“7 nl 4 , ‘ de aI w o b, , Lg ve t a i a y: $, ne ee 5 7 ; » ia H y e. ral ‘ ‘ we wt ’ Gy ay ps ; deg tee ff F % purge Ber ise , a 3g 28 Phat Lea cies JN ra oy Ag >) , a Ve) a, " > sf f oe fase . TF, z te a ee ‘3 , i , f ‘ . , ‘ ih re iAP ' . i] Sk. ye ‘ ’ 4 a MA ‘ sa itaiat ¢ { Ho) dies hoe Tey Py ere 4 i iJ , ‘ ’ a . a i ee Te. a fs. , me qe : . ‘er F ‘ Ret ‘ ’ ’ Bee 3,4 ‘ Ar ii aa ‘ PRA, ‘t eater ip . f H . in is p° ae | , f be 4 a eae A 7 “4 4% eet a if . Q + rr Bee ‘ p mw 44 4 -y 2 s? io 4% K’, MS ae a , oy Lar ye ‘Ty ee wh ob WOE dee Mok PRD. fF +o 8 44 ya Pe ss t i re oe) > nos ‘ ’ . | Hote i ' ‘ eres , e ieee. Bow : ‘ ' mane « O89 *, aa: pre. Lary fy 8 PS ¥ ant 6 vbdss. 45 4 Oe ei ea RULE So ee w p Care aa) G4 ha A ds eS a 7 CLS FO ee, eM te Be 4.9 » Ha wee a bd * y . ian ® . ‘ oan DO A Pe ee bye i
wii des
~ ha
Raaghey ie tat 3
sty ] tale
y bs, i ‘ 1 an yt % 2 Se Nay Wek:
Mt (oe einen? “teh
aay?
iting i taredly ve 8a abst soe eS er’ ys 0 eg 8 Ae , . ways,
yy
*\ J emi k ¢
spt
Blin S cote Dek
. ci aa tants
Uy we d Way dee’ 4-3 4
s a et aye
ci
at de ne apts ya Shi rik
Ba Bs Bat PF phe & bel Hoes } Ft eta 5 re t rea site
mS Rae
gait * 3 ye
wy:
* Todewe eat a Fe Ee Otro mely ath ve
Ae
ep)
li
<x
ae
mM
oa WuGI7 LIBRARIES
a
7J
>
AO
Mm
op) TUTION NOILNLILSNI
x =
Or Fe NE ye Z
a Pd vwugi7 LIBRARIES
Zz
wn
om
oof
es
a.
Oo
ee TUTION _ NOLLOLILSNI
Oo
=
=)
a
load
dp)
z WudtT LIBRARIES
, <
Fee.
oO
Bote CSO
au
E
S
9)
TUTION NOILN
YY Lid [8 «a = fo mM ny
VY¥dI1 LIBRARIES | aa oO AJ > Pa) m 3 oP)
TUTION NOILNILILSNI
Ce
Y 2
. > “att Pa
vugi7 LIBRARIES
“ S$
amd a w ocean ia. Bein he ee hae a ud a ul De es. a a oc hae. 4 . res a c S - + =f 4 = ina) — ma. oe © ; N om S) ——= oO = aes < bl meal me SMITHSONIAN INSTITUTION NOILNLILSN a2) ; wo © aa Oo a — ow — 5 yt 2 5 2 5 2 i Mb » = Aly a - GY, a = ra w”) Vij mM wn ri - < ” = (ap) = NVINOSHLINS S3IYVYEIT LIBRARIES INSTITUT Zz n im wee ” Pa in <x = Pas 4) = <x Ro WR z =! oe SXSS 4 > . 2° EN 2 Z = > = : > = a u) = Ww Sic ae 79) i pM ITHSONIAN INSTITUTION NOILNLILSNI NVINOSHLINS Saluva ul ue us 2 fe CC . a oc Bi wo a e < a <x cc = — = a om Z rs) . 5 pe a ag ee | z ad NVINOSHLINS S3SIYVYUSIT LIBRARIES ihe z - z * say (@) — oO — a QS = 9) 5 &) 2 WY E = . m XS eb m S | m W = w = wn SMITHSONIAN INSTITUTION NOILNLILSNI NVINOSHLINS SSIYVY = < = l 4 < = : : 2 §%: : S © ea o Zz > , Yh Vad = = Pa = >” = >" a 79) Zz n =z iy uf NVINOSHLIWS LIBRARIES INSTITU | a = a Jo ane fad a a a a < ps oc Cc ~ = a 4 pad oO = CO. — oO es oO av O = ad za hat rs | _ITESOMIAN INSTETUTION NOILALILSNI NVINOSHLINS S31 uVug a hy S S g . é ee ae — 2) N mas 5 a ae D S an ce = ; sR pe oe = Ee raat - ‘7 = 0 nes Le mi o - by to i¢2) a ah i z. i OSHS ea lyvuag Moti BRARI ES SMITHSONIAN ~ INSTITU < = a < hee = = NS ‘ ee as ee S > SNS fe) x Bip o). = 5 : < O yy. Ue ae O = = Z “py a Z ss = > i = a sae Ss 7’ ” ae ep) * z W | Pp MITHSONIAN _ INSTITUTION NOILALILSNI NVINOSHLINS S3JIYWVUB a = Bee wy ii Ls eutHtSON, tis VS ¥A4 m = AN os uJ ASAD MPI! sir YSSZIN — VSL ” RNG — (Safer
{ | ( ‘ { { (
\\
we x \
- .
RARIES SMITHSONIAN
NOILNLILSNI NOILNLILSNI NOILALILSNE LIBRARIES
INSTITUTION NOILNLILSNI Saiuvugi7 LIBRARI
Saruvugit LIBRARIES
INSTITUTION INSTITUTION INSTITUTION S3Jtuyvydiy
INLILSNI
NVINOSHLINS SAIYVYSIT_ LIBRARIES
2 = 3 a < Bos =| Age = a a wi Z 2 D i i = aS = = mS > @& = : Ss > Gj cr as te 1 ee a ” *. oH a. > R IES SMITHSONIAN _ INSTITUTION NOILNLILSNI NVINOSHLIWNS SaIlu¥vVudi7_LIBRARI = 5 Z 2 & 4 2 Wee 3 . = : = Jie c me c = = a ~ ve ~ 2 i Z O = = ca = as | ae a! s Zz ~~ aay z ISNI_NVINOSHLINS S3IYVYUSIT LIBRARIES SMITHSONIAN INSTITUTION NOILNLILS ae = os = Ay eB S a a = se Ee = a > a es ae z i a - — i Ee . = = w = wn * = wn = ep) = wn =
NVINOSHLINS S31YVua!I] LIBRAR!
INSTITUTION NOI
i
NWINOSHLINS SJIYVYa!
SMITHSONIAN
SMITHSONIAN NVINOSHLIWS SMITHSONIAN NVINOSHLINS
ILSNI_ NVINOSHLIWS LIBRARIES SMITHSONIAN INSTITUTION NOILNMLILS
. \ NI
NOILALILS
NOILNLILSNI
NOILALILSNI LIBRARIES
1ES SMITHSONIAN INSTITUTION NOILNLILSNI
NVINOSHLINS S31u¥yvugi7 LIBRARI
SAIYVvudl)
INSTITUTION INSTITUTION INSTITUTION
INSTITUTION. NOILNLIL
NVINOSHLINS S3INVYGIT_ LIBRARIES NVINOSHLINS SAINVUdIT LIBRARIES
SMITHSONIAN SMITHSONIAN SMITHSONIAN NVINOSHLIWS
DF THE SOUTH AFRICAN MUSEUM
CAPE TOWN >
42
VOLUME 64
y ra
ANNALS OF THE SOUTH AFRICAN MUSEUM ANNALE VAN DIE SUID-AFRIKAANSE MUSEUM
VOLUME 64
THE TRUSTEES OF THE DIE TRUSTEES VAN DIE SOUTH AFRICAN MUSEUM SUID-AFRIKAANSE MUSEUM CAPE TOWN
KAAPSTAD
1974
The ANNALS OF THE SOUTH AFRICAN MUSEUM
are issued in parts at irregular intervals as material becomes available
Obtainable from the South African Museum, P.O. Box 61, Cape Town
Die ANNALE VAN DIE SUID-AFRIKAANSE MUSEUM
word uitgegee in dele op ongereelde tye na beskikbaarheid van stof
Verkrygbaar van die Suid-Afrikaanse Museum Posbus 61, Kaapstad
OUT OF PRINT/UIT DRUK I, AN 3, 5» I ))p BiG 2s 5» Dele), 5 (1-2, 5» 1) 6(1, t—p.1-); 7(1-3); 8, g(1-2), 10(1), I1(1-2, 5, 7, t.-p.l.), 24(2), 27, 31(1-3), 33
Price of this part/Prys van hierdie deel R18,00
Trustees of the South African Museum © _ Trustees van die Suid-Afrikaanse Museum 1974
Issued|Uttgegee SEPTEMBER 1974
ISBN 0 949940 49 6
Printed in South Africa by In Suid-Afrika gedruk deur The Rustica Press, Pty., Ltd., Die Rustica-pers, Edms., Bpk., Court Road, Wynberg, Cape Courtweg, Wynberg, Kaap
149
SLUDIES ON VERTEBRATE PALAEONTOLOGY
A collection of invited scientific papers
presented to
LIEUWE DIRK BOONSTRA
Palaeontologist at the South African Museum, 1927-72 in recognition of his contributions to
Karroo palaeontology
CONTENTS
Lieuwe Dirk Boonstra .. Scientific papers of L. D. Boonstra In reminiscent vein, by S. H. HAUGHTON
The braincase and associated structures of the cotylosaur reptile Procolophon triconiceps Owen, by T. S. Kemp : ue
On the source of therapsids, by E. C. OLson A new genus of gorgonopsid from East Africa, by F. R. PARRINGTON The inner ear of Gorgonops (Reptilia, Therapsida, Gorgonopsia), by D. SIGOGNEAU
Etude du crane d’un jeune specimen de Moschorhinus kitchingi Broom, 1920 (? Tigrisuchus simus Owen, 1876), Therocephalia, Pristerosauria, Moschorhinidae d’Afrique australe. (Remarques sur les Moschorhinidae et les Whaitsiidae), par C. H. MENDREz
A new dicynodont ancestor from the eee Ecca (lower Middle Permian) of South Africa, by T. H. Barry a : Le a. is d ay
The skull and mandible of a new cistecephalid dicynodont, by M. A. CLUVER ..
A new dicynodont (Reptilia, Pe ee from ad zone one of South Africa, _by N. Horton m .. : d :
The alleged synonymy of Lycorhinus and Heterodontosaurus, by A. CHARIG & A. W. CROMPTON The tarsus of theropod dinosaurs, by S. P. WELLEs & R. A. Lone
Alfred Sherwood Romer
Aquatic adaptation in reptiles— primary or secondary, by A. S. Romer ..
Migrations of the Multituberculata and the late Cretaceous connections between Asia and North America, by Z. KIELAN-JAWOROWSKA
Dérive des continents et paléontologie: quelques réflexions, par J.-P. LEHMAN ..
Page
71
117 137
157 167 191 219
221
231
245
1) oe
_ ; in . aq i on - rat Se an Pte 7 77 4 ¥ 2 | pot: ¢ { i xX
NEW GENERIC NAMES PROPOSED IN THIS VOLUME
Cephalicustriodus Parrington, 1974 Cistecephaloides Cluver, 1974 Eodicynodon Barry, 1974 Kombuisia Hotton, 1974
Watongia Olson, 1974 ..
ge oP Seige:
Oe:
eo
LIEUWE DIRK BOONSTRA |
LIEUWE DIRK BOONSTRA
The contributions to this volume of the Annals of the South African Museum are dedicated to Dr Lieuwe Dirk Boonstra, who in January 1972 retired as Head of the Department of Palaeontology at the South African Museum.
Born in Volksrust, Transvaal, in 1905, Dr Boonstra completed his school education in Durban in 1923, and commenced his studies at the University of Stellenbosch, where he qualified with the degrees B.Sc. (1926) and M.Sc. (cum laude) (1928). In 1927 he was appointed as Assistant in Palaeontology at the South African Museum for a period of three years but in 1930, having completed a D.Sc. thesis on the osteology and myology of the pareiasaurian hind-limb, he accepted the post of Palaeontologist at the South African Museum, a position he held until his retirement. During his distinguished career, Dr Boonstra published numerous scientific papers and books, as well as many popular articles. His awards include the Queen Victoria Scholarship of Stellenbosch University (1933) and the Havenga Prize in Natural Science and Technology (Biology Section) of the South African Academy of Science and Art in 1959.
For the greater part of 45 years Dr Boonstra was in sole charge of the South African Museum’s collection of Karroo vertebrates, and it was through his personal efforts in the field that this collection has grown, from the nucleus established by R. Broom and S. H. Haughton to the extensive and representa- tive record that it is to-day.
Boonstra’s collecting trips in the early years of his career were a far cry from the four-wheel drive vehicle and caravan luxury of to-day. With only very limited funds, and carrying his equipment and supplies in a rucksack, he started on his survey of the vast Karroo on foot, relying on friendly farmers for assistance in crating and dispatching any specimens he found. A donkey- drawn cart was later a decided improvement, but many years were to pass before the comparative luxury of even a second-hand truck could be afforded.
In spite of these difficulties, Boonstra’s flair for collecting laid the basis for some extremely significant faunistic and palaeoecological research, and his collections are a record of thoroughness and an ability to see value in every identifiable specimen. Happily, the area on which Dr Boonstra concentrated most of his research was the Tapinocephalus zone, lowermost part of the Karroo succession. Although notorious for the difficulties it imposes on collector and preparator alike, this zone is central to the problems surrounding the origins and early radiation of the mammal-like reptiles. Using the wealth of morpholo- gical and faunistic information available to him through his own collections, and supplementing this with the results of personal examination of related material in the United States, Britain and Russia, Dr Boonstra was able to prepare a series of major contributions dealing with therapsid origins and early interrelationships.
Ann. S. Afr. Mus. 64, 1974: 1-2,
2 ANNALS OF THE SOUTH AFRICAN MUSEUM
But the study of fossil vertebrates encompasses more than the establish- ment of phylogenetic ‘family trees’, and in the early 1950s Dr Boonstra began incorporating his knowledge of the skeleton and inferred musculature of several important Tapinocephalus zone reptiles into a series of striking, life-size recon- structions. These groups, popular as they are, serve a wider purpose than public appeal—as recent trends in research show, more and more attention is being given to a hitherto neglected aspect of palaeontology, namely, the inter- action between the animal and its environment, with particular emphasis on probable behavioural and physiological activities. Taken together, Dr Boon- stra’s morphological, functional and faunal analyses of the early Karroo reptiles have set the scene for an important new approach to our understanding of the extinct vertebrate assemblages of the Karroo, and it is perhaps fitting that his work in this field is now forming a framework for several wide-ranging research projects. It can fairly be said that Boonstra set the pace, and that it is now up to a succeeding generation to continue the course.
28.
29.
SCIENTIFIC PAPERS OF L. D. BOONSTRA
1929
. Pareiasaurian studies. Part I. An attempt at a classification of the Pareiasauria based on
skull features. Ann. S. Afr. Mus. 28: 79-87. (With S. H. Haucuton) Pareiasaurian studies. Part III. On the pareiasaurian manus. Ann. S. Afr. Mus. 28: 97-112. Pareiasaurian studies. Part IV. On the pareiasaurian pes. Ann. S. Afr. Mus. 28: 113-122. 1930
A contribution to the cranial osteology of Pareiasaurus serridens (Owen). Annale Univ. Stellen- bosch (A) 8 (5): 1-18.
Pareiasaurian studies. Part V. On the pareiasaurian mandible. Ann. S. Afr. Mus. 28:
261-280. (With S. H. Haucuton) Pareiasaurian studies. Part VI. The osteology and myology of the locomotor apparatus. A. Hind limb. Ann. S. Afr. Mus. 28: 297-367. (With S. H. Haucuton) 1932
A note on the hyoid apparatus of two Permian reptiles (pareiasaurians). Anat. Anz. 75: 77-81.
Paleobiologiese beskouinge oor ’n uitgestorwe reptielgroep (Pareiasauridae). S. Afr. 7. Sct. 29: 487-494
Pareiasaurian studies. Part VII. On the hind limb of the two little-known pareiasaurian genera: Anthodon and Pareiasaurus. Ann. S. Afr. Mus. 28: 429-435.
Pareiasaurian studies. Part VIII. The osteology and myology of the locomotor apparatus. B. Fore limb. Ann. S. Afr. Mus. 28: 437-503.
. The phylogenesis of the Pareiasauridae. A study in evolution. S. Afr. 7. Sci. 29: 480-486.
1933
The geographical distribution of the pareiasaurs within the Karroo basin. S. Afr. 7. Sci. 39: 433-435:
*n Metode in gebruik vir die opgrawing van die pareiasauriérs. S. Afr. 7. Sci. 30: 436-440. 1934
Additions to our knowledge of the South African Gorgonopsia preserved in the British Museum (Natural History). Ann. S. Afr. Mus. 31: 175-213. The cervical vertebrae of a gorgonopsian (Aelurognathus tigriceps). Ann. Mag. nat. Hist. (10)
14: 53-57.
. A-contribution to the morphology of the Gorgonopsia. Ann. S. Afr. Mus. 31: 137-174. . A contribution to the morphology of the mammal-like reptiles of the suborder Theroce-
phalia. Ann. S. Afr. Mus. 31: 215-267.
On an aberrant gorgonopsian Burnetia mirabilis Broom. S. Afr. F. Sci. 31: 462-470. Pareiasaurian studies. Part [X. The cranial osteology. Ann. S. Afr. Mus. 31: 1-38. Pareiasaurian studies. Part X. The dermal armour. Ann. S. Afr. Mus. 31: 39-48. Pareiasaurian studies. Part XI. The vertebral column and ribs. Ann. S. Afr. Mus. 31: 49-66.
1935
. A note on the cynodont, Glochinodontoides gracilis Haughton. Am. Mus. Novit. 782: 1-6.
A note on the synonymy of the two deinocephalians, Dinophoneus ingens Broom and Jonkeria pugnax (Broom). S. Afr. F. Sci. 32: 329-331.
On a pareiasaurian reptile from South Africa, Bradysaurus whaitsi. Am. Mus. Novit. 7770: 1-4. On some South African reptiles of the suborder Therocephalia preserved in the American Museum of Natural History. Am. Mus. Novit. 7771: 1-12.
On the South African gorgonopsian reptiles preserved in the American Museum of Natural History. Am. Mus. Novit. 7772: 1-14.
Die versamelings van fossiele as basis van die paleontologiese studie. S. Afr. 7. Sci. 32: 332-340.
1936
The cranial morphology of the titanosuchid deinocephalians. Bull. Am. Mus. nat. Hist. 72: 99-116. ;
Some features of the cranial morphology of the tapinocephalid deinocephalians. Bull. Am. Mus. nat. Hist. 72: 75-98.
Ann. S. Afr. Mus. 64, 1974: 3-5.
ANNALS OF THE SOUTH AFRICAN MUSEUM
1938 On a South African mammal-like reptile, Bauria cynops. Palaeobiologica 6: 164-183.
. A report on some Karroo reptiles from the Luangwa Valley, Northern Rhodesia. Q. Jl
geol. Soc. Lond. 94: 371-384.
. Report on some Karroo reptiles from the Upper Luangwa Valley, north-eastern Rhodesia.
[Abstract.] Abstr. Proc. geol. Soc. Lond. 1340: 72-73.
1939 ’n Eeu van palaeontologiese ondersoek in die Karoo. S. Afr. 7. Sci. 36: 73-88.
1940
. Twee nuwe ragitome labirinthodontiérs. Tydskr. Wet. Kuns (n.r.) 1: 195-198.
1946
. Report on some reptilian fossils from Gunyanka’s Kraal, Busi Valley. Proc. Trans. Rhod.
scient. Ass. 41: 46-49. 1947
. Notes on some Stormberg fossil bones from Basutoland. In: stocKLEy, G. M. Report on the
geology of Basutoland: 94-95. Maseru: Basutoland Government. 1948
. Miljoene jare gelede in die Karoo. Johannesburg: Voortrekkerpers.
On the anomodont reptiles from the Tapinocephalus-zone of the Karoo system. In: ROYAL SOCIETY OF SOUTH AFRICA. Robert Broom commemorative volume: 57-64. Cape Town: Royal Society of South Africa. (Special publication. )
1950
Voorlopige lys van terme in die paleontologie (Vertebrata). Engels—Afrikaans. Pretoria: Suid- Afrikaanse Akademie vir Wetenskap en Kuns.
1951
. Beplanning vir ons museumwese. Tydskr. Wet. Kuns (n.r.) 11: 164-169.
Kurze Notiz itiber den Schadel der Dinocephalen-Gattung Keratocephalus F. v. Huene. Neues Fb. Geol. Palaont. Mh. 11: 341-343.
1952
. Agnosaurus gen. nov.: ’n nuwe geslag van die deinocephaliérs. Tydskr. Wet. Kuns (n.r.) 12:
242-245,
. Further observations on the type-skull of Struthiocephalus whaitsi. Ann. Mag. nat. Hist. (12) 5:
455~459-
. Die gorgonopsiér-geslag Hipposaurus, en die familie Ictidorhinidae. Tydskr. Wet. Kuns (n.r.)
12: 142-149.
. A new deinocephalian from the Karroo. Ann. Mag. nat. Hist. (12) 5: 988-989.
’n Nuwe soort van tapinocephalide deinocephaliér: Struthiocephalus akraalensis sp. nov. S. Afr. F. Sci. 48: 247-248.
*n Nuwe tapinocephalide, Riebeeckosaurus longirostris gen. et sp. nov. Tydskr. Wet. Kuns (n.r.) 12: 246-2409.
’n Nuwe titanosuchiérsoort (Anteosaurus abeli). Tydskr. Wet. Kuns (n.r.) 12: 150-151. On. a new tapinocephalid deinocephalian. Ann. Mag. nat. Hist. (12) 5: 509-511.
. Reconstructing life-sized models of prehistoric reptiles. Bull. S. Afr. Mus. Ass. 5: 179-183. . Struthiocephaloides: *n nuwe genus van mormosauride tapinocephaliérs. Tydskr. Wet. Kuns
(n.r.) 12: 237-241. ’n Uiters interessante nuwe deinocephaliér, Avenantia kruisvleiensis, gen. et sp. nov. Tydskr. Wet. Kuns (n.r.) 12: 250-255.
1953
. The cranial morphology and taxonomy of the tapinocephalid genus Struthiocephalus. Ann.
S. Afr. Mus. 42: 32-53.
. The gorgonopsians, Aelurognathus microdon and Hipposaurus boonstrai, reconstructed. Ann. S.
Afr. Mus. 42: 29-31.
. The lower jaw articulatory region in some pristerognathid therocephalians. Ann. S. Afr.
Mus. 42: 54-63.
. A new scaloposaurian genus. Ann. Mag. nat. Hist. (12) 6: 601-605. . A note on some rhynchosaurian remains from Tanganyika Territory. Ann. S. Afr. Mus.
42: 1-4.
———
87.
88.
SCIENTIFIC PAPERS OF L. D. BOONSTRA 5
. A report on a collection of fossil reptilian bones from Tanganyika Territory. Ann. S. Afr.
Mus. 42: 5-18.
. A suggested clarification of the taxonomic status of the South African titanosuchians.
Ann. S. Afr. Mus. 42: 19-28. 1954
. The cranial structure of the titanosuchian: Anteosaurus. Ann. S. Afr. Mus. 42: 108-148. . The Karroo fauna and the origin of mammals. Jn: DART, R. A., ed. Africa’s place in the human
story: 11-14. Johannesburg: S.A.B.C.
. Paranteosaurus gen. nov.: a titanosuchian reptile. Ann. S. Afr. Mus. 42: 157-159. . The pristerognathid therocephalians from the Tapinocephalus zone in the South African
Museum. Ann. S. Afr. Mus. 42: 65-107. A scaloposaurid from the Tapinocephalus zone. Ann. Mag. nat. Hist. (12) 7: 153-157.
. The smallest titanosuchid yet recovered from the Karroo. Ann. S. Afr. Mus. 42: 149-156.
1955
. The girdles and limbs of the South African Deinocephalia. Ann. S. Afr. Mus. 42: 185-326. . Struthiocephalellus: a new deinocephalian. Ann. S. Afr. Mus. 42: 180-184.
1956 Die navorsing van die fossiele van die Karoo-lae. Tydskr. Wet. Kuns (n.r.) 16: 135-138. Pareiasauriér-studies. Deel XIII. ’n Lewensrekonstruksie van Bradysaurus seeleyi. Ann. S. Afr.
Mus. 43: 171-172.
. The skull of Tapinocephalus and its near relatives. Ann. S. Afr. Mus. 43: 137-169.
1957
. The moschopid skulls in the South African Museum. Ann. S. Afr. Mus. 44: 15-38.
1962
. The dentition of the titanosuchian dinocephalians. Ann. S. Afr. Mus. 46: 57-112.
1963
. Diversity within the South African Dinocephalia. S$. Afr. 7. Sci. 59: 196-206.
Early dichotomies in the therapsids. S. Afr. 7. Sci. 59: 176-195.
1964 The girdles and limbs of the pristerognathid Therocephalia. Ann. S. Afr. Mus. 48: 121-165.
1965 The girdles and limbs of the Gorgonopsia of the Tapinocephalus zone. Ann. S. Afr. Mus. 48:
237-249.
. The Russian dinocephalian Deuterosaurus. Ann. S. Afr. Mus. 48: 233-236. . The skull of Struthiocephalus kitchingi. Ann. S. Afr. Mus. 48: 251-265.
1966 The dinocephalian manus and pes. Ann. S. Afr. Mus. 50: 13-26.
. The girdles and limbs of the Dicynodontia of the Tapinocephalus zone. Ann. S. Afr. Mus.
50: I-11. 1967
. An early stage in the evolution of the mammalian quadrupedal walking gait. Ann. S. Afr.
Mus. 50: 27-42.
. Langs verskillende weé (Pareiasauria en Dicynodontia). S. Afr. 7. Sci. 63: 201-206.
1968
. The braincase, basicranial axis and median septum in the Dinocephalia. Ann. S. Afr. Mus.
50: 195-273.
. The terrestrial reptile fauna of Tapinocephalus zone age and Gondwanaland. S. Afr. 7. Sct.
64: 199-204. 1969
. The fauna of the Tapinocephalus zone (Beaufort beds of the Karoo). Ann. S. Afr. Mus. 56: 1-73. . The terrestrial reptile fauna of Tapinocephalus-zone age and Gondwanaland. Jn: INTERNA-
TIONAL UNION OF GEOLOGICAL SCIENCES. International symposium on the Gondwana stratigraphy and palaeontology 1st (Buenos Aires, 1967). Gondwana stratigraphy 2: 327-330. Paris: Unesco. 1971
The early therapsids. Ann. S. Afr. Mus. 59: 17-46.
1972
Discard the names Theriodontia and Anomodontia: a new classification of the Therapida Ann. S. Afr. Mus 59: 315-338.
IN REMINISCENT VEIN By
S. H. HAuGHTON
The first fossil reptilian remains from the Karroo system in southern Africa to be studied scientifically were collected by Andrew Geddes Bain, the ‘father of South African geology’, during the course of his labours as a road-maker and engineer. The story of his enthusiasm for geological observations, fired by a perusal of Lyell’s classic Principles, of his abortive attempts to house a fossil collection in Grahamstown, and of his association with Professor (later Sir) Richard Owen has been told more than once and needs no repetition here.
Early studies of fossil material collected by Bain and others in the latter half of the nineteenth century were concentrated mainly on osteological description with a view to assessment of the taxonomic position of the bones in the animal kingdom. To each specimen was assigned a generic and specific name and a place in a recognized family and order. The total amount of material available for study was, by modern standards, small and some of it was decidedly of a scrappy nature, scarcely warranting selection as the types of new genera and species, at least in the eyes of modern students. Nevertheless, study of the relevant literature shows that, by the end of the year Ig11, descriptions had been published of about 100 new genera containing some 168 new species of Karroo Reptilia by authorities such as Owen, Cope, Lydekker, Seeley, Weithofer, and Broom. Most of the types of these forms were housed outside South Africa.
When the present writer first arrived at the South African Museum from England in September 1911, to take up the post of geologist-palaeontologist, he was instructed by the then Director, Dr L. A. Péringuey, to pay attention to vertebrates from the Karroo strata, the somewhat small collection of bones in the Museum being at that time under the honorary curatorship of Dr R. Broom, who was not resident in the city but had arranged for any new acquisi- tions acquired by the Museum to be sent to his home for study and description.
At that time, the Museum’s most enthusiastic amateur collector was the Rev. (later Archdeacon) J. H. Whaits, the Anglican rector of Beaufort West, who was recompensed financially by the Museum for out-of-pocket expenses incurred by him on his collecting expeditions. A number of the specimens collected by Whaits and sent to Broom for study never reached the Museum, but were acquired from Broom by the American Museum of Natural History in New York where they now lie. Broom prepared for that Museum a descriptive catalogue of the specimens there housed.
Quite early in his service with the South African Museum, the writer was given the opportunity of spending some time under the hospitable roof of Mr W. Musto, owner of the farm Dunedin, Beaufort West, lying north of the Nieuweveld escarpment, and of searching for fossil reptiles there. ‘Searching’
Ann. S. Afr. Mus. 64, 1974: 7-9.
8 ANNALS OF THE SOUTH AFRICAN MUSEUM
meant solitary walks from and to the house, looking for likely outcrops and, if finds of bone were made, the personal transport of the unearthed material to the house for packing and preparation for dispatch to Cape Town. Among the finds on this expedition was a fairly complete skeleton, with skull, of a pareiasau- rian in a mudstone below a slightly overhanging sandstone whose excavation took most of the working hours of three days and whose transport to the house, in sacks, proved somewhat tiring to the writer. This find, after preparation at the Museum with the primitive tools then available, was described in 1913 as the type of Parezasuchus peringuey: Broom & Haughton.
From then until his appointment to the post of Senior Geologist on the Geological Survey of the Union, with an office in the Museum grounds, the writer was given a number of opportunities of field-collecting, mainly of Karroo reptiles. Much hospitality was tendered to him by interested farmers, whether he arrived by cart, drawn by one horse, or by a donkey-drawn fitted covered wagon kindly placed at his disposal by the Geological Survey office in Cape Town. Attention was paid particularly to outcrops of the Tapinocephalus zone in the Prince Albert and Beaufort West districts, of the Cistecephalus zone in the Graaff-Reinet-Murraysburg area, and of the Red Beds in the Herschel district. Among the kindly hosts whose names have been perpetuated in the specific designations given to some of the numerous reptilian types collected may be mentioned the late Mr W. van der Byl of Abrahams Kraal, Prince Albert district (who employed a very keen-eyed shepherd, Ou Jan, who located many of the specimens coming from that farm in his daily care of the flocks and who remembered the positions of most of his finds), the late Rev. J. H. Whaits, then of Beaufort West, the late Mr B. Read, then Head of the Teachers’ Training College at Bensonvale in the Herschel district, and the late Mr B. Cullingworth of Kromme Spruit store in Herschel. These, and many others, turned the otherwise arduous task of ‘tramping the veld’, digging and transport- ing fossils back to base into enjoyable experiences.
Broom’s transactions with the American Museum of Natural History led to a decision by the South African Museum authorities to divorce him from his titular position on the Museum staff, and thereafter the writer was left alone to develop and, if thought desirable, to name and describe the finds that he had added to the collection. In those days, release of the bones from their enclosing matrix was accomplished with the primitive equipment of manual hammer and chisels of varying sizes and carried out in a galvanized iron shed, which also acted as an office. Under these circumstances, little more could be achieved but the removal of bones from their enclosing, and often very hard, matrix and to display their external features. Modern methods of preparation, especially of the internal structure of skulls, were not available and, for much of the time, the research workers had to do the preparatory cleaning single- handed.
At the end of 1920, the writer was appointed to the staff of the Union Geological Survey with an office in the cottage in the Museum grounds, but was
IN REMINISCENT VEIN 9
authorized to maintain an association with the palaeontological section of the Museum. Not long afterwards, the Museum Trustees appointed Mr (now Dr) L. D. Boonstra, a graduate of Stellenbosch University, as a palaeontologist to the Museum with the special task of working on Karroo fossil vertebrates. From the time of his appointment until the writer’s transfer to Pretoria at the end of 1933, we worked in close collaboration, particularly on the fairly large collection of pareiasaurian material that had accumulated, in an attempt to systematize the nomenclature of this herbivorous group. What we devised in this regard, primarily basing ourselves on the external characters of skulls and teeth, has not proved wholly acceptable to later investigators; but that is not an unusual fact in the whole field of fossil reptilian nomenclature.
In the course of more than forty years of service with the Museum Dr Boonstra has devoted himself to collecting fossil reptilian remains especially from the Lower Beaufort beds, to the recovery of the bones from the matrix surrounding them, to their study as osteological specimens, and to the conditions of life and the forms of the animals of which the bones were the framework. In each of these areas of work and study his contributions are of outstanding value. The opportunities that he made of studying South African (and other) materials in museums in the United States and various European countries were used to the full in enabling him to become an expert in his chosen field, while the restorations which were made under his supervision and placed in the exhibition halls of the Museum have aroused a public interest far greater than that which were ever accorded to collections of bone specimens set out in glass cases.
To the knowledge that is being acquired of the kinds of continental animals and of the environment in which they lived and died during the Karroo period, Dr Boonstra’s decades of study and of papers that enshrine many of his findings have contributed in great measure. For this, all students in the subject, both present and future, owe and will owe him a debt of gratitude.
THE BRAINCASE AND ASSOCIATED STRUCTURES OF THE COTYLOSAUR REPTILE PROCOLOPHON TRIGONICEPS OWEN
By T. S. Kemp University Museum and Department of Koology, Oxford (With 1 plate and 5 figures)
CONTENTS PAGE Introduction ; : : ; : : 3 II Description of the braincase and associated bones . en 3 Discussion. : : : i ; : : 19 Summary . : 6 : ; 5 ; se P24) Acknowledgements : : : i ; - PG References. : : : : ; : a 625 Abbreviations : i : : : : ay es INTRODUCTION
It is an extraordinary and unexplained fact that while the skull of the cotylosaurian reptile Procolophon is quite common in Lower Triassic Karroo deposits of South Africa, the braincase is almost always missing. Thus the description by Watson in 1914 of most of the skeleton, including the braincase, has remained practically the sole source of information about this part of the skull, although a figure and a very brief description of the ventral aspect has been published by Broili & Schréder (1936). Watson’s description showed great insight for its time and remains, indeed, surprisingly modern. However, it lacks adequate illustration and certain aspects of the braincase were not visible in his material.
This present study is based upon a single, particularly large specimen of Procolophon trigoniceps, consisting of the skull, lower jaws and anterior part of the postcranial skeleton. It was collected by Dr F. R. Parrington in 1964 from Klipfontein, Bethulie, South Africa and forms part of his collection. Dr Parring- ton’s collection has since been donated to the Museum of Zoology, Cambridge University where this particular specimen is housed, under the catalogue number T 1026.
Only the skull has been prepared so far. The matrix is a hard, red, fine sandstone which is barely affected by acetic acid. Thus preparation has been entirely mechanical, using a Burgess ‘Engraver’. The dorsal roof of the skull has been weathered away with almost surgical precision, making it possible to prepare the braincase and the palate from the dorsal aspect as well as from the more conventional views. The postero-lateral part of the right side of the skull is also missing although the braincase of that side is complete. Distortion has
II
Ann. S. Afr. Mus. 64, 1974: 11-26, 1 pl., 5 figs.
Ps ANNALS OF THE SOUTH AFRICAN MUSEUM
been slight and consists mainly of a degree of dorso-ventral crushing of the occipital region.
DESCRIPTION OF THE BRAINCASE AND ASSOCIATED BONES
Supraocciprital
The supraoccipital has suffered slightly from the dorso-ventral crushing but its form is clear. A narrow dorsal process runs postero-dorsally to make contact with the posterior edge of the skull table. It is triangular in section with a sharp median posterior edge and a flat antero-dorsal face. Ventrally the supra- occipital roofs over the posterior part of the brain cavity by means of a pair of wing-like lateral processes. Each of these processes is fairly thin (about 3 mm)
i v. fl.pt aoe
q. ra.pt Fig. 1. Procolophon trigoniceps CUMZ T 1026. Occipital view of skull, restored. x 1. For abbreviations see page 25.
but wide and lies in an approximately horizontal plane. The distal end contacts the prootic bone in front, the opisthotic extensively and the exoccipital pos- teriorly, and each of these respective sutures is in the form of a flush abuttment. A laterally directed ridge crosses the upper surface of each lateral process, in continuation of the sharp postero-dorsal edge of the paroccipital process. The anterior edge of the supraoccipital is slightly upturned and sharp.
Exoccipital
As a result of the slight downward displacement of the supraoccipital bone, the exoccipital of each side has been pushed a small degree backwards. On the right side the contacting faces of the supraoccipital and exoccipital have been exposed and are seen to be flat, buttressing surfaces of small extent. Below this suture the exoccipital is a narrow pillar, triangular in section, with a flattened lateral-facing surface. It is in contact with a posterior-facing facet
BRAINCASE OF PROCOLOPHON TRIGONICEPS 13
of the opisthotic until, about half-way down the bone, they separate to form the large jugular foramen. Below the foramen contact between the exoccipital and the opisthotic is regained briefly so that the jugular foramen is entirely surrounded by those two bones.
The ventral region of the exoccipital expands, particularly laterally, so that it forms a broad base resting upon the postero-lateral corner of the basi- occipital. As described by Watson (1914), two hypoglossal foramina lie near the ventral edge of the internal (intracranial) surface of the bone, but no external opening can be discerned and thus this specimen must be assumed to differ from Watson’s in that the hypoglossal foramina open directly into the jugular canal.
Bastoccipital
The occipital condyle is formed exclusively from the basioccipital and it is incipiently double, consisting of two low convexities separated by a medial cleft which together form a kidney-shape. Three ridges surround the ventral surface of the condyle, a pair of lateral ones and an anterior, transverse one. From the two points where these ridges meet, a pair of arm-like processes, separated from one another by a deep cleft, run forwards and slightly down- wards to meet the parasphenoidal tubera and indeed to participate in the con- struction of these tubera.
The side of the basioccipital is vertical and high and makes flush contacts with the exoccipital and the para-basisphenoid. A process of the opisthotic overlies it postero-dorsally. A sharp longitudinal ridge runs backwards across the lateral basioccipital surface from the level of the tubera and turns slightly dorsally just in front of the exoccipital, finally to be continued across the base of the latter bone right to the anterior margin of the jugular foramen.
The dorsal surface of the basioccipital is constricted posteriorly by the exoccipitals but further forwards it forms a broad floor to the braincase, slightly concave from side to side. A low median ridge is present.
