0

150

3,36

8,03

4,44

47,824

987,83

6221,0

185

3,36

8,01

4,43

59,466

989,11

6186,0

220

3,35

7,98

4,41

71,108

990,48

6151,0

TRANSITION ZONE (MIDDLE MANTLE OR ASTHENOSPHERE)

220

3,43

8,55

4,64

71,115

990,48

6151,0

265

3,42

8,64

4,67

86,497

992,03

6106,0

310

3,48

8,73

4,70

102,027

993,61

6061,0

355

3,51

8,81

4,73

117,702

995,22

6016,0

400

3,54

8,90

4,76

133,520

996,86

5971,0

400

3,72

9,13

4,93

133,527

996,86

5971,0

450

3,78

9,38

5,07

152,251

997,90

5921,0

500

3,84

9,64

5,22

171,311

998,83

5871,0

550

3,91

9,90

5,37

190,703

999,65

5821,0

600

3,97

10,15

5,51

210,425

1000,38

5771,0

600

3,97

10,15

5,51

210,426

1000,38

5771,0

635

3,98

10,21

5,54

224,364

1000,38

5736,0

670

3,99

10,26

5,57

238,334

1001,43

5701,0

LOWER MANTLE

670

4,38

10,75

5,94

238,342

1001,43

5701,0

721

4,41

10,91

6,09

260,783

1000,63

5650,0

771

4,44

11,06

6,24

282,927

999,85

5600,0

771

4,44

11,06

6,24

282,928

999,85

5600,0

871

4,50

11,24

6,31

327,623

998,36

5500,0

971

4,56

11,41

6,37

372,852

996,98

5400,0

1071

4,62

11,57

6,44

418,606

995,73

5300,0

1171

4,67

11,73

6,50

464,882

994,67

5200,0

1271

4,73

11,88

6,56

511,676

993,83

5100,0

1371

4,78

12,02

6,61

558,991

993,26

5000,0

1471

4,84

12,16

6,67

606,830

993,01

4900,0

1571

4,89

12,29

6,72

655,202

993,14

4800,0

— 527 —

1671

4,95

12,42

6,77

704,119

993,69

4700,0

1771

5,00

12,54

6,82

753,598

994,74

4600,0

1871

5,05

12,66

6,87

803,660

996,35

4500,0

1971

5,10

12,78

6,91

854,332

998,59

4400,0

2071

5,15

12,90

6,96

905,646

1001,56

4300,0

2171

5,20

13,01

7,01

957,641

1005,35

4200,0

2271

5,25

13,13

7,05

1010,363

1010,06

4100,0

2371

5,30

13,24

7,09

1063,864

1015,80

4000,0

2471

5,35

13,36

7,14

1118,207

1022,72

3900,0

2571

5,40

13,47

7,18

1173,465

1030,95

3800,0

2671

5,45

13,59

7,23

1229,719

1040,66

3700,0

2741

5,49

13,68

7,26

1229,741

1048,44

3630,0

2741

5,49

13,68

7,26

1269,742

1048,44

3630,0

2771

5,50

13,68

7,26

1287,067

1052,04

3600,0

2871

5,55

13,71

7,26

1345,619

1065,32

3500,0

2891

5,56

13,71

7,26

1357,509

1068,23

3480,0

OUTER CORE

2891

9,90

8,06

1357,510

1068,23

3480,0

2971

10,02

8,19

1441,941

1050,65

3400,0

3071

10,18

8,36

1546,982

1028,04

3300,0

3171

10,32

8,51

1651,209

1004,04

3200,0

3271

10,46

8,65

1754,418

980,51

3100,0

3371

10,60

8,79

1856,409

955,70

3000,0

3471

10,73

8,92

1956,991

930,23

2900,0

3571

10,85

9,05

2055,978

904,14

2800,0

3671

10,97

9,16

2153,189

877.46

2700,0

3771

11,08

9,27

2248,453

850,23

2600,0

3871

11,19

9,38

2341,603

822,48

2500,0

3971

11,29

9,48

2432,484

794,25

2400,0

4071

11,39

9,57

2520,942

765,56

2300,0

4171

11,48

9,66

2606,838

736,45

2200,0

4271

11,57

9,75

2690,035

706,97

2100,0

4371

11,65

9,83

2770,407

677,15

2000,0

4471

11,73

9,91

2847,839

647,04

1900,0

4571

11,80

9,98

2922,221

616,69

1800,0

4671

11,87

10,05

2993,457

586,14

1700,0

4771

11,94

10,12

3061,461

555,48

1600,0

4871

12,00

10,18

3126,159

524,77

1500,0

4971

12,06

10,24

3187,493

494,13

1400,0

5071

12,12

10,30

3245,423

463,68

1300,0

5149,5

12,16

10,35

3288,503

440,03

1221,5

INNER CORE

5149,5

12,76

11,02

3,50

3288,513

440,02

1221,5

5171

12,77

11,03

3,51

3300,480

432,51

1200,0

5271

12,82

11,07

3,53

3353,596

397,39

1100,0

5371

12,87

11,10

3,55

3402.383

362,03

1000,0

5471

12,91

11,13

3,57

3446,764

326,45

900,0

5571

12,94

11,16

3,59

3486,665

290,68

800,0

5671

12,98

11,18

3,61

3522,024

254,73

700,0

5771

13,01

11,20

3,62

3552,783

218,62

600,0

5871

13,03

11,22

3,64

3578,894

182,39

500,0

5971

13,05

11,23

3,65

3600,315

146,04

400,0

6071

13,06

11,24

3,65

3617,011

109,61

300,0

6171

13,07

11,25

3,66

3628,956

73,11

200,0

6271

13,08

11,26

3,66

3636,131

36,56

100,0

6371

13,08

11,26

3,66

3638,524

0,0

— 528 —

The isostatic equilibria

With ail ils mountain ranges, ils continents and ils océans the Earth is of course far from

hydrostatic equilibrium and a much better approximation is that of isostatic equilibrium. The

idea of isostatic equilibrium cornes from the large différences of viscosity and rigidity bctween

the lithosphère and the asthenospherc. Vertical equilibrium of the lithosphère is achieved much

more easily and fasler than the horizontal one. The upwards motion of Scandinavia, after the

recent melting of the large Quaternary glacier, is about one meter per ceniury and corresponds

to re-equilibrium on the lime scale of a few millenia only.

The theory of isostatic equilibrium requires the définition of a “level of compensation”,

the asthenospherc, a few hundred kilomelers below lhe surface. Above Ihis level only vertical

equilibrium is required. Wc thus arrive at the usual pictures with a ihick lithosphère under the

continents, especially lhe mountain ranges, and a ihinner one below ihc océans (Figs 4C. D).

Figure 4C corresponds to the isostatic equilibrium of a non-rolaling Earth. Ils mean sea-level

is of course aimo.st spherical and we will see the interest of this equilibrium below. Figure 4D

corresponds to the isostatic equilibrium of the Earth with ils true rotation. The main feature of

the surface is its oblateness (the différence between the équatorial and polar radii is 21.4 km).

The mean sea-level is not exactly a “geoid” i.e. the cquipolential surface of level zéro (because

of sea currents and of variations of température and salinity) but these two surfaces remain

within meters to eaclî other and they are close to the référencé ellipsoîd, as shown in Figure 1.

THE ROTATING AND NON-ROTATING SET OF AXES

For the study of the Earth’s rotation and of its inner motions it is cu.stomary to use the

Tisserand sel of axes with origin al the center of mass and rotation such that the remaining

velocities give to Earth a zéro total angular momenlum (Legros 1987).

However we are specially inleresied in lhe relative Poles-Earlh^crusl motions and we are

less interested in the oceanic currents and the motions of the Earth’s mantle. Hence we will

choo.se as üxyz axe.s the reference framc ITRF (International Terrestrial Référencé Frame) that

is more concretely relalcd to the crust (Fig. 5; Arias et al. 1995).

We will al.so use a non-roiaiing right iri-rectangular set of axes OXYZ (Fig. 6), with the

same origin O. It is the geocentric équivalent of the reference frame ICRF (International Celestial

Reference Frame; Arias et al. 1995). The relalivistic correction.s thaï are necessary between the

barycentric ICRF and its geocentric version are negligible in the présent study. We will define

the Earth’s axis and pôles by lhe rotation vector tu of Oxyz with respect to OXYZ. Its présent

module is:

(26) (û = 7.292 115 n5 Rd/s = présent rate of Earth’s rotation.

We will see below that the Earth’s angular momentum H remains always extremely near

to the direction of tu and thus, for the study of the large polar displacements, it will be sufficient

to study the évolution of H.