Para-basisphenoid
Although the basisphenoid cannot be distinguished from the parasphenoid by suture, the surface texture seems to indicate which part is dermal bone and which endochondral.
The paired tubera formed by the parasphenoid together with the basi- occipital are massive, narrow and are separated by a very deep cleft. From the tip of each tuber there runs forwards a rounded ridge which rapidly loses height until it merges with the swelling of the basipterygoid process. The basipterygoid process itself faces antero-laterally and fits into the deep socket in the pterygoid (as described by Parrington 1962) as an apparently kinetic joint. A very slender processus cultriformis arises medially between the basipterygoid processes and runs steeply upwards. Its ventral surface is convex from side to side and it narrows evenly to a sharp point.
14 ANNALS OF THE SOUTH AFRICAN MUSEUM
The huge size of the tubera may be appreciated in lateral view. The
posterior surface of each tuber is deeply hollowed out and, since the tuber |
expands laterally a little compared to the basioccipital, there are a pair of high, narrow concavities facing freely posteriorly and presumably being the sites of attachment of a pair of muscles. Above the level of the tubera, the parasphenoid runs postero-laterally as an otic process which reaches to the level of the fenestra ovalis, although it does not appear to have participated in the true margin of the fenestra as the rest of the margin is formed by unfinished bony lips which must have been covered by a cartilage layer in life. The parasphenoid has no such lip, and it is most probable that an unossified part of the basisphenoid lay immediately internal to the parasphenoid and formed the actual margin of the fenestra ovalis. The otic process of the parasphenoid is continued anteriorly as a broad, rounded ridge all the way to the basipterygoid process. The endo- chondral, basisphenoidal part of the complex rises as a delicate, less well- ossified vertical sheet of bone above this ridge. It forms the actual sidewall of the braincase and is in sutural contact with the anterior end of the prootic. However, between the prootic, the basisphenoid and the otic process of the
preult in pt vac
vi
' bpt. pr
Fig. 2. Procolophon trigoniceps CUMZ T 1026. Ventral view of the posterior part of the skull. x 1.
parasphenoid, the sidewall of the braincase is unossified, leaving a fairly large aperture anterior to the fenestra ovalis. The side of the basisphenoid gradually merges anteriorly with more compact, well-ossified bone until by the level of the basipterygoid process the parasphenoid part of the complex has apparently completely taken over.
BRAINCASE OF PROCOLOPHON TRIGONICEPS 15
The dorsal surface of the basisphenoid has been exposed completely on the left side of the skull. The posterior part continues the braincase floor for- wards from the basioccipital although these two bones are separated by a narrow unossified zone. The sides of the braincase too are formed from the basisphenoid, in continuity with the prootic. However, the sides converge and after a very short distance they turn medially to form the vertical dorsum sellae. This structure is thus composed exclusively of the basisphenoid bone. The sella turcica is a very large, deep pit, divided by a thin median septum in its
pr. cult in.car. f sel. tur a PT dor.sel /f S (co wide ee LR asp we M Q
BOC PRO OPIS Oe
SOC
Fig. 3. Procolophon trigoniceps CUMZ T 1026. Anterodorsal view of the posterior part of the skull. x 1.
posterior part. The more anterior part of the sella lacks the septum and its floor slopes quite steeply upwards to end abruptly at the base of the processus cultriformis. A medially directed spur of the basisphenoid slightly constricts the anterior part of the sella on either side.
As Watson (1914) suspected, a vidian canal pierces the basipterygoid process. The posterior foramen opens immediately behind and above the basi- pterygoid process and the anterior foramen is to be found above the anterior part of the process. That these two foramina do indeed connect with one another is shown by a break on the right side through the canal itself. An internal carotid foramen emerges on the left side of the sella turcica in the region referred to as the anterior part. Whether or not there is a corresponding foramen on the right side is unknown since this side has not been prepared (for the sake of the strength of the specimen).
The dorsal surface of the basipterygoid process is divided by a ridge into antero-dorsal and postero-dorsal faces. The ridge continues medially and antero- dorsally to form the lateral edge of the processus cultriformis. The upper surface of the processus itself is broadly concave from side to side.
16 ANNALS OF THE SOUTH AFRICAN MUSEUM
Opisthotic
The sutures between the opisthotic and the other bones of the braincase are all clearly distinguishable. The medial end forms a long suture with the supraoccipital, behind which the opisthotic turns to form a posterior facing facet which is in contact with the exoccipital, as described. The massive parocci- pital process is formed from both the opisthotic and the prootic. The opisthotic part runs horizontally and somewhat posteriorly and slightly ventrally, as a very well-developed process. The dorsal surface is deeply concave and it grad- ually widens laterally. It is marked off from the posterior surface by a high, narrow ridge which continues the line of a much less marked ridge on the supraoccipital. The posterior surface of the paroccipital process is smooth and bears a horizontal shelf along most of its ventral edge. ‘The upper part of the distal end of the posterior surface is overlapped by a process of the tabular bone. The ventral surface of the paroccipital process is smoothly rounded and is continuous with the ventral surface of the shelf referred to above. The distal end of the process ends freely as an unfinished bone surface that presumably remained cartilage-covered in life. The only direct contact of the opisthotic in the region of the distal end of the paroccipital process is with the tabular bone, so far as can be ascertained in the slightly damaged case of this specimen, by means of the overlap of its posterior surface noted above.
The anterior surface of the paroccipital process is smoothly concave from top to bottom, forming as it does the posterior wall of the middle ear cavity. The upper part is in contact with the prootic over much of its length as described below. Ventral to the prootic, the opisthotic sends a process antero-medially which contacts the basal region of the exoccipital and the side of the basioccipi- tal. The upper part of this process bears a lip of unfinished bone which makes the postero-ventral rim of the fenestra ovalis. The lower part contributes the posterior margin of the jugular foramen. Between the fenestra ovalis and the jugular foramen, the opisthotic process is strongly concave and carries a small foramen.
Prootic
The prootic meets the supraoccipital along a short suture at its postero- medial extreme. A second contact between these two bones occurs at a slightly more ventral level, because a horizontal shelf extends medially from the prootic into the brain cavity, about 1 mm below the dorsal edge of that bone. The shelf meets a similarly developed shelf arising from the antero-lateral face of the supraoccipital. It seems likely that the shelf marks the area of contact of the persistently cartilaginous taenia marginalis of the chondrocranium, as implied by Price (1935) for a similar structure in Captorhinus.
A slight ridge marks the medial part of the prootic, which forms the brain- case wall, from the paroccipital process of the prootic, which extends laterally in contact with the opisthotic. The inner part of the prootic is in the same plane as the adjacent opisthotic, but laterally it twists to form a horizontal shelf in
BRAINCASE OF PROCOLOPHON TRIGONICEPS iy
contact with the dorsal edge of the more or less vertical opisthotic. ‘Thus the prootic constitutes a definite roof over the middle ear cavity, extending as far laterally as does the opisthotic which forms the hind wall of the cavity. The distalmost part of the prootic actually contacts the quadrate ramus of the pterygoid although no suture seems to form.
p.pr
occ. cond BO EOC
j.f Fig. 4. Procolophon trigoniceps CUMZ T 1026.
Right lateral view of the braincase, with certain details added from the left side. x 2.
In front of the paroccipital process, the prootic expands to form a broad, thin sheet making the sidewall of the braincase. A lip of unfinished bone on the ventral edge constitutes the dorsal margin of the fenestra ovalis, while half- way up the bone is a foramen which presumably transmitted the facial nerve (VII). The dorsal edge of the prootic of both sides has been slightly damaged and obscured by the epipterygoid collapsing across it. It appears to bear a notch of some sort exactly where the epipterygoid lies, and if this is true, then it would be reasonable to suppose that such a notch marks the region of exit from the cranial cavity of the trigeminal nerve (V).
The internal surface of the prootic could not be widely exposed in this specimen. The anterior part is certainly well ossified but posteriorly, at the level of the paroccipital process, ossification seems to have been absent, in relation to the formation of the inner ear structures.
Stapes
The left stapes is perfectly preserved and is at least approximately in its correct position with the proximal end adjacent to the fenestra ovalis. It is surprisingly small, about 6 mm in length. The proximal end is swollen and roughly triangular in end-on view, and it narrows rapidly but smoothly into the very slender shaft. The distal end is expanded asymmetrically into a broad
18 ANNALS OF THE SOUTH AFRICAN MUSEUM
but very thin plate. The ventral surface (as oriented in the specimen) is slightly convex from anterior to posterior edges and the posterior edge continues the line of the shaft of the bone. Apart from the proximal surface, the whole bone is formed from well-ossified, finished bone and does not appear to have suffered any damage. Neither stapedial foramen nor dorsal process is present.
Quadrate
The narrow, kidney-shaped articulating surface of the quadrate condyle is oriented transversely with the convex edge at the front. It faces very slightly forwards and is gently concave from side to side. Above the condyle the medial part of the quadrate expands to form a very extensive but thin pterygoid wing, which is plastered on to the anterior surface of the quadrate ramus of the pterygoid and also contacts both the squamosal and the quadratojugal bones. In posterior aspect the pterygoid wing of the quadrate is hidden by the under- lying pterygoid meeting a large ventral extension of the squamosal, the latter itself also a thin sheet of bone in contact with the back of the quadrate. Only the ventral part of the quadrate is visible and is seen to form a suture with the lateral end of the ventral flange of the pterygoid (described below). There is no distinct boss on the quadrate in this region of this particular specimen, unlike the case of the specimen described by Parrington (1962).
Lateral to the condyles the quadrate has the form of a thick, stout process which buttresses powerfully against the quadratojugal, the latter bone extend- ing a little way medially behind the quadrate to further strengthen this contact. The quadratojugal foramen lies in the suture between the quadrate and the quadratojugal above the middle of the condyle. In posterior view the foramen is actually bounded medially by the ventral process of the squamosal.
Pterygoid
The deep sockets for reception of the basipterygoid processes appear to face postero-medially and to have the form of more or less hemispherical cups, as described by Parrington (1962). Immediately behind and lateral to them, the quadrate rami of the pterygoids arise as broad, thin sheets of bone. Each one is extensively overlain by the quadrate as described, and also by a long spur of the squamosal lying along the dorsal region. The postero-ventral surface of the pterygoid is, however, more or less completely exposed. It carries a well- developed crest along its antero-lateral margin which is horizontal at its front end but gradually twists until it is approaching a vertical plane by the time it sutures with the quadrate.
Epipterygoid
The epipterygoids of both sides are present but have been distorted so that they apparently lie more horizontally than in life. The base of each one is a thin, broad sheet of bone plastered against the lateral face of the pterygoid. The ascending process is narrow and typically reptilian.
BRAINCASE OF PROCOLOPHON TRIGONICEPS I9
Quadratojugal
The extraordinary postero-lateral extension of the quadratojugal of Procolophon is of course well known. The medial connections of the bone are with the quadrate as described and also with the ventral process of the squamo- sal behind the quadrate The posterior face is sculptured in a manner suggestive of superficial dermal bone and not at all in keeping with attachment of muscles or other soft parts.
Squamosal
The small component of the squamosal that lay on the skull roof has weathered away but two internal processes remain. An anterior process overlies the dorsal edge of the quadrate ramus of the pterygoid, and a ventral process, in the form of a flat sheet, runs down over the back of the quadrate as described earlier by both Watson (1914) and Parrington (1962). It is in contact with the medial edge of the quadratojugal for most of its height. From the dorso-medial part of the squamosal where the two processes meet, the bone turns medially and runs along the anterior face of the occiput, presumably to the parietal. This part is completely covered behind by the tabular.
Tabular
Again, most of the dorsal part of this bone has been lost by weathering. Anteriorly the tabular forms a strong buttressing suture with the squamosal, a connection which would have been obscured had the skull roof been present. From here the tabular curves ventrally becoming thinner, and medially it covers the dorso-lateral part of the paroccipital process. The lateral part is produced as the triangular tabular ‘horn’, which bears a distinct groove on its ventral surface. The groove runs from the distalmost tip of the bone along its posterior edge, in an antero-medial direction, to disappear at the level of the ventral occipital process of the tabular.
DiscussIon
As members of the cotylosaur reptiles, the procolophons are involved in the debate about the interrelationships of early tetrapods, despite their relatively late occurrence, and in this context the most important regions of the skull are the middle ear, the braincase (particularly the otic region) and the pattern of bones in the temporal region (see Panchen 1972 for a recent review). Since the latter part of the skull of procolophons is so specialized while the former two parts have hitherto remained poorly known, it is not surprising that the phylo- genetic position of this group is still unclear. Olson (1965) succinctly indicated the general possibilities as twofold. In the first place it is possible that the pro- colophonids, with or without their putative relations the pareiasaurs, arose directly from the seymouriamorph amphibian complex independently of the captorhinomorphs. This opinion has been held tentatively by Parrington
20 ANNALS OF THE SOUTH AFRICAN MUSEUM
(1962) on the basis of the temporal bones, and by Daly (1969). In the second place, the procolophons may be a very early divergence from primitive but recognizably captorhinomorph reptiles, a view associated with Romer (1968) and, less confidently, Carroll & Gaskill (1971).
We may now consider what bearing the present study has upon this problem.
Middle ear
The position of the fenestra ovalis is typically reptilian and there is no development of a tube-like extension of the parasphenoid and otic bones carry- ing the fenestra laterally as in seymouriamorphs (White 1939). The lateral expansion of the parasphenoid at the level of the middle ears is in fact no more extensive than in Captorhinus (Price 1935) or Sphenodon for example.
The position of the tympanic membrane, always assuming it was present, is problematical. Since the time of Watson (1914) stress has been laid on the otic notch of Procolophon and its allies, by which was meant the posterior margin of the quadrate along with the corresponding concavity in the superficial dermal bones. In fact the situation is less simple because in the procolophonids secondary extension of the tabular and the quadratojugal has resulted in a large semicircular notch in the posterior margin of the skull roof, which has a marked but superficial similarity to the otic notch of the labyrinthodonts. This secondary, dermal notch does not reflect at all closely the form of the under- lying primary otic notch. While Romer (1956, 1968) uses the term ‘otic notch’ for the primary notch, Colbert (1946) and Parrington (1962), among others, use the term for the secondary notch, and the implication is that these two usages of ‘otic notch’ refer to two different possible positions for the tympanic membrane.
There are good reasons for doubting that the tympanum lay within the secondary, dermal notch. Particularly, the stapes of the present specimen lies approximately in its natural position with its swollen footplate more or less in contact with the fenestra ovalis, yet its undamaged distal end lies a relatively huge distance from the dermal notch (Figure 2; Plate 1). In order to reach the notch, a cartilaginous extrastapes some 15 mm in length would have been necessary. ‘This represents 250% of the length of the stapes and seems most unlikely on mechanical grounds, and there is no modern reptile known which has such an arrangement (Baird 1970).
The actual edges of the dermal notch give no indication that they sup- ported any part of a tympanic membrane. The edges of the tabular and the squamosal are smoothly rounded, while that of the quadratojugal is sculptured in the manner of the rest of the superficial dermal bones, which does not suggest the presence of soft tissues. It would seem therefore that the tympanum did not lie within this dermal notch, although it remains possible that the notch marks the outer opening of an external auditory meatus which led internally to the ear drum.
BRAINCASE OF PROCOLOPHON TRIGONICEPS 21
PuaTE 1. Procolophon trigoniceps CUMZ T 1026.
Stereophotographs of the skull, natural size. Above, ventral and slightly posterior view; below, occipital view. (Photographs by Mr William Lee.)
P47. ANNALS OF THE SOUTH AFRICAN MUSEUM
If the stapes is to be taken as a guide, the tympanum must have been somewhere around the distal end of the paroccipital process, which in turn is posterior to the medial part of the quadrate. There are no markings on the bones of this region so the precise position can only be guessed, but in any case this represents a generally reptilian, post-quadrate position.
The stapes itself is specialized from the presumed primitive condition by the absence of both stapedial foramen and dorsal process. Regarding the latter, it is possible that the expanded distal end of the bone represents a broad dorsal process, although there are no serious indications of this. It is difficult to see how such a process could have contacted the paroccipital process, for example. The generally small size, delicacy and abrupt expansion of its foot- plate all give the stapes a rather ‘sauropsid’ appearance (Olson 1966; Baird 1970).
The tympanic cavity too, as far as can be judged from its osteological features, has the arrangement of a ‘sauropsid’ reptile like Sphenodon or the lizards. The manner in which the prootic contributes to the roofing of the cavity at the level of the top edge of the paroccipital process, and the extensive anterior limit formed by the quadrate ramus of the pterygoid, support such a comparison, which can be extended, too, to Captorhinus.
Thus the evidence of the middle ear structure points towards a true rep- tilian affinity of Procolophon, rather than to a particular relationship with the seymouriamorphs, despite certain difficulties in the form of the stapes and the absence of a clear-cut indication of the position of the tympanum. In fact, it would not be surprising for loss of the tympanum to have occurred, a common phenomenon amongst modern squamates (Baird 1970).
Braincase
The form and relations of the paroccipital process are often used to dis- tinguish between labyrinthodonts (including seymouriamorphs) and reptiles. In the labyrinthodonts the paroccipital process runs dorso-laterally and con- tacts the tabular bone at its distal end, thus forming a brace between the skull roof and the basis cranii. The supraoccipital is usually unossified. In contrast, the reptilian paroccipital process typically runs laterally to contact the suspen- sorial (squamosal-quadrate) region of the skull, bracing of the roof to the basis cranil involving a well-ossified supraoccipital bone.
In the case of Procolophon, the paroccipital process has a perfectly reptilian form and it extends laterally towards the suspensorial region; and a well- developed supraoccipital bone, closely similar to that of Sphenodon for example, is present. The difficulty is that the distal end of the paroccipital process is in contact with the tabular bone, and fails to reach the squamosal or quadrate, and thus the technical reptilian requirement is not met. However, the temporal region of the Procolophon skull is extremely modified from the primitive tetrapod condition, and in particular the quadrate bone is much further forwards, the squamosal reduced and the tabular (if indeed this bone is correctly homologized)
BRAINCASE OF PROCOLOPHON TRIGONICEPS 23
greatly enlarged and extended on to the occipital surface. It is not safe therefore to assume that the failure of the paroccipital process to contact the squamosal and/or the quadrate is anything more than a further manifestation of the extensive remodelling of this part of the skull.
The processus cultriformis of the parasphenoid of Procolophon is very much reduced compared to that of Seymouria (White 1939) or Kotlassia (Bystrow 1944), a feature in which it resembles all typical reptiles.
Like the middle ear, the structure of the braincase is essentially reptilian but shows certain peculiarities, in this case the relation of the distal end of the paroccipital process to the rest of the skull.
Primitive procolophons
The strongest arguments for relating procolophons directly to seymouria- morphs are based on certain forms which combine some of the procolophonid
SOC
Fig. 5. Nyctiphruretus acudens.
Skull in ventral and occipital views to show braincase and stapes. Natural size. (Redrawn from Efremov 1940.)
specializations with seymouriamorph characters. The most important is Nyctiphruretus (Fig. 5) from the Upper Permian of Russia (Efremov 1940) which shows shortening of the jaw length, reduction of the skull table coupled with loss
24 ANNALS OF THE SOUTH AFRICAN MUSEUM
of the supratemporal, relatively huge orbits and an enlarged pineal foramen. These Procolophon-like characters are however associated with a seymouria- morph-like braincase in which the paroccipital processes extend dorso- laterally to make firm contact with the tabulars at the postero-lateral corners of the skull. The tabulars do not extend down the occipital surface. Also like the seymouriamorphs, the processus cultriformis is unreduced. On the other hand, the middle ear is peculiar for there is no otic notch in the rear margin of the skull roof and the stapes runs ventro-laterally from the fenestra ovalis to a firm contact with the quadrate, and thus this part of the skull resembles neither Procolophon nor the seymouriamorphs. Clearly convergence has occurred. If Nyctiphruretus does link Procolophon directly to the seymouriamorphs, then Procolophon has achieved a reptile-like braincase and ear independently of captorhinomorphs. If Nyctiphruretus and Procolophon are related and are descended from true reptiles, then the braincase of Nyctiphruretus has converged upon the structure of the seymouriamorph braincase. And if Procolophon is a captorhinomorph derivative, while Nyctiphruretus is a seymouriamorph, then the similarities between these two genera have arisen independently, pre- sumably in relation to the adoption of a similar mechanical jaw function. As long as it is held that braincase and middle ear structure are the best guides to lower tetrapod phylogeny, the latter hypothesis will remain the most attractive.
None of the other supposed primitive procolophons help in this problem. Nycteroleter, also from the Russian Upper Permian (Efremov 1940), is even more seymouriamorph-like with its dorsal otic notch. The Upper Permian form Owenetia, from South Africa (Broom 1939), has a skull roof which appears to be ideally primitive procolophonid but unfortunately nothing is yet known about the details of its braincase.
The recent description of a Lower Permian tetrapod Acleistorhinus by Daly (1969) as a primitive procolophon has been doubted by Carroll & Gaskill (1971) and receives no support from the present work. The extraordinary structure of the braincase and stapes of Acleistorhinus only serve to highlight the unremarkably reptilian-like form of the corresponding structures in Procolophon.
SUMMARY
A detailed, bone by bone description of the braincase, stapes, pterygo- quadrate and parts of the quadratojugal, squamosal and tabular of a large specimen of Procolophon trigoniceps is given.
The middle ear, paroccipital process and processus cultriformis are all basically reptilian in form, and point towards a captorhinomorph ancestry of procolophons, rather than their direct derivation from seymouriamorphs.
Nyctiphruretus is problematical and may resemble the procolophonids by convergence, being itself a seymouriamorph.
—
BRAINCASE OF PROCOLOPHON TRIGONICEPS 25
ACKNOWLEDGEMENTS
My thanks must go particularly to Dr F. R. Parrington, F.R.S., for his continued interest and encouragement, and for the initial loan of the specimen. Dr K. A. Joysey has kindly allowed me to retain the specimen on loan from Cambridge University Museum of Zoology. I am grateful to Mr William Lee for undertaking the photography associated with this work.
REFERENCES
Bairp, I. L. 1970. The anatomy of the reptilian ear. In: GANs, c. & PARSONS, T. S., eds. Biology of the Reptilia. 3: 193-275. London; New York: Academic Press.
Brom, F. & ScHRODER, J. 1936. Beobachtungen an Wirbeltieren der Karrooformation. xxi. Uber Procolophon Owen. Sher. bayer. Akad. Wiss. 1936: 239-256.
Broom, R. 1939. A new type of cotylosaurian, Owenetta rubidgei. Ann. Transv. Mus. 19: 319-321.
Bystrow, A. P. 1944. Kotlassia prima Amalitzky. Bull. geol. Soc. Am. 55: 379-416.
Carro_i, R. L. & GasxiLz, P. 1971. A captorhinomorph reptile from the Lower Permian of Europe. 7. Paleont. 45: 450-463.
Co.BerT, E. H. 1946. Hypsognathus, a Triassic reptile from New Jersey. Bull. Am. Mus. nat. Hist. 86: 229-274.
Daty, E. 1969. A new procolophonoid reptile from the Lower Permian of Oklahoma. 7. Paleont. 43: 676-687.
EFREMoV, J. A. 1940. Die Mesen-Fauna der permischen Reptilien. Neues 7b. Miner. geol. Paldont. (Abt. B) 84: 379-466.
Otson, E. C. 1965. Relationships of Seymouria, Diadectes and Chelonia. Am. Zool. 5: 295-307.
Otson, E. C. 1966. The middle ear— morphological types in amphibians and reptiles. Am. ool. 6: 399-419.
PANcHEN, A. L. 1972. The interrelationships of the earliest tetrapods. In: Joysry, K. A. & KEMP, T. S., eds. Studies in vertebrate evolution: 65-87. Edinburgh: Oliver & Boyd.
PaRRINGTON, F. R. 1962. Les relations des cotylosaures diadectomorphes. Colloques int. Cent. natn. Rech. scient. 104 (Problémes actuels de paléontologie): 175-185.
Price, L. I. 1935. Notes on the braincase of Captorhinus. Proc. Boston Soc. nat. Hist. 40: 377-386.
Romer, A. S. 1956. Osteology of the reptiles. Chicago: University of Chicago Press.
Romer, A. S. 1968. Notes and comments on vertebrate paleontology. Chicago; London: University of Chicago Press.
Watson, D. M. S. 1914. Procolophon trigoniceps, a cotylosaurian reptile from South Africa. Proc. zool. Soc. Lond. 1914: 735-747.
Waite, T. E. 1939. Osteology of Seymouria baylorensis Broili. Bull. Mus. comp. Zool. Harv. 85:
325—409. ABBREVIATIONS
BOC — basioccipital occ. cond—occipital condyle bpt.pr—basipterygoid process OPIS — opisthotic BSP— basisphenoid ot. pr. psp— otic process of the parasphenoid cond — condyle p. pr—paroccipital process dor. sel—dorsum sellae pr. cult— processus cultriformis EOC— exoccipital P— parietal EPI—epipterygoid PRO — prootic fen. ov—fenestra ovalis PSP —parasphenoid in. car. f—internal carotid foramen PT— pterygoid in. pt. vac—interpterygoid vacuity p.t.fpost-temporal fenestra j.f—jugular foramen pt. w. q—pterygoid wing of the quadrate
26 ANNALS OF THE SOUTH AFRICAN MUSEUM
Q— quadrate
Q J—quadratojugal
qj. f— quadratojugal foramen
q. ra. pt— quadrate ramus of the pterygoid sel. tur—sella turcica
sh—shelf for cartilaginous taenia marginalis SOC—supraoccipital
SQ —squamosal
STA — stapes
TAB-— tabular
V —notch for the trigeminal nerve vid — opening of the vidian canal
v. fl. pt—ventral flange of the pterygoid VII—foramen for the facial nerve
ON THE SOURCE OF THERAPSIDS By EVERETT C. OLson University of California, Los Angeles (With 2 plates, 5 figures and 1 table)
CONTENTS
PAGE Introduction : ; : 5 ; ; 27 A North American perecuonccn : : : 28 Chronofaunas in the mid-continental limited States : : 37
Significance of North American therapsids The Chickasha gorgonopsian ‘ : : : ‘ 41 The San Angelo therapsids ; : : : : 42 Interpretation and summary é ‘ ; : : 43 References P P : 5 ; 3 ; : 45
INTRODUCTION
The very fragmentary evidence of therapsids in North America during the Permian has provided interesting but uncertain information concerning the geographic distribution of this order and the times of origin of different evolutionary lines. Many questions were raised in a general report (Olson 1962), but few of them have been touched on more than lightly since that time. They were summarized as far as possible in my recent book (Olson 1971: 641-655). Some aspects of the problems have been treated by Chudinov, in particular in Sigogneau & Chudinov (1972).
Discovery of vertebrates in the Chickasha Formation, Guadalupian of Oklahoma (Olson 1965), has added information. A better understanding of the vertebrates of the Cutler and Abo Formations, Wolfcampian of northern New Mexico and south-western Colorado, through the work of Vaughn (1966, 19694, b) and Lewis & Vaughn (1965) has given new insights into the signifi- cance of this faunal complex. Together these studies have produced substantial evidence of faunal assemblages which evolved largely independently of those of south-western New Mexico, Texas, and Oklahoma. The latter have been grouped into what has been termed the Permo-Carboniferous Chronofauna; the former assemblages have been termed the Caseid Chronofauna (Olson 1971: 830, fig. 145E). Only recently has it been determined that the Caseid Chronofauna has any relationship to the origin of therapsids.
Study of a new specimen found during the summer of 1972 in the Chickasha Formation revealed that it is not, as informally reported at the time of dis- covery, a somewhat therapsid-like sphenacodontine, but rather that it is a very primitive gorgonopsian. The implications of this find are, of course, manifold and form an important part of the substance of this paper. Following a descrip-
27
Ann. S. Afr. Mus. 64, 1974: 27-46, 2 pls, 5 figs, 1 table.
28 ANNALS OF THE SOUTH AFRICAN MUSEUM
tion of the new animal, they are considered in relationship to the broader problem of the source of therapsids.
The figures in this paper were drawn by Mrs Kathryne Bezy and the photographs were taken by Mr Rick Lassen. The study was supported by a National Science Foundation Grant, GB 31283X.
A NortH AMERICAN GORGONOPSIAN
Class REPTILIA Subclass SYNAPSIDA
Order THERAPSIDA Suborder GORGONOPSIA Family Gorgonopsidae
Watongzia gen. nov.
Diagnosis: A very primitive, sphenacodont-like gorgonopsian reptile. Estimated body length 1,22 to 1,52 metres. Skull with large preparietal bone. Ten to 12 simple, recurved, conical cheekteeth, diminishing in crown height from anterior to posterior; a large, anterior, canine-like tooth.
Vertebrae with centra relatively short (see measurements, Table 1), deeply amphicoelous and sharply keeled. Neural arch with a deep pit on either side above the diapophysis. Neural spine slender, height about equal to 4
TABLE 1
Measurements of Watongia meieri gen. et sp. nov. in millimetres. VERTEBRAE I 2 3 4 | Length, centrum, basal .. cf es £ 21 20 — a Total height, anterior .. ie ee fe 85 — =e — | Height neural spine, anterior .. i te 43 = = = | Width posterior zygapophyses ne ae 26 BF = = | Length diapophysis, dorsal he 1 oe — 12 12 14 LIMB BONES IU, PW DW WS Humerus .. fa Es ce 56 OR 65 — 16* Radius a oe os te a); ae 89 18 19 10** Ulna se he See a as Md Woe 2 33 30 17
Abbreviations: L, maximum length; PW, maximum proximal width; DW, maximum distal width; WS, minimum shaft width. *—with ventral surface dorsal; **— anterior view; *** — including olecranon process.
of total vertebral height. Diapophysis short, stocky and with heart-shaped articular surface. Posterior cervical and anterior dorsal ribs with tuberculum somewhat reduced and not fully separated from capitulum.
Scapula moderately high and narrow; cleithrum present but very slender. Humerus lightly structured with well-defined shaft and angle between two ‘blades’ of about 60 degrees. Carpus with very large, pisiform element; distal
ON THE SOURCE OF THERAPSIDS 29
5 present and not fused with distal 4. Phalangeal formula 2?453, with distal phalanges rather short and unguals strongly claw-shaped. The name is after the town of Watonga, Blaine County, Oklahoma.
Watongia meieri sp. nov.
Holotype: UCLA VP 3132. Part of the frontal-parietal region of the skull, including the dorsal part of the orbital margin and fragments of the maxilla with teeth in place. Three anterior dorsal vertebra with fragments of others. Left scapula and part of right; left clavicle; part of left cleithrum; and part of interclavicle. Several ribs and numerous gastralia. Right and left fore limbs with carpus and partial complement of phalanges.
Horizon and locality : Chickasha Formation, level of middle part of Flowerpot Formation, El Reno Group, Guadalupian, Upper Permian. From fossil site BC 7 (Olson 1965), NW. 4, NE. 4, Sec. 33, T. 18 N., R. 11 E., Blaine County, Oklahoma. The species name is given for Mr Meier of Hitchcock, Oklahoma, on whose land the specimen was found.
Description: ‘The specimen was found somewhat scattered in red, sandy shale. It lay near the base of a gently dipping deposit formed in a stream channel. Only the anterior part was found; the skull had weathered out and much of it had disintegrated and disappeared. At about the same level in this site have been found specimens of Rothianiscus and Cotylorhynchus.
1. Skull: Figure 1B, Plate IA, B. Part of the skull roof and fragments of the maxilla are all that have been preserved. The most significant feature is the presence of a large preparietal bone in the skull roof. There were about 12 teeth in the postcanine series. ‘The exact number is uncertain for some of the fragments cannot be certainly placed in the series. The teeth appear to have decreased regularly in crown height from anterior to posterior. A single ‘canine’ is present. Its height is estimated as 20 to 25 mm, somewhat more than twice the height of the longest postcanine. Skull length is estimated to have been about 24 to 26 cm on the assumption that proportions were more or less those of Dimetrodon.
2. Vertebrae: Figure 2A, B, Plate IID. Three anterior dorsal vertebrae and fragments of others have been recovered. Many of the features are those found in sphenacodonts. Neural spines, however, are not elongated or heavy as in sphenacodontines. Centra are strongly keeled, deeply amphicoelous and not elongated. Measurements in Table 1 show the major dimensions. Trans- verse processes are relatively short and thick and terminate in stout, heart- shaped articular surfaces. Above the diapophysis on each side, as in sphena- codontines, the neural arch has a deep pit on its lateral surfaces. A small inter- centrum is present on the anterior margin of one of the vertebrae.
3. Ribs: Figures 1 EF, 2D; Plate IIE. The ribs are slender and gently
curved. Comparisons of the shapes of the heads with those of sphenacodonts and gorgonopsians are shown in Figure 2D. The structure of the head is some-
30 ANNALS OF THE SOUTH AFRICAN MUSEUM
what more like that in gorgonopsians, with the tuberculum reduced and the bifid character much less well expressed than in sphenacodonts. Tooth-pick shaped gastralia were found scattered throughout the deposit in the vicinity of the specimen, Plate ITA, B.
--
7 (oee----
Fig. 1. Watongia meteri gen. et sp. nov.
A. Reconstruction of the left lower limb, based on data from both right and left. B. Skull frag- ment, showing midline, orbital margin and preparietal element. C. Right humerus, dorsal. D. Right humerus, ventral. E. Dorsal rib, head missing. F. Anterior dorsal rib. Abbreviations: FR, frontal; M, midline; PA, parietal; PO, post-orbital; PP preparietal. All x 4.
ON THE SOURCE OF THERAPSIDS 31
4. Shoulder girdle: Figure 3A, B. A somewhat broken left scapula, part of the right scapula, a partial left clavicle, a cleithrum and part of the inter- clavicle provide the information on this part of the skeleton. The scapula is
PLATE I. Watongia meieri gen. et sp. nov. A. Fragments of jaws and teeth in medial aspect. Lower set roughly in order in jaw. B. A frag- ment of the dorsal platform of the skull showing the preparietal bone. See Fig 1B for identification of elements. Midline of skull to top of page and anterior to the right. Both x 1.
32
ANNALS OF THE SOUTH AFRICAN MUSEUM
Fig. 2.
A. Anterior dorsal vertebra of Watongia meieri gen. et sp. nov.: 1. posterior, 2. lateral, 3. anterior. B. Anterior dorsal vertebra of Watongia meieri gen. et sp. nov.: I. posterior, 2. lateral, 3. anterior. C. Manus: 1. Aelurognathus (Boonstra 1934), 2. Watongia, 3. Dimetrodon (Romer 1956). D. Heads of anterior dorsal ribs: 1. Ctenospondylus, 2. Dimetrodon, 3. Watongia, 4. gorgonop- sian. E. Ulna: 1. Dimetrodon (Romer 1956), 2. Watongia, 3. Gorgonops (Broili & Schroeder 1935), 4. Aelurognathus (Boonstra 1934), 5. Lycaenops (Colbert 1948). F. Humerus: 1. gorgonopsian, 2. Watongia, 3. Dimetrodon (Romer 1956). G. Radius: 1. Dimetrodon (Romer 1956), 2. Watongia, 3. Aelurognathus (Boonstra 1934). H. Knoxosaurus (Olson 1962). A and B x 4
2°
ON THE SOURCE OF THERAPSIDS 33
PuaTE II. Watongia meieri gen. et sp. nov.
A. Right lower limb and carpus. Reconstructed part of ulna in white based on impression in the
enclosing matrix. B. Left fore limb including carpus and manus. Under side of limb with digits
flexed. Radius absent. C. Right humerus in ventral aspect. D. Anterior dorsal vertebra, from left side. E. Anterior cervical rib. All x 3.
34 ANNALS OF THE SOUTH AFRICAN MUSEUM
high, rather narrow, and the cleithrum is small. The most distinctive feature of the scapula is a strong rugosity on the inner surface adjacent to the area of attachment of dorsal part of the clavicle. This extends inward over about 4 of
Fig. 3. Watongia meiert gen. et sp. nov.
A. Reconstruction of right shoulder girdle. B. Left scapula and part of cleithrum, inner side. C. Interclavicle, preserved parts and suggested relationships in reconstruction. D. Clavicle, left inner and outer aspects. All x 4.
ON THE SOURCE OF THERAPSIDS 35
the scapular blade and carries dorsally in a slightly depressed area marked by vertical striae. Apparently this represents an area of insertion of ligaments from the clavicle, although the area seems excessively extensive for such liga- ments. It is possible that it marks the insertion of tendons of a large, levator scapulae which invaded the inner surface of the scapula. Again, the extent is considerably greater than would be expected. Various pelycosaurs and primi- tive therapsids have a somewhat similar small area of striated markings in this vicinity, but the development is much less. The functional significance in the new animal remains uncertain.
The clavicle and interclavicle, as shown in Figure 3C, D, fall well within the range of structures found among the pelycosaurs and primitive therapsids and have few unusual features. At the level of flexure between the shaft and ventral blade of the clavicle the anterior and inner margin of the bone has a strongly rugose surface, much like that on the inner surface of the scapula. Presumably this also was the site of attachment of ligaments.
5. Fore limbs: Figures 1A, 2C, E, F, G; Plate IIA, B, C. Most features are well shown in the figures and plates. Comparisons with sphenacodonts and gorgonopsians are shown in outline in Figure 2C and measurements are given in Table 1. It will be seen from these that in general the limb elements are intermediate between sphenacodonts and gorgonopsians, with the humerus, radius and ulna somewhat more like comparable bones in the latter and the manus closer to that of sphenacodonts.
The humerus is lightly built. The right humerus is well preserved and the left is crushed. The two blades in the right make an angle of about 60 degrees with each other. The deltopectoral crest has a well-developed boss for the insertion of the pectoralis muscle but is rather weak distally. The ectepicondyle is strong but the supinator process, which is not well preserved in either humerus, appears to have been weak. The entepicondyle is strong but is not greatly expanded. An entepicondylar foramen is present, but is not visible in dorsal aspect. No ectepicondylar foramen has been identified. The area in which it would be, were it present, is damaged and its existence is not out of the question.
The radius is slender, curved and slightly twisted along the long axis. It is much more like the radii of primitive therapsids than those found in sphena- codonts. The ulna is thick relative to its length. The olecranon process is moderately developed and the distal end of the ulna is broad. Except for the last feature, the ulna resembles that of various gorgonopsians, Figure 2E. It is not basically sphenacodont in its principal features.
The carpus, metacarpus and phalanges are preserved, mostly intact as shown in Plate IIA, B. The carpus is basically sphenacodontine in character. Distals 4 and 5 are not fused. Fusion is indicated in all figures of the gorgonop- sian carpus. Were it not for the presence of a remnant of distal 5 in Watongia, this interpretation would undoubtedly have been made, for the shape of distal 4 is essentially that of the supposedly fused element in various primitive therapsids.
36 ANNALS OF THE SOUTH AFRICAN MUSEUM
The pisiform is extremely large and presumably the tendon of the flexor muscu- lature was unusually strong. This is in marked contrast to the known feet of gorgonopsians and the element is as strong as or stronger than in Dimetrodon.
The phalangeal formula apparently was of the primitive 23453 pattern, although the second digit has not been identified. There is only moderate reduction of the more distal phalanges, much as in Dimetrodon and the gorgonop- sian, Aelurognathus, Figure 2C.
6. Systematic position: Watongia is placed with the Gorgonopsia primarily upon the undeniable presence of the preparietal. ‘This element is present only among gorgonopsians and anomodonts and its form in the latter is quite different from that in the gorgonopsians. It is, of course, possible that it was developed independently in Watongia and gorgonopsians but, in view of the many other structures in which Watongia tends towards the gorgonopsian con- dition, it seems unlikely that this is the case.
The specimen shows a mosaic of sphenacodont and gorgonopsian charac- ters. In it we find one of the rare instances of an almost ideal intermediate between two major groups, but one in which a definitive feature makes placement in one or the other plausible. Although solving an otherwise annoying taxonomic dilemma, the presence of the preparietal does not detract from the essentially intermediate nature of the Watongia.
The family Gorgonopsidae, as revised by Sigogneau (1970) has 22 genera which, while coherent in fundamental characteristics, have considerable variation in skull form, dentitions, and development of limb structure. The Russian gorgonopsians, Amalitzkia and Inostrancevia (Pravoslavley 19274, }), not treated by Sigogneau, fall within the gorgonopsids. Sigogneau recognized two subfamilies, Gorgonopsinae and Rubidgeinae. Major differences separate the two and it is to the former that Watongia bears close resemblance.
Much of the work on gorgonopsian systematics has treated skull structure and dentitions. The postcranium has not been studied for most genera so that the range of variation is not well known. Most of what is available is the result of work by Broom (1930), Broili & Schroeder (19354, 6), Boonstra (1924, 1953; 1963, 1965), Pravoslavlev (19274, b), Colbert (1948) and the summary by Sigogneau (1970). Little is known of the postcranium of the oldest specimens from South Africa. Most generalized of the better known materials from later times appears to be Aelurognathus from the Endothiodon and Cistecephalus Zones. Gorgonops (Broili & Schroeder 19354, b), from the base of the Endothiodon Zone is at much the same level, and Lycaenops (Colbert 1948) is much more advanced.
Watongia, although more primitive than any genus heretofore assigned, can be included in the family Gorgonopsidae and the subfamily Gorgonopsinae without seriously modifying the morphological limits given by Sigogneau.
ON THE SOURCE OF THERAPSIDS 37
CHRONOFAUNAS IN THE Mip-CoNTINENTAL UNITED STATES
The persistence and coherence of ecological systems during the late Pennsylvanian and early Permian of the United States is being increasingly documented. The vertebrate portions of these temporally persistent ecological systems have been termed chronofaunas, or more precisely vertebrate chrono- faunas (Olson 1952, 1958). A useful way to view and study the evolution of particular phyla is in the chronofaunal context, and that is the approach used in the remainder of this paper.
The excellent record of lowland, deltaic vertebrates in North America is the basis for what has been termed the Permo-Carboniferous Chronofauna (Olson 1971: 645-654). This is characterized by an aquatic base of the food chain, with the large carnivore Dimetrodon standing at the peak of the pyramid. It is best known from ‘Texas, Oklahoma and southern New Mexico, Figure 4. It ranges through the Wolfcampian and Leonardian, Figure 5. In the eastern
COLORADO ©)
®
NEW MEXICO
Fig. 4. Sketch map showing areas that have yielded specimens of Caseid Chronofauna as cited in text.
1. Lower Permian (Leonardian) of Texas. 2. Area of San Angelo Formation of Texas from which fossils have come. 3. Area of fossiliferous Chickasha beds of Oklahoma. 4. Northern New Mexico site. 5. The south-western Colorado site.
38 ANNALS OF THE SOUTH AFRICAN MUSEUM
part of the United States, in Ohio, Pennsylvania, and West Virginia, a some- what similar chronofauna exists. It is less well known and was partially isolated from the Permo-Carboniferous Chronofauna. European Autunian deposits, although showing little temporal continuity, indicate the existence of a generally similar system in the Old World during the very early Permian and, from what little is known, the early Permian vertebrates in many parts of the world were quite similar.
Less well known, but slowly becoming clearer, is a second chronofauna of the Mid-Continent area of the United States. This is the complex called
S.W. COLORADO N. NEW MEXICO TEXAS OKLAHOMA U SUSAR
\SHEEVO) DOG CREEK FM. DOG evita BLAINE FM. BLAINE FM. FLOWER POT FM. FLOWER POT FM
Zit OCHER ZI SAN ANGELO FM. DUNCAN FM.
GUADALUPIAN KAZANIAN
hai HENNESSEY FM.
VALE FM.
KUNGURIAN
ARROYO FM. ee ani
LEUDERS FM.
LEONARDIAN
CLYDE FM.
BELLE PLAINS FM. WELLINGTON FM.
ADMIRAL FM. (WICHITA)
PUTNAM FM.
CUTLER - ABO
FMS. MORAN FM.
WOLFCAMPIAN
PUEBLO FM.
ir G) ac us = a uw o < ui Oo ac © x a Oo re a og Ld zl O o =) fe) a G) < E x 2 =
Fig. 5. A stratigraphic chart of the pertinent geological subdivisions and their temporal relationships.
ON THE SOURCE OF THERAPSIDS 39
the Caseid Chronofauna (Olson 1971). It existed in somewhat more upland circumstances (see Olson & Vaughn 1970) and is characterized by varanopsids, caseids, or caseid-like, primitive edaphosaurians, and the absence of Dimetrodon. Its basic characteristics relate to a food chain in which terrestrial reptiles, caseids and captorhinids, occupy the role of secondary producers. Predators include varanopsids, and haptodontine-sphenacodontine pelycosaurs, but not Dimetrodon.
Its earliest manifestation in the Permian record is in the Cutler and Abo Formations of northern New Mexico and south-western Colorado, Figure 4. Both the known faunas and the sedimentation support the conclusion that this is not a truly deltaic fauna, but one that lived somewhat more in uplands, and in the case of the Colorado fauna, near to highlands (Vaughn 19696). ‘This earliest phase has not produced caseids, although possible relatives, such as Oedaleops (Langston 1965) are known. Sphenacodon is present, Aerosaurus, very close to Varanops, is present in New Mexico, and in Colorado a haptodontine, Cutleria, has been found (Lewis & Vaughn 1965).
The youngest representation of this chronofauna is in the Chickasha Formation of Blaine County, Oklahoma. It is from this assemblage that the new gorgonopsian, Watongia, has come. In the same beds occur the caseids, Cotylorhynchus and Angelosaurus, the varanopsid, Varanodon, a captorhinid, Rothian- iscus, and Fayella, a highly terrestrial dissorophid amphibian (Olson 1965, 1972). Both the Cutler-Abo and the Chickasha assemblages include other genera. The number of truly aquatic forms is low in both instances, although some are present.
It has long been apparent that the therapsids arose from among the pelycosaurs and largely if not entirely from the Sphenacodontidae. It is also evident that they did not arise from the principal sphenacodontid of the Permo- Carboniferous Chronofauna, Dimetrodon, or any of its less well-known associates. Dimetrodon is an excellent morphological forerunner in many respects, but it is too highly specialized, especially in the vertebral spines, to be close to the actual ancestry. The same applies to Sphenacodon of the New Mexico assemblage.
Earlier (Olson 1962) I suggested that it might be necessary to seek the ancestry of the therapsids among the Lower Permian haptodontines. At the time these were known only from Europe. This interpretation was based primarily upon the basically primitive nature of this group, which is an assem- blage of unspecialized sphenacodontids, grouped for convenience by Romer & Price (1940). Sigogneau & Chudinovy (1972) show this group as basal to various primitive therapsid lines, with Old puerta therapsid evolution being the source of the later therapsid stocks.
Increasing evidence of close association of the Old World and New World Permian, both on physical and biological grounds, brings into serious question the idea of an exclusively Old World origin of therapsids. Likewise, similar chronofaunal structures in different regions suggest the possibility that similar evolution may have produced therapsid lines in more than one area and from
40 ANNALS OF THE SOUTH AFRICAN MUSEUM
more than one haptodontine-sphenacodontine line. Lewis & Vaughn (1965) describing a haptodontine, Cutleria, from south-western Colorado, noted its significance in this respect. Also haptodontines are now being identified in the rocks of the Upper Pennsylvanian of North America.
Evolution leading to therapsids presumably passed through the hapto- dontine to the sphenacodontine level and thence to the therapsid level. Dimetro- don and Sphenacodon, as well as other less well-known sphenacodontines, are examples of the intermediate level, but are somewhat specialized offshoots. The presumed sphenacodontines from which therapsids arose are not known. It appears likely that they evolved under the general conditions shown by the Caseid Chronofauna. Only where faunas have been sampled in the areas in which they lived, in the Cutler-Abo Formations and the Chickasha Formation, have representatives of a haptodontine-generalized sphenacodontine lineage been found. What we see for the most part are offshoots living under different conditions to which they are specifically adapted, the adaptations being the specializations that exclude them from therapsid ancestry.
The limitations of sampling, both faunal and geographic, are severe. Within these it is reasonable to suppose that the Caseid Chronofauna, and probably others like it, represent a somewhat upland complex in which hapto- dontines evolved into sphenacodontines and thence to primitive therapsids. The gorgonopsid line represented by Watongia is the result of this process. No remains of one of the principal groups of carnivores, the haptodontine-sphena- codontine lineage, have been found in the ‘erratic’ remnants of the Caseid Chronofauna. This term ‘erratic’ is used to refer to occurrences of remains of genera and species that are foreign to the normal assemblages of animals and not associated with normal constituents of the in situ chronofauna. That these carnivores were a part of the chronofauna in spite of their absence in the erratics is reasonable under the concept that, within the complex, herbivores out- numbered the predators many times over. Where the chronofauna is well known, representatives of the lineage have been found. Still, of course, existence of such a line between the Wolfcampian and the Guadalupian is conjectural.
Beyond the fairly large assemblage from northern New Mexico and south- western Colorado, in the Cutler and Abo Formations, and the Chickasha Formation of Oklahoma, the evidence of the Caseid Chronofauna comes entirely from erratics. In the present instance such genera and species occur within the geographic range and sediments of the Permo-Carboniferous Chronofauna, but are not a part of it. Such interpretations, of course, require extensive information about the primary faunas and are reliable only after thorough and repeated samplings have been made.
The principal evidence for the Caseid Chronofauna may be summarized
as follows: 1. Cutler-Abo Formations (Wolfcampian) a. Northern New Mexico. Large faunal assemblage, including Aerosaurus (varanopsid), Sphenacodon, Oedaleops (possible caseid relative), various aquatic, semi-aquatic and terrestrial amphibians and reptiles.
ON THE SOURCE OF THERAPSIDS 4I
b. South-western Colorado. Somewhat similar, less extensive assemblage with a more ‘upland’ fauna. Cutleria present, varanopsids not reported. 2. Lower part of Vale Formation (Clear Fork, Leonardian, Texas) Small pocket, about 1,52 by 3,04 and 0,91 metres thick. With Varanops, Casea, and Cacops (unique dissorophid). Association with other types of vertebrates reported, intimacy uncertain. Some Permo-Carboniferous Chronofaunal elements present. None of 3 genera listed above occurs in ‘normal’ assemblages of the Arroyo or Vale Formations of Texas or Oklahoma. 3. Upper part of Vale Formation (Clear Fork, Leonardian, Texas) a. Two specimens of Casea in single nodule, no other genera closely associated (Olson 1954, 1958). b. Palate of large captorhinomorph with 8 rows of teeth (Olson 1956). 4. Middle part of Choza Formation (Clear Fork, Leonardian, Texas) Partial skeleton of Casea, probably complete at deposition (Olson 1954). Not asso- ciated with other vertebrates. Not in usual Choza assemblage. 5. Hennessey Formation (Clear Fork, Leonardian, Oklahoma) Many skeletons of Cotylorhynchus (caseid). Apparently floated in (Stovall, Price & Romer 1966; Olson 1968). One tooth of varanopsid character, associated with washed-in scrap of Captorhinikos (Olson 1970). 6. San Angelo Formation (Guadalupian, North Central Texas) A mixed assemblage with elements from Caseid Chronofauna, and also Permo- Carboniferous Chronofauna (association uncertain). In addition, ‘advanced ele- ments’ of unknown origin. 7. Chickasha Formation (Guadalupian, North Central Oklahoma) An assemblage with Varanodon (varonopsid), Watongia, Rothianiscus and caseids, Cotylorhynchus and Angelosaurus. Some ‘erratic’ elements, as Diplocaulus, a member of the Permo-Carboniferous Chronofauna.
SIGNIFICANCE OF THE NORTH AMERICAN THERAPSIDS
I. THE CHICKASHA GORGONOPSIAN
The new genus, Watongia, sheds some light upon the problem of therapsid origins, but it also adds confusion to an already complex situation. It goes without saying, of course, that the data are insufficient for firm conclusions and that new information, which comes bit by bit, can merely point towards a better understanding. Currently new discoveries continue to suggest a complex rather than simple transition from pelycosaurs to therapsids.
It is assumed in the following discussion that Watongia arose from a hapto- dontine-sphenacodontine lineage in the context of the Caseid Chronofauna. Temporally it is slightly younger than the San Angelo therapsids (Olson 1962) and probably slightly older than the Ocher (Yezhova) therapsids from the early Kazanian of the U.S.S.R., Figure 5 (see Chudinov 1960, 1965; Efremov 1956; Olson 1962).
Watongia is such an excellent intermediate between the sphenacodontines and gorgonopsids that it could in fact represent the actual transitional stage. In this role it represents the only known instance of such an intermediate. The suggestion of Kemp (1969, 1972) that the gorgonopsians represent a distinct line, separate from therapsids, is supported, at least as far as the Gorgonopsidae are concerned. Parallelism, of course, cannot be ruled out, but with current evidence it seems an unnecessary complication.
42 ANNALS OF THE SOUTH AFRICAN MUSEUM
Watongia may be more or less at the developmental level of Phthinosuchus and Lotitanosuchus of the Russian early Kazanian. Because of the absence of comparable parts, except as the fragments of the skull of Watongia can be matched with the skulls of the other two, little direct comparison can be made.
The relationships of the Russian genera to each other and to Biarmosuchus have been the subject of much speculation and are not yet resolved. Most recently Sigogneau & Chudinov (1972) have related eotitanosuchids to gor- gonopsids and biarmosuchids to ictidorhinids. So interpreted, these two families of the Gorgonopsia (following Sigogneau 1970) diverged very early, probably at what can be considered a sphenacodontine level. Boonstra (1963) considered Eotitanosuchus and Phthinosuchus as relict representatives of an earlier stock, but not themselves on the line to other therapsids. Kemp (1969) showed the phthinosuchids as a side branch of the gorgonopsian line.
Biarmosuchids show many structural resemblances to ictidorhinids and the association seems reasonable. If Watongia is, as argued in this paper, a very primitive gorgonopsid, then it follows that Phthinosuchus and FEotitanosuchus grouped here as Phthinosuchidae, cannot be on the line to this family. Rather, the family originated directly from a sphenacodontine ancestry, without such intermediates.
Watongia most closely resembles the less specialized members of the sub- family Gorgonopsinae, forms such as Galesuchus from the Tapinocephalus Zone and Aelurognathus from the Endothiodon and Cistecephalus Zones. It is, how- ever, decidedly more primitive than any of these. It differs in many ways from members of the subfamily Rubidgeinae (Sigogneau 1970). The rubidgeines are characterized by the absence or near absence of the preparietal bone and by the exclusion or near exclusion of the frontal from the orbital margin. Although many of their features are gorgonopsian, their relationship to the gorgonopsines is not clear. Watongia has little direct bearing upon this problem, other than that it is a very primitive gorgonopsine, the subfamily within which the origin of the rubidgeines may lie, and that it has a preparietal and a frontal which makes broad contact with the orbit.
2. THE SAN ANGELO THERAPSIDS
In 1953 (Olson & Beerbower) and in 1962 (Olson) several therapsid-like reptiles were described from the San Angelo Formation of Texas. Intermittent searches in these beds, both in old sites and in new areas, have not added materially to what was known by 1962. The remains which have been recovered are tantalizing but in many ways unsatisfactory because of their incompleteness. It is, however, possible to say with confidence that in North America at the time of deposition of the San Angelo Formation there existed a wide array of non- pelycosaurian synapsids which show affinities to members of several groups of therapsid reptiles from the Ocher region and Copper Sandstones of the U.S.S.R. In 1962 these were associated as follows:
ON THE SOURCE OF THERAPSIDS 43
Phthinosuchids: Gorgodon, Knoxosaurus, Steppesaurus
Deinocephalians (Eodinocephalians) : Driverta, Mastersonia, T appenosaurus Brithopodidae: Eosyodon
Venjukovoidea: Dzmacrodon
These assignments still appear to be the most reasonable possible for these genera. Both the San Angelo and the Ocher assemblages indicate that extensive radiations had taken place by very early Guadalupian-Kazanian times. Neither in Europe nor in North America is there a trace of the direct source of these forms, except the probable ultimate pelycosaurian origin from very early Per- mian haptodontines. The only possibility, now that the association of Watongia and caseids is known, is that some of the genera assigned to the phthinosuchids are actually gorgonopsids, close to Watongia and with a source in the Caseid Chronofauna. The most likely candidate is Anoxosaurus, Figure 3H, which has two large canine teeth and small postcanines, somewhat a gorgonopsian pattern but one also not greatly different from that of phthinosuchids.
Except for this rather remote and indeterminable possibility, the San Angelo and Ocher forms have no known sources. We may speculate that there existed one or more unknown chronofaunal types from which they arose and that these existed outside of areas where deposition and preservation are currently known. The San Angelo complex appears to include a variety of ecological types and probably a considerable mixing of animals from different adaptive zones. Elements of the Caseid Chronofauna are present. ‘The only sphenacodont is Dimetrodon, from the Permo-Carboniferous Chronofauna. It is known from specimens from two widely separated sites.
Close resemblances between some of the San Angelo genera and those from the upper Permian of the Soviet Union indicate similar sources. The postcrania of Driveria and Mastersonia are of the same general type found in Estemennosuchus from Ocher. ‘They seem somewhat more primitive. The various parts of the skull, jaw and postcranium of Eosyodon show close resemblances to these elements in Syodon, from Isheevo, but are clearly more primitive. The supposed phthinosuchids seem to be at about the level of the Ocher Eotitanosu- chus, but resemblances are at best vague. Dimacrodon is clearly trending in the general direction of venjukovoids, but the differences are marked. In addition to the therapsids, Kahneria of the San Angelo is very similar to Hecatagomphius of the U.S.S.R. as far as the lower jaw and dentition are concerned. Likewise, Cotylorhynchus and Ennatosaurus, caseids, have many features in common, although there are differences in cheek teeth and in skeletal proportions.
INTERPRETATION AND SUMMARY
Therapsids from the San Angelo Formation of Texas and now from the Chickasha Formation of Oklahoma have added a new geographic dimension to the analysis of the very early, primitive therapsids. The strong resemblances between some of the North American and Russian genera, therapsids, caseids
44 ANNALS OF THE SOUTH AFRICAN MUSEUM
and captorhinids, indicate the existence of close geographic ties between the areas. The general circumstances may be interpreted somewhat as follows. We may suppose that the areas that have yielded fossil remains of terrestrial vertebrates lay in a continuous continental belt. This area as a whole probably was tropical to subtropical, lying more or less along the equator (see Olson & Vaughn 1970). Over the area existed a series of different adaptive zones, or habitat types. These differed from each other in general faunal and environ- mental characteristics. Similarities of environments within each of the several zones, as well as the differences between the zones, depended primarily upon topographic elevation and the amount and distribution of rainfall. Each of the adaptive zones was occupied by faunas which were fairly similar throughout but which had distinct local differences. In general, it appears, the topographi- cally lowest adaptive zone lay near the seas or large bodies of water in evaporite basins and deposition was very broadly deltaic in nature.
It is probable that there were intermittent interconnections between the various local faunal units of each adaptive zone and, on occasion, between local units in different adaptive zones. This probably was accomplished pri- marily by particular species that developed capabilities of extending beyond the ecological system in which they originated. Patterns of crossing between systems probably were complex and it is likely that interchanges were accom- panied by speciation as migrants adapted to the new circumstances which they encountered. The integrity of the local faunal units through time, as far as their basic ecological structure is concerned, appears to have been strong. Marked changes in an area were the result of replacement, not of local faunal evolution.
The vertebrates of the local ecological units form the chronofaunas. We may envisage many of these existing over the broad continental area, with suites of chronofaunas with similar characteristics occupying similar environ- ments in different places. The ecological structure, although much the same within a single adaptive zone, appear to have been different in the several adaptive zones. In addition, the evolutionary levels were different between at least some of the zones.
This interpretation produces a model within which it is possible to view therapsid origins in ways which seem to give excellent fit to the rather meagre data now available.
From this model it follows that the development of haptodontine- sphenacodontine lines toward the therapsid level may have taken place in many areas. Relatively few of the evolving phyla would have passed the therap- sid threshold and most of those that did probably would have been only tem- porily successful. From the successful few came the major groups of therapsids for which we have an excellent fossil record.
Each of these major groups, to the extent that it is monophyletic, may be supposed to have originated in some local area—that is, within a particular chronofauna. Origins of the different groups probably took place at various places over the continental area and these may have been more extensive
ON THE SOURCE OF THERAPSIDS 45
than present records show. Species transfer between chronofaunas both within and between adaptive zones, and replacements of one type of chronofauna by another, from a different adaptive zone, undoubtedly had a part in producing the complex situations which we find in the record. Taphonomic mixing certainly tends to make more obscure an otherwise complicated record.
From the information available it seems safe to say that no therapsids arose in the Permo-Carboniferous Chronofauna and probably not in any chronofauna similar to it. We appear to have one instance in which a therapsid arose in the Caseid Chronofauna, this, of course, involving Watongia. During the times that these two chronofaunas existed there were others, those in which the types of therapsids recorded in the San Angelo Formation and the Ocher site originated. We have no first-hand knowledge of where or under what cir- cumstances these origins were taking place. What indirect evidence there is, from the San Angelo and from at least some of the Russian sites, shows that materials were transported into lowland depositional sites. Presumably the chronofaunas tapped by these samplings were more upland than any for which we have first-hand knowledge of areas of habitation. Whether such areas will be found preserved, of course, is an open question; but on the basis of present information, the prognosis is negative.
REFERENCES
BoonstrA, L. D. 1934. A contribution to the morphology of the Gorgonopsia. Ann. S. Afr. Mus. 31: 137-174.
Boonstra, L. D. 1953. The gorgonopsians Aelurognathus microdon and Hipposaurus boonstrat, reconstructed. Ann. S. Afr. Mus. 42: 29-31.
Boonstra, L. D. 1963. Early dichotomies in the therapsids. S. Afr. 7. Sci. 5g: 176-195.
Boonstra, L. D. 1965. The girdles and limbs of the Gorgonopsia of the Tapinocephalus zone. Ann. S. Afr. Mus. 48: 237-249.
Broiii, F. & ScHRODER, J. 19354. Beobachtungen an Wirbeltieren der Karrooformation. XIII. Uber die Skelettreste eines Gorgonopsiers aus den unteren Beaufort-Schichten. Sber. bayer. Akad. Wiss. 1935: 279-330.
Broiii, F. & ScHRODER, J. 19355. Beobachtungen an Wirbeltieren der Karrooformation. XIV. Ein neuer Vertreter der Gorgonopsiden-Gattung Aelurognathus. Sber. bayer. Akad. Wiss. 1935: 331-355-
Broom, R. 1930. On the structure of the mammal-like reptiles of the suborder Gorgonopsia. Phil. Trans. R. Soc. (B) 218: 345-371.
Cuupinov, P. K. 1960. Upper Permian therapsids of the Ezhovo locality. Paleont. Zh. 4: 81-94. (in Russian.)
Cuupinov, P. 1965. New facts about the fauna of the Upper Permian of the U.S.S.R. 7. Geol. 73: 117-130.
CoLBerT, E. C. 1948. The mammal-like reptile Lycaenops. Bull. Am. Mus. nat. Hist. 89: 359-404.
EFreMov, I. A. 1956. American elements in the fauna of Permian reptiles of the U.S.S.R.
_ Dokl. (Proc.) Acad. Sci. U.S.S.R. 111: 1091-1094. (In Russian.)
Kemp, T. S. 1969. On the functional morphology of the gorgonopsid skull. Phil. Trans. R. Soc. (B) 256: 1-83.
Kemp, T. S. 1972. Whaitsiid Therocephalia and the origin of cynodonts. Phil. Trans. R. Soc. (B) 264: 1-54.
Lancston, N. 1965. Oedaleops campi (Reptilia: Pelycosauria), a new genus and species from the Lower Permian of New Mexico and the family Eothyrididae. Bull. Texas Meml Mus.
9: 5-47.
46 ANNALS OF THE SOUTH AFRICAN MUSEUM
Lewis, G. E. & Vaucun, P. P., 1965. Early Permian vertebrates from the Cutler Formation of the Placerville area, Colorado with a section on Footprints from the Cutler Formation by Donald Baird. Prof. Pap. U.S. geol. Surv. 503-C: 1-50.
Oxson, E. C. 1952. The evolution of a Permian vertebrate chronofauna. Evolution 6: 181-196.
Otson, E. C. 1954. Fauna of the Vale and Choza: 7. Pelycosauria, Family Caseidae. Fieldiana, Geol. 10: 193-204.
Otson, E. C. 1956. Fauna of the Vale and Choza: 13. Diadectes, Xenacanthus, and specimens of uncertain affinities. Fieldiana, Geol. 10: 329-334.
Otson, E. C. 1958. Fauna of the Vale and Choza: 14. Summary, review, and integration of the geology and the faunas. Fieldiana, Geol. 10: 397-448.
Otson, E. C. 1962. Late Permian terrestrial vertebrates, U.S.A. and U.S.S.R. Trans. Am. phil. Soc. (n.s.) 52 (2): 1-224.
Otson, E. C. 1965. Vertebrates from the Chickasha Formation, Permian of Oklahoma. Circ. Okla. geol. Surv. 70: 1-70.
Otson, E. C. 1968. The family Caseidae. Fieldiana, Geol. 17: 225-349.
Otson, E. C. 1970. New and little known genera and species from the Lower Permian of Okla- homa. Fieldiana, Geol. 18: 359-434.
Otson, E. C. 1971. Vertebrate paleo zoology. New York: Wiley-Interscience.
Otson, E. C. 1972. Fayella chickashaensis, the Dissorophoids, and the Permian terrestrial radia- tions. 7. Paleont. 46: 104-114.
Otson, E. C. & Beerbower, J. R. 1953. The San Angelo formation, Permian of Texas, and its vertebrates. 7. Geol. 61: 389-423.
Otson, E. C. & Vaucun, P. P. 1970. The changes of terrestrial vertebrates and climates during the Permian of North America. Forma et Functio: 3: 113-138.
PRAVOSLAVLEV, P. A. 1927a. Gorgonopsidae. In: North Dvina excavations of Professor V. P. Amalitsky. 3: 1-118. Leningrad: Academy of Sciences S.S.S.R. (In Russian.)
PRAVOSLAVLEV, P. A. 19275. Gorgonopsia from the North Dvina excavation 1923. (Amalitzkia Annae gen. et sp. nov.) In: North Dvina excavations of Professor V. P. Amalitshy. 4: 1-20. Lenin- grad: Academy of Sciences S.S.S.R. (In Russian.)
Romer, A. S. & Price, L. I. 1940. Review of the Pelycosauria. Spec. Pap. geol. Soc. Am. 28: i-x, 1-538.
SIGOGNEAU, D. 1970. Revision systematique des gorgonopsiens sud-africains. Cah. Paléont. C.N.R.S.: 1-xil, 1-416.
SIGOGNEAU, D. & Cuupinoy, P., 1972. Reflections on some Russian eotheriodonts (Reptilia, Synapsida, Therapsida). Palaeovertebrata, 5: 79-109.
STOVALL, J. W., Price, L. I. & Romer, A. S. 1966. The postcranial skeleton of the giant Permian pelycosaur Cotylorhynchus romeri. Bull. Mus. comp. Zool. Harv. 135: 1-30.
TaqueEt, M. P. 1969. Premiére découverte en Afrique d’un Reptile Captorhinomorphe (Cotylo- sauria). C.r. hebd. Séanc. Acad. Sci., Paris (D) 268: 777-781.
Vaucuan, P. P. 1966. Comparison of the early Permian vertebrate faunas of the Four-Corners region and north-central Texas. Contr. Sci. 105: 1-131.
Vaucuan, P. P. 1969a. Early Permian vertebrates from southern New Mexico and their paleo- zoogeographic significance. Contr. Sci. 166: 1-22.
Vaucuan, P. P. 19695. Lower Permian vertebrates of the Four-Corners and the midcontinent as indices of climatic differences. Proc. N. Am. Paleont. Conv. pt. D: 388-408.
A NEW GENUS OF GORGONOPSID FROM EAST AFRICA By F. R. PARRINGTON University Museum of Koology, Cambridge (With 2 figures)
CONTENTS PAGE Introduction . ; : : 5 : 47 Description : ; : ; ; : 47 References : : 3 : ‘ : 51 INTRODUCTION
In his account of the theriodonts from the Ruhuhu Valley in south-west Tanzania Huene (1950) described a somewhat poorly preserved skull of a gorgonopsid as a new species of the genus Lycaenops, L. kingoriensis. ‘The illustra- tions of this form immediately recalled a much better specimen collected by the writer in 1933 (no. 42) from Stockley’s (1932) site B.4 at Katumbi vawili. This site is in Stockley’s so-called ‘Lower bone bed’ which has produced both Endothiodon and Cistecephalus, and which Charig (1963) has recommended should be called the Kawinga Formation. The site B.4 has yielded, among other things, the amphibian Peltobatrachus pustulatus Panchen (Panchen 1953) and the gorgonopsid Leontocephalus intactus Kemp (Kemp 1969).
The identification of gorgonopsid skulls is made difficult by reason of the fact that many names have been given to very poorly preserved and incom- plete specimens but the fact that the specimen called Lycaenops kingoriensis was immediately recognizable among the writer’s collection seems to establish it as a valid form. However the shape of the skull, which may be said to be triangular, is in marked contrast to the material referred to the genus Lycaenops where, by contrast, the skulls are comparatively narrow (e.g. L. ornatus Broom; L. kingwilli, see Sigogneau 1970). Sigogneau (1970) defines the genus Lycaenops as having “Crane étroit, museau long, plus haut que large’. Neither this author nor Watson & Romer (1956) accepted Huene’s reference to the genus, the former suggesting a possible reference to the genus Sycosaurus, the latter to the family Gorgonopsidae but without any reference to a genus.
DESCRIPTION
The specimen described here is most unusual in that it does not appear to have undergone any distortion. The teeth are missing, as are the quadrate complexes, and small parts have been lost by breakage but, on the whole, the
47
Ann. S. Afr. Mus. 64, 1974: 47-52, 2 figs.
48 ANNALS OF THE SOUTH AFRICAN MUSEUM
skull is exceptionally well preserved. It was prepared by means of automatic mallets before the acid techniques had been perfected.
The skull measures about 300 mm in length and about 235 mm at its broadest point and is, therefore, a moderately large form. Seen from above the
Fig. 1. A dorsal and B occipital views of Cephalicustriodus kingoriensis (Huene), gen. nov. X §. F, frontal; IP, interparietal; J, jugal; L lacrimal; MX, maxilla; N, nasal; OP, opisthotic; P, parietal; PF, postfrontal; PMX, premaxilla; PO, postorbital; PP, preparietal; PRF, prefron- tal; PSP, parasphenoid; SMX, Septomaxilla; SO, supraoccital; SQ, squamosal; T, tabular.
A NEW GENUS OF GORGONOPSID FROM EAST AFRICA 49
Fig. 2. A palatal and B side views of Cephalicustriodus kingoriensis (Huene), gen. nov. X $.
BO, basioccipital; ECT, ectopterygoid; PAL, palatine; PT, pterygoid; V,
as in Figure I.
vomer; other letters
50 ANNALS OF THE SOUTH AFRICAN MUSEUM
skull is markedly triangular in outline, the snout being broad, the jugals are flared out at their junction with the maxillae, and the parietal region and temporal vacuities are wide. In general the arrangement of the bones is straight- forward with a moderate preparietal and the pineal surrounded by a well- developed boss. Apart from the proportions the outstanding feature lies in the nasals, coarsely sculptured bones which meet along the midline at an angle of about go° for most of their length. In gorgonopsids in general (as far as is known) the snout is either rounded or flattened.
In occipital view the appearance is straightforward but the skull is very wide for its depth. The relationships of the bones are normal but it is note- worthy that the spike on the bottom of the squamosal, behind the recess for the quadrate complex, is clearly developed. This spike almost certainly fitted into a pit in the quadratojugal and helped to prevent the quadrate complex from moving too much in its recess in the squamosal (Parrington 1955).
The palate is well shown (Fig. 2A). The dentition must have been of normal gorgonopsid type there being sockets for five powerful incisors, a large canine (followed by a bony plug), and four diminutive postcanines. There are at least three small teeth on the ridge of the palatine but only one can be seen for certain on the pterygoid, this on the right side.
The fused vomers are clear and of characteristic form being expanded anteriorly but tapering posteriorly where they disappear above the palatines at about the level of the posterior borders of the internal nares. The importance of this feature, which is in contrast to the condition in Therocephalia, Bauri- amorpha and Cynodontia, in all of which the vomers grow backwards below the palatines and pterygoids, was first recognized by Boonstra (1934) who questioned the then widely accepted view that the cynodonts were descendants of the gorgonopsids, and was the first person to do so. Subsequent work on both the early cynodonts (e.g. Broom 1938a) and gorgonopsids (Kemp 1969) have established the fact that the gorgonopsids and cynodonts are not closely related, the latter being descendants of the Therocephalia.
Posteriorly no interpterygoid vacuity can be detected though one might be exposed if the specimen were subjected to acid treatment. Very marked incisures are present between the wings of the parasphenoid-basisphenoid complex and the basioccipital. Taken in conjunction with the unossified zone (Olson 1944) and the frequency with which the occipital component of the brain case is found separate in synapsid reptiles, the possibility of kinetism in these forms appears to be real. Movement of the anterior region of the skull about the lateral expansions of the supraoccipitals is a possibility, but only a closely sectioned skull, when reconstructed, would establish the matter.
The form of the zygomatic arch strongly suggests that there must have been ample room for an external component of the adductor muscle mass (Fig. 2A, B). This is shown also in Arctognathus breviceps and the forms Dinogorgon quinquemolaris, where the lower jaw is present, and Rubidgea atrox, also with the lower jaw in place (Sigogneau 1970). The writer has suggested that a masseter muscle ran
A NEW GENUS OF GORGONOPSID FROM EAST AFRICA 51
from the anterior region of the jugal (which is deep and thickened in gor- gonopsids) and the reflected lamina of the angular, thereby accounting for that peculiar structure (Parrington 1955) but Barghusen (1968) claims that only in cynodont theriodonts was this muscle evolved and Kemp (1969) has postulated in the gorgonopsids the presence of a zygomatico-mandibularis muscle between the posterior region of the arch and the reflected lamina, this in view of the special strengthening of the bone of the lamina.
Examination of Dr Sigogneau’s invaluable monograph on the gorgonopsids (Sigogneau 1970), and other sources, has failed to disclose at all a close match in form with the skuil from Katumbi vawili though similar broadening of the posterior region of the skull is present in a number of forms, notably Sycosaurus, Prorubidgea and Rubidgea, and it may well be allied to these forms. Also the meeting of the nasals at about a right angle along much of their length does not appear to have been reported elsewhere, the snout of gorgonopsids as a whole being either flattened or rounded in section.
There is some resemblance to Gorgonops torvus but this latter form has a proportionately much longer antorbital region and in occipital view the East African animal is much flatter.
In the new form the width across the paroccipital processes is approximately half the posterior total width, whereas in Sycosaurus it is distinctly more than half, in Dinogorgon quinquemolaris it is about three-fifths, and in Rubidgea atrox it is slightly less than half. Again in posterior view the depth of the occiput, taken to the bottom of the condyle, is about one-third of the total width, a character it shares only with Rubidgea atrox, the depth in Dinogorgon quinquemolaris being a little less than half and other forms being much deeper in proportion. The shape in Rubidgea atrox is quite different from that of the Ruhuhu specimen (Broom 19380).
In accordance with the foregoing discussion it is proposed to erect a new genus Cephalicustriodus, defined principally on the triangular shape of the skull, for Huene’s Lycaenops kingoriensis which now becomes Cephalicustriodus kingoriensis (Huene). It can probably best be placed in Sigogneau’s (1970) subfamily Rubidgeinae. The specimen described here is catalogued number T.891 in the University Museum of Zoology, Cambridge.
REFERENCES
Boonstra, L. D. 1934. Additions to our knowledge of the South African Gorgonopsia preserved in the British Museum (Natural History). Ann. S. Afr. Mus. 31: 175-213.
Broom. R. 1932. The mammal-like reptiles of South Africa and the origin of mammals. London: Witherby.
Broom, R. 1938a. The origin of cynodonts. Ann. Transv. Mus. 19: 279-288.
Broom, R. 1938. On a new family of carnivorous therapsids from the Karroo beds of South Africa. Proc. zool. Soc. Lond. (B) 108: 527-533.
Brink, A. S. 1960. A new type of primitive cynodont. Palaeont. afr. 7: 119-154.
Cuaric, A. J. 1963. Stratigraphical nomenclature in the Songea Series of Tanganyika. Rec. geol. Surv. Tanganyika, 10: 47-53.
52 ANNALS OF THE SOUTH AFRICAN MUSEUM
HueEneg, F. von 1950. Die Theriodontier des ostafrikanischen Ruhuhu-Gebietes in der Tiibinger Sammlung. Neues Fb. Geol. Paléont. Abh. 92: 47-136.
Kemp, T. S. 1969. On the functional morphology of the gorgonopsid skull. Phil. Trans. R. Soc. (B) 256: 1-83.
Manrten, A. A. 1959. Two new gorgonopsian skulls. Palaeont. afr. 6: 51-76.
Otson, E. C. 1944. The origin of mammals based upon the cranial morphology of the therapsid suborders. Spec. Pap. geol. Soc. Ann. 55: i-xi, 1-136.
ParRINGTON, F. R. 1955. On the cranial anatomy of some gorgonopsids and the synapsid middle ear. Proc. zool. Soc. Lond. 125: 1-40.