— 529 —

FiG. 5. — The Oxyz rotating set of axes (Iniernaiional

Terrestrial Référencé Frame). O is ihe Earth’s

center of nmss. Oz passes through lhe Conventional

International Origin (lhe point O of Figures 2 and

3). Oxz is along lhe Inlcniational Référencé Mer-

idian (Greenwich). Oxyz is a nghl and rectangular

trihedron.

Fig. 6. — The OXYZ non-rotating set of axes. O is the

Earth’s center of mass. OXY is lhe équatorial plane

of the Internationa! Celeslial System. OX is the

direction of the International Vernal Point.

THE EARTH’S INERTIA MATRIX M

The inertia matrix M relates the rotation vector œ and the angular momentum lî of rigid

bodies rotating about their center of mass:

(27) H = M (0.

This relation is valid at any time and can be expressed in any set of axes, for instance

Oxyz or OXYZ. The Earth is not exactly a rigid body but (27) will nevertheless be very useful

(see (63)).

— 530

We will call Ixy, etc. the moments and products of inertia in the Oxyz axes:

(28) I 2 = f x^ dm ; Ixy = 1 xy dm.

M M

Thus, when Oz is the main axis of largest inertia:

(29) C = IxV = j (x^ + y^)dm

and, with (I6)-(I9), when Oxyz is the rotating geocentric set of axes used:

(30) Ixy=MR2S2.2 ; Ixz = MR2C2,, VÎ"; Iyz = MR2S2,, .

The components of the inertia matrix M are then classically:

-Ixy

(31) M= -Ixy IxVz=

— Ixz — lyx

Hence the model Grim-4 (Reigber et al. 1993) gives in its axes with (20), (21):

rU-3.64024892n4 1.807445n6 0

(32) M = MR2 1.807445 n6 U-3.577 266 06 n4 0

0 0 U + 7.21751498n4

MR^U is the average Earth moment of inertia that is less accurately known (see (6)-(8)):

(33) MR^U = y (A + B + C); U = 0.329956± n6 .

Similarly the Goddard Earth Model 9 (Lerch et al. 1979) gives in its own axes, with (22):

Tu-3.640 1797 n4 1.80463n6

(34) M = MR‘ 1.80463 n6 U-3.577 334 I n4

2.7 n 10 5.24 n9

The différences between (32) and (34), a few n9 lhat is a few billionths, are essentially

caused by a différence of orientation of a few seconds between the corresponding set of axes

(that are both extremely near to Oxyz of Figure 5). Nevertheless there remains also an irreducible

différence of a few nlO, a différence related to the accuracy of présent measures.

However we need not only the présent expression of the matrix M but also its past and

future expressions under varions possible orientations of the polar axis. We will aiways use the

Oxyz set of axes of Figure 5, that is concretely t1xed to the EaUh's crust; and we will consider

the four equilibria of Figure 4 and the corresponding inertia matrices Ma, Mb, Mf;, Mp. We are

looking for Mo under various orientations of the polar axis and we know that:

Ma and Mc are, by définition, independent of the polar axis while Mb is a simple function

of this axis.

2.7nl0

5.24 n9

U+ 7.217 513 9 n4

— 531

The principle of small modifications allows us to write;

(35) Md - Mc = Mb - Ma + “second order”.

Finally we will also need the différence M - Md thaï remains very small at least as long

as there are no large variations of the Earth’s crust; each major variation being followed by a

period of reequilibrium (see below),

Notice that these variations will also modify Mc and thiis our model of Earth will use;

(36) M = Mc + (Mb ~ Ma) + R,

with:

The différence Mb - Ma is only a function of the direction and the modulus of the Earth’s

rotation vector co.

The matrix Mc will generally hâve only extremely slow variations.

The remainder R will generally be very small but it will be not negligible during and Just

after periods of large variations of Mc, such as periods of orogeny of large mountain ranges.

The matrices Ma, Mb and (Mb - Ma)

The sphcrical symmetry of the hydrostatic and non-rotating equilibrium leads to a very

simple matrix Ma:

(37) Ma=MR2

Ua 0 0

0 Ua 0

0 0 Ua

The average moment of inertia MR^ Ua is very near to MR^ U of (33) and a little smaller,

their relative différence is of the order of one miilionth.

The spheroidal symmetry of the hydrostatic and rotating equilibrium leads to a matrix Mb

that is a function of only four parameters; the smallest main moment of inertia Ab, the largest

main moment of inertia Cb and the two angles giving the direction of the polar axis.

(38)

V )

ap ;

0 ‘

( . 1 j

Mb = MR2

+ (Cb - Ab)

«P ;

P -T

( ^ J

ay; Py;

with:

( I ^

y2_-L

^ 3

V /J

( 39 )

MR^Ub = average moment of inertia = (2Ab + Cb )/3

Ua < Ub < U : these three numbers are almost equal

— 532 —

and;

(40)

(a, P, y) = components of the unit vector of polar axis in the rotating Oxyz set of axes

(a, P, y) = ~ ; a" + p^ + Y^ = 1 ; with en = rotation vector of Oxyz with respect OXYZ.

We can verify that in the présent State:

(41) Mb =

Ab 0 0

0 Ab 0

0 0 Cb

The values of Ab and Cb dépend upon the Earth model considered, such as the PREM Is

model of Table 1.

The depth of inner discontinuities are accurately known but the knowledge of inner densities

remains weak. Fortunately the influence of this ignorance remains small in our problem. Let us

assume for instance that we only know the following.

(42)

The total Earth mass M = 5.9737 p24 kg.

The équatorial radius R = 6 378 136 m and the mean radius Rm = 6 371 200 m

(including the volume of continents).

The average moment of inertia:

(2Ab + Cb )/3 = MR“Ub = MR"U = 0.329 956 MRI

The présent Earth angular momentum Hp = 5.859 89 p33 kg mVs

Hp = Ceo; with C = 0.330678 1VIR\ co = 7.292 115 n5 Rd/s.

The first layer (the water, essentially the océans) : density 1.02 g/eve? ;

global mean depth 2750 m.

The following limits on inner densities: 2.6 g/cm^<density< 15g/cm^.

The limits on (Cb - Ab) are then given by the two extreme cases with only three homogenous

layers the outer one being the océan.

Maximum of Cb-Ab:

(43)

radius (km)

6371.2

6368.45

3112.732

density (g/cm^)

1.02

4.266 843

\ Cb - Ab = 1.077 408 n3 MR^

— 533 —

Minimum of Cb - Ab:

radius (km) density (g/cm^)

6371.2

1.02

(44)

6368.45

2.6

5225.794

7.885 025

Cb - Ab = 1.070 805 n3 MR^.

In these conditions the realistic model with ten homogeneous layers presented in appendix 2

gives an accurate resuit, very near to the minimum (44):

(45) Cb-Ab = 1.070978 n3 MRI

These results (43)-(45) correspond to the présent value Hp of the Earth’s angular momentum

H. They would be practically proportional to H" for neighbouring values of H.

We are thus led to the following expression of Mb - Ma".

(46) Mb - Ma = MR^

(Ub-Ua)

1 0 0

0 1 0

0 0 1

+ 1.070978 n3^

' 2

“-3

V )

aP ;

ap ;

^ 1 ^

P^-:

^-3

Ub - Ua is of the order of n6 and we will see in the next two sections that this différence

is useless while the accuracy of (45) and (46) is superabundant.

The remaining matrix R

The “remaining matrix R” of (36) is composed of two terms: the second order terms of

(35) and the différence M - Md.

The second order terms arc of the order of (Mb - Ma) (Md - Mb)/MR‘ that is about n8 MR^,

i.e. less than M - Md. The différence M - Md is given by the discrepancies between the true Earth

and its isostalic equilibrium.

The discrepancy corresponding to the Chandler effect (.see above with (25)) is very small,

a few nlO MR^, but it will become of the order of the above terms for large disequilibria. The

discrepancy corresponding to re-equilibrium motions such as the recent upward motion of Scandi-

navia are of the order of a few n8 MR^, and their time scale is only a few millenia. The global

discrepancies can be larger but their time scale is only about one century (these discrepancies

hâve a time scale proportional to the inverse of their width).

— 534

Hence we can model the remaining matrix R as follows. During Ihc periods of lectonic

quiescence on Earlh the matrix Mc is quasi-constant and R is at most of thc order of n7 MR^.

If the Earth and the matrix Mc undergo large variations (i.e. variations larger lhan n7 MR^ per

millenium) the remaining matrix R can be larger than n7 MR‘ and can remain above Ihis level

for a few centuries after the period of large variations.

The matrix Mc

The matrix Mc is quasi-constant during the periods of tectonic quiescence on Earth and,

with (32), (36), (41), (46) we can obtain its présent value corresponding to a = (3 = 0; y = 1:

(47) Mc = M - (Mb - Ma) - R; R = O (n7 MR^).

Hence:

[Uc- 7.032 n6 1.807 n6 0 1

(48) Mc = MR2 1.807 n6 Uc-7.34n7 0 -r O (n7.MR‘)

0 0 Uc +7.766 n6

with:

(49) Uc = U - Ub -I- Ua = 0.329 956 ± n6.

The corresponding main moments of inertia are then:

Ac = (Uc-7.5I4n6±n7) MR^

(50) . Bc = (Uc-2.52n7±n7) MR2

Ce = (Uc + 7.766 n6 ± n7) MR^ .

Thus Ce is the largest main moment of inertia and its axis is along the présent Earth’s axis

of rotation. This classical situation of stability will be shown essential for the présent stability

of the Earth’s rotation.

The Earth inertia matrix M

Let us recalI that M is given by (36) that is by:

(51) M = Mc + (Mb - Ma) + R,

with (Mb - Ma) given in (46) where (a, p, y) is the unit vector of the Earth’s axis of rotation.

During the periods of tectonic quiet on Earth the matrix Mc is almost constant while R

remains smaller than n7 MR^. But R can be larger when Mc has variations faster than n7 MR^

per millenium.

The mean values U, Ua, Ub, Uc are ail within 0.329956 ± n6. They are thus less accurately

known but this will hâve no effect.

— 535 —

The présent value of M is given in (32) with an accuracy of a few nlO MR" except for U.

This allows us to obtain, with (46), the matrix M with almost the same accuracy for directions

(a, p, y) different from (0, 0, 1):

[

'U-3.640249n4

1.8074n6

■«2

(52) M = MR^

1.8074n6

U-3.577266n4

+ 1.070978 n3

U + 7.2175l5n4

Y^-l

with U = 0.329 956 ± n6.

THE ANGLE BETWEEN THE ROTATION VECTOR œ

AND THE EARTH’S ANGULAR MOMENTUM H

The vector co is the rotation vector of the rotating set of axes Oxyz of Figure 5 with respect

to the non-rotating set of axes OXYZ of Figure 6.

We will call tOi, Cû 2 , 0)3 the components of tn in Oxyz; hence from (40):

(53) ôl = (O)], ü) 2 , ( 03 ) = (ma. mp, my).

Let us consider an element dm of the Earth’s mass with a radius vector t^(t) in the Oxyz

rotating .set of axe^. This mass dm will hâve a relative velocity v^ with respect to Oxyz and an

absolute velocity V with respect to OXYZ.

(54)

\~7 —^ ^ ^

V = rox r + V .

In the analysis of v we find:

The motions rclated to plate tectonics (velocity; a few centimeters per year);

The general motion of Earth mantel (velocity: a few meters per year);

The quasi-periodic motions related to the tides of the crust (amplitude up to one meter;

periods from a few hours to a few years);

The motions of the océans and the atmosphère: quasi-periodic tidal motions, permanent

currents and w'inds, seasonal currents and winds, meteorology.

Let us call II the angular momentum of Earth.

(55) H = J dm?xV: M = mass of Earth.

Since V = mx we can write:

(56) iî=H„ + Hv

with, M being the Earth’s inertia matrix:

— 536 —

(57) Hco = J dmr'x (MX ?) = Mw

and:

(58) Hv=f dmr*x Ÿ.

The z-component of I^o) is about 5.859 89 p33 kg s ' and we can décomposé tïy into a

main quasi-periodic part Hqp given by the tides and a remaining erratic part Her given mostly

by the oceanic currents. For instance the famous “Pacific Equatorial Under-Current” gives a

componeni of about 1.5 p24 kg m^ s ‘ that is more tlian the combined effects of plate tectonics

and motions in the Earth's mantle. The Gulf Stream and the Kuro Shivo give, in other directions,

componcnls of the .same order of magnitude. The general tidal motions give a much larger com-

ponent, up to p27 kg m' s ', and we can thus model Hv by the following:

(59) Hv = Hqp + Her,

Hqp: quasi-periodic component, periods from a few hours to a few years,

modulus < p27 kg m^ s ',

-—k ‘y _ I

Her: erratic and long period component, modulus < 5 p24 kg m s' .

For the angle (Êf, œ ) we find:

The angle (H, Hp,) remains smaller than or equal to | Hv|/| Hp, | that is less than 2 n7 radian.

The angle (Hp,, given by:

I M = Mc + (MB-MA)-rR.

The main part of M is. by far, Mc, while (Mb - Ma) où is parallel to Où.

The component Rco inlroduces a new angle less than 3 n7 radian and the typical example

(48) giving the présent Mc shows that the angle (Mcôù, Cù) remains small, less than a few n5

radian, i.e. a few seconds of arc.

—> _ ^

Hence, at least during the periods of tectonic quiet on Earth, the angle (H, œ) remains

extremely small, smaller than 5" or 10". This limit is of course larger when the Earth’s crust

and the matrix Mc hâve fast variations but 1' or 2' seems a good upper limit for ail cases.

Thus, in order to know the motions of the pôles with respect to the Earth’s crust with an

accuracy of 1' or 2', it is sufficient to look for the évolution of the Earth angular momentum

H with respect to the set of axes Oxyz rotaling with Earth.

EVOLUTION OF THE EARTH’S ANGULAR MOMENTUM lî WITH RESPECT

TO THE ROTATING SET OF AXES OXYZ

Case of an isolated Earth

If the Earth were isolated its angular momentum îî would be constant in the non-rotating

axes OXYZ and its évolution with respect to the axes Oxyz rotating with Earth would be simple.

— 537 —

(61) 0=

fdH^

rdH^

V J

OXYZ

( y

+ coxH

Oxyz

that is:

(62)

= - cox

îî=i?

X (0 .

Oxyz

However, with (56), (57) (59):

H = H(ü + Hqp + Her

Her : erratic component smaller than 5 p24 kg

Hqp : quasi-periodic component smaller than p27 kg m s

Ho)= Mœ : modulus about 5.86p33 kg m^ s“'

(64)

and thus

= {M(0 + Hqp + Her) X tO .

Oxyz

The term lïerXü) is very small (an erratic polar displacement of less than 13 m per year).

The term l^pXco is larger but quasi-periodic with an average equal to zéro. Its average

effect in one month is already very small and its average effect in a few years is negligible.

The remaining term (Mô^ x co can be considered accurately with the présent expression of

M given in (52).

This expression gives a negligible polar displacement for the présent polar position, but it

also gives a polar displacement of almost 340 km per year for pôles at the worst position (This

large velocity is of course theoretical, see for instance the analysis presenled with Figure 10.)

This resuit emphasizes the importance of the notion of “stable equilibrium position” of the

pôles.

Let us neglect the Hqp and Her terms and let us use the décomposition (51 ):

(65) M = Mc + (Mb — Ma) + R.

dH .

The product [(Mb - Ma) (O] x m is always zéro and thus the main term of — in Oxyz is:

( 66 )

{McCû - 1 - Rœ -r Hqp + Her} X cû = {Mc(û} x co.

Oxyz

This équation, with constant Mc, also appears in the classic Euler-Poinsot theory of the

rotation of a free and invariable solid (with inertia matrix Mc) about its center of mass. Hence

the stable equilibrium positions of the pôles are on the main axis of largest inertia of Mc and

this is precisely the situation of Earth as shown by (48)-(50). The considération of outer torques

in the next section will not modify this resuit.

— 538 —

Influence of outer torques

The Earth is not isolated. it undergoes many torques.

The main torques are given by the attraction of outer bodies, especially the Sun and the

Moon. The other torques, such as the electromagnetic torque due to the solar electromagnetic

field, are much smaller (at most a few billionths of the main torques) and we will neglect them.

Let us first consider the solar gravitational torque Ts-

S is the Earth-Sun distance and S the Earth-Sun vector (Fig. 7).

■■>A ' V'

O /

Earth

Sun

Fig. 7. — Study of the solar gravitational torque Ts.

The elementary mass dm is at the radius vector r and the classical expression of the solar

tida] force Fs dm is:

(67)

Fs dm = ps dm

p'*

s-^

with:

( 68 )

Ps = 1.327 124 38 p20 m^ s ^ = solar gravitational constant

• P = S - r = vector from the mass dm to the Sun

P = lpl

Let us put:

(69)

projection of r on

Then:

(70)

Fs dm =

3 w - r^+ 0

— 539 —

and the solar gravitational torque Ts is:

(71) Ts=f r^xFsdm = ^^J 7xwdm.