SIGOGNEAU, D. 1970. Révision systématique des gorgonopsiens sud-africains. Cah. Paléont. C.N.R.S.: i-xii, 1-416.
StockLey, G. M. 1932. The geology of the Ruhuhu coalfields, Tanganyika Territory. Q. 7l geol. Soc. Lond. 88: 610-622.
THE INNER EAR OF GORGONOPS (REPTILIA, THERAPSIDA, GORGONOPSIA)
By D. SIGOGNEAU Institut de Paléontologie, Muséum National d’ Histoire Naturelle, Paris (With 6 figures)
I welcome the opportunity to express my admiration and thankfulness to Dr Boonstra, to whom the mammal-like reptiles and myself are greatly indebted.
It is to Watson that we owe the first tentatives toward elucidating the structure of the inner ear in mammal-like reptiles (cynodonts), but it was Olson (1944) who undertook a systematic study of this region of the skull in the main therapsid infra-orders. Olson’s work included two gorgonopsians: the type of Cyonosaurus longiceps, and a form not generically identified, but which he con- sidered as ‘more primitive than Cynariops robustus,’ the genus to which he had originally (1938a) assigned it. Thus it is only a modest supplement to his com- parative study that is presented here. The genus Gorgonops would be situated, according to the conclusion I reached in 1970, at a higher evolutionary level than the two forms studied by Olson.
The method employed here to study the specimen of Gorgonops (B.P.I. 277) was that of serial sectioning, such as was used by Olson (1944) and previously by Romer & Price (1940) for the study of the inner ear of Dimetrodon. Details of the process I used (in reality not sectioning, but a grinding of the specimen) will be given later in a publication on the ensemble of the skull; it should be mentioned here that the latter measured about 237 mm long, and that grinding proceeded at + mm intervals. ,
The preservation of the specimen turned out to be less satisfactory than was initially expected, and flattening of the posterior part of the skull is rather marked. The ear region is in reasonably good condition, but the defective ossification of this region renders its interpretation delicate and numerous uncertainties remain.
The inner ear is lodged, in Gorgonops, in the prootic, the opisthotic, the supraoccipital and the basioccipital. It is situated between sections 230 A— 300 A.!
An immediately noteworthy fact concerns the inclination, dorsally and anteriorly, of the ear, this being in correlation with the same slope of the basi- cranial axis (Fig. 1). The mesial wall of the ear is only ossified at the back; there does not exist, therefore, on the endocranial face a proper internal auditive foramen, but a vast opening which leads into the cavity of the bony ear in which was contained the membranous ear. Obviously, the eighth cranial nerve
1 The section numbers cited in this work refer to the left side, but in fact there is very little offsetting from one side to the other.
Do
Ann. S. Afr. Mus. 64, 1974: 53-69, 6 figs.
54 ANNALS OF THE SOUTH AFRICAN MUSEUM
OOOO an | OT || sn MRE nil l iH canal pineal cy. ren AT { 1
i y; Un yh j mn SEHR ARAL ti!
Sha A
| * Moe MRRARAY | ti | i
fosse flocculaire
| m
oreille interne
paroi mediale de
or Wi Vy a) \ ef ie ay iy I ear me yi hi Wf hile ih y Mi hi &\ ul AL
\\\ Rn mr SOAK ihe io
aegis fi Berne x50 Arn rT RRS CN Auraria ener HA } OC ; WS Sees Nth y PAP RPA Sape tty oot Es WNW Y AL itt! UH PULL Raypegagtl || F It Wy, uit > TO Z ddligo—- ‘s TSP 0 H = List tb badd dat ~ | Se y Wht nA ' Se / CAO ouverture du cang} Saccule Xl Pk iH At canal carotidien horizontal ™~ / ‘N 41cm a S/ N
Fig. 1. Longitudinal section of the posterior part of the skull of Gorgonops torvus (B.P.1. 277), reconstructed from serial sections (from Sigogneau 1970). X 1. f h: hypophyseal fossa.
also traversed this opening. The latter, in conjunction with the cavity of the bony ear, correspond to the vestibule of some authors, but it would be prefer- able, as noted by Romer (i955), to avoid this term originally applied to the upper part of the internal bony ear. The absence of ossification is such that the interior of the horizontal semicircular canal, contained in the lateral wall of the ear, is visible mesially; the canal, nearly totally lacking an internal wall, seems to open into the utriculo-saccular cavity. ‘The same absence of ossification is responsible for the opening of the vertical anterior semicircular canal into the bony floccular fossa.
Indeed, intruding between the elements of the inner ear was an enormous flocculus (o),? directed postero-laterally after having passed anteriorly under the anterior semicircular canal. This flocculus must have been remarkably developed, and even partially covered the crus communis canalium. Opening anteriorly into the cerebral cavity, the floccular fossa penetrates first the prootic, which surrounds it on three sides; at the same time its floor, oblique ventrally and toward the median line, progressively rises until it becomes horizontal. But soon the supraoccipital invades both the roof and the floor, and when the fossa closes mesially (section 278 A), it is nearly completely included in the supraoccipital ; it narrows progressively within this bone and terminates 1,5 mm behind its mesial closure.
* This and following numbers refer to features in the figures.
THE INNER EAR OF GORGONOPS 55
Fig. 2. Lateral view of the internal ear of Gorgonops torvus (B.P.I. 277), reconstructed from serial sections. X 2.
cl,: emplacement of the partition between the floccular fossa and the ear proper; f ov: fenestra ovalis; ju: jugular foramen; la: lagenar recess; z: perilymphatic cistern?
o: floccular fossa; 3: anterior ampulla; 3 dis: external ampulla; 4 and 10: anterior vertical canal; 5: horizontal canal; 6: utriculus; 7: ventral part of the sacculus; 8: 8 cc: crus communis? 11: posterior ampulla; 13: dorsal part of the sacculus.
ST: antero- posterior extension of the stapes.
Pprootique__—____
_— susoccipital
_—opisthotique
___basioccipital
Oreille gauche . Vue latérale
Fig. 2 bis. The same as Figure 2, showing the emplacement of the sutures.
56 ANNALS OF THE SOUTH AFRICAN MUSEUM
In the ear itself the three semicircular canals, the utriculus, the sacculus and the lagena were sought. The perilymphatic system certainly left traces, but they are difficult to identify.
The vertical anterior canal (4) is distinctly longer than the posterior canal; also, because of the inclination of the ear, the dorso-posterior slope of the canal is slight; it bends mesially to meet the vertical posterior canal. This anterior canal originates in a very short ampulla (probably the external ampulla) (3 bis), the latter being preceded by a small but deeper cavity inclined ventro- laterally (anterior ampulla) (3). Soon the ventral edges of the anterior canal approach each other (section 258 A), and it becomes enclosed within the thick- ness of the prootic, lateral to the floccular fossa. It rises progressively inside this bone, passes above the prootic partition which separates the floccular fossa from the internal ear, and attains thereby the roof of this fossa. When the supraoccipital invades this roof (c. 270 A), a curious phenomenon is observed: the orifice becomes double, with the lateral part (enclosed in the prootic) closing rapidly, and the mesial part (10) (situated first at the prootic-supra- occipital limit (Fig. 4, section 268 A), then in the supraoccipital alone) increas- ing in size and relaying the first canal; it then encircles the floccular fossa and mesially joins the posterior part of the ear. I originally saw, here, the replacement of the anterior canal by the posterior canal, especially since the situation is identical on both sides of the skull. But the utriculus is not visible at this level: we are above the floccular fossa in which the two canals open ventrally. An interpretation of this mesio-lateral doubling remains to be found; I have seen no example of it in the literature. For the present, I will consider that it is all the same anterior canal which continues until opening into the utriculus at section 285 A; this canal terminates 2 mm farther.
The posterior canal was much more difficult to identify. It is regarded here as being formed by the mesial half of the posterior otic cavity (Fig. 4, section 295 A), and is, in consequence, masked in lateral view by the other constituents of the latter; there would be no bony separation between the two. But Olson shows the posterior canal in the sections of ‘Cynariops robustus’ (1938a, fig. 7, sections A and D) as being more lateral than the vertical anterior canal. I have found nothing equivalent; in any case, this canal must constitute the most posterior part of the ear; as a final argument, I will add that the structure here identified as the posterior canal terminates, as it should, in what is con- sidered to be the posterior ampulla (11). This canal would be very short.
The horizontal canal (5), on the other hand, is easy to follow: it begins in the prootic in the same ampulla as the vertical anterior canal, and closely behind it (section 247 A). At first they are separated only by a fine crest, then after the definitive individualization of the latter, the horizontal canal pene- trates laterally, slightly sloping downwards; its floor becomes formed by the opisthotic (section 272 A) and, one millimetre farther, the roof becomes formed by the supraoccipital. It continues, totally enclosed in the latter bone, as far as section 285 A, still sloping downward but bending mesially, and terminates in
| | hi
THE INNER EAR OF GORGONOPS
Fig. 3. Cavity of the internal ear of Gorgonops torvus (B.P.I. 277) seen in mesial view; reconstructed from serial sections. X 2.
b m OP: mesial border of the opisthotic closing the ear mesially; bo: mesial ossified border of the lagenar recess; cl,: partition between the floccular fossa and the ear proper; cl,: prootic then opistho- tic partition closing cavity 9 mesially, and consti- tuting the lateral wall of the lagenar recess; c v a: opening of the vertical anterior canal in the floccu- lar fossa; g g: fossa situated under the cranial floor and corresponding perhaps to the geniculate ganglion; la: lagenar recess; r: level where the basioccipital replaces the basisphenoid as the dorso- mesial border of the lagenar recess; s PRO-BS: separation of the basisphenoid and the prootic allowing the passage of the lagena above the cranial floor; sl: slit prolonging the fenestra ovalis.
o: floccular fossa; 3: anterior ampulla; 3 bis: external ampulla; 5: horizontal canal; 6: utriculus; 8: see text p. 58; 9: ventral part of the sacculus? 13: dorsal part of the sacculus.
prootique______ a __ susoccipital
basisphenoide___= / __opisthotique
basioccipital___
Oreille droite. Vue mésiale
Fig. 3 bis. The same as Figure 3, showing the emplacement of the suiures.
58 ANNALS OF THE SOUTH AFRICAN MUSEUM
the posterior ampulla just behind the opening of the sacculus. Taken in its entirety, the horizontal canal is, therefore, inclined postero-ventrally and scarcely rises again at the rear; it is rather convex laterally. Its mesial wall is more or less ossified according to the location, but never completely.
The ampullae are not as clearly recognizable as one would expect; this is particularly true for the anterior and external ampullae. The ear begins with a minute recess (I mm by 2,5 mm) directed laterally and ventrally (Fig. 4a, section 240 A, 3); this should represent the anterior ampulla since it precedes the depression (1 mm by 1 mm) in which the anterior and horizontal canals originate and which should therefore be the external ampulla. The recess here interpreted as the posterior ampulla is, on the contrary, very well developed. In fact, its great antero-posterior extension created some doubt about this interpre- tation, and I at first thought it to be an element of the perilymphatic system. But its mesial situation, and especially its posterior connection with the hori- zontal canal, seemed to constitute convincing arguments in favour of the ampullar hypothesis; the vertical posterior canal then, would communicate with this ampulla dorsally and mesially with respect to the horizontal canal.
The crus communis canalium (8 cc), very mesially situated, starts also, in my interpretation, very posteriorly (section 282 A); it should open into the utriculus.
The central part of the ear presents a series of concavities which overlap each other latero-mesially. One of them (6) begins at the front of the ear and communicates there with the anterior and horizontal canals; it continues all the way to the posterior part of the ear where it receives, on its mesial side, the posterior canal; it should therefore represent at least the dorsal part of the utri- culus. Ventrally this utriculus communicates first (Fig. 4a) with the lagena(la), then (Fig. 4b) with that which is interpreted here as the sacculus (9), and finally with the posterior ampulla, from which it is isolated farther on by the posterior canal. The utriculus terminates very posteriorly (at about section 295 A). It is doubled mesially by a cavity (8), of which we also have only the roof (the latter being at first constituted by the prootic, then by the supraoccipital) and from which it is separated by a crest. This cavity, the most mesial element of the ear, begins about section 260 A and also continues all the way to the rear. I could not identify it with certainty; it probably could not be the endolymphatic sac, which would thus be intraotic.
Between the utriculus and the horizontal canal is a cavity still better delimited dorsally (13) as well as ventrally (7) and more prominent than the preceding ones, but of smaller dimensions (approx. sections 260 A-290 A), and that I have regarded as corresponding to the sacculus. Its floor is in fact divided (7 and 9) for a great part of its length, but it is difficult to know if the two cavities are parts of the same element. The horizontal canal opens directly into this cavity, which, as noted above, communicates dorsally and mesially with the utriculus. Antero-ventrally the pocket opens into the lagena (Fig. 4a,
- ea se
THE INNER EAR OF GORGONOPS 59
section 260 A); when the latter closes, a crest rises from the opisthotic isolating the sacculus from the posterior ampulla (Fig. 4b, sections 268 A and 278 A). This so-called sacculus goes farther dorsally than ventrally. ‘The interpretation given here for the sacculus is in opposition to that of Olson (1938a), where the saccular recess, indistinctly separable from the lagena, does not extend above the level of the horizontal canal, and to that of Romer & Price (1940), where the sacculus in Dimetrodon is situated below the level of the cranial floor. In this fea- ture in Gorgonops, I had thought to recognize an element of the perilymphatic system, but it would be surprising that such an element would be so dorsal, that it would intergrade with the parts of the ear properly speaking,’ and that it would be so distant from the fenestra ovalis.
Finally, the lagenar recess (la) constitutes an elongate tube sloping slightly antero-ventrally. Its anterior limit is difficult to determine (perhaps section 232 A), being situated in a very poorly ossified zone. The anterior part of the recess is lodged between the basisphenoid dorsally, the prootic dorso-laterally, and the basioccipital ventrally; then farther posteriorly (section 244 A), the opisthotic starts to form its lateral wall. Anteriorly, the lagenar recess is entirely below the level of the cranial floor. It opens laterally in a deep depression (Fig. 4a, z), the edge of which receives the stapes, and mesially in a small rounded recess interpreted as the receptacle for the geniculate ganglion (gg). When the prootic and basisphenoid separate, the dorsal part of the lagena passes above the level of the cranial floor, and soon (section 250 A) opens directly into the cerebral cavity. Only after the anterior third of the ear (section 260 A), and by the junction of the lateral and mesial parts of the opisthotic, is the lagenar recess completely closed laterally. It then narrows and terminates by a small circular fossa excavated in the opisthotic alone, in the floor of that which I have interpreted above as the posterior ampulla (Fig. 46, section 268 A). In its middle part this recess communicates dorsally with the utriculus and mesially with the sacculus.
Concerning the perilymphatic system, three elements have been sought: the cistern, or reservoir of liquid at the level of the upper, or vestibular, part of the ear; the duct which prolongs the cistern and parallels the lagena par- ticularly at the level of the basilar papilla (scala tympani); and finally the perilymphatic sac, the termination, in part at least extraotic, of the duct, lodged in what is improperly called the recessus scalae tympani.
On this subject a parenthetical explanation is perhaps not superfluous. In tetrapods, and with the appearance of the fenestra ovalis, the perilymphatic space which isolated, in fishes, the membranous ear from the bony ear becomes organized into a specialized system for the conduction of vibrations from the outside environment to the properly sensorial zone of the membranous ear, the basilar papilla of the lagena. This system is composed of, first of all, a
* However Romer & Price write (1940) with respect to Dimetrodon: ‘Cavities which are apparently associated with the perilymph system push outwards within the loops of the two vertical canals... .’
60 ANNALS OF THE SOUTH AFRICAN MUSEUM
Fig. 4a Fig. 4a. See legend opposite page.
roughly circular cavity, the perilymphatic cistern, applied to the lateral opening that is the fenestra ovalis,* and therefore against the stapes. From this cistern
“ Or foramen vestibuli, called thus because it is situated at the level of the vestibule of the ear. This is not true, however, in certain living reptiles, nor in the mammal-like reptiles, where it opens at the level of the lagena.
THE INNER EAR OF GORGONOPS 61
— ro (=
285 A 295 A Fig. 4b
Fig. 4. Some of the serial sections of the ear region of Gorgonops torvus (B.P.I. 277). xX 2. a) anterior part of the ear; b) posterior part.
Cc Vv p: posterior vertical canal; f ov: fenestra ovalis; g g: fossa corresponding per- haps to the geniculate ganglion; la: lagenar recess; r sc t: recessus scalae tympani? t ju: jugular canal; y: foramen corresponding perhaps to the endolymphatic duct; z: perilymphatic cistern?
o: floccular fossa; 2: fossa from which the ampullae will develop; 4 and to: anterior vertical canal; 5: horizontal canal; 6: utriculus; 7 and 9: ventral part of the sacculus; 11: posterior ampulla; 13: dorsal part of the sacculus.
BO: basioccipital; BS: basisphenoid; PRO: prootic; OP: opisthotic; SO: supra- occipital; ST: emplacement of the stapes.
The sutures have been omitted in the interest of clarity, although they are dis- cernible in the sections.
The dotted line corresponds to the median axis of the cerebral cavity.
62 ANNALS OF THE SOUTH AFRICAN MUSEUM
a tube leaves, the perilymphatic duct; in lower tetrapods the latter is applied against the basilar papilla for part of its length by the intermediary of the basilar membrane. Beyond this papilla it attains the lower posterior bony wall of the otic capsule, a wall that it traverses by the perilymphatic foramen (called in error by some authors the fenestra cochleae). From there it spreads out into another evagination, the perilymphatic sac. The latter, apart from its dorsal opening into the duct, opens mesially into the cerebral cavity (by an internal jugular foramen, so-called because it is utilized at the same time by the jugular or posterior cerebral vein) and extends into the subarachnoid lacunae. Laterally, it comes in contact with the tympanic cavity, where an opening
Canal. — | M. Lim. — Pet rs Soc: .—— M. Basil Stapes
cit. Ply! a la-~~ Soc. Ply.-
Fig. 5. Interrelationship of the peri- and endolymphatic systems in Sphenodon (from De Burlet).
pierces the wall of the otic capsule, the fenestra rotunda sensu lato, or pseudo- fenestra rotunda. Thus, not only are the vibrations transmitted by the stapes received by the perilymphatic liquid, which fills all of the system described, and conducted to the level of the basilar papilla where the sensorial cells receive the excitation, but the lower part of the system has, in addition, a compensating role necessitated by the incompressibility of liquids; thus a mesiad depressing of the tympanum results in a laterad depressing of the secondary tympanic membrane, which obstructs this pseudo-fenestra rotunda.
On this scheme a certain number of variations are realized. It is necessary, if not logical, to begin with mammals, because it is with them that the terms were first defined. In this class, then, the lagena has acquired such a length that it coiled up in a cochlea. The perilymphatic sac has become intraotic, and it is this structure (not the perilymphatic duct) that is applied against the basilar membrane. This perilymphatic sac has followed the coiling of the cochlea (or scala media) along each side, forming thereby a double tube: one part, the vestibular ramp, or scala vestibuli, is directly connected to the fenestra
THE INNER EAR OF GORGONOPS 63
ovalis; the other part, terminating at the secondary tympanic membrane (closing the fenestra rotunda sensu stricto), is the tympanic ramp or scala tympani. It follows, therefore, that, according to this definition, it is the scala tympani which is applied against the basilar papilla (in mammals called the organ of Corti).°
Sac. prcel
3: Membrane = Limitans.
D. Perilymph. Scala Tymp.——se—s
Fig. 6. Interrelationships of the peri- and endolymphatic systems in mammals (from De Burlet).
By homology, the part of the perilymphatic system which in reptiles forms the same support for the basilar papilla is also called the scala tympani,® even though it concerns the perilymphatic duct, and not, as in mammals, the peri- lymphatic sac. This difference leads to a dislocation of the terms: the bony cavity which in mammals encloses the scala tympani (that is, the whole peri- lymphatic sac) constitutes the recessus scalae tympani. By homology, but a somewhat perverse homology, the (incomplete) bony cavity which houses the
° It should be noted that, despite this development, the internal ear is entirely included in the periotic; this is also the case in living reptiles, but in mammal-like reptiles, as was seen above, not only the periotic but also the basioccipital participates in the bony internal ear. In cynodonts, however, this bone is reduced to a minimal participation.
6 This term is all the more improper in that the lagena is here short and straight, and that the duct is not parallel, but perpendicular to it; there are not, therefore, two ramps. In crocodiles the lagena is elongate and the ramps are discernible, but they remain straight.
64 ANNALS OF THE SOUTH AFRICAN MUSEUM
perilymphatic sac in reptiles has received the name of recessus scalae tympani, even though, let it be repeated, the scala tympani is here constituted by the duct and not by the perilymphatic sac, and is not, in consequence, contained in the ‘recessus scalae tympani’. Moreover, this recessus is in reptiles an extraotic and extracerebral space (enclosed either between the periotic and the supra- occipital, or within the exoccipital). Its embryological origin goes back to the metotic fissure, that is, the space left between the embryonic otic capsule and the base of the skull. This fissure, in the embryo, serves not only as an opening for the perilymphatic sac, but it forms also a passage for the cranial nerves IX, X, XI, and the jugular vein.’ In mammals (as well as in crocodiles and in birds, although by a different process) the recessus has become entirely closed and intraotic by the secondary formation of a bony floor. In the course of this process the openings have been preserved, but at the same time submitted to modifications which result in, for example, the lateral opening (or fenestra rotunda) not being entirely homologous to the opening of the same name in reptiles.
Many other homologies have been proposed. I have lightly touched upon the problem here, only wanting to explain the terms utilized.
To come back to Gorgonops, this cistern is probably indiscernible at the level of the fenestra ovalis, unless, ventrally and laterally to the lagenar recess, a non-ossified region between the basioccipital and the prootic (sections 230 A— 246 A) represents a trace of this cistern (Fig. 4a, z). The perilymphatic duct also remains indistinct, although a part of the cavity interpreted as the posterior ampulla may correspond to it. On the contrary, it has been thought possible to detect a trace of the perilymphatic sac in the form of a ‘recessus scalae tympanz1’: there exists (Fig. 4b, section 278 A), on the mesial side of the opisthotic partition closing the ear posteriorly, a depression which opens farther to the rear (section 284 A) into the jugular passage; the latter must therefore represent the whole of the metotic fissure (the jugular foramen plus the pseudo-fenestra rotunda).
Finally, concerning the endolymphatic sac, I have evoked the possibility that the depression (discernible in the prootic, then supraoccipital, partition) which separates the floccular fossa from the ear proper (8) could represent the trace of this sac. Another possibility is that the orifice which persists pos- teriorly (Fig. 4b, section 285 A, y), in the mesial wall of the bony ear, corre- sponds to the passage of the endolymphatic duct towards this sac, which would be, in that case, more mesial.
The fenestra ovalis is situated very anteriorly; moreover, it does not open at the base of the lagenar recess, but ventro-laterally with respect to it (Fig.-2). One can, therefore, consider that it has started to rise towards the vestibular part of the ear, where it lies in living reptiles and mammals, and where it is properly a ‘fenestra vestibuli’. Furthermore, whereas in these two cases it pierces the otic capsule, it is here included between the otic capsule and the
“In the adult, different cases are possible: the recessus scalae tympani can fuse with the jugular canal; it can also contain only the perilymphatic sac and the nerve IX.
THE INNER EAR OF GORGONOPS 65
base of the skull. Olson had already noted this last point in 1938, but the fenestra ovalis and the metotic fissure are, in Gorgonops, independent of each other. The anterior edge of the fenestra ovalis is no more distinct than that of the lagenar recess. Anteriorly it is bounded dorsally by the prootic, ventrally by the basioccipital. It almost closes (section 248 A) by the intervention of the mesial part of the opisthotic, but it is prolonged by a slit (2,5 mm long) as far as the complete junction of this mesial part with the lateral part of the bone, which results in the disappearance of the prootic from its roof (section 261 A). Behind its closure persists, in the opisthotic, a conical depression one millimetre long. The stapes covers all of the fenestra ovalis, and half of the slit. One might suppose that the two millimetres remaining behind the stapes (half of the slit plus the posterior fossa) were occupied by a cartilaginous expansion.
That which has been very tentatively interpreted here as the mesial orifice of the nerve VII is situated slightly in front of the level of the horizontal canal; the nerve would then cross the bone, somewhat obliquely towards the front and laterally, and would exit just in front of the ear (Fig. 1). The position of the geniculate ganglion would be visible ventrally and behind nerve VII, between the prootic and the basisphenoid (Fig. 4a, section 240 A, gg).
The nerves IX, X and XI must have exited by the jugular canal, which passes first between the opisthotic and the exoccipital then, farther toward the rear, inside the exoccipital (Fig. 4b, section 295 A); this canal is completely separated from the posterior part of the ear by a bony wall.
COMPARISON WITH OTHER GORGONOPSIANS
As has been noted above, the internal ear has already been studied in Cyonosaurus longiceps (Cistecephalus zone®) by Olson in 1937, and the same author in the following year sectioned a skull of ‘Cynariops’ (Endothiodon zone’) and reconstructed the membranous labyrinth. In 1944 he made several modifications in his reconstructions. The comparison of these with that proposed here for Gorgonops torvus (Endothiodon zone’) may indicate a certain evolution of this region within the gorgonopsians.?®
In all three cases a feeble degree of mesial ossification can be observed. The separation of the floccular fossa from the internal ear proper intervenes in the three cases also, but the partition is here parallel to the horizontal canal, whereas it forms with the latter an angle of about 65° in Cyonosaurus and 45° in ‘Cynariops’. One would witness here, if the proposed morphologic sequence is accepted, a dorsal migration of the floccular fossa, which, from an anterior position, would become dorsal.
The floccular fossa seems to have approximately the same depth in the three forms. Also it is partially enclosed in the supraoccipital.
8 Sensu Broom. ® In a previous study (1970), I considered that Cyonosaurus was the most primitive gorgonopsid,
that Gorgonops would be a rather evolved form, and that ‘Cynariops’ would be situated morpholo- gically between the two.
66 ANNALS OF THE SOUTH AFRICAN MUSEUM
The anterior semicircular canal is always longer than the posterior, but the difference in length is greater in Gorgonops and in ‘Cynariops’ than in Cyonosaurus. As for the individualization of the posterior canal from the crus communis, it is the least marked in Gorgonops. ‘The horizontal canal would be more convex here than in the genera studied by Olson, a fact which, according to him, would constitute a primitive character.
The anterior and external ampullae seem to have been well individualized, in Cyonosaurus at least; the external ampulla is not figured in ‘Cynariops’, and the difficulty met with in distinguishing both of them in Gorgonops has already been commented on; they seem to have been, in any case, smaller here.
The comparison becomes more difficult for the utriculo-saccular region, due to the fact that my interpretation differs from that of Olson. The latter considered that the sacculus was entirely situated in the lower part of the ear, indistinct from the lagena, with the ensemble forming what he calls the sacculo- cochlear recess. I have, on the contrary, suggested that the cavity, above and below the horizontal canal and mesial to it, could hardly represent anything but the sacculus, whereas the crus communis and the utriculus would be less well individualized. A side by side comparison of the material is needed to resolve this matter.
The lagenar recess is even longer here than in Cyonosaurus; 1t was very short in ‘Cynariops’. It is very mesially situated in Gorgonops and oriented ventro- anteriorly, whereas it is directed towards the rear in Cyonosaurus and becomes vertical in “Cynariops’ and a little more anterior than in the preceding genus. One can therefore confirm the tendency noted by Olson, wherein the lagena would migrate anteriorly and mesially in an evolutionary sequence. Gorgonops would then be situated above the two others in this regard.
The fenestra ovalis, instead of being, as it is here, antero-lateral with respect to the lagena, is definitely ventral in the two genera studied by Olson, and even oriented a little posteriorly in Cyonosaurus. This situation, like that of the lagena, makes of the ear of Gorgonops an evolved structure with respect to that of the two preceding genera.
But, on the contrary, the supposed position of the orifice for nerve VII, slightly in front of the horizontal canal, does not prolong the sequence sketched according to Olson by the two other gorgonopsians; in his view this orifice should migrate toward the rear in an evolved form. This point will be taken up further on.
As in Cyonosaurus and ‘Cynariops’, the jugular foramen is isolated from the internal ear by a bony wall, although this foramen is less posteriorly situated here.
In 1938 Olson indicated the trace in ‘Cynariops’ of an endolymphatic duct, leading to an endolymphatic sac situated against the suture separating the prootic and the supraoccipital, behind the floccular fossa. But in 1944 he interpreted this opening as the ‘dorsal entrance of the sinus utricularis into the periotic’ and as indicating ‘the level of the junction of the utriculus with the crus communis canalium’. Whatever may be the interpretation of this structure,
| ;
THE INNER EAR OF GORGONOPS 67
it seems that it agrees with that cited above in the mesial wall of the ear of Gorgonops.1°
From this short comparison it can be seen, first of all, that the region of the internal ear shows a definite homogeneity in the three gorgonopsians discussed. But one can also discern indications of variability, even though it is difficult to decide in which direction the evolution occurred. To determine this, it would be necessary to methodically explore the infra-order. Neither does one know if this evolution was harbinging that was undergone by the more evolved theriodonts in their tentatives toward mammalness, or if, like the prootic bone, the ear of Gorgonopsidae evolved more or less against the trends seen in the other theriodonts.
On this topic, and in conclusion, some of the evolutionary tendencies presented by the therapsid internal ear can be envisaged. These were essentially brought out by Olson (1944).
According to this author, the relationship of the ear length to the length of the skull diminishes when the latter increases in absolute value. The specimen of Gorgonops, larger than all the theriodonts described by Olson, fully confirms this point of view, with a proportion of 0,06, which is lower than all the indices obtained by him. The relationship of ear width over length is very ‘gorgonop- sian’ with 0,06, but it does not appear possible to follow a progression through- out the therapsid series. On the other hand, the height of the ear would make this specimen the most evolved of all, with 0,1 (the cynodonts, however, could not be evaluated in this respect).
The feeble degree of mesial ossification seems general, from the pelycosaurs to the cynodonts, with no sign of progress even in the latter. In most of the forms also, the anterior vertical canal apparently opens, by the same lack of ossifica- tion, into the floccular fossa. Finally, and paradoxically, it is only in Dimetrodon and the gorgonopsians that the jugular canal is separated by a bony wall from the labyrinthic cavity.
If the floccular fossa was shallow in pelycosaurs and the first anomodonts, it is, however, difficult to discern a precise evolutionary direction of this struc- ture in the therapsid series, and still less to recognize a relationship between the fossa and the flocculus. One can say that it is in the three gorgonopsians, the therocephalian B of Olson, and in the cynodonts (Olson and Simpson), that the fossa is the deepest. On another point it seems to me that one can follow the migration of the flocculus from in front of the internal auditive foramen (anomo- donts, gorgonopsians A and B of Olson), toward the dorsal side (Pristerodon, Gorgonops, therocephalians and cynodonts), and then toward the rear (mammals). But Dzmetrodon adds a disturbing element to this sequence with a flocculus placed above the internal ear."
According to Olson, the inequality between the anterior and the posterior
70 And with that noted by Simpson (1933) above and behind the floccular fossa in a cynodont. 11 This constant reference to the specialized form that is Dimetrodon is hardly justified, except that it constitutes the only source of information we have for the pelycosaurian ear.
68 ANNALS OF THE SOUTH AFRICAN MUSEUM
vertical canals increases as one rises stratigraphically through the anomodontid series. This is not valid in the gorgonopsids, since, by this trait also, Gorgonops differs from the other forms and attains the degree reached by the more advanced and later theriodonts. On the other hand, the canals are generally more dis- tinct from the crus communis than they are here.
The lagena is contained in the prootic dorsally and in the basioccipital ventrally, an arrangement that began with Captorhinus. But the participation of the basioccipital diminishes progressively (cynodonts) as the ear rises in the periotic to attain a mammalian position. In this regard, Gorgonops occupies a primitive position; on the contrary, the migration of this structure anteriorly is as advanced in this genus as it is in the known cynodonts, and the mesial migration is even more marked. The fenestra ovalis would have followed the migration of the lagena.
As expressed by Olson, one could expect that the orifice of nerve VII would migrate toward the rear progressively with the increase in the general degree of mammalness. Unhappily, the available facts hardly confirm this supposition, since Olson’s gorgonopsians A and B are in this respect ‘evolved’, and Gorgonops and the cynodonts lag behind.
It is supposed that the nerves IX, X and XI exited through the jugular canal in all forms. Let us keep in mind that this canal is isolated from the ear only in pelycosaurs, gorgonopsians and the cynodont studied by Simpson. The dicynodont ear described by Cox (1962) agrees, on the contrary, with the more general condition.
The nerve XII leaves by a single orifice in pelycosaurs, anomodonts, and gorgonopsians, and by two orifices in cotylosaurs, certain therocephalians and cynodonts.
In summary, defining a primitive stage for the therapsid ear as well as a general evolutionary direction is hardly easier than it is for the gorgonopsians alone. It is probable that each infra-order tried the ascension independently, by beginning at nearly the same low level. But there exists an obvious necessity to considerably extend our knowledge of each of these infra-orders before we can validly appreciate their positions with respect to each other, even as concerns only the internal ear region. This is all the more desirable since the latter is a very stable region, unaffected by secondary adaptations, and which, in consequence, should contribute to our understanding of the acquisition of the mammalian stage.
REFERENCES
Acnew, J. D. 1958. Cranio-osteological studies in Dicynodon grimbeeki with special reference to the sphenethmoid region and cranial kinesis. Palacont. afr. 6: 77-107.
Barry, T. H. 1967. The cranial morphology of the Permo-Triassic anomodont Pristerodon buffaloensis with special reference to the neural endocranium and visceral arch skeleton. Ann. S. Afr. Mus. 50: 131-145.
Brink, A. S. 1955. A study on the skeleton of Diademodon. Palaeont. afr. 3: 3-38.
I eee
=== a
igs t- ~
THE INNER EAR OF GORGONOPS 69
Broom, R. 1912. On the structure of the internal ear and the relations of the basicranial nerves in Dicynodon, and on the homology of the mammalian auditory ossicles. Proc. zool. Soc. Lond. 1912: 419-425.
Burtet, H. M. DE 1934. Vergleichende Anatomisches des stato-akutischen Organs. Jn: BOLK, L. et al., eds. Handbuch der vergleichenden Anatomie der Wirbeltiere. 2 (H.2): 1293-1374.
Cox, C. B. 1962. A natural cast of the inner ear of a dicynodont. Am. Mus. Novit. 2116: 1-6.
Estes, R. 1961. Cranial anatomy of the cynodont reptile Thrinaxodon liorhinus. Bull. Mus. comp. Kool. Harv. 125: 165-180.
Otson, E. C. 1937. The cranial morphology of a new gorgonopsian. 7. Geol. 45: 511-524.
Otson, E. C. 19384. The occipital, otic, basicranial and pterygoid regions of the Gorgonopsia. JF. Morph. 62: 141-175.
Orson, E. C. 19385. Notes on the braincase of a therocephalian. 7. Morph. 63: 75-86.
Otson, E. C. 1944. Origin of mammals based upon cranial morphology of the therapsid sub- orders. Spec. Pap. geol. Soc. Am. 55: 1-xi, 1-136.
Romer, A. S. 1955. The vertebrate body. 2nd ed. Philadelphia; London: Saunders.
Romer, A. S. 1957. Osteology of the reptiles. Chicago: University Press.
Romer, A. S. & Price, L. W. 1940. Review of the Pelycosauria. Spec. Pap. geol. Soc. Am. 28: I-x, 1-538.
SIGoGNEAU, D. 1970. Révision systématique des gorgonopsiens sud-africains. Cah. Paléont. C.N.R.S.: i-xil, 1-416.
Simpson, G. G. 1933. The ear region and the foramina of the cynodont skull. Am. 7. Sci. 26: 285-204.
ETUDE DU CRANE D’UN JEUNE SPECIMEN DE MOSCHORAINUS KITCHINGI BROOM, 1920 (? TIGRISUCHUS SIMUS OWEN, 1876), THEROCEPHALIA PRISTEROSAURIA MOSCHORHINIDAE
D’AFRIQUE AUSTRALE
(Remarques sur les Moschorhinidae et les Whaitsiidae’) Par CHRISTIANE HELENE MENDREZ Institut de Paléontologie, Muséum National d’Histoire Naturelle, Paris
(Avec 7 planches, 12 figures et 1 tableau)
PLAN Introduction 5 : : ‘ ; ‘ ‘ : : 5 3 72 Description : : ; : 2 : : : : : : 72 I. Crane : s , . B : , : : : 73 A— rea daestiene , : : : : : : : i 73 . Face dorsale ; : ; : ; : : : : 73 2. Occiput . ; : P é ; : : : : 79 3. Joue ; : ‘ : : : : : : : 79 i 4. Palais : : 5 é : : : : ; : 80 | B—Endocrane . 83 | C—Principales fosses et ouvertures ‘de la région postérieure du crane 86 II. Mandibule ‘ : . : : ‘ : : : : 88 III. Splanchnocrane ; : : : : : F i 5 Core) IV. Dentition ‘ : ‘ : : B ; ‘ ; : 94 | Discussion. : 94 I. Coneeane sae one les Missdtodbiniées et i daw ates Euniilles - LOO | A—Moschorhinidae et Whaitsiidae 5 : : : : 2 TOO B— Moschorhinidae et Annatherapsididae : : ; : tO? II. Composition des Whaitsiidae et des Moschorhinidae . ; : 3) aelO3 A—Les Whaitsiidae : ; : : : ‘ : : : 103 | 1. Moschowhaitsiinae .. ‘ : : : 5 : 5 ROM / 2. Whaitsiinae : ‘ : : A ; : 5 5 OM, B—Les Moschorhinidae 3 : , : 4 3 : : 105 | 1. Annatherapsidinae . 5 : 2 : é ; 2) 105 ) 2. Moschorhininae : ; ‘ : : : : LOO 3. Euchambersiinae F ; ; . 109 C— Genres exclus des Whaitsiidae et Moschorhinidae : : LOO Conclusion ; 3 : : : ‘ ef. 10 Liste des abréviations des noms fe salllsetions ; : : ; é : III | Résumé : : : : . : ; ; : ‘ : : 112 Abstract , : : ; : : ‘ i : : ‘ : 112 Zusammenfassung : ‘ : ‘ ; : F : : : 112 Resumen 3 : E : ; : , : : ‘ : ; 113 Remerciements ; : : : : ; : 5 : : f 113 Littérature citée . ; : s F 2 ; : : é : 114 | 1 Bien que le nom de genre Whaitsia tombe en synonymie avec Theriognathus qui lui est antérieur, il ne nous a pas paru nécessaire de remplacer le nom de famille Whaitsiidae, consacré ! par l’usage, par celui de Theriognathidae. | 7! Ann. S. Afr. Mus. 64, 1974: 71-115, 7 pls, 12 figs, 1 table.