Earth mass S M

Let us call u = (Ui, U 2 , U 3 ) the unit vector of the direction of the Sun in the Oxyz set of

axes.

(72)

uf + U2 + U3 = I.

Hence, with the approximation of (71):

(XUi +yU2 + ZU3),

—> 3lis

(73) Ts=^

[ dm

^xO

fu, ^

U 2

\ /

that is:

(74)

(U|U3lxy - UiU2lxz + U2U3lyV + (U3 - U2) lyz)

< (U|U2lyz - U2U3lxy + U|U3lz^_x^ + (uf - U3) Ixz)

(U 2 U 3 IXZ — UiU3lyz + UiU2lx'-y^ + (U2 - uf) Ixy)

In the set of axes Oxyz used the components U|, U 2 , U 3 and the torque Ts hâve very fast

daily variation. However, if we only consider the outer torque Ts, the équations (61), (62) be-

come:

(75) Ts =

(76)

V J

OXYZ

V J

+ (0 X H,

Oxyz

^ dH ^

= H X CO + Ts-

Oxyz

The tcrm H x co lias already been analysed in the previous section. Its main part is

(Mc œ) X (0 équivalent lo (Mc co) x co .

The daily variations of Ts hâve little importance of course, and we essentially need the

daily averaged value of Ts.

We will a.ssume that the Oxyz rotation vector œ, that is also practically the Earth’s rotation

vector, is in the direction of Oz or in its immédiate vicinity. Since the orientation of Oxyz is

arbitrary this condition can always be satisfied after, if necessary, a suitable modification of the

Oz direction.

The component U 3 is then fixed, or very slowly variable, while U| and U 2 essentially hâve

a daily sinusoidal évolution.

The daily averaged values of U|U 2 , U 1 U 3 and U 2 U 3 are almost zéro while these of uf and

U 2 are (1 - u^)/ 2 .

— 540 —

Hence, with these Oxyz axes;

V « J

Notice that with the same Oxyz axes:

(78) Hx ô) = (Mœ) X to = ca‘

These two vectors appearing in (76) hâve the same direction but the second is much larger:

at least 40 000 times larger. The effect of the solar gravitational torque Ts is negligible and this

conclusion remains true even if we take into account the final terni that was neglected in (70).

The same analysis can be donc, with a similar conclusion, for the lunar gravitational torque

and for ail outer torques undergone by Earlh.

Notice that this conclusion would nol be true for an analysis of the évolution of the angular

momentum H with respect lo the non-rotating set of axes OXYZ. In this set of axes the com-

ponents of the unit vector u, equal to —, would not hâve a fast daily variation and we would

thus reach the motions of the pôles with respect to the stars: nutation and procession.

^yz

—> 3|ts 1 — 3 u5

(77) Tg dailyaveraged~ ^3 2

MODIFICATIONS OF THE EARTH’S INERTIA MATRIX B Y THE OROGENY

OF A LARGE MOUNTAIN RANGE

It is essential to take the isostatic equilibrium into account in the analysis of the variations

of the Earlh’s inertia matrix. This equilibrium reduces very' much the possible variations that

nevertheless are not ai ail negligible.

Let us consider the orogeny of the Himalaya range. That range is classically considered as

the resuit of the collision between the plates of Asia and India (Fig. 8).

Beiween the two plate.s the northward motion of India pushes the Earth's crust partiy upwards

and partiy downwards (because of the isostatic compensation); while behind India we hâve the

opposite transformation; a ihin oceanic crust takes the place of a thick continental crust. These

two opposite évolutions combined with the i.sotatic equilibrium reduce to almost nothing (about

a billionth) the variations of the average Earth's moment of inertia that is (A + B + C)/3. However

this is not true for each moment of inertia and al go for the products of inertia.

Let us for instance consider the product of inertia lyz.

Let us call v the local volumic mass, g the local intensity of gravity and h the local altitude.

If D is the depth of the compensation level, the isostatic equilibrium can be written along

any vertical:

541

Fig. 8 . — The Asia-India collision and the orogeny of Himalaya.

J. + 50 km

(79) V g dh = Pc = pressure at the compensation level

-D

(above 50 km, in the stratosphère, the volumic mass v is negligible).

Hence along any vertical the passage from the volumic mass vi(h) to the volumic mass

V 2 (h) must satisfy the following isostatic condition:

(80) J (V2-V|) g dh = 0.

The product of inertia lyz is defined by:

( 81 ) Iyz = J V y Z dx dy dz.

Earth'svolume

That is, with the ordinary geocentric spherical coordinates r, (p, L defined in (13):

y = rcos (p sin L ; z = rsin(p

dx dy dz = r^ cos (p dr dtp dL

Iyz = J V r"* cos‘tp sin tp sinL drdtp dL

Earth’s volume

and the variation of the product of inertia lyz is then:

( 83 ) Alyz = (Iyz)2 - (lyz)i =1 (v2 - V|) r"* cos^ tp sin tp sin L dr dtp dL.

Earth's volume

We must of course take the isostatic condition (80) into account along any vertical.

This condition gives along any radial direction:

(V 2 - vi) g cos e dr = 0

— 542 —

with;

\ R = équatorial radius =6 378 136 m

(85) I Rp = polar radius = 6 356 751.3 m

[ e = angle between lhe vertical and the radial directions.

The angle e remains everywhere smaller than 13' and its cosine remains between 1 and

0.999992.

With V 2 - V] = 0 when r < Rp — D and with the définition of triple intégrais this équation

(84) leads to:

(86) j (V 2 - vi) g cos e dr. cos^ cp sin (p sin L dtp dL = 0.

Earth's volume

In a small dtpdL pyramid starting from the center of mass the coefficient of (\2 - vi)

cos^ tp sin tp sin L drdtpdL is r"* in (83) and g cos e in (86).

These Iwo coefficients vary in opposite directions when r varies in the range

Rp-D<r<R + 50 km

(even if g cos £ is there airnost constant as shown in Table 1).

It is this variation of the ratio r‘’/g cos E in terms of r that is the source of the variations

of the Earth's moments and products of inerlia. Without it the Earth’s inertia matrix would be

constant for ail States of isostalic equilibrium and, since there exist isostatic equilibria with a

symmetry of révolution about the polar axis, ail isostatic equilibria would then hâve equal équa¬

torial main moments of inertia instead of the présent différence B - A = 7.26 n6 MR^ (with M = mass

of the Earth).

Which value can we give to the différence (vn —vi)?

In lhe orogeny of the Himalaya range and its vicinity the seabed (average depth 3.8 km)

has been pushed up over a large région.

That région - Himalaya range, Tibet plateau, Hindu-Kush, Karakoram, Pamir, Kunlun,

Chamdo, Quinghai, etc. - has an area of about 10 millions of square kilometers and an average

altitude of 2800 m (The average altitude of ail Asia is 1000 m while that of Europe is only

300 m).

Thus over that région that we will call “H” we can estimate Aly,, with:

I = 2600 kg/m^ between the altitudes 0 and 2800 m

I V 2 -V 1 = 16(X)kg/m^ between the altitudes -3800 m and 0.

Of course, becau.se of (84) that is of the isostatic equilibrium, the différence (v 2 —Vi) must

be négative somewhere below, around the bottom of average continental plates at depths of

about 300 km.

Hence for the Himalaya région and its surroundings, the région “H”, we can write with

(83), (86) and with an arbitrary constant K:

(88) Aly,, = J (v 2 - vi) dr d 9 dL cos^ tp sin tp sin L [r"* - K g cos ë].

— 543 —

We will choose K such that the factor (r'* - K g cos e) is zéro in the middle of the depths

where (V2-V1) is négative. Hence for these depths the intégral Aly, will be negligible.

We can ai.so neglect the small variations of g cos e and since these variations are in the

opposite direction of the variations of r*^ our computation will give a lower bound of Aly^.

Hence, with K g cos e = ro:

r Rm+ 2800 m i d f t

(89) AIyz = J (V 2 -V|) dr (r'* - ro) J cos" (p sin cp sin L dtpdL

Rm-SSOO m H

with:

RM = mean Earth radius=637l km

ro = Rm - 300 km

j(v 2 -V|) = 2600kg/m^ for Rm < r < Rm + 2800 m

I = 1600 kg/m^ for Rm - 3800 m < r < Rm

J cos tp dtp dL = (area of H)/R^ = p7 kmV[6371 km]"

cos (p sin tp sin L = yz/r^ : with an average value of 0.45 in H .