72 ANNALS OF THE SOUTH AFRICAN MUSEUM
INTRODUCTION
Le crane étudié ici appartient au plus petit spécimen connu actuellement du genre Moschorhinus (la possible synonymie avec Tigrisuchus est discutée plus loin). L’espéce-type de Moschorhinus est M. kitchingi; Kitching (1968: 67) a regroupé dans cette espéce les spécimens jusqu’alors considérés comme M. esterhuyseni et M. natalensis et laissé a part M. warreni. Celui-ci, tout comme M. minor, est également a intégrer dans M. kitching: (? Tigrisuchus simus). Matériel: crane, fragments de mandibule et d’éléments post-craniens, recueillis par l’équipe Crompton (A. W.), Ewer (R.F.), Griffin (B. S.) et Malan (M. E.), en janvier 1959.
Collection: South African Museum (Cape Town, Afrique du Sud), n°K.118. Horizon: non indiqué. Localité: Skerpioenkraal, Middelburg, Province du Cap, Afrique du Sud.
Seul le crane est étudié ici.
DESCRIPTION
Le crane du genre Moschorhinus (? Tigrisuchus) est remarquable par la briéveté et la massivité de son museau. Le spécimen 8.A.M. n°K.118 ne fait pas exception a la régle. I] présente également une constriction postcanine trés peu marquée. Les orbites sont relativement petites et les fosses temporales vastes. La créte sagittale est apparemment peu élevée.
Les principales mesures du crane (reconstruit) de S.A.M. les suivantes:
n°K.118 sont
160 mm environ 122 mm environ
Longueur du crane Largeur maximale
Largeur intertemporale safratianelle
Largeur interorbitaire minimale ;
Distance entre le bord antérieur des padres lai # le sovd annem Ae Vorbite
Distance entre le bord eae dies arGmeedl lntives et ie Bord ance de ie fenétre temporale :
Largeur du museau au niveau de le consiisiian santionatne
Hauteur du museau au méme niveau
Hauteur maximale de Vocciput ..
Largeur maximale de l’occiput ..
Longueur totale de la denture maxillaire £2 a gauche ‘a droite
Longueur totale de la rangée postcanine J x gauche a droite
14 mm 36 mm environ
66 mm environ
95 mm environ 62 mm environ 45 mm environ 58 mm environ 122 mm environ 24 mm environ 27 mm environ II mm environ 14mm environ
Figures: Les figures sont des reconstitutions. Le léger aplatissement dorso-
ventral a été corrigé et les régions manquantes d’un cété (région postérieure droite, cété gauche de la surface dorsale du museau, processus transverse du ptérygoide gauche, . . .) ont été reconstituées par symétrie. Tandis que l’arrieére- crane est bien conservé a gauche, les dessins de détail représentent cette région
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 73
retournée comme s'il s’agissait d’un cété droit pour faciliter une comparaison ultérieure avec les autres Thérocéphales.
I— CRANE A—Crane dermique
Le crane de ce spécimen est assez bien conservé malgré l’érosion du cote gauche de la face dorsale du museau, des arcades zygomatiques (dont subsiste une empreinte du cété gauche), des barres sousorbitaires et d’une partie des arcades postorbitaires.
1. Face dorsale (Fig. 1; Pl. 1)
Les deux prémaxillaires (Pmx, Figs 1, 3; Pls I, III A, B, VII B), soudés sur la ligne médiane, sont presque complets, hormis le processus nasal dont il ne subsiste que la base. Pres de celle-ci, on observe dorsalement le foramen prémaxillaire postérieur (f.pmx.p, Figs 1, 3; Pl. 1) et antérieurement les foramens prémaxillaires dorsaux (f.pmx.d, Figs 2, 3; Pl. VII B), au nombre de trois du cote droit, le mieux conservé. Selon Brink (1958: 46-47), chaque pré- maxillaire de B.P.I. n°F.1713/M.295 ne présenterait qu’un foramen prémaxil- laire dorsal. Le nombre d’orifices de sortie des ramifications du rameau frontal de la branche ophthalmique du nerf trijumeau semble donc assez variable: le maximum connu serait trois. Le degré de recouvrement par le maxillaire est ici incertain du fait de l’érosion de ces régions.
Chaque prémaxillaire porte cinq incisives.
Le septomaxillaire (Smx, Figs 1, 3; Pls I, III A, B, VII A) est large et bien développé sur la face. La région de la fosse maxillo-septomaxillaire est abimée ici, mais chez un autre spécimen (B.P.I. n°F.2205/M.193) est représentée une configuration comparable a celle d’ Akidnognathus décrite par Brink (1960: 156, fig. B, C, D; 164), de Regisaurus décrite par Mendrez (1972a: 192-196, fig. 6) et dun Whaitsiidae, Tii.n°K.65 (observation personnelle). Il est a noter que chez les différents Thérocéphales ot cette fosse a pu étre observée en deétail, sa structure est comparable a celle que Sigogneau (1970): 26, fig. 1) a étudiée chez un Gorgonopsien Icitidorhinidé: Rubidgina sp., B.P.I. n°F.3924. Si érosion de la partie antérieure du museau a détérioré la région de la fosse maxillo-septomaxillaire, elle a mis en évidence le canal septomaxillaire (c.smx, Pls Til, VII B).
Le maxillaire (Mx, Figs 1, 3; Pls I, III) est élevé et participe largement a la barre sousorbitaire; ce que Brink (1958: 47) remarquait chez le spécimen B.P.I. n°F.1713/M.295. La limite postérieure du maxillaire n’est ici pas connue en raison de l’érosion partielle des barres sousorbitaires. La région de contact entre maxillaire et préfrontal (Pfr) est également endommagée, mais, a droite, l’empreinte du préfrontal montre que ce contact, classique chez les Théro- céphales, était loin d’étre aussi réduit que chez le spécimen B.P.I. n°F.2205/ M.1092, tel que Brink l’a représenté (1958: 41, fig. 16-A; 42, fig. 16-B). Ce
74 ANNALS OF THE SOUTH AFRICAN MUSEUM
dernier spécimen est en réalité trés dissymétrique et si sur le coté droit endom- magé le contact entre le maxillaire et le préfrontal est bref, il est par contre important sur le cété gauche (observation personnelle). La surface du maxillaire
LEP
Y Le i Tai
NGS
i ANS
m TCT = | | | | | | I | | |
NN
pr.q.Pt ‘ ; Vo+V3 Se a i a IN |
\W
pr.ept.Sq — Ny
Q— f s bi fo.paroc.Op
Fig. 1. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118. Vue dorsale du crane (x #).
Boc, basioccipital; cr.ptorb, créte postorbitaire; cr.sag.Fr, créte sagittale du frontal; cr.sag.Pa, créte sagittale du pariétal; Eoc, exoccipital; Ept, épiptérygoide; fla, foramens lacrymaux; (f-mg), foramen magnum; f.pi, foramen pinéal; f.pmx.p, foramen prémaxillaire postérieur; f-pt.- paroc, foramen ptérygo-paroccipital; fe.temp, fenestra temporalis; fo.mx.-smx, fosse maxillo-septo- maxillaire; fo.paroc.Op, fosse paroccipitale de l’opisthotique; /fo.sorb, fosse sousorbitaire; F7, frontal; Ifa, interpariétal; Ju, jugal; La, lacrymal; Mx, maxillaire; Na, nasal; nar, narine; Op, opisthotique; (orb), orbite; Pa, pariétal; Pfr, préfrontal; Pmx, prémaxillaire; pr.ept.Sq, processus épiptérygoide du squamosal; pr.proot.Sq, processus prootique du squamosal; pr.q.Pt, processus carré du ptérygoide; Proot, prootique, Pt, ptérygoide; Ptorb, postorbitaire; Q, carré; Smx, septomaxillaire; Soc, supraoccipital; Sq, squamosal; St, stapes; Tab, tabulaire; Tr, trans- verse; V2 + V3, lieu de sortie des branches maxillaire et mandibulaire du nerf trijumeau.
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 75
Smx_—_ #
fo.mx-SmMx_. 2B
#
Ptorb_ —- =
fetemp — <a
Sq—
f.pt-paroc Proot Ec o¢ soc Oe item:
PLANCHE I. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118, vue dorsale du crane (x $).
Ept, épiptérygoide; Foc, exoccipital; fipmx.p, foramen prémaxillaire postérieur; / pt.-paroc, foramen ptérygo-paroccipital ; fe.temp, fenestra temporalis; fo.mx.-smx, fosse maxillo-septomaxillaire; Fr, frontal; Ipa, interpariétal; Fu, jugal; La, lacrymal; Mx, maxillaire; Va, nasal; nar, narine; orb, orbite; Pfr, préfrontal; Pmx, prémaxillaire; Proot, prootique; Pt, ptérygoide; Ptorb, post- orbitaire; Q, carré; Sang, susangulaire; Smx, septomaxillaire; Soc, supraoccipital; Sg, squamosal ; Tab, tabulaire; 77, transverse.
76 ANNALS OF THE SOUTH AFRICAN MUSEUM
{i
by. NE
7 prmas.Op _ f.pf.-paroc
Fig. 2. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118. Vue ventrale du crane (x #).
Boc, basioccipital; C.s, canine supérieure; cr.ch, crista choanalis; cr.v.-it.Pi, créte ventro-inter- médiaire du ptérygoide; cr.v.-m.Pt, créte ventro-médiane du ptérygoide; cr.v.-m.Vo, créte ventro- médiane du vomer; Foc, exoccipital; Ept, épiptérygoide; f ju, foramen jugulaire; fmx.-p, foramen maxillo-palatin; f.pmx.d, foramen prémaxillaire dorsal; f.pt.-paroc, foramen ptérygo-paroccipital ; f.v.c.sorb, foramen pour les vaisseaux du canal sousorbitaire; f.vo, foramen voméro-nasal ; fe.temp, fenestra temporalis; fo.C.i, fosse pour la canine inférieure; fo.ipt, fosse interptérygoidienne; fo.sorb, fosse sousorbitaire; J}, premiére incisive supérieure; J?, deuxiéme incisive supérieure; J°, cin- quiéme incisive supérieure; 72, deuxiéme incisive supérieure de remplacement; Ju, jugal; La, lacrymal; Mx, maxillaire; o.ch, ouverture choanale; 0.¢.c.c.i, orifice externe du canal de la carotide interne; Op, opisthotique; orb, orbite; P.-Bsph, complexe para-basisphénoide; Pal, palatin; (PC*), alvéole dela premiére postcanine supérieure; PC?, deuxiéme postcanine supérieure ; (PC*), alvéole de la troisitme postcanine supérieure; pC, précanine; Pfr, préfrontal; Pmx, pre- maxillaire; pr.bsph.Pt, processus basisphénoide du ptérygoide; pr.mas.Op, processus mastoide de l’opisthotique; pr.mx.Pmx, processus maxillaire du prémaxillaire; pr.g.Op, processus carré de l’opisthotique; pr.q.Pt, processus carré du ptérygoide; pr.tr.Pt, processus transverse du ptery- goide; pr.vo.Pmx, processus vomérin du prémaxillaire; Proot, prootique; Ptorb, postorbitaire; Q, carré; q.P.-Bsph, quille du complexe para-basisphénoide; Qj, quadratojugal; r.Psph, rostre du parasphénoide; Sg, squamosal; St, stapes; sut.itvo, suture intervomérine; 77, transverse; iub.sph.-oc, tuberculum spheno-occipitale; Vo, vomer.
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 77
~ cr.v-m.Pt m Pt
a Ea on Spe BEE. fo.sorb ~ Si i aan ———
Se eee — f0.ipt
wecceeneeme mmm
5 a crv-itPt ~
a — fe.temp
q.P-Bsph _ = P.-Bsph — tub.sph.-oc . Sy -Qj z= Sq Qe a a Figeh a”
PLANCHE II. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K. 118, vue ventrale du crane (Xx $).
Boc, basioccipital; C.z, canine inférieure; C.s, canine supérieure; cr.ch, crista choanalis; cr. v. -it. Pt, créte ventro-intermédiaire du ptérygoide; cr.v.-m.Pt, créte ventro-médiane du _ ptérygoide; cr.u.-m.Vo, créte ventro-médiane du vomer; Foc, exoccipital; f.ju, foramen jugulaire; f.mvx.-p, foramen maxillo-palatin; fv.c.sorb, foramen pour les vaisseaux du canal sousorbitaire; f.vo, foramen voméro-nasal; fe.temp, fenestra temporalis; fo.C.i, fosse pour la canine inférieure; fo. ipt fosse interptérygoidienne; fo.sorb, fosse sousorbitaire; J1, premiére incisive supérieure; J?, deuxieme incisive supérieure; J°, cinquiéme incisive supérieure; i2, deuxiéme incisive supérieure de rem- placement; La, lacrymal; Mx, maxillaire; 0.ch, ouverture choanale; Op, opisthotique; P.-Bsph, complexe para-basisphénoide; Pal, palatin; (PC), alvéole de la premiére postcanine supérieure; PC?, deuxiéme postcanine supérieure; (PC*), alvéole de la troisitme postcanine supérieure; pC, précanine; Pmx, prémaxillaire; Proot, prootique; Pt, ptérygoide; Q, carré; g.P.-Bsph, quille du complexe para-basisphénoide; Qj, quadratojugal; Sg, squamosal; swi.itve, suture inter- vomerine; St, stapes; Tr, transverse; tub.sph.-oc, tuberculum spheno-occipitale; Vo, vomer.
78 ANNALS OF THE SOUTH AFRICAN MUSEUM
est ‘grélée’ et marquée de trés petits foramens pour les diverses branches cuta- nées du nerf alvéolaire supérieur.
La briéveté des nasaux (Na, Figs 1, 3; Pls I, III) est en relation avec celle du museau. II est a noter cependant que la suture fronto-nasale est située en arriere du niveau du bord antérieur des orbites.
Le lacrymal (La, Figs 1, 3; Pls I, III) est haut et bien étendu sur la face, en avant des orbites. Par contre, il participe peu a la face externe de la barre sousorbitaire.
Le préfrontal (Pfr, Figs 1, 3; Pls I, III) est tres développé, tant vers avant que vers l’arriére. Son bord antérieur présente un important contact avec le maxillaire (contact comparable a celui indiqué par Brink (1958: 44, fig. 18; 46, fig. 20) chez le spécimen B.P.I. n°F.1718/M.295). L’extension postérieure du préfrontal est telle qu’il forme plus de la moitié du bord supérieur de Vorbite, ce qui s’observe également chez les autres specimens de Moschorhinus (? Tigrisuchus) .
Les frontaux (Fr, Fig. 1; Pl. I) sont relativement brefs mais larges. Leur participation au bord supérieur de l’orbite est assez faible en raison de l’exten- sion des préfrontaux.
La partie antérieure du postorbitaire (Ptorb) est courte, tandis que son expansion postérieure est importante. Son processus latéral, long et massif, participe de fagon importante a lépaisse arcade postorbitaire.
On n’observe ni postfrontal ni prépariétal.
Les pariétaux (Pa, Figs 1, 3; Pls I, III) sont assez épais. Du fait de la relative brieveté de la créte sagittale (cr.sag.Pa, Fig. 1) et de sa division précoce en crétes postorbitaires (cr.ptorb, Fig. 1) mousses, la partie antérieure des parié- taux forme chez les divers spécimens de Moschorhinus (? Tigrisuchus), une plate- forme triangulaire a sommet postérieur. La créte sagittale du frontal (cr.sag.Fr, Fig. 1) est ainsi séparée de fagon marquee de celle du pariétal. Chez les Anna- therapsididae de U.R.S.S., Annatherapsidus Kuhn, 1961 ( = Anna Amalitzky, 1922 préoccupé) (P.IN. Mw. n°2005/1993) selon Vjushkov 1955 (fig. 1) et Chthonosaurus Vjushkov, 1955 (P.IN. Mw. n°521/1-1) également selon cet auteur (p. 147 et fig. 5) la situation est voisine de celle existant chez Moschorhinus (? Tigrisuchus): ‘Du cété dorsal, dans leur partie antérieure, ils (les pariétaux) ont la forme d’un triangle équilatéral dont la pointe dirigée vers larriére se transforme en une créte sagittale étroite et longue . . .’.
Chez le Whaitsiidae d’U.R.S.S., Moschowhaitsia (P.IN. Mw. n°1100/20) un trés petit creux existe a la place de cette plateforme. Chez les Whaitsiidae africains du genre Theriognathus Owen, 1876 (syn.: Alopecopsis Broom, 1920, Aneugomphius Broom & Robinson, 1948—celui-ci étant un sujet trés jeune a crétes peu marquees présente un aspect différent de celui des adultes—, Hyeno- saurus Broom, 1935, Moschorhynchus Broom, 1936, Notaelurops Broom, 1936, Notosollasia Broom, 1925, Whaitsia Haughton, 1918), la créte sagittale parictale ne se divise qu’au niveau de la suture fronto-pariétale en deux crétes postorbi- taires aigués soulignées par les concavités du frontal de part et d’autre de la
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 79
créte sagittale du frontal qui ici, est en continuité avec celle du pariétal.
La région intertemporale est assez étroite, comme il est de régle chez les Thérocéphales Pristerosauria.
Le foramen pinéal (f.f7) est ici une fente extrémement é€troite située anté- rieurement. Chez le spécimen B.P.I. n°F.1713/M.295, Brink (1958: 43) fait la méme remarque. Chez Annatherapsidus (cf. Vjushkov 1955: 135) et Chthonosaurus (Vjushkov 1955: 147) ce foramen serait méme pratiquement fermé. Chez Chthonosaurus, un léger ‘ourlet’ des pariétaux indiquerait son emplacement. Chez Moschowhaitsia, ce foramen manque complétement selon Tatarinov (1963: 78), tandis que chez les Whaitsiidae africains, ce foramen pinéal subsiste.
2. Occiput (Figs 4, 9; Pls IV A, B)
Comme chez de nombreux Thérocéphales et en particulier chez les Whaitsiidae, le pariétal contribue de facgon assez importante a la face occipitale, au-dessus de l’interpariétal (Jpa) et d’une partie des tabulaires (Tab).
L’occiput est large et moyennement élevé, et la créte occipitale aigué.
Sur la face occipitale, Pinterpariétal est ovalaire, assez élevé par rapport a sa largeur. Il participe en outre, entre le pariétal, le supraoccipital et le squamosal, a Pangle postéro-médial de la fenétre temporale (Fig. 3).
Le tabulaire, plus large que long, est totalement exclu du bord supérieur de l’ouverture postérieure de la fosse posttemporale (fo.pttemp) par une barre épaisse formée de l’association du processus postéro-dorsal de l’opisthotique (pr.p.-d.Op) et du processus intermédiaire du squamosal (pr.it.Sq). Il est égale- ment exclu du bord latéral de cette fosse par le processus mastoide du squamosal
(pr.mas.Sq, Fig. 9).
geepoue ic. 2° PL ITT)
Du jugal (Ju) ne subsiste qu’une empreinte du cété gauche et un fragment mince sous la partie antérieure de l’orbite droite. Il s’étend vers l’avant a peine plus loin que le bord antérieur de l’orbite.
Le squamosal (Sq, Figs 1, 3, 4, 6, 7, 8, 9; Pls I, III, 1V, V) est bien con- serve a gauche; il ne manque que le processus jugal. Antérieurement (Fig. 6), le squamosal participe largement a la paroi postérieure de la fenétre temporale (fe.temp) :
— Le processus dorsal (pr.d.Sq, Figs 3, 6, 7, 8; Pl. V) recouvre le pariétal latéralement; son extrémité est brisée, mais son empreinte montre qu’il n’atteignait pas l’angle postéro-interne de la fenétre temporale. Entre la base de ce processus et le bord supérieur du processus intermédiaire (pr.it.Sq) est pincée l’extrémité externe de l’interpariétal.
— Le processus intermédiaire est donc ici en contact non seulement avec le prootique et le supraoccipital mais encore d’une part, dorsalement, avec linterpariétal et d’autre part, ventralement, avec le processus postéro-dorsal de l’opisthotique qui V’isole du bord supérieur de l’ouverture antérieure de la fosse posttemporale.
80 ANNALS OF THE SOUTH AFRICAN MUSEUM
AW YAY Z)
\\\
NZ \ {" AI lu
G4,
= |
AP | \
Ny |
@w a
AS
x ‘a i a Mp 2 \\ “ Se il
| | Tr prtr. a | a -Beph f.v.c.sorb ! | CrV-m. Pt _ pr.bsph.Pt cr.v-it.Pt
eo)
=
ie)
A no) a SS
Fig. 3. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118. Vue latérale gauche du crane (x 4).
Boc, basioccipital; C, canine; cr.v.-it.Pt, créte ventro-intermédiaire du ptérygoide; cr.v.-m.Pt, créte ventro-médiane du ptérygoide; Hoc, exoccipital; Ept, épipterygoide; f.pmx.d, foramen pré€- maxillaire dorsal; f.pmx.p, foramen prémaxillaire postérieur; f-v.c.sorb, foramen pour les vais- seaux du canal sousorbitaire; fo.mx.-smx, fosse maxillo-septomaxillaire; Fr, frontal; J1, premiere incisive supérieure; [>, cinquieme incisive supérieure; Jpa, interpariétal; Fu, jugal; La, lacrymal; Mx, maxillaire; Wa, nasal; nar, narine; P.-Bsph, complexe para-basisphénoide; Pa, pariétal; Pal, palatin; pC, précanine; PC?, deuxieme postcanine supérieure; Pfr, préfrontal; Pmx, pré- maxillaire; pr.bsph.Pt, processus basisphénoide du ptérygoide; pr.d.Sq, processus dorsal du squamosal; pr.etp.Sq, processus épiptérygoide du squamosal; pr.it.Sqg, processus intermédiaire du squamosal; pr.mas.Op, processus mastoide de Vopisthotique; pr.q.Op, processus carré de Popisthotique; pr.g.Pt, processus carré du ptérygoide; pr.tr.Pt, processus transverse du ptéry- goide; Proot, prootique; Ptorb, postorbitaire; Q, carré; q.P.Bsph, quille du complexe para- basisphénoide; Qj, quadratojugal; Smx, septomaxillaire; Soc, supraoccipital; St, stapes; Tr. transverse.
— Le processus antéro-ventral (pr.a.-v.Sq) est caractérisé par sa situation nettement en-dessous du bord supérieur du processus paroccipital de Popistho- tique et par sa subdivision en deux processus: |’un, le processus prootique (pr.proot.Sq), fusionné avec le processus central du prootique (pr.c.Proot), autre, le processus épiptérygoide (pr.ept.Sq) dont il ne subsiste ici que la racine. Cette subdivision du processus antéro-ventral existe également chez les Whaitsiidae. |
Le quadratojugal (Q 7) n’est représenté que par un petit fragment (Figs os 3; Pls III, V, VII A).
4. Palais (Figs 2, 5; Pls II, VII A) |
La structure du palais est une des différences essentielles entre Moscho- rhinidae et Whaitsiidae, les premiers ayant un palais nettement plus primitif |
| CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 81
soc< = = CsSinx f.ju eee tennx oe Subs _ Pmx Eoc/ A N' Blech “Hus eae fal .-DS V.C. E Pt TGakee Gace Fr Ptorb Pa Pfr Li | ai | ' fe) 4cm. swe ESS)
| Pt Ept P-Bsph Pt
/ PLANCHE III. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118; A, vue latérale droite du crane (x 4%); B, vue latérale gauche du crane (x $).
| Boc, basioccipital; C.s, Canine supérieure; c.smx, canal septomaxillaire; Foc, exoccipital; Ept, épiptérygoide; f. ju, foramen jugulaire; fmx, foramen maxillaire; fpmx.d, foramen prémaxillaire dorsal; f-v.c.sorb, foramen pour les vaisseaux du canal sousorbitaire; fo.mx.-smx, fosse maxillo- septomaxillaire; Fr, frontal; J+, premiére incisive supérieure; /°, cinquie€me incisive supérieure; ju, jugal; La, lacrymal; Mx, maxillaire; Na, nasal; nar, narine; Op, opisthotique; P.-Bsph, | complexe para-basisphénoide; Pa, pariétal; Pal, palatin; pC, précanine; Pfr, préfrontal; Pmx, | prémaxillaire; Pt, ptérygoide; Ptorb, postorbitaire; Q, carré; Smx, septomaxillaire; Soc, supra- occipital; Sg, squamosal; 77, transverse.
82 ANNALS OF THE SOUTH AFRICAN MUSEUM
que les seconds, tant dans sa région antérieure que dans sa région postérieure.
Les prémaxillaires (Pmx) sont brefs médialement mais trés étendus latérale- ment. Le processus maxillaire (pr.mx.Pmx, Fig. 2) est, ainsi que la partie anteé- rieure du maxillaire, creusé d’une fosse en grande partie aveugle pour la canine inférieure (fo.C.z, Fig. 2; Pl. II) en continuité avec l’ouverture choanale (o.ch, Fig. 2; Pl. II). Cette derniére est réniforme, son bord latéral étant concave du fait du renflement de la paroi médiale du maxillaire et du palatin au niveau de la canine et des premiéres prémolaires. Ce renflement peut: soit étre di au grand diamétre de la canine, soit correspondre a un début de palais secondaire osseux. Dans la derniére hypothese, il est a noter que cette région est plus primitive chez Moschorhinus (? Tigrisuchus) que chez les Pristerognathidae Pristerognathus et Ptomalestes, telle qu’elle fut décrite par Mendrez (1972): 2960-1961, figs A, B). En effet, chez le premier, la crista choanalis (cr.ch, Fig. 2; Pl. II) est moins marquée a Vavant de ce renflement.
Les vomers (Vo, Fig. 2; Pls II, VII) ne sont fusionnés que de facon trés incomplete: un reste de suture subsiste a avant de la plaque anteérieure, interchoanale et une suture bien visible (sut.ztvo, Fig. 2; Pl. II) sépare sur toute leur longueur les deux éléments qui forment la plaque postérieure qui s’étale sur les palatins et les ptérygoides. Comme chez les Thérocéphales primitifs, et en particulier les Annatherapsididae, la plaque antérieure et la plaque pos- térieure du vomer sont situées a peu prés sur le méme plan, tandis que chez les Whaitsiidae, Vinclinaison de la plaque postérieure est plus marquee et la plaque antérieure apparait comme plus ventrale. Une autre caractéristique qui fait rapprocher Moschorhinidae et Annatherapsididae est l’étalement de la plaque antérieure ‘en éventail’ notait Brink (1958: 48) chez Moschorhinus (? Togrisuchus) B.P.I. n°F.1713/M.295, ‘en rame’ observait Vjushkov (1955: 138) chez Annatherapsidus P.IN. Mw. n°2005/1933. Chez les Whaitsiidae de PU.R.S.S., de Tanzanie et de l’Afrique du Sud, par contre, cette plaque est un élément cruciforme dont la partie antérieure est convexe ventralement, les branches latérales plates et recouvertes par des processus émis par les maxillaires et la région basale postérieure réduite a une étroite créte.
Chaque vomer est percé a l’avant d’un foramen voméro-nasal (f.v0, Fig. 2; Pl. II) bien développé.
La suture avec les ptérygoides est large, a fortes indentations.
Les ptérygoides (Pt, Figs 2, 3, 5; Pls I, I1) bordent l’angle postéro-meédial de la fosse sousorbitaire (fo.sorb, id.). Leurs crétes ventro-intermédiaires (cr.v.- it.Pt, id.), a inverse de ce qui est connu chez les Annatherapsididae (cf. Vjushkov 1955) et les Whaitsiidae (cf. Mendrez 1965) ne semblent pas porteuses de dents (Kemp 1972: 32, considére que la perte des dents du palais est une des caractéristiques des Whaitsiidae, ceci est contraire aux observations de Mendrez 1965), mais ces régions étant altérées, on ne peut étre absolument certain de ce fait. Comme chez certains Scaloposauridae et les Annatherapsididae Annatherapsi- dus (cf. Vjushkov 1955; fig. 3) et Chthonosaurus (Vjushkov 1955: 153, fig. 6), le processus carré (pr.g.Pi, Figs 2, 3, 4, 5, 8, 9, 10, 11; Pls Il, V) et leqene-
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 83
cessus transverse (fr.tr.Pt, id.) sont reliés par ‘une fine lame osseuse’, comme le notait Vjushkov (1955). Chez les Whaitsiidae, cette lame n’existe pas. La relation entre les processus basisphénoides (pr.bsph.Pt, id.), dont les bords sont ourlés, est comparable a celle décrite par Vjushkov (1955) chez Chthonosaurus : ces deux processus ne sont pas jointifs sur la ligne médiane et sont séparés par le rostre du parasphénoide (7.Psph, Figs 2, 5; Pl. V). Chez Moschorhinus (? Tigrisu- chus) S.A.M. n°K.118, cette région a pu étre dégagée assez profondément et on peut ainsi observer deux petits processus (pr.psph.Pt, Fig. 5; Pl. V) qui pincent ce rostre, dans la partie antérieure de la fosse interptérygoidienne (/o.7ft, Figs 2, 5; Pls II, V). De ce fait, cette derni¢re forme deux fentes étroites de chaque cote de la poutre constituée par le rostre du parasphénoide et les pro- cessus parasphénoides des ptérygoides (pr.psph.Pt). Chez Annatherapsidus par contre, les processus basisphénoides des ptérygoides sont jointifs sur presque toute leur longueur et ne laissent a l’avant qu’une petite fosse (Vjushkov 1955). Chez les Whaitsiidae, hormis Moschowhaitsia chez qui subsiste une étroite fente antérieure séparant partiellement les processus basisphénoides, les processus basisphénoides sont jointifs sur toute leur longueur, ventralement par rapport au rostre du parasphénoide.
lesepalatins, (Pal, Pigs 2, 3, 5; Pls Il, Ill, V, vip, bien développés, recouvrent les processus antérieurs des pterygoides jusqu’a la moitié de la longueur des fosses sousorbitaires. Ces derniéres sont vastes comme chez la presque totalité des Thérocéphales, hormis les Whaitsitidae de Tanzanie et d’Afrique du Sud et le Moschorhinidae Euchambersia.
Les transverses (Tr, Figs 2, 3, 5; Pls I, II, III, V) ou ectoptérygoides ont leur partie postérieure é¢rodée: aussi est-il impossible de préciser le degré exact de recouvrement des apophyses transverses des ptérygoides. Le foramen des vaisseaux du canal sousorbitaire (f.v.c.sorb, Fig. 2; Pl. II) est trés grand.
B—Endocrane
La base de l’endocrane est particulicrement bien exposée en vue ventrale (rrgss2s05> Pls Il; V).
Seuls les processus basiptérygoides du complexe para-basisphénoide (pr.bspt.P.-Bsph) sont recouverts par les ptérygoides. La quille de ce complexe (q.P.-Bsph) apparait beaucoup moins importante que chez les Whaitsiidae africains. Elle est endommagée chez les Whaitsiidae et Annatherapsididae russes. Le rostre du parasphénoide est trés robuste et son extrémité antérieure est recouverte par les petits processus parasphenoides du ptérygoide (pr.psph.Pt). Pratiquement a la limite quille-rostre, a la base de la quille, s’ouvrent de chaque cété de celle-ci les orifices externes du canal de la carotide interne (0.e.c.c.t.). Le complexe para-basisphénoide participe, pour une faible part, semble-t-il, et de facon trés superficielle, au bord de la fenétre ovale, en partie endommagée. Chez les Whaitsiidae sud-africains, il en parait exclu (observation personnelle). Les tubercula spheno-occipitalia (tub.sph.-oc), formés moitié par le
84 ANNALS OF THE SOUTH AFRICAN MUSEUM
basioccipital (Boc) recouvert par le complexe para-basisphénoide, moitié par le basioccipital seul, sont volumineux. A l’avant de ces tubérosités, le complexe para-basisphénoide est beaucoup moins é€troit que chez les Whaitsiidae africains.
L’épiptérygoide (Eft, Figs 1, 2, 3, 6, 11; Pls I, 111 B; Vip Vil A esmamme plaque aplatie moyennement €largie, a petite apophyse postérieure (a.p.Ept), et qui ressemble assez a l’épiptérygoide du générotype d’ Annatherapsidus (obser- vation personnelle). Cet os est dans les deux cas bien moins élargi que chez les Whaitsiidae africains. Le bord postérieur de l’épiptérygoide de ceux-ci présente en outre un processus supplémentaire (le processus prootique de l’épiptérygoide) qui recouvre le prootique et est lui-méme recouvert par l’apophyse postérieure. Celle-ci en outre s’étend assez pour entrer également en contact avec le proo- tique (observation personnelle). Il est possible que les branches maxillaire (V2) et mandibulaire (V7) du trijumeau soient séparées. Il faudrait que des spécimens de Moschorhinus (?Tigrisuchus) présentent cette région en meilleur état pour que l’on soit str de l’homologie de ces processus. Le processus postéro- ventral (pr.p.-v.Ept), élevé, devait étre reli¢é au squamosal, car le processus ventral de celui-ci est, comme nous l’avons vu plus haut, divisé en deux pro- cessus, lun soudé au processus central du prootique (r.c.Proot), autre se dirigeant vers |’épiptérygoide. Une situation analogue a pu étre observée chez des Whaitsiidae africains. Le cavum epiptericum (ca.ept, Fig. 6) est important.
Les os otiques
Comme lauteur l’a observé chez un Scaloposauria, Regisaurus Mendrez, 1972 (F.R.P. n° 1964/27 in Mendrez 1972a: 201) et chez des Pristerosauria
ipa | Pa
wl) EA Whe a] D
Ii i SZ SSS
pr.mas.Op
O ‘1cm. i __f
Fig. 4. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118. Vue occipitale du crane (X #).
Boc, basioccipital; Eoc, exoccipital; f-ju, foramen jugulaire; f.mg, foramen magnum; fo.paroc.Op, fosse paroccipitale de l’opisthotique; fo.pttemp, fosse posttemporale; Ipa, interpariétal; P.-Bsph, complexe para-basisphénoide; Pa, pariétal; pr.mas.Op, processus mastoide de l’opisthotique; pr.q.Op, processus carré de l’opisthotique; pr.q.Pt, processus carré du ptérygoide; Q, carré; Soc, supraoccipital; Sg, squamosal; St, stapes; Tab, tabulaire.
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 85
A Q_icm. ipa fo.pttemp Pa cc aed | pr.p-dUp pr.it.Sq | Teeny fmg
fo.paroc.Op) (fo.p x Pp
pr.mas.Op pr.tr.Pf vi | } — Bae ef | st | | \ Ec tt.orl.moy Boc | f.ju | pr.a-v.Sq q.P-Bsph O icm
B
PLANCHE IV. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118; vue occipitale du crane; AG Xo: chv.)5B, (OG 8).
Boc, basioccipital; Eoc, exoccipital; f.ju, foramen jugulaire; fmg, foramen magnum; (fo.paroc.Op), fosse paroccipitale de l’opisthotique; fo.pttemp, fosse posttemporale; Jpa, interpariétal; Mx, maxillaire; Op, opisthotique; Pa, pariétal; pr.a.-v.Sg, processus antéro-ventral du squamosal; pr.it.Sq, processus intermédiaire du squamosal; pr.mas.Op, processus mastoide de l’opisthotique; pr.p.-d.Op, processus postéro-dorsal de l’opisthotique; pr.g.Op, processus carré de l’opisthotique; pr.tr.Pt, processus transverse du ptérygoide; Q, carré; q.P.-Bsph, quille du complexe para- basiphénoide; Soc, supraoccipital; Sg, squamosal; St, stapes; Tab, tabulaire; tt.orl.moy, toit de Voreille moyenne.
86 ANNALS OF THE SOUTH AFRICAN MUSEUM
Whaitsiidae, Theriognathus (B.P.I. n°F.100/M.141) et Pristerognathidae, Ptomalestes Boonstra, 1954 (S.A.M. n°11942)?, le prootique (Proot) et Popistho- tique (Op) sont bien individualisés (Figs 6, 7, 12) ; ce qui est visible de la suture qui les sépare correspond a ce qui a été déja décrit chez Regisaurus (Mendrez 1972). —
Le prootique est en partie endommagé et, de ce fait, le processus antéro- ventral n’a pu étre observe. Cet os apparait massif, a région dorsale nettement plus ossifiée que chez Regisaurus. Le prootique de Moschorhinus (? Tigrisuchus) comme celui de Theriognathus présente le méme nombre de processus que celui de Regisaurus tandis que, chez Ptomalestes, un processus centro-ventral supple- mentaire est situé juste sous le processus central (observation personnelle).
L’aspect général de lopisthotique ne différe pas essentiellement de celui de Regisaurus: le processus postéro-dorsal (pr.p.-d.Op) est moins allonge, il est également moins recouvert par le processus intermédiaire du squamosal (pr.it.Sq) et par conséquent est apparent en vue antérieure au-dessus de Pouver- ture de la fosse posttemporale (fo. pttemp). Le processus paroccipital (pr. paroc.Op est particuliérement massif et est creusé d’une vaste fosse paroccipitale (fo.paroc.Op). Sur la face ventrale, entre le processus carré (pr.g.Op) et le pro- cessus mastoide (pr. mas. Op), la fosse qui correspondrait a l’extrémité latérale du toit de l’oreille moyenne (tt.orl.moy) est peu profonde.
Le supraoccipital (Soc), trapu, apparait largement dans la fenétre tem- porale (fe.temp, Figs 3, 6 7; Pls III A, VII A). En vue occipitale (Fig. 4; Pl. IV), ses processus latéraux sont brefs et leur limite correspond a peu pres au niveau du bord médial de la fosse posttemporale. Chez Theriognathus également, ces processus sont courts tandis que chez Annatherapsidus, ils sont plus allongés.
L’exoccipital (Hoc) est aussi important et participe pour une bonne part au. bord supérieur du foramen magnum (f.mg, Fig. 4; Pls I, IV) tout comme au condyle occipital (cdy.oc). Avec lopisthotique, il élimine le basioccipital du plancher du foramen jugulaire (f. ju, Figs 4, 9; Pl. IV).
Le basioccipital (Boc) participe au plancher du foramen magnum et forme avec les exoccipitaux un condyle tripartite (Fig. 4; Pl. IV). Sa participation au bord de la fenétre ovale est assez réduite (Figs 2, 10; Pls II, V).
Du carré (Q, Figs 1, 2, 3, 6, 10; Pls II, IV, V) ne subsiste qu’un fragment comportant le processus ptérygoide (pr.pt.Q).