We thus obtain finally for the H région;

(91) Aly^ = 4.28 p32 kg.m^= 1.76 n 6 MR^

To this variation of the product of inertia lyz in the région H we must add the variation

obtained behind the Indian plate where a thin oceanic plate takes the place of a thick continental

plate.

The signs of (vi- V|) arc there opposite but so are the signs of the factor costp sintp sinL

with a négative sin tp. Hence the two variations Alyz hâve the same sign and their sum is about

3 n 6 MR^, which is of course a rough estimation. (Notice the essential rôle of the factor cos tp

sin tp sin L that is yz/r^. Without that factor the computation would give the variation of the

average moment of inertia (A + B + C)/3 and we hâve seen that this variation is negligible; its

two components, before and behind the Indian plate, are almost opposite.)

The Earth’s crust can easily bear such a dissymrnetry, even under the condition of isostatic

equilibrium; today it bears the larger différence B - A = 7.26 116 MR^ between its two équatorial

main moments of inertia. But the Earth's rotation reacts rapidiy to this non-zero produel of

inertia ly, as we will see in the next section.

Variations of the Earth’s moments of inertia of the same order of magnitude, or even a

litlle larger, are obtained in the outstanding study of RICARD et al. (1993). However these authors

consider that the main source of these variations is the motion of downgoing slabs during sub¬

ductions.

Whatever the source of these variations they imply large polar displacements.

— 544 —

REACTION OF THE EARTH’S ROTATION TO A DISSYMMETRY

OF THE EARTH’S INERTIA MATRIX

The Earth’s reaction to a dissymmetry of its inertia matrix is of course a very complex

phenomenon. We hâve aiready considered thaï reaction and we can at least notice the following

points.

The équations (61 )-(66) and (75)-(78) are oblained by the application oIThe general tbeorems

of Mechanics. They are rigorous and can be applicd to any type of body: rigid. elastic, plastic,

viscous. fluid, mixed, etc. These équations are in agrecment wiih the Euler linearized équations,

and this was of course a necessary condition, but they are valid even for a viscous Earth and

for a large drift of the pôles. They are of course the only équations that can be written for a

viscous Earth withoul a detailed rheology study. The main discussion is that of the relative

importance of the four terms of (66) that is:

(92)

= {Mcœ + Rro + Hqp + Her) X (0 = I Mcco) X œ.

Oxv/

For a long-term uniform rotation about a fixed Earth’s axis the main term of (92) is, by

far, the term Mcü) and this shows that the stable Earth’s rotation is about the main axis of largest

inertia of matrix Mc. Notice thaï this rotation corresponds also. very naturally. to the minimum

of mechanieal energy for a given angular momentum.

This rotation is precisely that given by the Earth's model Grim4 (REIGBER ei al. 1993)

with ils coefUcients given in (20). It is also e.stremely near to the rotation given by the Goddard

Earth's model 9 (LERCH ei al. 1979) with its coefficients given in (22): there is then less than

100 m belween the pôles and lheir positions of stable equilibrium, This pre.scnt Earth’s situation

is obviously a very sirong indication.

The cxi.sicnce of a large équatorial bulge of course opposes a sirong résistance to the motion

of the pôles towards a new position of stable equilibrium after the orogeny of a large mountain

range; and il is precisely because the Earth is viscous (on the long term) that its pôles can move

towards the new position of stable equilibrium with successive isostatic deformations of the

équatorial bulge.

In the discu.ssion of the motion given by (92) the three small terms can bccome more im¬

portant when the rotation is nol a pure rotation about a fixed Earth’s axis, cspecially the Rca

term that is relaled to the successive deformations of the équatorial bulge. However these three

small terms correspond to losscs of energy and the.se losses are large for non-stable rotations.

The march towards stabilily is then unavoidable even if the path followed by the pôles dépend

very much on the Earth’s properties (Fig. 9).

Let us look more carefully al the influence of the équatorial bulge Lei us assume for instance

that at some initial time tu the Earth has a stable rotation about the axis leading from the city

of Madrid to ils antipode in the North island of New Zealand, and let us assume that suitable

orogenies givc to its crust a new shapc corresponding to the présent matrix Mc of the “non-

rotaling Earth in isostatic equilibrium" (Fig. 4C). Hence the lheoretical stable rotation becomes

about the présent North Pole-South Pôle axis. This is of course a very artitlcial assumption but

it will allow to give a better idea of the reaction of the Earth’s rotation.

— 545 —

Fig. 9. — March of the North pôle towards a new position N 2 of stable equilibrium afler a variation of the Earth’s crust. A,

march for û very viscou.s Earih. B. march lor a very fluid Eanh. In the absence of a detailcd rheulogy siudy il is impossible

lo guess the palh thaï will be l'oilowcd, however the large losses of energy corresponding lo non-pure stable rotations will

lead more or Icîts rapidly lo ihc vicinity of lhe stable equilibrium.

At the initial time to the Earth has a large équatorial bulge corresponding to lhe Madrid-New

Zealand axis and very similar to the présent équatorial bulge. This bulge stabilizes the Earlh’.s rotation

very much and. since il is about 300 limes larger than lhe considea'd Earth’.s modifieations, lhe

variations of the Earlh’s rotation are inilially weak. Under the condition of isostatic equilibrium the

initial Earth's inertia matrix M is given in (52) with the direction cosines a. p, y corresponding lo

lhe geocenlric direction of Madrid (aM = 0,76189; Pm =-0.04949; Ym =0.64581).

If the Earth were rigid aller in its rotation would be an Euler-Poinsot motion corresponding

to the inertia matrix M: the North pôle would remain in the vicinity of Madrid and starting from

Madrid il would describe a small "polhode" thaï is then almosl a circle with center near lhe

small city of Torrelaguna at lhe point T (Fig. 10). That point T corresponds to lhe main axis

of smallest inertia, it is only 44.0 km north of Madrid and 6.3 km east of Madrid. The motion

of the North pôle would be counter-clockwise along the polhode with a period given by (25)

that is about 300 days.

If the Earth were perfectly elastic the motion would be similar but with a longer period

(Chandler effect), a period of 400 or 450 days.

However the Earth is neilher rigid nor perfectly elastic and we can expect lhe two following

effects.

Notice that between a pôle at M (Madrid) and at the opposite point N (Fig. 10) the equi¬

librium sea-level has variations reaching almost 300 meters. The Earth’s elasticity reduces these

variations to about 200 meters, but ncvcrthcless they are sufficiently large to give a great com-

plexity to the Earth's rotation and the corresponding large losses of energy lead lo a North Pôle

at a medium position in the vicinity of T.

Meanwhile the condition of isostatic equilibrium slightly modifies lhe équatorial bulge. The

local isostatic re-equilibria, as the présent upward motion of Scandinavia, hâve a rather short

relaxation time (a few millenia for Scandinavia) and the global isostatic re-equilibria, as the

modification of the équatorial bulge, hâve a shorter relaxation time of one or two centuries only

(because this relaxation time is roughiy proportional to the width of the affected zone).

— 546 —

Hence after one or two centuries the équatorial buige corresponds to a North Pôle in the

vicinity of T, instead of M (Fig. 10) and the Earth's inertia matrix M given in (52) must now

use the direction cosines ar, Pr, Tt instead of Um, Pm. Ym- This matrix M gives a main axis of

largest inertia corresponding to a point in the vicinity of N, instead of T, and thus another move

of the pôle towards north must be expected in the next centuries for exactly the same reasons.

Fig. 10. — Hypoihesis of a rigid Earlh; an example of polhode (cune described

by the North Pôle). The period of ihis motion is about .“^OO days.

This évolution is repeated again and again as long as the axis of the équatorial buige differs

from the main axis of largest inertia, that is until the arrivai of the pôles at their position of

stable equilibrium and there is no such position before the présent position pôles.

Thus, in spite of the existence of a large équatorial buige. the viscosity of the Earth and

its tendency to the isoslatic equilibrium allow a more or Icss fast move of the two pôles towards

their positions of stable equilibrium, positions that correspond to the main axis of largest inertia

of the "non-rotating Earth in isostatic equilibrium" (Fig. 4C).

Of course the complexity of the Barth’s inner motions, with ils elasiicity, its plasticity and

its viscosity, as well as the large transgressions and régressions of sea-lcvcl that must bc expected

during the phases of variable rotation and of isostatic re-equilibria forbid to oblain a simple

description of the évolution.