C—Principales fosses et ouvertures de la région postérieure du crane
La fenétre ovale (fe.ov, Figs 2, 10; Pls II, V) est ici masquée par le stapes. Il est cependant possible de remarquer que ce sont essentiellement le prootique et l’opisthotique qui forment la majeure partie de son bord. Le basioccipital et le complexe parabasisphénoide y participent pour une bien plus faible part, et le
2 Cette derniére observation est a l’inverse de celle faite par Olson (1944: 11-15, fig. 6C) sur le “Thérocéphale’ qui semble étre un Pristerognathidae également. I] faut donc admettre soit que les Pristerognathidae présentent une région otique diversement consolidée, soit que la technique des sections sériées transversales fait disparaitre la suture entre prootique et opisthotique (celle-ci se trouve effectivement dans un plan pratiquement transversal). La méme remarque serait a faire chez les Scaloposauria 4 propos de la différence existant entre Ictidosuchops (in: Crompton 1955: 165-167, figs 6 A, B, C) et Regisaurus.
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 87
complexe parabasisphénoide seulement trés superficiellement. I] est a noter que chez Theriognathus (B.P.I. F.100/M.141), ce dernier os est complétement exclu du bord de la fenétre ovale qui, dans ce cas, est bordée, de fagon égale, par seulement les trois autres os (observation personnelle).
La fosse posttemporale (fo. pttemp. Figs 2, 6, 7, 9; Pls IV, VI), qui s’ouvre postérieurement et vers le haut, est vaste. I] s’agit ict comme chez la plupart des Thérocéphales d’une ‘fenétre’ plut6t que d’une ‘fosse’, car l’ouverture postérieure et l’ouverture antérieure coincident en partie. Chez le Pristero- gnathidae Pitomalestes au contraire, ouverture antérieure, minuscule, est décalée médialement par rapport a ouverture postérieure et, en vue occipitale, on ne voit a travers cette derniére que le processus intermédiaire du squamosal.
L’ouverture antérieure (Fig. 7; Pl. VI) est bordée essentiellement par Popisthotique avec une faible participation du squamosal dans l’angle latéral et une encore plus faible participation du prootique dans l’angle médial. Ces os sont bien visibles en vue antérieure étant donné la faible hauteur de la paroi antérieure du foramen ptérygo-paroccipital (Fig. 6). Chez Thertognathus au contraire, cette paroi, plus élevée, masque le bord inférieur de [ouverture antérieure de la fosse posttemporale. En outre, la participation du prootique au bord supérieur est moindre (observation personnelle).
L’ouverture postérieure (Figs 2, 9; Pl. IV) est uniquement bordée par Popisthotique et le squamosal. Les processus intermédiaire et mastoide de ce dernier sont trés réduits par rapport a ceux de Theriognathus; d’autre part, tandis que, chez ce dernier, le squamosal forme la majeure partie du bord dorsal, Pangle latéral et prés de la moitié du bord ventral de louverture postérieure de la fosse posstemporale, chez Moschorhinus (? Tigrisuchus) il n’en forme que la moitié du bord supérieur et la plus grande part de l’angle latéral.
La fosse paroccipitale (Figs 2, 9; Pl. IV) est vaste. Cette fosse existe également chez Regisaurus et Theriognathus mais est totalement absente chez Ptomalestes (observation personnelle). Elle existe également chez Annatherapsidus (cf. Vjushkov 1955: 137) et Chthonosaurus. I] serait intéressant de savoir si cette fosse est homologue de celle observée par Olson (1968: 243) chez les Pelyco- sauria Caseidae et décrite comme suit: ‘. . . Sur la face dorso-latérale de lopisthotique, se trouve une fosse profonde qui aboutit par un sillon a la fenétre posttemporale .. .’?
Le foramen ptérygo-paroccipital (fipt.paroc, Figs 1, 2, 6, 10; Pls I, VI) différe de celui de Regisaurus et de Theriognathus par la moindre hauteur de sa paroi antérieure formée par le processus antéro-ventral du squamosal et le processus central du prootique.
Le foramen magnum (Figs 4, 9; Pl. IV) est bordé essentiellement par les exoccipitaux. Le supraoccipital ne forme qu’une partie du bord dorsal tandis que la participation au bord ventral est limitée 4 une bande étroite.
Le foramen jugulaire (Figs 4, 9; Pl. IV) est bordé exclusivement par Pexoccipital et lopisthotique, comme chez Theriognathus.
88 ANNALS OF THE SOUTH AFRICAN MUSEUM
II —MANDIBULE
Des fragments de mandibule sont inclus dans un bloc contenant également des restes du squelette postcranien.
Le dentaire est surtout remarquable par son trés fort angle mentonnier tandis que chez Theriognathus, cette région forme une pente douce. Paradoxale- ment l’aspect de la région mentonniere est ainsi plus proche de celle de Moscho- whaitsta que de celle d’Annatherapsidus.
en sut.itvo \ CRVENMIVOs5 ee ee eee Ww . y Se ae Bs, \\ , y SZ \ \\ / HSCF Da \\ ff —-— Pal \ \ / \ si SEY? ) . | \\ S WN . 1 \ \ /, Ww SS oy j crvsii Ppt er ee a \\ Fa Ja SS WS ey \S¥4 LY, = = Vs \ gy = = a |) Chv-m: Pt ee ee SS a = BES = =] epee aN Gy y \\\ SS — f \
= DPB
EZ
ee
—————— TeEROHAESD
——
fh) ——————
MH
Sate KW
_— eer = Vga (308 SES:
~S WG SS
S
ae ((« SS
LA —
—/ / —/)
/ } i), y Z, ' Z Z LZ J; Sz ==
m
\\ yy
\ WN 1,
ZS SOM
if ______ Ng ___prbsph pt
" \\
SRY SS
O 1cm.
|
Fig. 5. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118. Détail de la fosse interptéry- goidienne en vue ventrale (x 5/3).
cr.v.-it.Pt, créte ventro-intermédiaire du ptérygoide; cr.v.-m.Pt, créte ventro-médiane du ptéry- goide; cr.v.-m.Vo, créte ventro-médiane du vomer; /fo.ipt, fosse interptérygoidienne; fo.sorb, fosse sousorbitaire; 0.¢.c.c.i, orifice externe du canal de la carotide interne; Pal, palatin; pr.bsph.Pt, processus basisphénoide du ptérygoide; pr.bspt.P.-Bsph, processus basiptérygoide du complexe para-basisphénoide; pr.psph.Pt, processus parasphénoide du ptérygoide; pr.q.Pt, processus carré du ptérygoide; pr.tr.Pt, processus transverse du ptérygoide; g.P.-Bsph, quille du complexe para- basisphénoide; r.Psph, rostre du parasphénoide; sut.itve, suture intervomérine; Vo, vomer.
I a TIE
—_— —- —
SPECIMEN DE MOSCHORHINUS KITCHINGI BROOM 89
cr.it.Pt fo.sorb | cr.v-m.Pt Pal
, Pr. tr.Pt pr. bsph.Pt
ge q.P-Bsph oe fe.ov \ = <— , = \_ ‘pr.q.Op “ | N tub.sph,-oc / / VA \ f.ju ! tt.orl.moy Boc / pr.in.Op \ pr.mas.Op Eoc pr-bspt.P-Bsph O 1cm. B
PLANCHE V. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118; vue ventrale de la partie postérieure du crane. A, (X 055, env.); B, (x 1).
Boc, basioccipital; cr.it.Pt, créte intermédiaire du ptérygoide; cr.v.-m.Pt, créte ventro-médiane du ptérygoide; Foc, exoccipital; f.ju, foramen jugulaire; fe.ov, fenétre ovale; fo.ipt, fosse inter- ptérygoidienne; fo.sorb, fosse sousorbitaire; 0.e.c.c.i, orifice externe du canal de la carotide interne ; Pal, palatin; pr.bsph.Pt, processus basisphénoide du ptérygoide; pr.bspt.P.-Bsph, processus basiptérygoide du complexe para-basisphénoide; pr.in.Op, processus interne de |’opisthotique; pr.mas.Op, processus mastoide de l’opisthotique; pr.psph.Pt, processus parasphénoide du ptéry- goide; pr.q.Op, processus carré de l’opisthotique; pr.g.Pt, processus carré du ptérygoide; pr.ir.Pi, processus transverse du ptérygoide; Proot, prootique; Q, carré; q.P.-Bsph, quille du complexe para-basisphénoide; Qj, quadrato-jugal; 7.Psph, rostre du parasphénoide; Sg, squamosal; Si. stapes; 77. transverse; tt.orl.moy, toit de l’oreille moyenne; tub.sph.-oc, tuberculum spheno-occipitale.
go ANNALS OF THE SOUTH AFRICAN MUSEUM
II] —sPLANCHNOCRANE
Le stapes (S#, Figs 2, 4, 7, 9, 10; Pls II, IV, V), en forme d’haltére, n’a pu étre dégagé complétement. Il n’y a pas de foramen stapédial visible en vue ventrale ou en vue occipitale. Ce foramen est pourtant vaste chez le Théro-
——— — Pa sa ad Z —_pr.d.Sq - Pd ZL02-—\pa i ll - a | = Soc yy ff fe pr.it.Sq
Za \ Be: => = se) | wm - 4 _ ap.Ept | arc.zyg_ Ze | Te SSS ji __pr.p-d.Proot |
rec.sq— eee ais eee aH ~\ DD uy \ eo. : , x y/ 5 See pr.p.-v.Proo , SE eae os + BAS ZZ _Ik prc.Proot | ce) \ SoS = fi pr.p.-v.E pt | Lj, NN SS : Z=X (ca.ept) pr.q.Pt_
i (ll SSS Z oe ee pr-o-v.Ept |
ii a aM ——__rg.basProot |
pr.a-v.Sq—
pr.pt.Q_
pr.ept.Sq
| l U Pal a ue) > a9) =<
On aterm: pr.paroc.Op tub.sph-oc Fig. 6. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118. Vue antérieure du mur pos- térieur de la fosse temporale droite, reconstruite par symétrie (X 5/3). |
a.p.Ept, apophyse postérieure de l’épiptérygoide; arc.zyg, arcade zygomatique; (ca.cb), cavité cérébrale; (ca.ept), cavum epiptericum; fo.pttemp, fosse posttemporale; Jpa, interpariétal; /.d.Ept, { lame dorsale de l’épiptérygoide; P.-Bsph, complexe para-basisphénoide; Pa, pariétal; par.a.- v.Ept, processus antéro-ventral de l’épiptérygoide; pr.a.-v.Sg, processus antéro-ventral du squa- mosal; pr.bsph.Pt, processus basisphénoide du ptérygoide; pr.c.Proot, processus central du prootique; fr.d.Sq, processus dorsal du squamosal; pr.ept.Sq, processus épiptérygoide du squa- mosal; pr.it.Sq, processus intermédiaire du squamosal; pr.p.-d.Op, processus postéro-dorsal de l’opisthotique; pr.p.-d. Proot, processus postéro-dorsal du prootique; pr.p.-v.Ept, processus postéro-ventral de l’épiptérygoide; pr.p.-v.Proot, processus postéroventral du prootique; pr.paroc.- Op, processus paroccipital de Vopisthotique; pr.proot.Sq, processus prootique du squamosal; pr.pt.Q, processus ptérygoide du carré; pr.g.Op, processus carré de l’opisthotique; pr.g.Pt, pro- cessus carré du ptérygoide; q.P.-Bsph, quille du complexe para-basisphénoide; rec.sq, recessus squamosal; rg.bas.Proot, région basale du prootique; Soc, supraoccipital; St, stapes; tub.sph.-oc tuberculum spheno-occipitale ; XX’, plan de la section parasagittale en Fig. 8.
CRANE D'UN JEUNE MOSCHORHINUS KITCHINGI gi
Yb
SQ
WE
f.ven
pr.a-v.Ept Ptorb
fe.temp ~~
\ f.pt-paroc
\ 0 fern pr.a.-v.Sq
orate B
PLANCHE VI. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118; vue dorso-latérale de la fosse temporale gauche. A (xX 0,55 env.); B, (Xx 1).
a.p.Ept, apophyse postérieure de l’épiptérygoide; Eoc, exoccipital; Ept, épiptérygoide; f pt.-paroc, foramen ptérygo-paroccipital; f.ven, foramen veineux; fe.temp, fenestra temporalis; (fo.paroc.Op), fosse paroccipitale de l’opisthotique; Fr, frontal; Jpa, interpariétal; Pa, pariétal; pr.a-v.Ept, processus antéro-ventral de l’épiptérygoide; pr.a.-v.S¢q, processus antéro-ventral du squamosal ; pr.c.Proot, processus central du prootique; pr.d.Sq, processus dorsal du squamosal; pr.7t.Sq, processus intermédiaire du squamosal; pr.q.-v.Ept, processus postéro-ventral de |’épiptérygoide; pr.p.-v. Proot, processus postéro-ventral du prootique; pr.g.Pt, processus carré du ptérygoide; Proot, prootique; Ptorb, postorbitaire; Soc, supraoccipital; Tab, tabulaire; V2 + V3, lieu de sortie des branches maxillaire et mandibulaire du nerf trijumeau.
92 ANNALS OF THE SOUTH AFRICAN MUSEUM
céphale B décrit par Olson (1944: 41; fig. 12 A, E). A ce propos, Cys (1971: 124) a fait la remarque suivante: ‘A part le Thérocéphale B (Olson 1944: 39-41), tous les Thérocéphales et Bauriamorphes chez lesquels le stapes est connu n’ont pas de foramen stapédial. Les Cynodontes et les Gorgonopsiens cependant, possedent un foramen stapédial’. Les relations exactes du stapes et carré sont impossibles a connaitre ici.
Sa Pa pr.p-d.Op, 3 ___ Bed'sq MS ; Ps a) Ipa ey ae A |. 7a - LY . _ 2 Soc ee pr.it.Sq SS pr.p.-d.Proot J pr.p.-v.Proot __________pr.c.Proot | ae:
tub.sph-oc
Fig. 7. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118, Vue antérieure du cété droit de l’occiput, reconstruit par symétrie (x 5/3).
arc.zyg, arcade zygomatique; fo.pttemp, fosse posttemporale; Jpa, interparictal ; lv. fe.ov, levre de la fenétre ovale; P.-Bsph, complexe para-basisphénoide; Pa, pariétal; pr.a.-v.Sg, processus antéro- ventral du squamosal; pr.c.Proot, processus central du prootique; pr.d.Sq, processus dorsal du squamosal; pr.it.Sq, processus intermédiaire du squamosal; pr.p.-d.Op, processus postéro-dorsal de Vopisthotique; pr.p.-d.Proot, processus postéro-dorsal du prootique; pr.p.-v.Proot, processus postéro-ventral du prootique; pr.paroc.Op, processus paroccipal de l’opisthotique; pr.q.Op, processus carré de l’opisthotique; g.P.Bsph, quille du complexe para-basisphénoide; rec.sq, recessus squamosal; rg.bas.Proot, région basale du prootique; Soc, supraoccipital; St, stapes; tub.sph.oc-, tuberculum spheno-occipitale.
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 93
Ept fven Proot ipa \ if eh
Mx Pal.g NO pf, | eet ns
/ .
z — Qj 1D -—pr.a-v.Sq Ps -0. ——UPp Pmx — i =fju Vo—-- =|E0C.g EIEOC:Gt \Boc cs ul " \ ve f.mx.-p Pal.dt + dt pr-tr.Pt.g \ slt ‘Proot i r \ \ c.smx x pr.tr.Pt.dt q-P.-Bsph f.pmx.d — Sq ihe — fo.mx.-smx J ——La du. > ms QO ‘4cm
PLANCHE VII. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.1 18; A, vue latéro-ventrale gauche du crane (x #); B, vue antérieure du museau ( $).
Boc, basioccipital; C.1, canine inférieure; C.s, canine supérieure; c.smx, canal septomaxillaire; Eoc.dt, exoccipital droit; Eoc.g, exoccipital gauche; Ept, épiptérygoide; f.ju, foramen jugulaire; f-mx, foramen maxillaire; fimx.-p, foramen maxillo-palatin; f.pmx.d, foramen prémaxillaire
droit; f.ve, foramen veineux; fo.mx.-smx, fosse maxillo-septomaxillaire; J', premiére incisive | supérieure; /°, cinquiéme incisive supérieure; [pa, interpariétal; Fu, jugal; La, lacrymal; Mx, ; maxillaire; Va, nasal; nar, narine; Op, opisthotique; Pa, pariétal; Pal.dt, palatin droit; Pal.g, | palatin gauche; pC, précanine; Pfr, préfrontal; Pmx, prémaxillaire; pr.a.-v.Sq, processus antéro- ventral du squamosal; pr.qg.Pi, processus carré du ptérygoide; pr.tr.Pt.dt, processus transverse } du ptérygoide droit; pr.tr.Pt.g, processus transverse du ptérygoide gauche; Proot, prootique; | Q, carré; g.P.-Bsph, quille du complexe para-basisphénoide; Qj, quadrato-jugal; s/.t, selle turcique; Smx, septomaxillaire; Soc, supraoccipital; Sg, squamosal; T7.dt, transverse droit; Tr.g., transverse gauche; Vo, vomer.
04. ANNALS OF THE SOUTH AFRICAN MUSEUM
oe f.pt.-paroc | fo.paroc.Op oe _pr. proot.Sq pr.paroc.Op —- — p-q-Pt O 1cm.
Fig. 8. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118. Coupe parasagittale, selon l’axe XX’ de la Fig. 6 (Xx 5/3).
S-pt.-paroc, foramen ptérygo-paroccipital; fo.paroc.Op, fosse paroccipitale de lopisthotique; fo.ptiemp, fosse posttemporale; Pa, pariétal; pr.d.Sq, processus dorsal du squamosal; pr.it.Sq, processus intermédiaire du squamosal; pr.p.-d.Op, processus postéro-dorsal de Vopisthotique; pr.paroc.Op, processus par- occipital de l’opisthotique; fr.proot.Sq, processus prootique du squamosal; pr.q.Pt, processus carré du ptérygoide; St. stapes;
Tab, tabulaire.
IV — DENTITION
La formule dentaire est 5 I, 1 pC, 1 CG, 3 PC. Les Moschorhinidae pré- sentent la réduction de la série postcanine si courante chez les Pristerosauria et qui atteindra son maximum chez les Whaitsiidae et chez Euchambersia. La formule antécanine est intéressante par le fait qu’elle est encore relativement primitive avec 5 I et 1 pC. Cette formule se retrouve chez Moschowhaiisia (cf. Tatarinov 1963: fig. 2) et chez Annatherapsidus (observation personnelle). Vjushkov (1955) considére par contre que ce genre ne présente que 4 I et 1 pC, mais reconstruit le museau de Chthonosaurus avec 5 I et 1 pC. Chez les Whaitsiidae africains la précanine a disparu, mais deux canines en croc a fonctionnement alternatif subsistent (cf. Brink 1948: 31-34).
DiscussIon
Par son palais, et ses régions pariétale et épiptérygoidienne, Moschorhinus semble se rapprocher plus des Annatherapsididae d’U.R.S.S. que des Whaitsii- dae, mais il s’en éloigne par l’aspect de son dentaire essentiellement. Les caractéres qui nous ont paru intéressants a retenir sont les suivants:
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI
a |
— 91)9S0} OTNOsnuTUT oun .p s9p3091d STR SNYJDUGOLAY J, ep InjI0 op UISIOA soi} yOodse e—
jesnf 9] 1ed sopur10F JUOUUO] [91] UISsa
sossred9 yuNTIOUUDAOUL
sonbsew sutures -jsod U01}911]su09 Be — ouusAoul Inonsuo] sp—
(gL :&961
AOUIIVIET, JO) 7's Snjoyasajst) & JUOZ eT & juoyeambg9 9.139-nod IMs1I9dNs Udt1e}7e J, ‘AJ 9u0z
:Inotedns ustwi9g
oyney — gyej91s1ed-oj U0 INNS LT
JURAY |] & JULUSIO}e —
jesnf 9] zed sopuist07 JUOUI9T [91] Uassa
JUIUIOULYe pp 910A Ud
sonbsrew sutueo -jsod uoro11jsu09 e— suudAoul ImMonsuol op —
“45 snpoyqasajsy) & JUOZ
:oyeaysne onbrayy — UONeCUIIO |
esuIMey :o1uezue y —
:adMorsgdns uoIuIag
DISJIDYMOYISOPT snyjoudorsay J, ssa onbryy AVGISLIVHM
oyney JUsUIOUUDAOW —
[eqguied 9] enb 931n09 snjd —
oITeT]IXeUL np uoled -edronaed 93.10F soy & sassred9
senbieur nod suruvs -jsod uod11}su09 e—
1INOO —
SNANDSOMSUT 49 snyvyqgas0jquqy % SOUOZ saMortsigdns uowIag
(snyansiio27, ¢) SNUIYLOYISOJAT
geaysne onbrayy
avd
-INIHYOHDOSOW
juowsiqeqoid [e19ueg é —| 9ney JUSWIOUUOAOUI—} anoyneYy g[ejg1ied [eiotred jeqgrsed g[ei18¢es aj onb 931n09 snjd—]} = 9] onb 934n090 snjd—} anonsuo'T 94919 SSUO]T sjorq xXNeUOLT
jesnf oy 1ed saguri0fF
yusumaye}0} onbsoid
soyqey so
¢ — gsuo]e zosse juowls;qeqoid —
uolseye [, “AJ 9u0z :Inorsgdns ustwI9g
jesnf oy red saguri0j; yuoursye10} onbsoid
suusAour Inassted9,p suruvojsod
UOT}ITIISUOD Op sed — suusAour
iMmonsuo] sp —
UdTIVJe Ty, ‘AT 9UOZ :dnowiodns usrulIag
snpisqvsayjvuup = snanpsouoyjy’) puUup ssa sepipisdesoyjyeuuy = aAVGCINNV
I OVATEVL
SOITEIQIOSNOS sorIEg
soi1ej1q.10}s0d. yo sonbyyewoshz sopro1y
neasnyjy
ANNALS OF THE SOUTH AFRICAN MUSEUM
oureyppceur -O19WI0A juOd 3] red sogied9s
sgnbieur usIq
‘oxreyxeur -o19UI0A yUOd 9] red sogiedgs
sonbsreur ustq
xnasso oirepuodses srieyed op inqep Syuose1d
97muT}U0D us
SOUIULD Sop NVIATU 3] JueAL,] B sed queUusio}e,u ‘ossnoul
sonbreut qyuowouUsAOUL
93muruoS us
yeueoyo DOYIIO 39 SOUTULD SOSSO}
Seed SUDUDOYI D4S119
QITE][IXVUI-OI9WIOA 49%]U09
i, | | | | | SD
ojerod -UI9} J1]QUOJ eT sUep gsodxo jusuIOUUeAOUT
oje10duis 91]QUO] VT suep gsodxo JUSUIOS Ie]
oouIge onied uo 4s9 UOIS9I 91199 red ‘oruL9p
ew oye10dui9} 91}Q9UNF el & uoledionsed
es | fa | |
JIQeiszea IIe} Op o}uIz
9110119 9309} oun % jUepuUOdso1109
jesuid UdUIeIOF
—— ee rrernnnEnEEEEEnEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEeEEee! i | |
ou10joie]d 91399 op souasqe
(gL :€Q61
AOULIEIL], *J9) “7's Snyoy gargs) @ UOZ eT & yuoyeambs 2339-ynod Imorigdns usts1e7e T, ‘AI 2u0z
samotigdns usrutiog
a | | ES | |
a |S |
our10jo}e[d 97199 op douasqe
1s snyoyqasajs1y) @ QUOZ
soyeajsne onbuyy —
UOT} CULIO TF
esuImMey :o1uezue Ty, —
sdnorgdns uow19g
DISJIDYMOYISOPAT snyjougorsay J, ‘sswaA onbiisy AVGIISLIVHM
gyeises 93919 e] WP JULAL] & IIILT -nsueis} sULI0JO} ed
SNINDSOMSCT @ 49 snyoyg2s0jqocy @ SXUOZ ;Inorigdns ustulIog
(snysnsusiy, ¢) SNUIYLOYISOIT
gjeajsne onbripy
avd -INIHYOHDOSOW
Q[eIWISes 93910 e] op JULAL] & IIe] -nsuers} su107;9}e[d
UdTIe}VT, ‘AT 2UOz :amotigdns ustulag
sninpsouoyjy")
gyeises 93919 el Wp JULAL] & IITeT -nSuels} oWUII0JO}e[d
uarie}e J, SAT 9U0Zz :anori9dns ustulIog
snpisqosayjouup = puupy
‘Ssan
sepripisdesoy}euuy = aVCINNV
oamori9}ue uoIs91
Dil
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI
Cue |
9710119 $21] 97U9F 9uN Ud JUBA | & S9119ANO
Q[PAO 9I]QU9F e] 9p prog np njoxo —
0104191990 -ouaygs pjna4aqny Sap JURAL,] @I101}9 $91] —
oprou -9ydseied np 9)s01 ne j10ddei red JUDULI[VIJUIA SOQUIIIT
o[BAO 91]QUIJ LV] 9p prog ne yuowe|qrey odrion1zed —
01j0j1q1990-ouayds vjna -199N] SOP JUCAL | & 310.139 JUsUIOUUIAOU —
Q[VAO 91]QUIJ V] Op 2p p10q ne odionsed —
p1jpj1qi990-ouayqs pnd -19QN] SOP JURA | B 10139 JUOWISUUAAOUW —
Q[VAO 91}QUDJ PT op p10q nv odrons1ed —
xXNOUTUINIOA
$9J} Tor yUOs mb
p1jnj1qi990-ouaygs 0nd
-19QN] SOP JUCAR | B 310179 yUSUIDUUOAOW —
oplousydsiseq-ered oxo[durory
JULAL | B JUDUIOTNIS $9}.19ANO
————————————————— OO (oe | |__|
sojyussqe — sjtqutol yUsUIITe10} onbsoid saptou -9ydsiseq snssao0i1d — Osnosso SULT oun sed s91jo1 uou asI9ASUeI} snssod01d 39 91189 snssa00i1d —
sopuvis
UIIOJIONAD amatigjue onbeyd e—
o[esiop
snjd osjtered y1e7
e, mb 90 ‘“sonbieut
aimor19}sod onbeyd e] op uosreurpour —
9110139 19 onsuoT SULIPZUI UOIS9I B
soyuosgid —
sjqutof saprou -9ydsiseq snssoooi1d —
OSNossO OUP]
oun sed sotjo1 uou
dsIOASUeI} SNsso001d 39 91189 snssoo01d —
soyuosqe yuswo1edde —
$911209 soplou -9ydsiseq snssao0i1d — dsIOASUeI} Snsso00i1d 39 91189 snsso901d 3T JULI[II ISNISsO DUILT oul oun p sduasgid —
soyuosg1d — s911@99 $91} Soplou -9ydsiseq snssoooid — dsIOAsSUeI} snssa901d 39 9119 snssa00id 9] JULT[OI OSNIsso OUP] sug sun ,p sdues91d —
souusrpiosh19}d
soyuosgid — syuop syurof
yuoWoT[I171ed sopiou
-9ydsiseq snssoo01d —
dSIBASULI] snssov01d
32 91189 snssooo1d oF
JULI[II ISNISSO JUILT
ouy oun_,.p s0ues91d —
soprosA191q
—_e—————————e—e ee eee") oO — | ses ___...__._._._._.............._._.____.___!___..__.________..__.._...
soy[NuU & solINp9.1 op
ULIOFIONAD oinoti9j3ue onbey;d e—
o[esiop
snjd osjrered 3123
ey nb 99 ‘ognbieur
oinot19}sod onbe|d e] Op UosreurpouI—
9110139 19 onsuOT QULIPPUL UOIdAL &
sopuvis
OITE][IXCUL 9]
DIA 19"}UOD suLS
*{l@JUIAP Ud 99TR19 oimoti9jue onbeyd e—
ueyd ouigut 9] ans soid nod e
ainot193sod onbeyd 49
oimori9jue onbeyd e—
oBIR] 19 9AIIq. JULIPPUL UOIS9I
sopuvis
c
c
sopurisd SOITL}IGAIOSNOS SOSso.T
OITET[IXCU OT DAB 19%]UOD sues *TIWJUDAQ Ud 991219 oimoi9}ue onbejd e— SIOQUIOA, ued ouigur 9] ans soad nod ve aimori9}sod onbeyd 49 oimorigyue onbeyd e—
OBIT ‘ansuoy] JusWIsUUOAOUL QULIPZU UOISZI &
SOIT] [IXCUIDIG
ANNALS OF THE SOUTH AFRICAN MUSEUM
a Vv 34 4Q3ee""4 "*~* eon” ”
Cre
rr (| eee
——— | |
(gL :S961
AOULILIET, JO) 7's snjyoygarajsy') @ DUOZ e] B quoyeamb9 9139-ynod Inotigdns ustieye J,
‘AI 9u0z
:Inotigdns usluiiog
| | | |
sio1] un 1nod undvyo ‘[eqI1dr1d90Iseq yo onbrnoyjsido ‘anbrj01d :so
sto1} red s9p10q
sjoiq
xNv19}e] snssad01d eB —
anbnooid snssas01d un 39 d1not19}sod
oshydode sun e —
fa) Fo 02) B) JUOUI9}1OJ $9.1}
‘orneide onbeyd ua —
7's snpoyqasajst) @ JUOZ
:o[erysne onbrayy —
UO] eUulIOy
eSUIMeY :o1Uezue y, —
:anoriodns uswiiog
DISJIDYMOYISOPT snyjousoisay J, ‘sswn onbigy AVATISLIVHM
oprousydsiseq-eied axo[duroo 32 ye1Id19090 -Iseq 9[QUIZsUd |
red 9urz0j 1Ue}9 S191} o1jne J £s1o1) un nod unoeyo ‘onbrjoyjstdo | yo onbrjoo.d 9]
>So o1jenb red a9pi0q
sjouq
xXNV1I91"] snssod01d e —
oanori9}sod
osAydode oun ve —
fe) F864) quoWIoUUDAOUL
‘arye[de onbejd ua —
SNANDSOLISUT @ 49 snjoygas0jqoq & Sou0z :morigdns usIulIEg
(snysnsusiy, ¢) SNULYLOYISOPAT
gyejsne onbriyy
avd -INIHYOHDSOW
SOUIL}IOOUI JUOS SO SoI]Ne Soy DIAB S9JDVXO SUOTILIOI SOS JO UOTSII 93999 suep 9seurlu0pus
$21} 3s9 [eIdrID90I1seq 3] Juepuodso oprousydsiseq
-eied oxo[dwiod 32 [eqdroso1seq ‘onbrj0y} -stdo ‘onbrj}00.1d :so o1yenb red a9pi10q
WOs 9]]9, nb a,quuos |r
sjorq
UdIe}Z J, “AT 9UOZ : Motigdns USIULIOg
oprousydsiseq
-eied oxoyduios 3a ‘jeqdio01seq ‘onbr} -oystdo ‘onbrnooid >So oujenb zed a9pi0q
—_—__rrr | | fa
spsuo][e Zzasse xnv191"] snssooo0i1d e —
o9seul
-WIOpUd 189 UOIS91
9}}99 STRUT JINOTI
-93sod asAydode oun ye A ]imb ayquzas [I —
S1B1e]9
qusuIDUTO AOU ‘arne[de onbeyd ua —
UsTIeIeT, “AT 9u0z : Motigdns UdIULIOg
snpisqosayyouupy = snanvsouoy}y’) puup Sswn sepipisdeisyieuuy = AVCINNV
Q[VAO 91}9U9,T
[e1dios0vidng
aprosA19}d1dq
99
Lore ) a = li 2) B — mM 5 L fy) & 9 9 SOUTULI}SOg a | | O_O y z 3 I I I souluery) (—, Ce a I Oo I I I quIUuLozIg mM —— |S eee | O_O O s osAyduths ea el op oseq eT g o19T[ NBII é eg o}uod op oanjdni e o[Nqrpueyy oe gnbieutr 39 90nop 93u0Ed ua gnbieu so7} steur ‘sonop oyued ua 7 JoUUO}UIU o[Suv Y | DIQIUUOJUDU UOISZ1 & JaIUUO}UNU o[sue 91QTUU0}USUL UOTS91 & ES eS I Sn (Se eer ee es ee eer al Oa et) eee A (gL :€961 "1's 2 AOUIIEIET, JO) “7S | snzoydasajst) e au0z a snjoygasajs1) @ SUOZ ey eB :9[eajsne onbrayy — SNAnvsouseTy @ 49 uolieye y, SAT 9u0z ualieyey, “AT 2u0z O quoyeainbg 9139-ynod UOI}CULIO,T snjoyqas0jquq e Sauoz : IMotigdns ustwWIIg :amori9dns UustulIog dmorigdns ustiejey, | esuUIMey :oruezuey—| :anotigdns ustwiiag SAI 2u0z :amori9dns ustur1og :anorigdns usrwi1og (snymnsusty ¢) Snpisqosayjouupy = DISIIDYMOYIS OPT snyjoudorsay J, SNUIYLOYISOTAT snnosouoyjy’) puup ‘sswa onbiyy gyeqisne onbiyy ‘sswa AVG seprpisdesoyyeuuy =
AVCTISLIVHM “-INIHYHOHOSONW AVCINNV
— - ern
100 ANNALS OF THE SOUTH AFRICAN MUSEUM
I. COMPARAISON ENTRE LES MOSCHORHINIDAE ET LES DEUX AUTRES FAMILLES
A—Moschorhinidae et Whaitsiidae
D’aprés le tableau précédent on remarque que les Moschorhinidae différent des Whaitsiidae d’Afrique d’une part, par la rétention de caractéres structuraux primitifs et d’autre part, par des éléments correspondant a des insertions musculaires. La rétention de caractéres structuraux primitifs touche surtout la région du palais (absence de contact voméro-maxillaire correspon- dant a un début de palais secondaire, et donc fosses canines et orifice choanal en continuité, crista choanalis bréve et mousse, présence de grandes fosses sous- orbitaires), l’'endocrane (épiptérygoide plus étroit, incomplete séparation des sorties des branches maxillaire et mandibuiaire du nerf trijumeau, fosse inter- ptérygoidienne ouverte, fenétre ovale bordée par quatre os) et la denture maxillaire (rétention d’une précanine et de quelques postcanines, bien que réduites). Les différences d’attaches musculaires s’observent au niveau des ptérygoides, des pariétaux et du complexe para-basisphénoide. Cependant, les
Soc fo.pttemp Tab
| fastelal ori eesee | pr.p.-d.Op i pr.it-Sq al vis
(Be A
SN = =
TTT
==) — tt-orl.moy
a =
<2
Z = A
SS =
pr.paroc.Op lv.fe.ov
pr.q.Pt
1cm. tub.sph.-oc
Fig. 9. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118. Détail de la partie inférieure du coté droit de locciput, reconstruit par symétrie (X 5/3).
Boc, basioccipital; cdy.oc, condyle occipital; Eoc, exoccipital; f.ju, foramen jugulaire; fmg, foramen magnum; fo.paroc.Op, fosse paroccipitale de l’opisthotique; fo.pttemp, fosse posttemporale; lv. fe.ov, levre de la fenétre ovale; m.ant.pr.paroc, mur antérieur du processus paroccipital; pr.in.Op processus interne de l’opisthotique; pr.it.Sqg, processus intermédiaire du squamosal; pr.mas.Op, processus mastoide de l’opisthotique; pr.mas.Sq, processus mastoide du squamosal; pr.p.-d.Op, processus postéro-dorsal de l’opisthotique; pr.paroc.Op, processus paroccipital de l’opisthotique; pr.q.Op, processus carré de l’opisthotique; pr.q.Pt, processus carré du ptérygoide; Q, carré; q-P.-Bsph, quille du complexe para-basisphénoide; Soc, supraoccipital; Sg, squamosal; St, stapes; Tab, tabulaire; tt.orl.moy, toit de l’oreille moyenne; tub-sph.-oc, tuberculum spheno-occipitale.
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI IOI
Whaitsiidae ont conservé deux canines tandis que les Moschorhinidae n’en présentent plus qu’une. Par ailleurs, la mandibule des Moschorhinidae pré- sente une région mentonniére trés anguleuse tandis que celle des Whaitsiidae africains est en pente douce. Moschorhinus (? Tigrisuchus) différe également de Theriognathus par un museau court a constriction postcanine peu marquée, a arcades zygomatiques et postorbitaires et a barres sousorbitaires épaisses. I] s’en rapproche par contre par l’aspect du bord de la fenétre ovale. Les Moscho- rhinidae différent moins du Whaitsiidae d’U.R.S.S. dans la mesure ot celui-ci a conservé certains caractéres primitifs tels qu’une grande fosse sous-orbitaire, la présence d’une précanine et de postcanines. Ils s’en rapprochent également par l’épaisseur des arcades et la forme du menton.
I] faut noter que les palais des Moschorhinidae et des Whaitsiidae ne diffe- rent pas seulement par la rétention ou la disparition de caractéres primitifs
N
NS. % WN \Y
RZ GYUb—>F—Ix : pr.bspt.P.-Bsph__S\\!_' —SS \ Uy = = : N \
ZN W q.P-Bsph__— 2) “4 i
pr.bsph.Pt ee
P-Bsph_{S=_ ° pr.c.Proot— ax \\ XS
AW See z =\\ YS —Ppr. otS 4 = Ua
\ = tub.sph-oc <- a ae, MLN oe
un \ \
j
®
4
|
| YY,
‘/ if
of
pr.in.Op f.ju
coe Yas f.pt-paroc| Tab = ‘e) 1cm Prparoc.Op pr.mas.Op \ Sq tt-orl.moy
Fig. 10. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118. Détail de la vue ventrale de la partie postérieure du cété droit du crane, reconstruit par symétrie (x 5/3). Boc, basioccipital; Hoc, exoccipital; Ept, épiptérygoide, f.ju, foramen jugulaire; /-pt.-paroc, foramen ptérygo-paroccipital; P.Bsph, complexe para-basisphénoide; pr.bsph.Pt, processus basisphénoide du ptérygoide; pr.bspt.P.-Bsph, processus basiptérygoide du complexe parabasi- sphénoide; pr.c.Proot, processus central du prootique; pr.in.Op, processus interne de l’opistho- tique; pr.mas.Op, processus mastoide de l’opisthotique; pr.p.-v.Proot, processus postéro-ventral du prootique; pr.paroc.Op, processus paroccipital de l’opisthotique; pr.proot.Sg, processus prootique du squamosal; pr.g.Op, processus carré de l’opisthotique; pr.q.Pt, processus carré du ptérygoide; Proot, prootique; Q, carré; q.P.-Bsph, quille du complexe para-basisphénoide; Sg, squamosal; St, stapes; Zab, tabulaire; tt.orl.moy, toit de Voreille moyenne; tub.sph.-oc, tuberculum spheno-
occipitale.
102 ANNALS OF THE SOUTH AFRICAN MUSEUM
mais encore par une spécialisation différente: il en est ainsi de aspect en éventail du vomer des Moschorhinidae.