Our only solid éléments are the general theorems of Mechanics, explicited in (6l)-(66) and

(75)-(78). and the tendency towards the State of minimum energy (for the given value of the Earth's

angular momentum) because of the large losses of energy associated to non-pure rotations.

This section allows us to give an approximate value of the relaxation time of the march

of the pôle towards its position of stable equilibrium. If Bc and Cç are the two largest main

moments of inertia of the “non-rotating Earth in isostatic equilibrium” the relaxation time is

roughly proportional to [Cc-Bcj ’ and it is about 100000 years when the différence Cc-Bc

is a few n6 MR‘.

A variation of moments and products of inertia of the order of n6 MR* (that is a millionth

of the product MR^) requires a very powerfui orogeny, as shown in the previous section. Such

— 547 —

an orogeny will usually take millions of years and then the pôles will follow very nearly and

very slowly their position of stable equilibrium.

Notice that the drift of the pôles can be large, for instance the orogeny of the Himalaya

range (analysed above, Fig. 8) has modified Ac, Bc, Ce, Mc by about 3 n6 MR^, even if we

only consider isostatic situations. This must be compared lo the présent Cc-Ac (fe. 1.53 n5

MR^) and Ce - Bc (8.02 n6 MR*). A modification of about 20“ in the position of the North

Pôle (from Siberia to the présent position) was likely if tt was not offset by other orogenies.

(The présent matrix Mc is given in (48) and its (-lyz) components are zéro. The main effect

of the Himalaya orogeny is an increase of the product of inertia ly,. of about 3 n6 MR^ and

hence the (-ly,) components were about 3 n6 MR" before the orogeny. The corresponding matrix

Mc had a main axis of largest inertia (cigenvector of the largest eigenvalue) oriented toward

the point 72" North and 83" East near the présent mouth of the Yenisei river.)

Also notice that. with the différence Cc-Bc having the large value 8.ü2 n6 MR^, the Earth

presently has a ralher good stability without immédiate threat of large di.sequilibrium. However

with a much smaller différence Ce - Bc (and the hydrostatic equilibrium gives Ac = Bc = Ce)

the position of the pôles would be very unstable. This was perhaps the case at some past periods.

CONCLUSION

The study of the Earth’s rotation and of the di,splacemcnl of its pôles over geological time

is a complex but also very beautifui undertaking implying many domains of science from

astronomy, mechanics and space dynamics to geology, geophysics, rheology and seismology.

The positions of pôles at the Earth's surface are usually very near to their positions of stable

equilibrium which are given by the main axis of largest inertia of the “non-rotating Earth in

isostatic equilibrium". The corresponding relaxation time is usually about 100000 years.

The stable equilibrium positions dépend very much on the variations of the Earth’s crust

and the pôles hâve certainly moved much more than the different lithospheric plates. Large oro¬

genies and/or large subductions can move the pôles by tenths of degrees within a few million

years. We can hope that plate tectonics and paleogeography will be able to yield an accurate

description of the successive polar shifts.

Acknowledgements - Remerciements

Je suis heureux de pouvoir remercier ici Messieurs Hervé Lelièvre ei Martin Pickeord du Laboratoire

de Paléontologie du Muséum national d'Histoire naturelle qui m’ont donné l'idée de ce travail et la possi¬

bilité de le publier. Je lien.s aussi beaucoup à exprimer ma gratitude à mes collaborateurs de l'Office

National d'Etudes et de Recherches Aérospatiales, M. Nguyen Van-Nhan et Bruno Christophe dont l aide

m’a été très précieu.se. Je remercie enfin vivement M. Jean-Louis Le MoiiEt,, Jean-Paul Poirier et J.

CalmuscK' de rin.stitui de Physique du Globe de Paris, M. Hilaire Lecros de l'Institut de Physique du

Globe de Strasbourg;, M. François Barlier de l’Observatoire de la Côte d’Azur et MM. Bruno Morando,

Jacques Laskar, Olivier NÉRON DE StiRGY et Pierre Bretagnon du Bureau des Longitudes qui m'ont

transmis un très grand nombre de renseignements indispensable.s.

Manuscript suhmitted for publication on 19 Seplemher 1995; accepted on 31 January 1996.

REFERENCES

Arias E. R, Charlot P., Feissel M. & Lestrade J.-F. 1995. — The extragalactic référencé System of the

International Earth Rotation Service, I.C.R.S. Astronomy and Astrophysics 303: 604.

Balmino g., Reigber Ch. & MOYNOT B. 1978. — Le modèle de potentiel gravitationnel terrestre GRIM2.

Annales de Céophssique 34 (2); 55-78.

Bureau Des Longitudes 1984. — Encyclopédie scientifique de l'Univers. La Terre, les eaux, l’atmosphère.

Gauthier Villars (eds).

Bursa m. 1992. — Parameters of common relevance of Astronomy, Geodesy and Geodynamics. In C. C.

TscHF.RNINO (ed.). Ceodetic Eulletin, The Ceodesist's Handhook. 66 (2).

Carev s. W. 1958. — The tectonic approach lo continental drift: Symposium on Continental Drift. Hobart:

177-355.

— 1976. — The Expanding Earth. Developments in Geotectonics 10. Elsevier Scientific Publishing Company.

Amsterdam, Oxford, New York, 488 p.

Dziewonski a. m. & ANDERSON D. L. 1981. — Preliminary reference Earth model. Physics of the Earth and

Planetary Interinrs 25; 297-356.

HiLGENBERC O. 1933. — Vom Waehsenden Erdhall. Berlin, 50 p.

Jeans J. H. 1917. — Memoirs of the Royal Astronomical Society 62.

LaskaR J. 1993. — La Lune et l'origine de l'homme. Pour la science, hors série, août: 14-21.

Laskar J. & Robutel p. 1993. — The chaotic Obliquity of the Planets. Nature 361: 608-612.

LASKAR J., JOUTEL F. & Robutel P. 1993. — Stabilization of Earth’s Obliquity by the Moon. Nature 361:

615-617.

LEGRO.S h. 1987. — Sur quelques prohlèmes de dynamique planétaire. Thèse de doctorat. Université Louis Pasteur

de Strasbourg, Institut de Physique du Globe, Laboratoire de Géodynamique, 16 septembre.

Le MOUEL J.-L. & COURTILLOT V. 1981. — Coxe motions, electromagnetic core mancle and variations in the

Earih's rotation; new consiraints. Physics of the Earth and Planetary Interiors 24: 236-241.

Lerch F. J., Klosko s. m.. Laubscher R. E. & Wagner C. A. 1979. — Oravity model improvement using

OEOS-3 (GEM-9 and 10). Journal of Geaphysical Resarch 84; 3897-3915.

LICHIEN.STEIN L. 1933. — Gleichgewirhtsifiguren rotierender Flussigkeiten. Leipzig.

Love A. E. H. 1911 — Some Prohlems of Geodynamics. Cambridge University Press.

March.AL c, 1968. — Figures d’équilibre séculairement stables des masses fluides hétérogènes en rotation.

Ruiletin astronomique, série 3, 3 (3); 341-360.

Marsh J. G., Lerch F. J.. Putney B. H., Pelsentreger T. L.. Sanchez B. V.. Klosko S. M., Patel G. B.,

ROBBiNS J. w.. Williamson r. g., Engelis t. l., Eddy w. F., Chandler n. l., Chinn d. s.. Kapoor

S., Rachi.in K. F,., RRAAT/ L. fi. lü Pavi.I.*! fi. C. 1990. — The GfiM-T2 Gravitational Model. Journal

of Geophysica! Research. December 10. 95 (B 13); 22043-22071.

Owen h. g. 1981. — Océan floor spreading evidence of Global Expansion. In S. W. Carey (ed.). The Expanding

Earth. .4 Symposium at University of Sydney, February 10-14, University of Tasmania.

— 1984. — The Earth is expanding and we don’t know why. New Scienti.st November 22.

— 1989. — La Terre en expansion. 107' Congrès de l'Association Française pour l’.Avancement des Sciences,

Orléans 23-25 novembre.

Poincaré H. 1952. — Formes d'équilibre d’une masse fluide en rotation. In GAUTHIER Villar.ü (ed.). Œuvres

de H . POINCARÉ. Paris 7.14-219.

Reigber C., Schwintzer P. Barth W., Massmann F. H., Raimondû J. C., Bode A., Ll H., Balmino G.,

Biancale R.. Moynot b., Lemoine J. M., Marty J. C., Barlier F & Boudon y, 1993, — Grim4-

Cl-C2p: Combination .solutions of the global Earth gravity field. Sun’eys in Geophysics 14; 381-.393.

Ricard Y., SPADA G, & SaBADINI R, 1993. — Polar wandering of a dynamic Earth. International Geaphysical

Journal 113: 284-298.

Spada g., Ricard Y. & Sabadini R. 1992. — Excitation of true polar wander by subduction. Letter to Nature.

Nature 360: 452-454.

— 549 —

APPENDIX 1

STABLE HYDROSTATIC EQUILIBRIUM OF A ROTATING PLANET

AND SPHEROIDAL SYMMETRY

Stable hydrostatic equilibrta

A rotating planet has a négative gravitational energy (it requires energy to disperse ail its

parts) and a positive kinetic energy of rotation. The sum of these two energies is the (négative)

mechanical energy while the stable hydrostatic equilibria correspond to the local minima of that

mechanical energy under the following conditions.

The planet is isolated, very far front other celeslial bodies.

The total angular momentum H of the planet is given and cannot be modified.

The planet is composed of undilatable and incompressible fluids. Notice that this third con¬

dition is not restrictive since for ordinary tluid planets, composed of dilatable and compressible

fluids, any stable hydrostatic equilibrium corresponds also to a particular stable hydrostatic equi-

librium in the above meaning.

The general properties of stable hydrostatic equilibria are as follows:

The planet has no inner motion, it rotâtes uniformly as an invariable solid about its main

axis of largest inertia that we will call Oz, with O being the center of mass. Oz is of course

in the direction of the angular momentum H and we will use an ordinary rotating set of axes

with Ox along the main axis of smallest inertia and Oy along the main axis of mean inertia.

The équatorial plane Oxy is a plane of symmetry (LICHTENSTEIN symmetry. Lichtenstein

1933) and on straight lines parallel to Oz the density increases, or is at least is non-decreasing,

when I z I decreases.

It seems that ail stable hydrostatic equilibria hâve a second plane of symmetry: the meridian

plane Oxz (with the corresponding density property). However. this property has only been par-

tially demonstrated (see for instance condition (A5) below).

The SPHEROIDAL SYMMETRY

A planet with spheroidal symmetry not only has the above LICHTENSTEIN symmetry about

the équatorial plane but also has a symmetry of révolution about the polar axis Oz. Furthermore

on Z = constant planes the density is non-decreasing when we approach Oz.

Domain of spheroidal symmetry

For centuries, until the JACOBI counter-example given below, it was believed that ail stable

hydrostatic equilibria must hâve spheroidal symmetry. Nevertheless the domain of this symmetry

remains large.

— 550 —

Let us call:

(Al) ^

G the constantof Newton’slaw

M the total mass of the planet

P its mean density (p = M/volume)

0 ) its rateof rotation

D its “diameter”(largestdistancebetweentwopoints)

R its “radius” (largesldistancebetween the centerof mass O and the surface), hence

R<D<2R

A, B, C the mai n momenlsof inertia.with A < B < C

A = ^V(GM/( 0 ^ = average distance of “planetostationary” and “planetosynchronous” satellites.

The angular momentum H is equal to Cto and it is easy to verify that:

2C<A + B + C< 2MRI

For an homogeneous planet the figures with stable hydrostatic equilibrium are either oblate

MacLaurin ellipsoids or three axis JACOBi ellipsoids. Ail POINCARÉ figures are unstable (Jeans

1917).

The stable oblate MacLaurin ellipsoids hâve spheroidal symmetry, a ratio major axis/minor

axis in the range 1 to 1.71608 and a slow rotation: they exist only if

H < 0.23924 GiMsp's

(see the next appendix). The one-parameter family of stable three-axis Jacobi ellipsoids follows

that of MacLaurin ellipsoids without discontinuity. These Jacobi ellipsoids hâve a ratio major

axis / minor axis in the range 1.71608; 2.8983. (The ratio T 2 = (mean axis/minor axis) is almost

exactly given in tenus of the ratio ri = (major axis/minor axis) by the relation (rj -a)(r 2 -a) =

0.581415; with a= 0.953572.) They exists only when the number without dimension

I 5 X

H G 2 M 3 P6 is in the range 0.23924; 0.30684. Beyond the upper limit an homogeneous planet

has too fast a rotation and generally undergoes fission into two very unequal parts (Marchal

1968).

These results can be extended to heterogeneous planets and the following partial but already

very general results can be demonstrated.

Ail stable hydrostatic equilibria verifying:

(A2) D < A

hâve spheroidal symmetry (MARCHAL 1968).

Ail stable hydrostatic equilibria verifying:

(A3) 2MR^ + A + B + 5C < 2MA^

also hâve spheroidal symmetry (which implies of course A = B).

Ail stable hydrostatic equilibria verifying:

(A4) 2MR^ - A + 3B + 3C < 2MA^

— 551 —

hâve at least thc “tri-symmetry” aiready encountered for three-axis Jacobi ellipsoids. The ihree

main planes Oxy, Oxz and Oyz are planes of symmetry and the density is non-decreasing when

one moves towards these planes on a straight line parallel to one of the three axes.

Finally if:

(A5) 2MR- + 3A - B + 3C < 2IVIA‘

it is easy to demonstrate the “bi-symmetry” that was conjectured above for ail stable hydrostatic

equilibria.

Ail known stable hydrostatic equilibria with only bi-symmetry are very far from spheroidal

symmetry. They alvvays verify 4A < B.

For the Karth: D = 2 R and A = 6.61 R (= distance of geostationary satellites). The ineqiiality

(A2) is widely satisfied.

This condition D < A is satisfied by ail planets but Saturn (for which D = 1.07 A). For-

tunately Saturn, and ail other planets, verify (A3) and thus ail planets of the solar System hâve

an hydrostatic equilibrium with spheroidal symmetry.

APPENDIX 2

STUDY OF THE STABLE HYDROSTATIC EQUILIBRIUM OF AN EARTH MODEL

We hâve seen above that with only the information (42) we can aiready obtain very narrow

limits (43), (44) on the différence (Cb-Ab). A better Earth model than the limiting cases used

in (43). (44) will ihcn give (Cb — .Ab) with excellent accuracy,

Let us consider for instance the following ten homogeneous layer Earth model (Table 2)

that is a simplification of thc model presented in Burbau DES LONGITUDES 1984. Notice that

the third and fourth layers are given with cxtrcmciy accurate densities: these two layers corres¬

pond to the largest Earth zone, the lowcr mantle (55% of the volume, 49% of the mass); and

the two densities are chosen in ordcr to keep the following accuratcly known cléments of Earth:

the total mass and the average moment of inertia. These accurate densities are not at ail an

indication of the degree of our knowledge of Earth, they are there only to give us a better value

of the différence (Cr-Ab) that we are looking for,

We can then research the stable hydrostatic equilibrium of this model of Earth. A first

approximation is to consider that ail level surfaces are oblate ellipsoids. We will then need the

gravitational potential of an homogeneous oblate ellipsoid, that is of a MacLALIRin ellipsoid

aiready considered in appendix I. Let us call a the semi-major axis of this oblate ellipsoid (équa¬

torial radius), c its semi-minor axis (p olar radiu s). R ils average radius (with R^ = a'c), e the

eccentricity of ils mcridians (with c = a V ( I - e’) and p its density. With Oz in ihe polar direction

we obtain the following classical expression ot the inner gravitational potential U:

(A6) U = GM

3 . . x^ -E y- -E Z" 2z“

T — Arcsm e + - --E -

2ae 2R-’

4R’

F(e)

— 552 —

G - constant of Newton’s law

M = massof theellipsoid = 47tR'^p/3 = 47ta^cp/3

(A7) L = R(i -e ^)-6 ; c = R(1 -e^)7

F(e) = ^- 1 -^ V(1 -e‘) • Arcsin e = 76 "+ ^

3 rri ^ , 2 -J 2x44 2x4x6^

V(1 -e‘) • Arcsin e = t e‘+ 7 —r + -—z—- e® + ...

5 5x7 5x7x9

(For an isolated MacLaurin ellipsoid the rate of rotation co is given by the uniformity of

the total potential U - 0.5 co" (x^ + y~) along the surface. This leads to:

co^ = 27iGp [2e^ + (2e^- 3) F(e)] /3.

This oblate MacLaurin ellipsoid is stable if and only if:

(3e + lOe^) V(1 -e^) > (3 + 8e^ - Se"^). Arcsin e; that is e < ef= 0.81267001...

The lasl stable MacLaurin ellipsoid, obtained for e = Cf, is also the first Jacobi ellipsoid.)