B—Moschorhinidae et Annatherapsididae
On voit que la plupart des caractéres qui éloignent les Moschorhinidae des Whaitsiidae les rapprochent au contraire des Annatherapsididae. Par cer- tains caractéres, ils se rapprochent plutot de Chthonosaurus: par Vaspect de la fosse interptérygoidenne et du supraoccipital; mais les caractéres qui les rapprochent de Annatherapsidus sont plus nombreux: briéveté des frontaux, aspect de la crista choanalis, dimensions de la quille du complexe para-basi- sphénoide. Il faut noter toutefois que la région antérieure du palais de Chthono- saurus n’est pas connue. |
Les Moschorhinidae different des deux genres d’Annatherapsididae par la briéveté du museau, la présence d’un foramen pinéal, peut-étre absence de dents ptérygoidiennes et la faible participation du complexe para-basisphénoide au bord de la fenétre ovale.
Moschorhinus (?Tigrisuchus) differe également d’Annatherapsidus par des caracteres qui ne sont pas connus chez Chthonosaurus tels que la brieveté de la region médiane des prémaxillaires sur la face ventrale et aspect anguleux
ig Id-Ept
ma
a pr.a.-v.Ept
Fig. 11. Moschorhinus kitchingi, Moschorhinidae, S.A.M. n°K.118. Vue latérale de l’épiptérygoide droit, reconstruit par symétrie (x 5/3). a.p.Ept, apophyse postérieure de |’épiptérygoide; fiven, fora- men veineux; l.d.Ept, lame dorsale de l’épiptérygoide; Pa, pariétal; pr.a.-v.Ept, processus antéro-ventral de |’épiptéry- goide; pr.p.-v.Ept, processus postéro-ventral de l’épiptéry- goide; pr.q.Pt, processus carré du ptérygoide; proot, prootique.
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 103
de la région mentonniére. La différence d’aspect de cette derniére région est moins grande entre Annatherapsidus et Theriognathus.
II—cOMPOSITION DES WHAITSIIDAE ET DES MOSCHORHINIDAE
Une discussion détaillée concernant chacune de ces familles devant paraitre ultérieurement, nous ne présenterons ici qu’un bref résumé.
A—Les Whaitsiidae
Parmi les Therocéphales, les Whaitsiidae ont l’exclusivité d’un palais secondaire osseux rudimentaire, a vomers cruciformes en contact avec les maxillaires dont les processus vomerins sont suivis d’une crista choanalis marquee. Ils ne partagent qu’avec les Euchambersiidae la tendance a la disparition des fosses sousorbitaires et des postcanines. I] faut noter cependant que la tendance a la réduction de la série dentaire maxillaire postcanine est une caractéristique commune aux [hérocephalia Pristerosauria dont ces deux familles font partie. D’autres caracteres ont été cités dans la diagnose établie par Tatarinov (1963: 76-77); soit ‘un crane bas et large, un museau a forte constriction postcanine, Pabsence de postpariétal’. Chez les Whaitsiidae la fosse interptérygoidienne tend a se fermer et la canine a disparaitre. En outre, cette famille conserve
pr.c.Proot Soc | ld weal yi éc.d ite nee 2 | eee Ge __-Ept coe ee Pie Ne cs pr.p.-d.Proot —— 3: =o eee (fo.pttemp) ____ wwe = ——~—a.p-Ept Pr.p-v.Proot_—_—___ eS = Z —_>~~ pr.a-d.Proot (fo.paroc.Op)—__ 12 = Le of Co~ Sy prprootSq—__£&. Kage Sif Ws 2rv3 pr.paroc-Op— ZA i a __— pr.a-v.Proot lv.fe.ov__ EE NN |
@) icm.
Fig. 12. Moschorhinus kitchingit, Moschorhinidae, S.A.M. n°K.118. Vue latérale du prootique droit, reconstruit par symétrie (x 5/3).
a.p.Ept, apophyse postérieure de 1’é piptérygoide; éc.d, échancrure dorsale; Ept, épiptérygoide; f-ven, foramen veineux; (fo.paroc.Op), fosse sanacorsiialle de l’opisthotique ;( fo.pttemp), fosse soxtienjsarlle: l.d., lame dorsale; lv.fe.ov, lévre de la fenétre ovale; P.-Bsph, complexe para- pasion x pr.a.-d.Proot, processus antéro-dorsal du prootique; /r.a.-v.Proot, processus antéro-ventral du prootique; pr.c.Proot, processus central du prootique; pr.it.Sqg, processus intermédiaire du squamosal; pr.p.-d.Op, processus postéro-dorsal de l’opistho- tique; pr.p.-d.Proot, processus postéro-dorsal du prootique; pr.p.-v.Proot, processus postéro-ventral du prootique; pr.paroc.Op, processus paroccipital de Vopisthotique; pr.proot.Sg. processus prootique du squamosal; rg.bas, région basale; s/.t, selle turcique; Soc, supraoccipital; V2 + V3, lieu de sortie des branches maxillaire et mandibulaire du nerf trijumeau.
104 ANNALS OF THE SOUTH AFRICAN MUSEUM
deux canines a fonctionnement alternatif. Malheureusement la région posteé- rieure du crane des Whaitsiidae Moschowhaitsiinae n’est pas connue, ce qui limite la définition de la famille.
En l’état actuel de nos connaissances, les Whaitsiidae se réduisent a deux genres correspondant aux deux sous-familles créées par ‘atarinov.
1. Moschowhaitsinae
Cette sous-famille se distingue des Whaitsiinae par la présence d*‘une précanine’ et de ‘nombreuses postcanines’ (7 chez le seul genre connu), |’épais- seur moyenne des arcades zygomatiques et postorbitaires, la ‘grande taille des fosses sousorbitaires’, la persistance d’une étroite ouverture dans la partie antérieure de la ‘fosse interptérygoidienne’, absence de foramen pinéal et de dents ptérygoidiennes; un seul spécimen est connu, il serait nécessaire d’en avoir plusieurs pour étre certain de la validité de ce dernier caractere. D’autre part, la mandibule est caractérisée par un angle mentonnier fortement marque. (Les caractéres indiqués entre guillemets faisaient partie de la diagnose et de la liste comparative établies par ‘Tatarinov (1963: 78).)
Moschowhaitsia vjuschkovt ‘Tatarinov: Permien supérieur d’U.R.S.S., partie superieure de la zone IV (ou Tatarien); ‘équivalent possible de la zone a Cistecephalus d’ Afrique du Sud’ selon Tatarinov (1963: 78).)
2. Whaitstinae
Cette sous-famille est caractérisée par absence totale de pré et postcanine (il n’y a aucune raison de penser que le spécimen considéré jusqu’a présent comme le type de Alopecopsis ait possédé, a un plus jeune stade, des postcanines a chaque machoire), laffinement des arcades postorbitaires et zygomatiques, la réduction ou la disparition des fosses sousorbitaires (il peut y avoir chez le méme spécimen une fosse réduite d’un cété et pas de fosse de autre), la fermeture de la fosse interptérygoidienne chez l’adulte, la présence de foramen pinéal et de dents ptérygoidiennes. La mandibule présente une région mentonniére en pente douce.
D’autres caractéristiques ont pu étre observées dans cette sous-famille dans une région qui n’est pas connue chez les Moschowhaitsiinae. Ainsi, les Whaitsiinae ont lexclusivité, parmi les Thérocéphales connus, d’un épiptéry- goide extrémement élargi dont le bord postérieur présente a la fois une apophyse postérieure et un processus prootique. D’autre part, la fenétre ovale des Whait- siinae n’est bordée que par trois os: le prootique, l’opisthotique et le basioccipi- tal, a parts égales.
Theriognathus microps Owen: Permien supérieur d’ Afrique.
Theriognathus Owen, 1876 est connu en Afrique orientale et australe:
— ‘Tanzanie: zone a Cistecephalus s.l. (cf. Huene 1950: 105, 124), Kawinga Formation (cf. Charig 1963: 52).
— Afrique australe: zone a Cistecephalus s.l. (existe dans la zone a Cistecephalus s.§., mais est particulitrement abondant dans la zone a Daptocephalus selon
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI I05
Kitching (information orale, septembre 1970).
Broom (1932: 90) fut le premier a4 reconnaitre l’appartenance du type aux Whaitsiidae.
B—Les Moschorhinidae
Jusqu’ici nous avons traité des ‘Moschorhinidae’ et des “‘Annatherapsididae’ comme de deux familles indépendantes; cependant, étant donné le nombre de leurs caractéres communs, il nous semble préférable de les considérer comme deux des trois sous-familles (la troisitme étant les Euchambersiinae) de Moscho- rhinidae au sens large. A ce sujet, il est intéressant de noter que Watson & Romer (1956: 72) considéraient les Annatherapsididae comme des “Theroce- phalia incertae sedis (cf. Whaitsiidae primitifs)’ 4 une €poque ot ce dernier terme recouvrait a la fois les Whaitsiidae et les Moschorhinidae d’Afrique.
Les Moschorhinidae sont essentiellement caractérisés par |’élargissement en éventail de la plaque antérieur du vomer. Ils présentent en outre une plate- forme triangulaire sur la face dorsale du pariétal, a l’avant de la créte sagittale, un €épiptérygoide moyennement élargi, une fenétre ovale entourée de quatre os: le prootique, lopisthotique, le basioccipital et le complexe para-basisphénoide, ces deux derniers de facon moindre. La constriction du museau 4 l’arriére des canines est tres peu ou pas marquée. La fosse interptérygoidienne est ouverte. Le processus carré et le processus transverse du ptérygoide sont reliés par une fine lame osseuse. Il n’y a pas de postfrontal. Cette famille conserve la précanine, tandis qu’elle ne posséde qu’une canine fonctionnelle.
1. Annatherapsidinae
Les Annatherapsidinae se distinguent des deux autres sous-familles de Moschorhinidae par la longueur du museau, l’absence de constriction post- canine, la formule dentaire maxillaire la plus compléte (1 pC, 1 C, 6 PC). C’est dans cette sous-famille que la forme en éventail de la plaque antérieure des vomers est la moins marquée et la lame entre le processus carré et le pro- cessus transverse des ptérygoides la plus large. Ils different des Moschorhininae par la présence de dents ptérygoidiennes, l’absence de foramen pinéal, la participation a égalité des prootique, opisthotique, basioccipital, complexe para-basisphenoide au bord de la fenétre ovale, des barres sousorbitaires presque totalement formées par le jugal. De plus, ils se distinguent, des Eucham- bersiinae (outre les caractéres de spécialisation de ceux-ci) par la grande taille des fosses sousorbitaires.
La comparaison entre les deux genres d’Annatherapsidinae est incompléte du fait de l’érosion de la partie antérieure du museau de Chthonosaurus.
Annatherapsidus petri Kuhn: Permien supérieur d’U.R.S.S., zone IV (Tatarien).
Annatherapsidus Kuhn, 1961 différe de Chthonosaurus par un museau de longueur moyenne, des arcades zygomatiques et postorbitaires d’épaisseur
106 ANNALS OF THE SOUTH AFRICAN MUSEUM
moyenne, des frontaux courts, absence de trace du foramen pinéal, une crista choanalis mousse, des processus basisphénoides des ptérygoides jointifs dans leur partie postérieure, ne laissant la fosse interptérygoidienne ouverte qu’a avant. La quille du complexe para-basisphénoide est relativement peu importante. Les processus latéraux du supraoccipital sont assez allongés.
En outre, la région médiane des prémaxillaires est moyennement longue, large et la région mentonniére de la mandibule est en pente douce bien qu’a rupture de pente a la base de la symphyse.
Chthonosaurus velocidens Vjushkov: Permien supérieur d’U.R.S.S., zone IV (Tatarien). }
D’aprés l’aspect de ce qui reste du museau de Chthonosaurus Vjushkov, 1955, celui-ci devait posseder un museau assez long. Les arcades zygomatiques et postorbitaires de faible épaisseur, des frontaux longs, une trace de foramen pinéal sous l’aspect d’un rebord ourlé des pariétaux, une crista choanalis moyenne- ment marquée, des processus basisphénoides écartés sur toute leur longueur laissant une fosse interptérygoidienne complétement ouverte. La quille du complexe para-basisphénoide est relativement élevée. Les processus latéraux du supraoccipital sont courts.
2. Moschorhininae
Les Moschorhininae sont caractérisés par la robustesse du museau, une trés faible constriction postcanine, une formule dentaire maxillaire en régression (1 pC, 1 C, 3-5 PC). L’élargissement de la plaque antérieure des vomers en éventail est accentuce; la lame entre le processus carré et le processus transverse des ptérygoides est de taille moyenne ou réduite. Les dents ptérygoidiennes paraissent absentes. La quille du complexe para-basisphéenoide est moyenne ou peu importante. Le frontal est court et le foramen pinéal est présent chez les spécimens autres que Promoschorhynchus chez lequel cette région n’est pas connue. Chez Moschorhinus (?Tigrisuchus) et Promoschorhynchus le prootique et lopisthotique forment chacun un tiers du bord de la fenétre ovale (cette région n’est pas connue chez Cerdops). Les fosses sousorbitaires sont grandes.
Promoschorhynchus platyrhinus Brink: Permien supérieur d’Afrique australe, zone a Cistecephalus s.l. (plus précisément zone a Daptocephalus selon Kitching, information orale, septembre 1970).
Promoschorhynchus Brink, 1954 est le genre de Moschorhininae dont le nombre de postcanines est le plus grand (5). Son museau est moyennement long et ses arcades zygomatiques et postorbitaires peu importantes. Le maxil- laire participe fortement a la barre sousorbitaire. La crista choanalis est bien marquée, tandis que la plaque antérieure des vomers, moyennement élargie, s'unit a des processus vomérins des prémaxillaires assez longs et larges; cette situation est trés voisine de celle que l’on observe chez Annatherapsidus (ceci a pu étre observé sur le type du genre B.P.I.n°F.484/M.192, aprés nouvelle prépa- ration et sur un autre spécimen plus complet R.C.n°116). La fosse interptéry- goidienne est ouverte sur toute sa longueur. Les processus carré et transverse
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 107
des ptérygoides forment un angle trés aigu; la lame unissant ces deux processus est réduite. L’angle mentonnier est bien marqué, beaucoup moins toutefois que chez Moschorhinus (? Tigrisuchus).
Moschorhinus (? Tigrisuchus) kitchingt Broom: Permien supérieur d’Afrique australe. Zone a Cistecephalus s.l. en général, plus précisément zone a Daptoce- phalus selon Kitching (information orale, septembre 1970); se perpétue dans la zone a Lystrosaurus (cf. Brink 1958: 42). Cette derniére zone est classiquement considérée comme étant la base du ‘Trias inférieur, mais certains auteurs (Cosgriff 1965: 89; Anderson 1970: carte n°5) considérent qu’elle forme la par- tie terminale du Permien; ceci d’apreés la faune.
Moschorhinus Broom, 1920 est caractérisé par un nombre réduit de post- canines (3), un museau court, des arcades zygomatiques et postorbitaires trés épaisses. Le maxillaire participe fortement a la barre sousorbitaire. La crista choanalis est mousse; la plaque antérieure des vomers est fortement élargie et unie a avant au bref processus vomérin des maxillaires. La fosse interptéry- goidienne est ouverte sur toute sa longueur. Les processus carré et transverse des ptérygoides forment un angle aigu peu accentué et sont unis par une lame osseuse de dimension moyenne.
L’angle mentonnier est fortement marqué.
Problemes de synonymie avec Moschorhinus
Tigrisuchus stmus Owen. Permien supérieur d’Afrique australe. Horizon incertain. Zone a Cistecephalus s.l. selon Watson & Romer (1956: 61).
Tigrisuchus fut placé suivant les auteurs tant6t parmi les Gorgonopsiens, tantot parmi les Thérocéphales et tout récemment (cf. Sigogneau, 1970a: 375-376) parmi les Theriodontia incertae sedis. Boonstra fut le premier (1934: 260-261) a le considérer comme un Whaitsiidae s./. Malheureusement le spéci- men type de Tigrisuchus Owen, 1876 (B.M.N.H. n°R. 1721) n’est qu’un frag- ment antérieur de museau, brisé postérieurement au niveau des canines, et ventralement, suivant un plan frontal, un peu plus bas que le bord inférieur des septomaxillaires, au ras de deux foramens prémaxillaires dorsaux. Ce fragment est cependant suffisant pour que !’on puisse noter son aspect globuleux, le degré d’étalement du septomaxillaire et la disposition de ses foramens ainsi que l’aspect de la région antérieure du palais. Celle-ci est identique a celle des divers spéci- mens de Moschorhinus: les prémaxillaires sont brefs sur la ligne médiane et unis a la plaque antérieure des vomers fortement élargie en éventail. Un léger doute quant a la synonymie persiste en raison du nombre des dents antérieures a la canine: dans le maxillaire on observe quatre sections d’incisives et du cdté gauche l’extréme pointe d’une alvéole; par contre, il n’y a pas trace de précanine et la suture entre prémaxillaire et maxillaire est trés proche de la canine. Seules, des sections frontales de museaux de ‘Moschorhinus’ pourront donner des renseignements sur la profondeur de l’alvéole de la précanine (probablement moindre que celle des incisives et canine). Si c’est le cas, il faudra considérer Trigisuchus et Moschorhinus comme synonymes.
Cerdops burghert Broom. Permien supérieur d’Afrique australe. Zone a
108 ANNALS OF THE SOUTH AFRICAN MUSEUM
Cistecephalus s.l. (plus précisément, zone a Daptocephalus selon Kitching, informa- tion orale, septembre 1970).
Ce genre a déja été placé (bien qu’avec doute) parmi les Whaitsiidae s./. Romer (1966: 373 L).
Cerdops Broom, 1948 se rapproche beaucoup par la briéveté de son museau et la réduction du nombre de ses postcanines (3) de Moschorhinus (? Tigrisuchus). Cependant, il semble que le nombre des dents antérieures a la canine soit différent: 5 dents correspondant a 4 I et 1 pC ou a 5 I et o pG, selon les avis. Les arcades zygomatiques et postorbitaires sont moyennes. La barre sousorbi- taire est presque exclusivement formée par le jugal qui en exclut pratiquement le maxillaire (le spécimen de Moschorhinus (? Tigrisuchus), B.P.I. n°F. 1713/M.295 montre dans cette région un épaississement du jugal qui n’est pas sans rappeler celui de Cerdops, mais toutefois le maxillaire participe largement a cette barre). La partie postérieure de la plaque antérieure des vomers montre qu'elle s’étale en éventail. La crista choanalis est moyennement marquée. Les processus basi- sphénoides des ptérygoides jointifs dans leur partie postérieure, ne laissent, comme chez Annatherapsidus, que la partie antérieure de la fosse interptéry- goidienne ouverte. Les processus carré et transverse des ptérygoides semblent former un angle moyennement aigu (ceci compte tenu de |’écrasement latéral du spécimen) et sont reliés par une petite lame osseuse. L’angle mentonnier n’est pas trés marqué en raison de la courbure de la partie antérieure de la symphyse.
Kitching (information orale, avril 1966) pense que ce specimen devrait étre considéré comme une forme jeune de Moschorhinus (? Tigrisuchus). En raison des quelques caractéres cités plus haut et surtout de ignorance dans laquelle nous sommes quant a l’aspect de la partie antérieure du palais et par conséquent des rapports entre prémaxillaires et vomers (ceux-ci constituant dans ce groupe un critére essentiel), il nous semble prudent de laisser ce genre subsister pour le moment.
Hewittia albanensis Brink. Permien supérieur d’Afrique australe. Zone a Cistecephalus s.l. (plus précisément zone a Daptocephalus selon Kitching informa- tion orale, septembre 1970).
Hewittia Brink, 1958 est également représenté par un seul spécimen fragmentaire: partie antérieure du crane brisée a l’arriére des orbites. Kitching (information orale, juillet 1972) considére que Hewtttia est un synonyme de Moschorhinus. En faveur de cette hypothése on peut noter (seulement d’aprés les figures 14 A-B et 15 et la description de Brink 1958: 37-40, le spécimen A.M.G. n°4206 étant introuvable): la réduction des postcanines en nombre et en taille et la forte angulation mentonniére. Cependant le museau de Hewnttia est sensiblement plus allongé et étroit que chez Moschorhinus (? Tigrisu- chus), quatre postcanines sont présentes, la barre sousorbitaire est peu épaisse, les nasaux tres fortement resserrés en leur milieu, les fosses sousorbitaires sont relativement petites et l’aspect de la plaque antérieure des vomers incertain. De plus, selon la figure 14~A, les lacrymaux auraient un assez large contact
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 109
avec les nasaux; ceci est complétement opposé a ce qui est représenté figure 15. Dans ces conditions, il est préférable de ne tirer aucune conclusion tant que le spécimen n’a pas été retrouve.
3. Euchambersiunae
Les Euchambersiinae sont les seuls Thérocéphales présentant une canine a cannelure et, a avant de l’orbite, une profonde fosse sur la face externe du maxillaire et du lacrymal; la face externe du maxillaire est en outre marquée de fines rigoles dans la région de la canine. Ils partagent avec Moschorhinus (? Tigri- suchus) la massivité du crane, en particulier l’aspect tres lourd du museau, et la faible longueur des processus vomérins des prémaxillaires. La formule dentaire est encore plus réduite que chez ces derniers puisqu’il n’y a pas de postcanine. A emplacement de celles-ci se trouve une subdivision de la fosse citée plus haut. L’élargissement de la plaque antérieure des vomers est extréme. Cette plaque est marquée de légéeres ondulations au niveau de son contact avec les prémaxillaires. La lame entre le processus carré et le processus transverse des ptérygoides est trés étroite; par contre une lame extroversée, trés dévelop- pée unit le processus basisphénoide et le méme processus transverse. Les processus basisphénoides sont unis de sorte que la fosse interptérygoidienne n’est visible que dans sa partie antérieure. I] ne semble pas qu’il y ait de dents ptérygoidiennes. Les fosses sousorbitaires sont extrémement réduites. Les arcades zygomatiques et postorbitaires sont gréles mais complétes (observation personnelle sur un second spécimen, B.P.I. n°F.4009/M.384). L’extension des préfrontaux vers Parriére est encore plus poussée que chez Moschorhinus (?Tigrisuchus) et le frontal est presque totalement éliminé du bord supérieur de lorbite. Ce frontal est bref. La région intertemporale est élargie. Il n’y a pas de foramen pinéal visible. La créte sagittale pariétale est basse. La région dorsale de l’épiptérygoide est plus étalée que chez les autres Moschorhinidae connus. La fenétre ovale semble bien entourée par quatre os mais son bord étant fortement érodé, il est impossible d’apprécier dans quelles proportions le prootique, l’opisthotique, le basioccipital et le complexe para-basisphénoide y participent.
La mandibule n’est pas connue.
Cette sous-famille n’est représentée que par un seul genre:
Euchambersia mirabilis Broom. Permien supérieur d’Afrique australe. Zone a Cistecephalus s.l. (plus précisément zone a Cistecephalus s.s.—c’est-a-dire partie inférieure de la zone a Cistecephalus s.1.—selon Kitching, information orale, septembre 1970).
Le genre Euchambersia Broom, 1931 n’est actuellement représenté que par deux spécimens.
C—Genres exclus des Whaitsiidae et des Moschorhinidae
Hofmeyria atavus Broom. Permien supérieur d’Afrique australe. Zone a Cistecephalus s.l. (zone a Cistecephalus s.s., selon Kitching, information orale, septembre 1970).
IIo ANNALS OF THE SOUTH AFRICAN MUSEUM
Hofmeyria Broom, 1935 posséde un museau lourd, une région intertempo- rale élargie et une créte sagittale pariétale basse, caractére que l’on retrouve chez Euchambersia mais aussi chez certains ‘Ictidosuchidae’ (terme employé ici selon la définition de Brink 1965: 35). Hofmeyria a également en commun avec les ‘Ictidosuchidae’ les caractéres suivants: des crétes ventro-intermédiaires des ptérygoides dentigéres et trés allongées, un foramen pinéal de grande taille et surtout une plaque antérieure des vomers étroite. D’autre part, Hofmeyria présente deux canines sur chaque maxillaire. Ce genre serait a rapprocher de Ictidostoma hemburg: Broom tel que l’a décrit Brink (1960: 169-171, fig. 38 Jen, 1b, (Or))e
Mairotenthes digitipes Attridge. Permien supérieur d’Afrique australe. Zone a Cistecephalus s.l.
Murotenthes Attridge, 1956, apres la récente révision de Cys (1971: 124), est assigné par cet auteur aux “Nanictidopsidae, famille primitive des Bauria- morphes telle qu’elle est définie par Watson & Romer (1956: 74) ou aux Ictidosuchidae, famille primitive des Scaloposaures telle qu’elle est définie par Brink (1965)’. Ce spécimen présente un caractére qui n’a encore jamais été observé chez un Thérocéphale qu’1l soit de la lignée des Pristerosauria ou de la lignée des Scaloposauria, a savoir la présence de dents ptérygoidiennes sur la créte ventro-médiane des ptérygoides (p. 122, fig. 1), en plus de la présence assez courante de dents sur les crétes ventro-intermédiaires des mémes os.
Proalopecopsis haughton Brink & Kitching. Permien supérieur d’Afrique australe. Zone a Cistecephalus s.l. (zone a Cistecephalus s.s., selon Kitching, information orale, septembre 1970).
Proalopecopsis Brink & Kitching, 1951, serait a placer, selon Kitching (information orale, juillet 1972), parmi les Ictidosuchidae tels que ceux-ci ont été définis par Brink en 1965.
La révision des ‘Ictidosuchidae’ n’étant pas terminée, ces diverses attribu- tions a cette famille sont a considérer comme provisoires.
CONCLUSION
I] semble maintenant qu’il faille admettre que les ressemblances notées chez le Whaitsiidae d’U.R.S.S. et les Moschorhinidae sud-africains correspon- dent a la conservation de caractéres ancestraux (présence de pré et postcanines, fosses sousorbitaires grandes) ou a une caractéristique générale a la lignée Pristerosauria (tendance a la réduction des dents postcanines).
Les différences dans la structure du palais nous paraissent plus importantes et éloignent ces deux familles l’une de l’autre et nous fait abandonner la sub- division des Whaitsiidae, adoptée par Tatarinov (1963: 77), en trois sous- familles: Whaitsiinae, Moschowhaitsiinae et Moschorhininae. I] semble préférable de conserver (bien qu’en en modifiant la composition) les familles des Whaitsiidae et des Moschorhinidae redéfinie pour la premiére et créée pour la seconde par Brink (1958: 40). Les Whaitsiidae seraient subdivisés en Mos-
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI III
chowhaitsiinae d’U.R.S.S. (Moschowhaitsia) et Whaitsiinae d’Afrique (Therio- gnathus); les Morschorhinidae seraient subdivisés en Annatherapsidinae d’U.R.S.S. (Annatherapsidus et Chthonosaurus), en Moschorhininae d’Afrique australe (Promoschorhynchus, Moschorhinus (? Tigrisuchus), Cerdops, ?Hewittia) et Euchambersiinae également d’Afrique (Euchambersia). Il est curieux de noter que dans les deux familles, c’est chez les formes d’U.R.S.S. que le foramen pinéal disparait tandis qu’il persiste chez les formes africaines (Euchambersia mis a part). Par contre, c’est dans des sous-familles sud-africaines que, a Vintérieur des deux familles, on observe la disparition des postcanines et la fermeteure de la fosse sousorbitaire.
Dans l’ensemble, les formes russes (mise a part la fermeture du foramen pinéal) paraissent plus primitives que les formes africaines. Elles pourraient étre plus proches de la souche commune. A ce sujet il est intéressant de noter que Annatherapsidus fait partie du ‘complexe a Pareiasaure’ de la zone IV @U.R.S.S., a propos duquel Chudinov (1965: 128) n’exclut pas ‘la possibilité qu il ait été placé quelque peu trop haut dans la section du Permien supérieur @U.R.S.S. et qu’en réalité il corresponde a un horizon plus ancien que la zone a Cistecephalus’. Cette opinion est en faveur d’une migration a une €poque soit antérieure a la zone a Cistecephalus s.l., soit équivalente a sa base pour les Moschorhinidae. Rappelons qu’il a di en étre de méme pour les Whaitsiidae que lon trouve deja dans la partie inférieure de la zone a Cistecephalus (zone a Cistecephalus s.s. de Kitching) et qui s’épanouirait dans le sommet de cette zone (ou zone a Daptocephalus de Kitching) a tel point que cet auteur hésitera entre la dénomination de ‘zone a Daptocephalus’ et ‘zone a Whaitsia’ (voir synonymie ci-dessus). Les Whaitsiinae semblent étre parvenus en Afrique australe plus t6t (zone a Cistecephalus s.s.) que les Moschorhininae (zone a Daptocephalus). A propos de la faune de la zone a T apinocephalus, qu’1l compare a celle de la zone II du Permien supérieur d’U.R.S.S., Boonstra (1968: 203) admet que la ‘faune sud-africaine serait une faune immigrante originaire de Russie’. I] en est de méme, en partie tout au moins, de la faune de la zone a Cistecephalus s.l. (puis de celle de la zone a Lystrosaurus du moins en ce qui concerne les Moschorhini- nae). Ces mouvements de faune témoignent d’au moins deux périodes de contact entre Laurasia et Gondwanaland au Permien supérieur. Mais, comme le notait Haughton (1953: 30), aucun document paléontologique ne permet de connaitre, a ’heure actuelle, quelles voies ont emprunté ces immigrants entre les différentes régions d’U.R.S.S. dont ils sont originaires et la région Tanzanie- Nyasa.
LisTEs DES ABREVIATIONS DES NoMS DE COLLECTIONS
A.M.G. Albany Museum, Grahamstown, Afrique du Sud.
Bere aE. 2 sf NIN.. .. Bernard Price Institute for Palaeontological Research, Johannesburg, Afrique du Sud. F = numéro du catalogue de fouille. M = numero du catalogue de musée.
LEZ ANNALS OF THE SOUTH AFRICAN MUSEUM
Pe Rae 4 a .. Collection du Dr F. R. Parrington, Museum of Zoology, Cambridge.
PUNE ee ie .. Paleontologitchesklyy Muzei, Moscou, U.R.S.S. ReGss es Be H .. Rubidge collection, Wellwood, Graaff-Reinet, Afrique du Sud.
S.A.M. ne ae .. South African Museum, Cape Town, Afrique
du Sud. IG Tee 58 A .. Institut und Museum fiir Geologie und Palaon-
tologie der Universitat, ‘Tubingen, République Fédérale Allemande.
RESUME
L’anatomie du crane d’un jeune Moschorhinus (? Tigrisuchus) est étudiée ici, tout particuli¢rement les régions du palais et de l’arriere-crane. La relation existant entre les Whaitsiidae russes (Moschowhaitsiinae) et les Whaitsiidae africains (Whaitsiinae) est notée briévement. Les ressemblances entre les Annatherapsididae russes et les Moschorhinidae sudafricains conduit a les inclure dans une méme famille, les Moschorhinidae, qui sont redéfinis. Cette famille comprend une sous-famille russe (Annatherapsidinae) et deux sous- familles sudafricaines (Moschorhininae et Euchambersiinae). I] est suggéré que les faunes ancestrales des Whaitsiinae d’une part et les faunes ancestrales des Moschorhininae d’autre part aient migré de la Laurasie vers le continent de Gondwana avant ou au début de la zone a Cistecephalus s_l.
ABSTRACT
The skull anatomy of a young Moschorhinus (? Tigrisuchus) is discussed with special reference to the palate and posterior part of the skull. The relationship between the Russian Whaitsiidae Moschowhaitsiinae and the African Whaitsiidae Whaitsiinae is briefly discussed. The similarities between the Russian Annatherapsididae and the South African Moschorhinidae lead to their inclusion in the family Moschorhinidae, subdivided into Russian Annatherapsidinae and South African Moschorhininae and Euchambersiinae. It is suggested that the forms leading to the Whaitsiinae on one hand and those leading to the Moschorhininae on the other hand migrated from Laurasia towards Gondwanaland before or at the beginning of the Cistecephalus zone s.1.
ZUSAMMENFASSUNG
Die Anatomie des Schadels eines jungen Moschorhinus (? Tigrisuchus) wird hier untersucht, speziell die des Gaumens und des Hinterschadels. Die Bezie- hungen zwischen den russischen Whaitsiidae (Moschowhaitsiinae) u. den afrikanischen (Whaitsiinae) werden kurz erwahnt. Die Ahnlichkeit zwischen den russischen Annatherapsididae und den siidafrikanischen Moschorhinidae
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI 113
fiihrt dazu, sie in eine Familie der Moschorhinidae zusammen zu fassen; der eine neue Definition gegeben wird.
Zu diesser Familie gehéren eine russische Unterfamilie (Annatherapsi- dinae) und zwei siidafrikanische (Moschorhininae und Euchambersiinae).
Es ist angedeutet, dass die Urstamme der Whaitsiinae, einerseits und der Moschorhininae andereseits vor oder am Beginn der Cistecephalus-Zone s.l. von Laurasien in den Gondwana Kontinent eingéwandert sind.
RESUMEN
La anatomia del craneo de un joven Moschorhinus (?Tigrisuchus) esta estudiada aqui, muy particularmente en las regiones del paladar y del craneo posterior. La relacién existente entre los Whaitsiidae rusos (Moschowhait- sinae) y los Whaitsiidae africanos (Whaitsiinae) esta indicada brevemente. Las semejanzas entre los Annatherapsididae rusos y los Moschorhinidae sud- africanos conducen a incluirlos en una misma familia, los Moschorhinidae, que estan definidos de nuevo. Esta familia comprende una subfamilia rusa (Annatherapsidinae) y dos subfamilia sudafricanas (Moschorhininae y Eucham- bersiinae). Esta sugerido que las formas ancestrales de los Whaitsiinae por una parte, y las formas ancestrales de los Moschorhininae por otra parte, han migrado desde Laurasia hasta el continente de Gondwana antes o al principio de la zona de Cistecephalus s.l.
REMERCIEMENTS
Jai le plaisir de remercier ici les responsables et les chercheurs des différen- tes institutions scientifiques sud-africaines qui m’ont accueillie; il faut nommer en premier lieu le Dr J. T. Barry, le Dr L. D. Boonstra et le Dr M. Cluver, du South African Museum de Cape Town ot l’essentiel de cette étude a été effectué. Mon travail de comparaison a été facilité par le Dr S. H. Haughton, le Dr A. S. Brink et le Dr A. R. I. Cruickshank, du Bernard Price Institute for Palaeontological Research de Johannesburg. Ma dette de reconnaissance est particuliérement grande vis-a-vis de Dr J. W. Kitching pour ses renseignements sur la répartition stratigraphique des spécimens, les intéressantes discussions et Paide infatigable et stimulante qu’il m’a apportée dans la préparation des spécimens, tout particuli¢rement (en ce qui concerne cet article) de Promos- chorhynchus. Je remercie également le Dr S. H. Rubidge et sa famille pour le prét de leur magnifique collection.
En Europe, j’ai eu accés a diverses collections. Au British Museum (Natural History) je fus accueillie par le Dr A. J. Charig et son assistant Mr C. Walker, et au Museum of Zoology de Cambridge par les Drs F. R. Parrington, K. Joysey et I’. Kemp. Au Geologisch-Palaontologisches Institut de Tiibingen je fus recue par les Prs F. von Huene, O. H. Schindewolf et A. Seilacher et le Dr F. Westphal. I] me fut également possible d’étudier les spécimens du Paleontolo- gitcheskij Muzei de Moscou et mes recherches y furent facilitées par les Drs
II4 ANNALS OF THE SOUTH AFRICAN MUSEUM
L. P. Tatarinov et P. K. Tchudinov. A Paris, j’ai bénéficié des conseils des Prs J. P. Lehman et R. Hoffstetter et de Vhospitalité du Dr Ch. Blanc.
Je dois 4 Mr C. Gow, alors au South African Museum, ainsi qu’a Mr R. Croucher, du British Museum, la délicate préparation de spécimens, et a Mr N. Eden, du South African Museum, de précieux moulages qui m’ont grandement aidée dans la préparation des reconstitutions. Il m/’est difficile d’exprimer toute ma gratitude a Melle J. Crapart pour les trés beaux dessins, exécutés avec talent et patience, qui illustrent ce texte; son esprit d’observation m’a souvent obligé a approfondir certains points de mon travail. Je dois encore a Mr N. Eden et a Mr D. Serrette les photographies. Mme J. Seurin m’a aidée dans l’ingrat travail de dactylographie.
Sans l’aide constante de deux de mes collegues, Mr B. Battail et le Dr J. C. Rage, ce travail n’aurait pu étre terminé a temps.
Que tous veuillent bien trouver ici expression de ma reconnaissance.
Les différentes Missions ont été accomplies grace a des crédits accordés par le Centre National de la Recherche Scientifique et le Ministére des Affaires Etrangeéres de la République frangaise.
LISTE DES OUVRAGES CITES
AMALITZKy, V. P. 1922. Diagnoses of new forms of vertebrates and plants from the upper Permian of North Dvina. Bull. Acad. Sci. St. Petersburg (6) 16: 329-340. (En russe.)
ANDERSON, H. M. & ANpDERson, J. M. 1970. A preliminary review of the biostratigraphy of the Uppermost Permian, Triassic and Lowermost Jurassic of Gondwanaland. Palaeont. afr. 13 (suppl.): 1-22.
ATTRIDGE, J. 1956. The morphology and relationships of a complete therocephalian skeleton from the Cistecephalus zone of South Africa. Proc. R. Soc. Edinb. (B) 66: 59-93.
Boonstra, L. D. 1934. A contribution to the morphology of the mammal-like reptiles of the suborder Therocephalia. Ann. S. Afr. Mus. 31: 215-267.
Boonstra, L. D. 1954. The pristerognathid therocephalians from the Tapinocephalus-zone in the South African Museum. Ann. S. Afr. Mus. 42: 65-107.
Boonstra, L. D. 1968. The terrestrial reptile fauna of Tapinocephalus-zone-age and Gondwana- land. S. Afr. FJ. Sci., 64: 199-204.
Brink, A. S. 1954. On the Whaitsiidae, a family of therocephalian mammal-like reptiles. Trans. R. Soc. S. Afr. 34: 43-59.
Brink, A. 8S. 1958. Notes on some whaitsiids and moschorhinids. Palacont. afr. 6: 23-49.
Brink, A. S. 1960. On some small therocephalians. Palaeont. afr. 7: 155-182.
Brink, A. S. 1965. A new ictidosuchid (Scaloposauria) from the Lystrosaurus-zone. Palaeont. afr. 9: 129-138.
Brink, A. S. & Kircuinc, J. W. 1951. Some theriodonts in the collection of the Bernard Price Institute. Ann. Mag. nat. Hist. (12) 4: 1218-1236.
Broom, R. 1920. On some new therocephalian reptiles from the Karroo Beds of South Africa. Proc. zool. Soc. Lond. 1920: 343-355.
Broom, R. 1925. On some carnivorous therapsids. Rec. Albany Mus. 3: 309-326.
Broom, R. 1931. Notices of some new genera and species of Karroo fossil reptiles. Rec. Albany Mus. 4: 161-166.