It is then easy to obtain the total gravitational energy Eq of the System of the ten ellipsoidal

layers. Notice first that we can consider this System as a System of ten homogeneous concentric

ellipsoids (with a spatial superposition of these ellipsoids); the k* ellipsoid having the density

47C -1

8k = Pk - pk +1 and the mass Mk = — Rk Ôk .

Then, with the fonction F(e) of (A7):

Rf 3-ef(l+F(ek)) 15 (l-e ^)6

1 <j<k< 10

(l-e?)3

Rk Ck

Arcsin Ck

10 T T 1

3G y Mk (1 - Ck) 6

5 , Rk ek

Arcsin Ck

We also need the kinetic energy of rotation Ec:

(A9) Ec = :

4 X MkR^I-eb'

H = Earth’s angular momentum = 5.859 893 762 p33 kg m^ s '

Cb = polar moment of inertia of the System of the ten homogeneous ellipsoids

(AlO) 2 I

= - X MkRg(l-e^)-I

k=l

— 553 —

The mechanical energy E = Eg+Ec is then a function of 22 known quantities (G, H, the

ten Mk, the ten Rt) and ten unknown quantities (the eccentricities Ck). The minimisation of the

mechanical energy E with respect to the ten eccentricities gives the stable hydrostatic equilibrium.

These minimising eccentricities are given in the fourih column of Table 2; they increase from

Cl to eiu and correspond to the oblateness: ^ I - V 1 -efo 1/299.8115.

The corresponding différence (Cb - Ab) between the main moments of inertia is the fol-

lowing (as aiready presented in (45)).

(Ail)

Cb-Ab=j X MkR-k

k = l

(1 -eg)-J-(l

= 2.602 623 p35 kg m^ = 1.070 978 n3 MR‘.

ek)3|=^ X MkREe^(l-e^)

k = l

The true stable hydrostatic equilibrium has level surfaces that are only quasi-ellipsoids.

These quasi-ellipsoids are slightiy depressed at mid-latiiudes but the dépréssions are extremely

small. The largest one, at the Earth's surface, reaches the proportion 6.7 n7 only of the corres¬

ponding radius, that is only 4.3 melers. The corresponding modification of the différence Cb-

Ab is negligible, it doesn't modify the digits given in (Ail).

Table 2. Study of an Earih model with ten homogeneous layers.

Notice 1: the radius Rio lakcs the volumes of continents into account.

Notice 2 : the accuracy of pi and pj is artificial and doesn't reflect our uncertain knowledge of Earth. These densities of the

two main layers (corresponding to the lower mande) allows us to keep the two essential parameters: the total mass

M = ^ Mk = 5.973 69S 985 p24 kg and the average moment of inertia = ~ X MkR£ = 8.018 385 949 p37 kg m" (the.se

two parameters thcmselves hâve a superabundant accuracy which allows at Icast a direct comparison with the extreme cases

presented in (43) and (44) and which leads to the best pos,sible value of the différence (Cb - Ab) we were looking for).

average radius

(in km)

density of the layer

(in g/cm3)

mass Mk (in 10*^ kg)

Mk = ^ Jl R? (Pk- pk+l)

eccentricity of

the meridian ellipses

(after minimisation of

the mechanical energy)

Ro = 0

Ri = 1221.5

12.90

Ml = 0.015 192 250

ei = 0.069 801 515 54

Rz = 3480

10.91

Mz = 0.974 226 590

ez = 0.072 035 569 37

Rs = 4845

5.391 340 009

Ms = 0.461 997 078

es = 0.075 994 548 44

Ra = 5701

4 421 568 826

Ma = 0 435 367 701

ea = 0-079 157 866 65

Rs = 5971

3.86

Ms = 0.338 854 626

es = 0.080 133 114 66

Re = 6151

3.48

Ms = 0.107 230 536

es = 0.080 790 375 46

R? = 6346.6

3 37

M? = 0 503 280 205

67 = 0.081 515 276 99

Rs = 6356

2.9

Ms = 0.322 672 185

es = 0.081 550 310 60

Rg = 6368.45

2.6

Mg = 1.709 412 701

eg = 0.081 596 868 85

Rio = 6371.2

pio

1.02

Mio = 1.104 975 112

eio = 0.081 607 191 58

— 554 —

APPENDIX 3

ABOUT THE HYPOTHESIS OF EARTH EXPANSION

In 1912, A. Wegener présentée! his Theory of Continental Drift now siibsumed by the

Theory of Plate Tectonics. But this theory was so new and so audacious that it was discarded

by most scientists until the end of the sixties and the discovery of a possible motor: the motions

of convection in the Earth's manlle. These motions are envisaged as being produced by the beat

supplied essentially by the Earth's natural radioactivity.

Similarly, afler the pioneering works of O. HiLGENBERG (1933), A. Du roix, S. W. Carey

(1958, 1976), the theory of Earth Expansion has been promoted by H. G. OwEN (1981, 1984,

1989). This theory helps to explain a wide variety of geological features; however, in the absence

of a plausible explanation. it remains discarded by most scientists.

Nevertheless, let us consider Table 1 of inner Earth. and let us examine the two main layers,

i.e. the lower mantle (55% of Earth's volume and 49% of Earth's mass) and the outer core (16%

of Earth’s volume and 31% of Earth's mass).

The boundary beiween these two layers is particularly abrupt and well-defined, its position

is well-known and it séparâtes two media with vei^ different densities and very different seismic

velocities. Notice that the vclocity Vs of transversal waves is 7 km/s in the lower mantle and

0 in the outer core as if the lower mantle were solid and the outer core liquid.

In these conditions it is reasonable to envisage that the lower mantle and the outer core

are compo.sed of the same material under two different phases, the solid phase (lower mantle)

floating above the liquid phase (outer core) exactiy as in the polar océans in which the ice-cap

is floating on water.

If this is true, in a slowly cooling Earth. the outer core can progressively transform into

lower mantle and the volume of Earth will e.xpand slowly and continuousiy. The average density

of the outer core is 10.9 and that of the lower mantle is 4.9. If we keep these same average

densities during the transformation, the volume of Earth can expand from 70% of its présent

value (when ail the lower mantle was liquid) to 119% of its présent value (when ail the outer

core will be solid). The corresponding radii are 0.89 R and 1.06 R which seems in good agree-

ment with OWRN’s theory of Earth Expansion. If that slow transformation can be extended to

the transition zone (middle mantle). the initial volume of the Earth could hâve been as low as

58% of its présent value.

Date de distribution le 30 juillet 1996

Le fascicule n° I 1996 a été distribué le 29 février 1996.

LOUIS-JEAN

avenue d’Embrun, 05003 GAP Cedex

Tél. : 92.53.17.00

Dépôt légal : 547 — Juillet 1996

Imprimé en France

Mémoires du Muséum national d’Histoire naturelle

Série C - Sciences de la Terre

À partir de 1993, les Mémoires du Muséum national d'Histoire naturelle ne paraissent plus en quatre

séries séparées (A. B, C, D), mais en une seule série. Le nom de la revue devient Mémoires du Muséum

national d'Histoire naturelle (ISSN 1243-4442). La numérotation commence au tome 155.

164 Broutin, J. et al. 1995. La Flore fossile du bassin houiller de Saint-Étienne.

165 Marshall, L. G., Muizon Ch. de & Sigogneau-Russell D., 1995. Pucadelphys andinus

(Marsupialia, Mammalia) l'rom the Early Paleocene of the Santa Lucia Formation, Bolivia.

Bulletin du Muséum national d’Histoire naturelle

Les gisements de mammifères du Miocène supérieur de Kemikiitepe, Turquie. 4^ série, 16, 1994,

section C, n° 1, 240 p.

VIE’ Symposium International, Parc de Miguasha, Québec. Études sur les Vertébrés inférieurs.

Coordonné par Marius Arsenault, Hervé Lelièvre et Philippe Janvier, 4^ série, 17 (1-4), 1995,

section C, 530 p., ca. 200 figures.

Vente en France

MUSÉUM NATIONAL D’HISTOIRE NATURELLE

Sales Office (France excluded)

UNIVERSAL BOOK SERVICES

Éditions Scientifiques du Muséum, Paris

Diffusion Delphine Henry

57, rue Cuvier. 75005 Paris, France

T él.: [.33) (1)40.79.37.00

Fax : [33] ( I ) 40.79.38.40

Internet http://www.mnhn.fr

Dr W. B.\ckhuys

P.O. Box 321 2300 AH Leiden

The Netheri.ands

T el.: [311(71).5 17 02 08

Fax : [311 (71)5 17 18 56

Internet backhuys @ euronet.nl