Broom, R. 1932. The mammal-like reptiles of South Africa and the origin of mammals. London: Witherby.
Broom, R. 1935. On some new genera and species of Karroo fossil reptiles. Ann. Transv. Mus. 18: 55-72.
Broom, R. 1936. On some new genera and species of Karroo fossil reptiles, with notes on some others. Ann. Transv. Mus. 18: 349-386.
CRANE D’UN JEUNE MOSCHORHINUS KITCHINGI Lis
Broom, R. 1948. A contribution to our knowledge of the vertebrates of the Karroo Beds of South Africa. Trans. R. Soc. Edinb. 61: 577-629.
Broom, R. & Rosinson, J. T. 1948. On some new types of small carnivorous mammal-like reptiles. In: ROYAL SOCIETY OF SOUTH AFRICA. Robert Broom commemorative volume: 29-48. Cape Town: Royal Society of South Africa. (Special publication.)
Cuaric, A. J. 1963. Stratigraphical nomenclature of the Songea Series of Tanganyika. Rec. geol. Surv. Tanganyika 10: 47-53.
Cuupinoy, P. K. 1965. New facts about the fauna of the Upper Permian of the U.S.S.R. 7. Geol. 73: 117-130.
CoseriFF, J. W. 1965. A new genus of Temnospondyli from the Triassic of Western Australia. J. Proc. R. Soc. West. Aust. 48: 65-90.
Crompton, A. W. 1955. A revision of the Scaloposauridae with special reference to kinetism in this family. Navors. Nas. Mus. Bloemfontein 1: 175-200.
Cys, J. M. 1971. The palate and affinities of Mirotenthes (Reptilia: Theriodontia). 7. Paleont. 45: 122-125.
HaucutTon, S. H. 1918. Investigations in South African fossil reptiles and Amphibia. Some new carnivorous Therapsida with notes upon the brain-case in certain species. Part II. Ann. S. Afr. Mus. 12: 175-216.
Haucuton, S. H. 1953. Gondwanaland and the distribution of early reptiles. Trans. geol. Soc. S. Afr. 61 (annex.): 1-30.
HUuEngE, F. von 1950. Die Theriodontier des ostafrikanischen Ruhuhu— Gebietes in der Tiibinger Sammlung. Neues 7b. Geol. Paléont. Abh. 92: 47-136.
Kemp, T. S. 1972. Whaitsiid Therocephalia and the origin of cynodonts. Phil. Trans. R. Soc. (B) 264: 1-54.
Kitcuinc, J. W. 1968. On the Lystrosaurus zone and its fauna with special reference to some immature Lystrosauridae. Palaeont. afr. 11: 61-75.
Kusn, O. 1961. Die Familien der rezenten und fossilen Amphibien und Reptilien. Bamberg: Meisenbach.
MeEnpREZ, C. H. 1965. Sur les affinités des Whaitsiidae (Therocephalia). C. r. hebd. Séanc. Acad. Sct. Paris 261: 1044-1045.
MEnpDREZ, C. H. 1972a. On the skull of Regisaurus jacobi, a new genus and species of Bauriamorpha Watson and Romer 1956 (= Scaloposauria Boonstra 1953), from the Lystrosaurus-zone of South Africa. In: JoysEy, K. A. & KEMP, T. S. eds. Studies in vertebrate evolution: 191-212. Edinburgh: Oliver & Boyd.
MENpDREZ. C. H. 1972b. Premiéres ébauches d’un palais secondaire osseux chez les Reptiles mammaliens. C. r. hebd. Séanc. Acad. Sci. Paris (D) 274: 2960-2061.
Otson, E. C. 1944. Origin of mammals based upon cranial morphology of the therapsid sub- orders. Spec. Pap. geol. Soc. Am. 55: i-xi, 1-136.
Otson, E. C. 1968. The family Caseidae. Fieldiana, Geol. 17 (3): 1-331.
Owen, R. 1876. Description of the Reptilia of South Africa in the collection of the British Museum. 1-2. London: British Museum.
Romer, A. S. 1966. Vertebrate paleontology 3rd ed. Chicago: University of Chicago Press.
Romer, A. 8. & Price, L. I. 1940. Review of the Pelycosauria. Spec. Pap. geol. Soc. Am. 28: i-x, 1-538.
SIGOGNEAU, D. 1963. Note sur une nouvelle espéce de Scaloposauridae. Palaeont. afr. 8: 13-37.
S1GOGNEAU, D. 1970a. Révision systématique des gorgonopsiens sud-africains. Cah. Paléont. C.N.R.S.: i-xui, 1-416.
SIGOGNEAU, D. 1970). Contribution a la connaissance des ictidorhinidés (Gorgonopsia). Palaeont.
afr. 13: 25-38.
TaTARINov, L. P. 1963. Nouveau Thérocéphale de la fin du Permien. Paleont. Zh. 1963: 76-94. (En russe.)
VyjusHKov, B. P. 1955. Thérocéphales de l'Union Soviétique. Trudy paleont. Inst. 49: 128-143. (En russe.)
Watson, D. M. S. & Romer, A. S. 1956. A classification of the therapsid reptiles. Bull. Mus. comp. Kool. Harv. 114: 37-89.
‘ ae hve 2 Sern. wala ee | ea i. a aT os ' ,
ey : ; i? 7 ;
Ning ee bie ‘i ' D ' ine ‘2 - Fj . ae a ian
Sc gre
= ie
he
“>
A NEW DICYNODONT ANCESTOR FROM THE UPPER ECCA (LOWER MIDDLE PERMIAN) OF SOUTH AFRICA
By T. H. Barry South African Museum, Cape Town (With 13 figures in the text)
CONTENTS
PAGE Introduction . : ; ; 3 ‘ ‘ es lel 7, Stratigraphic and locality data . : , : 5 LRT Systematics . : : , ; : ‘ 5 UDs Cranial osteology . : : : : : 5 OT Summary : ; ; : ; : : 5 Gy) Acknowledgements . : . : é ; 5 BVA References : 3 , : : ; ; - Oe Abbreviations . : : : : : 2 2 136
INTRODUCTION
The South African Karroo (Fig. 1) is well known for the abundance and variety of reptilian forms found in its bedded shales. Of the four major consti- tuent series, viz. the Dwyka, the Ecca, the Beaufort and the Stormberg, the Beaufort Series has yielded the majority of fossils. The relative abundance and variety of forms found in the lowest layers of the Beaufort Series and the virtual total absence of fossil discoveries in the underlying Ecca Series led to the generally accepted view that the Ecca was barren.
This view was supported on tectonic evidence by many geologists, amongst them the noted South African geologist Alex du Toit (1948), who states that the absence of Lower Permian vertebrates, save fishes, from the deep and pre- sumably cold-water geosynclinal deposits, such as the ‘Green’ or ‘Blue Ecca’ facies of the South African Cape Province or the ‘Coal Measure’ deltas and swamps to the north east, was understandable, for whereas tropical to sub- tropical climatic conditions would seem to have ruled in the Northern Hemi- sphere during Permo-Carboniferous times, the land in the Southern Hemisphere was still recovering from a great Ice Age. He maintained further that it was only during the later Permian that silting, coupled with the initiation of the Cape Fold Ranges within the geosyncline, produced shallowing. By intercepting moisture from the ocean to the south, the Fold Ranges would have reduced precipitation over the interior basin, the drier climate producing sediments of a more terrestrial nature which would have enabled tetrapods to enter the southern section of the hollow. This, according to Du Toit, was the reason for the abundance of new life during the Lower Beaufort in the southern and central Karroo region, in contrast to its scarcity, in equivalent horizons, further north.
ng)
Ann. S. Afr. Mus. 64, 1974: 117-136, 13 figs.
ee
Hiaevziia LyOd
NOCNO1 LSVI6
Av dy
Sanduv & 0ONINNOT
ANNALS OF THE SOUTH AFRICAN MUSEUM
VOINAV
SWS a5aMOMisiOd
118
ee
A NEW DICYNODONT ANCESTOR FROM THE UPPER ECCA 119
The fact that Du Toit’s views regarding the climatic conditions which prevailed during Ecca and Beaufort times were generally accepted by geologists and palaeontologists alike was probably a contributing factor why such scant attention was paid to Broom’s claims that the terrestrial forms Eccasaurus priscus (Broom 19054, 1912) and Archaeosuchus cairncrossi (Broom 19056, 1909) came from Ecca beds. The announcement (Barry 1972) of the discovery by Mr Roy Oosthuizen of the farm Zwartskraal, Prince Albert district, of some 50 nodules containing fossil remains in beds which are Ecca (Lower Middle Permian) according to present stratigraphical definition, seemed to warrant the reopening of the ‘barren Ecca’ theory.
The southern or ‘Green’ facies in which the latest fossils were found was deposited within the Karroo Geosyncline. The strata occur south ofa line drawn from Calvinia (latitude 31°30’S) in the west to East London (latitude 33°S) in the east, and range in thickness from 2 000 metres near Prince Albert to over 3 000 metres north of Grahamstown. The beds are involved in the Cape foldings and consist of green, bluish, olive and variegated mudstones and argillaceous sandstones and of a blue-green and yellow fine- to medium-grained hard and sometimes quartzitic spheroidally-weathering sandstones, but without any coarse or conglomeratic kinds.
Because of the Cape folding the strata are in places tilted up vertically and even overturned towards the north as between Prince Albert and Klipplaat (Du Toit 1954). Duplication of exposures of the same beds are commonly found, thus complicating the determination of fossil horizons, but on site investigations of the layers and folding patterns at Zwartskraal show the fossil sites to lie within the Upper Ecca (Figs 2 and 3). Lower Beaufort exposure is limited to a narrow projection eastwards along the Sandriver valley with the fossils occurring to the north, south and east of this exposure.
= aa) < q 2 uu jaa Ne WwW z — 2) = (eo) uy E = < oO n oe < [ag oO Ww < & Sea \ a = FLO pas . SS N = ao. ‘ \ = 5 oe te SS Ww * 2 S A—B—'C D ore me t / , \ | | x | | HA |} : ZL y r
\ \ A \ \
\\ \ \\ \\ SS \\ SSS 0% OPN \ \| WAS SSSA
A— 116 m.
B— 508 m.
C— 955m. D—573m.
X— Fossil exposure
LE 7 \\ \\
Fig. 2. Cross-section of the strata at Zwartskraal, Prince Albert.
120 ANNALS OF THE SOUTH AFRICAN MUSEUM
As mentioned earlier, the first major deposits of reptilian fossils in the South African Karroo system occur in the Tapinocephalus zone of the Beaufort Series. This fauna is rich in numbers and established genera and already displays a diversified and balanced ecological composition (Boonstra 1969). Considering this, and the fact that the lithology does not show a sudden transition between the underlying Ecca and the Lower Beaufort, Barry (1972) contended that these factors were not consistent with the generally held view that climatic conditions during Upper Ecca times were such that they mitigated against the existence of land vertebrates during this period. |
This view would seem to be borne out by comparative osteological studies carried out on the fossil material which forms the subject of this paper.
BEAUFORT WEST
39’ 00" pean ANA
RIETBRON
PRINCE ALBERT ap ger WILLOWMORE
PRINCE ALBERT FoR sseteeoerone es Stee - WILLOWMORE
OUDTSHOORN
Fig 3. Geological map of the Zwartskraal area.
A NEW DICYNODONT ANCESTOR FROM THE UPPER ECCA 120
STRATIGRAPHIC AND LOCALITY DATA
Eodicynodon oosthuizen and other specimens still to be described came from Mr Roy Oosthuizen’s farm Zwartskraal in the Prince Albert District of the Cape Province, Republic of South Africa, and were collected by him between 1964 and 1970. A schematic cross-section of the area in which the fossil was found is shown in Figure 2.
The Zwartskraal material found to date includes the following:
(1) the skull of a new species, designated Eodicynodon oosthuizen
(2) portion of a basicranial axis
(3) the skull of a small dicynodontid
(4) the palate, quadrate and articular of a small endothiodontid
(5) eroded skull of a small dicynodontid
(6) postcranial skeleton of small therapsid
(7) snout of small dicynodontid
(8) many small unidentifiable pieces
With the exception of specimen no. 3, which was sectioned using the Croft Parallel Grinding method, all specimens were acid prepared using weak formic acid to remove the matrix.
SYSTEMATICS
Class REPTILIA
Order THERAPSIDA Suborder ANOMODONTIA Infra-Order DICYNODONTIA
Family Dicynodontidae Sub-Family Dicynodontinae Eodicynodon gen. nov.
Diagnosis: Skull small, Dicynodon-like in skull form and possession of a pair of canine tusks, but showing ancestral features. Vomers paired, lateral pterygoid processes strongly developed, premaxillaries partly paired.
Eodicynodon oosthuizeni sp. nov.
Holotype: Skull lacking anterior portion of snout and mandible. R.O.Z. No. 1 (collection of Mr Roy Oosthuizen).
Horizon and locality: Upper Ecca (Lower Middle Permian), Zwartskraal, Prince Albert District, Cape Province, Republic of South Africa.
Diagnosis: Same as for the genus. CRANIAL OsTEOLOGY
The skull lacks the mandible, the right temporal arch and the anterior portion of the snout (Figs 4 and 5). For an unknown reason the latter is a
P22 ANNALS OF THE SOUTH AFRICAN MUSEUM
feature common to all of the skulls found to date. Because of this, few longitudi- nal measurements could be taken but of these the ratio ‘anterior limit of the squamosal to pineal’ /‘pineal to anterior limit of orbit’ would seem to be of some significance. In Hodicynodon this is 1 : 3,6, compared with 1:1 or an outside limit of 1 : 2 ratio in all Dicynodon species measured. It is of interest to note that nearest ratio in the Dicynodontia would seem to be the 1 : 4 ratio of Lystrosaurus.
The interorbital width is less than the intertemporal width, giving a ratio of 3: 4. The postorbital is broad, sloping ventro-laterally at an angle of 35° to the top of the skull roof. The parietal becomes slightly narrower posteriorly, a postero-laterally directed flange extending posteriorly well beyond the medial termination of the bone.
mn
M/M 1
mM
M/M 1
MII
III
HU
U0 ee
Fig. 4. Eodicynodon oosthuizeni, dorsal view. Fig. 5. Eodicynodon oosthuizent, ventral view.
The pineal foramen is oval-shaped and situated on a slightly raised triangular platform (Fig. 6). The surface of the platform, and especially the V-shaped portion extending posteriorly from the rim of the pineal, is rugose. The rim itself projects above the platform.
A posteriorly directed process of the frontal reaches a point in line with the front of the pineal foramen. The postfrontal lies lateral to this process. The parietal is well developed, an anteriorly extending process reaching the level of the postorbital arch. The medianly situated preparietal is wedged in between
the frontals and parietals, the frontals meeting its anterior and anterolateral borders.
A NEW DICYNODONT ANCESTOR FROM THE UPPER ECCA 123
Although the configuration of most of the dermal bones of the skull roof would seem to resemble that of the genus Dicynodon, the preparietal does not participate in the formation of the rim of the pineal foramen, as is generally found in Dicynodon species. In Eodicynodon it lies 3 mm in front of the foramen, a distinct suture indicating the line along which the parietals meet anterior to the pineal foramen to separate the preparietal from the pineal foramen.
Although both Broom (1935) and Toerien (1953) have considered the relationship between the pineal foramen and the preparietal, it would seem advisable to re-analyse the position in the light of conditions found in Eodicyno- don. Broom (1935: 7) found that in Dicynodon grimbeeki “The pineal foramen is rather small and rounded and behind it a little median boss of bone. The preparietal lies in front of the pineal, but on the surface at least it it covered by the parietal, so that the posterior end of the preparietal is 3,5 mm in front of the foramen. A second skull has a similar condition. In one of the male skulls the preparietal nearly reaches the pineal. In others the preparietal forms the anterior wall of the foramen. In some of these the bone is a little weathered, and possibly a bridge of the parietal in front of the foramen has been weathered off. In the type skull the parietal forms the surface bone in front of the foramen, but it sends forward a process on each side of the preparietal.’ Toerien (1953: 54) re-examined Broom’s material and states that he found this deviation from the
gn Se -
LAC. FOR
Icom O.OT. PR
Fig. 6. Eodicynodon oosthuizeni, dorsal view.
124 ANNALS OF THE SOUTH AFRICAN MUSEUM
general dicynodont condition to be the case in only 4 specimens of D. grimbeeki, including the type, and concluded that the non-participation of the preparietal in the formation of the pineal foramen ‘is a rare feature in Dicynodons, but is not uncommon in some of the small Endothiodonts’.
Phylogenetically the preparietal makes its first appearance in the therap- sids. Even amongst them it is not universally present but confined to the Gorgo- nopsia and Dicynodontia. Huxley (1865: 10) was the first to describe this ‘,. . undivided bone wedged in anteriorly between the two parietals’ in a skull fragment from the Panchet Rocks in Bengal. He identified the fragment as belonging to the genus Dicynodon on the strength that in a specimen of Dicynodon murrayi and in D. lacerticeps, ‘. . . while the two parietal bones are separated by a median suture a single intercalary, or wormian, bone is interposed between them and the frontals, and bounds the parietal foramen, in just the same way as the single median bone limits it in the Indian fossil’ (p. 11). Seeley (1895) appears to be the first worker to have referred to this bone as the preparietal.
The preparietal retains a fairly constant position at the median intersection of the frontals and parietals which would seem to be conimensurate with its identification as a “Wormian bone’. Its relationship with the pineal foramen, which in both the ancestral Pelycosauria and Cotylosauria retains a fairly constant position well inside the borders of the parietals, is variable within the Gorgonopsia and the Dicynodontia.
In the majority of the Gorgonopsia the preparietal occupies a position anterior to the pineal foramen. The exceptions to the rule include Aloposaurus ?tenuis from the Cistecephalus zone in which ‘le grand preparietal touche l’orifice pineal’ (Sigogneau 1970: 111) and forms such as Sycosaurus laticeps (Haughton 1924), Sycosaurus brodiei (Broom 1941) and Leontocephalus cadlex (Broom 1940) similarly from the Cistecephalus zone, in which the preparietal is entirely absent. Sigogneau (1970) has interpreted this latter condition as secondary.
In the majority of Dicynodontia again, the preparietal is closely associated with the pineal foramen, participating to a lesser or larger degree in the formation of the rim of the foramen. The exceptions here include the Ecca form Eodicynodon, the Endothiodon zone forms Emydops arctatus (Owen 1876), Emydops longus (Broom 1921), Emyduranus platyops (Broom 1921), Emyduranus granilis (Broom 1936), Emydorhynchus formosus (Broom 1936), and the Cistecephalus zone form Emydorhinus sciuroides (Broom 1921), in all of which the preparietal lies anterior to the pineal foramen. In what is considered to be a secondary condi- tion the preparietal increases its participation in the formation of the borders of the pineal foramen. This trend is illustrated in Diictodon galeops (Broom 1913), Aulacephalodon peavoti (Broom 1921) and Aulacephalodon tigriceps (Broom 1932), all from the Cistecephalus zone in which the hind portion of the preparietal very nearly or completely surrounds the foramen. In others such as the genera Cistecephalus and Myosauroides minnaari (Broom 1941) from the Cistecephalus zone the preparietal is entirely absent.
Thus, in the Dicynodontia as in the Gorgonopsia, the primary position of
Spo ———
A NEW DICYNODONT ANCESTOR FROM THE UPPER ECCA 125
the preparietal would seem to be the one in which the preparietal is situated anterior to the pineal foramen. The increased participation of the bone in the formation of the borders of the pineal foramen in some forms and its absence in others can be considered to be stages in the later development of the skull roof.
Judging by its relationships with surrounding elements it would seem cer- tain that the pineal foramen retains a fairly constant position in the skull roof and that it is the preparietal which is subject to backward displacement. In forms such as Dicynodon woodwardi (Broom 1921), Aulacephalodon peavoti (Broom 1921) and Aulacephalodon haughtoni (Broom 1921), all from the Cistecephalus zone, this backward movement would seem to be accompanied by an increase in the relative size of the preparietal.
The postparietal(s?) occupies the dorsomedial area of the slanting occipital plate (Fig. 13). The name postparietal is here preferred to that of interparietal, usually used, as the bone is not situated between the parietals in this form. It also differs significantly from the usual dicynodont condition in that a medial postero-ventrally directed longitudinal ridge divides the bone into bilateral concave elements. The longitudinal ridge shows a pronounced midline groove or cleft extending the length of the bone. It is not possible to see whether this represents an entire or partial separation of the postparietal into two elements, but it would seem to reflect the ancestral paired condition of this element as found in Captorhinus and other cotylosaurs.
It is interesting to note that the postparietal is single in most pelycosaurs (Romer & Price 1940: 50). Although a well-developed ridge is absent, a longitudinal mound in the midline separates the concave areas on its sides. In most known Gorgonopsia the bone is also single but the median ridge is promi- nent, extending even on to the supraoccipital, the concavities following the general outward sweep of the tabulars and squamosals.
As mentioned earlier, the front portion of the snout has been eroded away.
ORB. SPH.
Fig. 7. Eodicynodon oosthuizeni, anterior view.
126 ANNALS OF THE SOUTH AFRICAN MUSEUM
However, part of the premaxillary is still present and shows signs of being paired. A hairline suture extends the length of the palatal face of the bone and is seen to continue dorsally into the short dorsally directed median process of the pre- maxillary (Figs 7 and 8). The two relatively thin plates thus formed represent the single dorsal premaxillary process commonly found in Dicynodon.
Paired premaxillaries are not known in any other member of the Dicyno- dontia. Its presence in the presumably related, but primitive, Russian forms Venyukovia and Otsheria and in the sphenacodont Pelycosauria would seem to point to the fact that the paired condition in Eodicynodon represents the ancestral condition. The fusion of the premaxillaries in all Beaufort Series Dicynodontia is a secondary evolutionary trend which has not been followed by the other synapsid groups.
The premaxillary ends postero-medially in a short palatal process which makes contact with the vomer (Fig. 8). A postero-laterally directed tapering extension continues some distance backward so that the premaxillary forms the anterior and antero-lateral borders of the choana. ‘The presence in Pelycosauria of a similar relationship would seem to point to a primitive condition, for in later dicynodonts the palatal face of the premaxillary increases in size thereby pushing the choanae further back. Robertia broomiana, a small endothiodont from the lowest division of the Tapinocephalus, which Toerien (1953) suggests is the most likely ancestor of Oudenodon, shows close resemblance to the Eodicynodon condition with the difference that, although the secondary palate is still short and the internal nares reach the level of the tusks, the anterior parts of the choanae are reduced to slits and therefore open functionally more posteriorly. In Eodicynodon the functional anterior limit of the choana lies in line with the canine tusk. Within the nasal cavity there is a longitudinal groove on either side of the median premaxillary septum which leads to the anterior notch of the choana. Camp & Wells (1956) suggested that a similar groove found in Placerias served to transmit the duct of Jacobson’s organ. A similar groove is found in Lystrosaurus (Cluver 1971).
Due to a pronounced overlap of the ventro-lateral portion of the palatal surface of the premaxillary by the maxillary the palatal surface of the former shows as relatively narrow (Fig. 8). In later dicynodonts such as D. grimbeeki, Pristerodon buffaloensis, etc., this relationship is reversed, the lateral border of the premaxillary lying ventral to the maxillary. The condition in Eodicynodon seems to be a retention of the pelycosaurian condition as in the latter the premaxillary is “deeply overlapped by the maxilla’ (Romer & Price 1940: 54). The increased contribution of the premaxillary to the composition of the secondary palate as seen in later dicynodonts would thus necessitate the development of a new premaxillary process covering the maxillary from below.
In all known dicynodonts the vomer is a median unpaired sheet of bone, bifurcated anteriorly and posteriorly. One of the most important features distinguishing E. oosthuizeni from other dicynodonts is that the vomer is dis- tinctly paired in this species (Fig. 8). Anteriorly the vomers form the inter-
A NEW DICYNODONT ANCESTOR FROM THE UPPER ECCA 1947)
PMX
=— --.
\
N HNO Om camry rN
VO
PAL
L.PRPT APR.PT. LVLPTV. QRPT ACI B.SPH. QJ QU oer SQ
OO PR F OV oc O.OT.
Fig. 8. Eodicynodon oosthuizeni, ventral view.
digitating contact with the premaxillaries as seen in Pristerodon (Barry 1967) but behind this area the palatal surface of the anterior third of the vomers is a pair of relatively broad, rounded ridges. Immediately behind this region the vomers curve upwards above the general palatal level of the palate, and the ridges now become thin and blade-like with a V-shaped groove developing between them. In this way the vomers form the anterior and antero-lateral borders of the interpterygoidal vacuity. As vomeral elements participate in the formation of the boundaries of this vacuity, the interpterygoidal vacuity in Eodicynodon technically becomes an intervomero-interpterygoidal vacuity (Fig. 8). This is also the case in most dicynodonts. In many Tapinocephalus zone dicynodonts the vomers form more than one half of the borders of the intervomero-interptery- goidal vacuity. The evolutionary development leading up to this condition has been dealt with in a separate paper (Barry, in press).
The vomers are continued dorsally to form a paired median septal sheet. Dorsally each half of this paired sheet flares out laterally in a horizontal plane, thus forming the median and dorsal walls of the internal nares. It is in contact laterally with the palatine and posteriorly with the pterygoid.
128 ANNALS OF THE SOUTH AFRICAN MUSEUM
Fig. 9. Eodicynodon oosthuizeni, stereoview.
Paired vomers reflect the primitive tetrapod condition, being present not only in sphenacodont pelycosaurs but also in the captorhinomorph cotylosaurs. The paired condition has also been recorded for the Russian Zone II form Venyukovia (Efremov 1940) which according to Watson (1948) occupies a position between the Dinocephalia, which similarly retain paired vomers, and Dicynodontia, in which the vomers have become fused. It is of interest to note that the vomers have become secondarily fused in only two of the main groups of the Therapsida, viz. the Dicynodontia and Gorgonopsia.
The palatine is developed along two planes; a rugose ventral, compara- tively broad kidney-shaped section extending along the general palatal line from just behind the canine tusk to a tapering point in line with the anterior limit of the inter-pterygoidal vacuity, and a dorso-medially curving section which meets the dorso-lateral flange of the vomer to form the lateral and dorsal walls of the internal nares (Fig. 8). Postero-medially the palatine meets the pterygoid in a long suture.
The ectopterygoid forms part of the outer edge of the palate and is wedged between the pterygoid, palatine and maxillary. Its ventral surface is as rugose as that of the palatine, from which it is separated by a deep longitudinal groove.
The rugosity of the ventral surfaces of the ectopterygoid and palatine could possibly be related to the clusters of small teeth on corresponding areas on the palatine, pterygoid and ectopterygoid in pelycosaurs such as Dimetrodon and Edaphosaurus. In Eodicynodon, the antero-ventral portion of the vomer is similarly covered with small ring-like protuberances resembling the small vomerine teeth in Edaphosaurus. In most gorgonopsians sparse clusters of small teeth occur on the palatine and pterygoid.
The pterygoid has 3 main processes extending from the median plate, viz. the anteriorly directed palatal ramus, the laterally directed lateral process and the postero-laterally directed quadrate ramus of the pterygoid. Postero-medially
A NEW DICYNODONT ANCESTOR FROM THE UPPER ECCA 129
the pterygoid meets the parabasisphenoid. As in other dicynodonts the pterygoid bears no teeth.
In Pelycosauria, the pterygoid possesses a long anterior process running horizontally in the palate to meet the vomer. The process meets its mate along their entire length. The anterior process has undergone a drastic reduction in length in Otsherta and Venyukovia but the processes still meet in the midline anterior to the interpterygoidal vacuity and before making contact with the vomer. In Eodicynodon as in all other dicynodonts the anterior processes do not meet in the midline but are separated to form the postero-lateral walls of the intervomero-interpterygoidal vacuity (Fig. 8). As shown elsewhere (Barry, in press) this development is not restricted to the dicynodont line but is also evident in the Dinocephalia.
In the region of the intervomero-interpterygoidal vacuity the pterygoid sends out a broad ventro-lateral projection, the lateral process (Figs 8 and 10). As the lateral process is strongly developed in the pelycosaurs and is absent in dicynodonts, it would seem to indicate the retention in Eodicynodon of a primitive character. Otsheria still shows a prominent lateral process but it is less well developed in Venyukovia.
The basicranial axis of Eodicynodon consists of the basioccipital, the basi- sphenoid and the dermal parasphenoid which is fused with the latter (Figs 10 and 11). The median sphenethmoid complex is situated anterodorsally to the parabasisphenoid. The paired carotid arteries enter the basisphenoid ventrally and emerge dorsally through a single canai posterior to the base of the para- basisphenoid rostrum (Fig. 12). Unfortunately the proximal portion of the parabasisphenoid rostrum is missing so that it is not possible to determine
FR
VA PROT. 7 4 Jd f pet LZ ‘ ‘a Es, SO EX.OCC QJ { Sea QU. ME PBSPH O.0T —————| lcm
Fig. 10. Eodicynodon oosthuizeni, lateral view, left side.
130 ANNALS OF THE SOUTH AFRICAN MUSEUM
S.OC
O.OT
Fig. 11. Eodicynodon oosthuizeni, lateral view, right side.
whether the basisphenoid and parasphenoid are indistinguishably fused throughout its length as would seem to be the case in dicynodonts and also in the pelycosaurs (Romer & Price 1940).
The parabasisphenoid arches upwards and forwards at an angle of 45° before levelling off to run parallel to the main axis of the skull. A short distance behind the interorbital plate the dorsal portion of the bone ends abruptly as if it butted against an unossified posterior portion of the interorbital plate (Fig. 11). It is not clear from the specimen whether the dorsal portion of the parabasi- sphenoid must be regarded as a dorsal process of the parabasisphenoid rostrum, whether it represents a separate septal ossification (the so-called presphenoid claimed for most dicynodonts), or whether it is an anterior extension of the basisphenoid within the parabasisphenoid rostrum. Although there is no sutural evidence of dorsal and ventral elements in the parabasisphenoid in Eodicynodon this possibly cannot be ruled out. Near similar conditions are found in pelycosaurs where according to Romer & Price (1940: 76) ‘. . . in some cases the basisphenoid terminates abruptly within the parasphenoid trough . . .’ just behind the posterior end of the orbitosphenoid. Barry (1967) found questionable indications in Pristerodon of a suture dividing the rostrum horizontally and states that the shape of the posterior part of the rostrum lends support to the conclusion that it consists of two fused elements, the rostrum parasphenoidale ventrally and either a rostrum basisphenoidale or rostrum praesphenoidale dorsally. The latter alternative depends upon whether the ossification in the trabecula communis represents an extension anteriorly into the trabeculae of the ossification of the basisphenoid or whether it is a separate ossification in the trabecula communis. It can also be pointed out that amongst recent reptiles, lizards have a rostrum parasphenoidale while Chelone has a rostrum basi-
A NEW DICYNODONT ANCESTOR FROM THE UPPER ECCA 131
sphenoidale. Judging by the flatness and slightly roughened condition of the dorsal surface of this process it would seem possible that the braincase wall was continued by cartilage above this point (Fig. 10).
Anterior to this process, the dorsal surface of the parabasisphenoid rostrum, which probably now represents parasphenoidal elements only, is deeply grooved to accommodate the lower end of the interorbital plate. Approximately half- way along the ventral border of the plate, the upwardly expanded median sides of the vomer take over the function of providing a groove for the plate (Fig. 11). It is not certain how far anteriorly the parasphenoid rostrum continues within the trough formed by the vomers.
The interorbital plate is roughly rectangular in shape and extends anteriorly from a point in line with the postorbital bar to the anterior rim of the orbit (Fig. 11). There are definite indications that the plate consists of two median ossifications separated by a vertical division. ‘The homology of the elements comprising the sphenethnoid complex and the terminology used by different authors varies considerably and has been dealt with by Barry (1967) and Cluver (1971). For reasons set out previously (Barry 1967), the posterior of these two ossifications will be referred to as the orbitosphenoid and the anterior ossification as the septosphenoid.
The median keel of the orbitosphenoid is roughly square in outline. Dorsally it bears two concave lateral wings which dorsally make contact with the longitudinal ridges on the inner surfaces of the frontals. The median longitudinal trough thus formed is not subdivided by bone. The trough ends anteriorly at the front margin of the orbitosphenoid. From this point forward, at least, the brain must have been subdivided to form the olfactory lobes as the median septosphenoid extends upwards beyond the level of the trough before expanding laterally to form two narrow, slightly convex wings. The latter projections probably covered the olfactory lobes from above.
The palato-quadrate complex in dicynodonts comprises two separated bones, viz. the quadrate, with its two ventrally directed limbs which form the medial and lateral condyles, and the epipterygoid. The latter is usually L- shaped, a thin, dorsally projecting columella, rising from a footplate which rests upon and partly covers the dorsal surface of the quadrate ramus of the pterygoid.
In Eodicynodon, the lateral condyle of the quadrate and part of the medial one have been eroded away, but in spite of this the quadrate does not appear to have marked differences from the general dicynodont condition. It is not clear whether the quadratojugal participates in the formation of the lateral condyle.
The epipterygoid is unusual in that its base or footplate is more expanded than in dicynodonts in general. As can be seen in Figure 11 the footplate extends from a point in line with the posterior surface of the dorsal wing of the vomer to the root of the quadrate ramus of the pterygoid where it covers more than half the dorsal surface of the quadrate ramus of the pterygoid. Unfortunately the latter is broken off so that it is impossible to determine the length of the quadrate ramus of the epipterygoid but judging by its breadth at the break it could have
132 ANNALS OF THE SOUTH AFRICAN MUSEUM
OL BiPELSr ial ea IVIPTV. Paley PT
x Pio CE Gi OU GRer ee
Fig. 12. Eodicynodon oosthuzeni, A. Dorsal view. B. Lateral view of palatoquadrate complex.
continued some distance posteriorly. At the break, the quadrate ramus of the epipterygoid has a distinct dorsally directed ridge, which probably represents the basal portion of the dorsally directed columella of the epipterygoid. A separate skull fragment featuring the basicranial axis confirms this view (Fig. 12).
Anterior to the base of the columella the footplate shows a broad medial expansion which stretches across the dorsal surface of the pterygoid and ends medially against the base of the parabasisphenoid (Fig. 12). The exceptional development of this part of the footplate of the epipterygoid is most unusual for dicynodonts. In the latter the footplate is only weakly developed, or not at all, and the columella is anterior to the base of the columella. In Kannemeyeria (Case 1934) and Daptocephalus (Ewer 1961) the footplate is continued anteriorly but as a much less extensive and narrower bone. Ewer (1961) suggested that this extension was possibly developed to compensate for the weakening of the middle region of the skull resulting from the narrowing of the parietal region. This interpretation does not hold true for Eodicynodon as the parietal region is relatively broad in this form. Being a dicynodont ancestor it is more likely that Eodicynodon would display the more primitive condition. The remarkable simi- larity of the Eodicynodon and Dimetrodon (Romer & Price 1940) footplates would seem to justify this conclusion. The anterior process found in Daptocephalus represents the retention in this form of a primitive feature rather than a later development. In all known Gorgonopsia too the epipterygoid still retains a long tapering anterior extension.
As in Pristerodon (Barry 1967) a well-demarcated groove is formed between the quadrate and quadratojugal, the latter being attached only at its ventral and dorsal extremities.
The occiput (Fig. 13) resembles a trapezoid with the oblique sides bowed gently outward. The cartilage bones of the region do not show many differences
A NEW DICYNODONT ANCESTOR FROM THE UPPER ECCA 133
PO. PAR.R PO.PAR
SOC EX.OCC
PERO
O.OT
O.OT.PR.
O.C
Fig. 13. Eodicynodon oosthuizent. Occipital view.
from the general Dicynodon condition. The borders of the supraoccipital are only faintly recognizable but it would seem that its ventro-medial border forms the dorsal rim of the foramen magnum in the usual way. From here it fans out dorsally and laterally to meet the postparietal and tabulars.
The line of fusion between the supraoccipital and exoccipital cannot be traced. The exoccipital forms the lateral rim of the foramen magnum and extends outward to the post-temporal fossa. From its base a flat roughly triangu- lar flange extends to cover the ventro-medial portion of the opisthotic. Although the sutures are indistinct it seems that the exoccipitals also form the lateral borders of the foramen magnum. The basioccipital forms the ventro-medial portion of the occipital condyle, it is continued anteriorly in the midline from where two ventro-laterally directed triangular extensions develop to cover the posterior surface of the prominent tubera, which form the bony casing of the fenestra ovalis and for the cochlear recess of the internal ear. In most dicyno- donts the tubera are described as being formed by the basioccipitals, but in Eodicynodon each structure is quite clearly formed by the basioccipital and the basisphenoid, with the opisthotic forming the postero-lateral rim.
The opisthotic, similarly, is a roughly triangular plate. Its dorsal boundary extends from the ventro-lateral corner of the foramen magnum to the ventral rim of the post-temporal fossa. A ventro-medially directed flange forms the postero-dorso-lateral rim of the prominent tuber. As in Pristerodon (Barry 1967) the opisthotic carries a strongly developed ridge-like opisthotic process which laterally tapers into a posteriorly directed point (Fig. 13). As stated in the afore- mentioned paper it is believed that this process is associated with the origin of the m. obliquus capitis below it, and the m. rectus capitis posterior lateralis et medialis dorso-medially to it, as still found in some of the recent turtles in which an analogous process occurs.
The post-temporal fossa is small, as in Tapinocephalus zone dicynodonts.
134. ANNALS OF THE SOUTH AFRICAN MUSEUM
The reduction in its size, in comparison with ancestral tetrapods, appears to be due to an increase in the size of the opisthotic. ‘The reduction in size of the post- temporal fenestra would give support to the views of Watson (1911, 1913), Cox (1959) and Cluver (1971) that the fenestra probably served as a passage for the v. capitis dorsalis, which in modern reptiles such as Sphenodon drains the occipital musculature and passes forward to enter the braincase in front of the auditory capsule. In Eodicynodon a shallow, antero-dorsally directed groove on the pro-otic could indicate the passage of the vein within the braincase. However, a second shallow groove-like depression on the pro-otic, similarly emanating from the post-temporal fenestra, runs antero-medially along the surface of the pro-otic. The latter groove and the general smoothness of the bone anteriorly and antero-laterally to the inside edge of the fenestra suggest that in the ancestral forms, with its wide fenestra, and possibly still in Eodicynodon, other elements apart from the v. capitis dorsalis passed through the fenestra.
SUMMARY
Unlike the Tapinocephalus zone (Lower Beaufort) beds of South Africa, which have yielded an abundance of diversified terrestrial forms, the Ecca beds which underlie them were considered to be devoid of such remains. This assumption was based on the view that these layers were deposited during a time when deep water conditions prevailed in the Karroo. It was strengthened by the fact that apart from two claims made by Broom, which were refuted, no terrestrial remains had been definitely reported from these beds.
The discovery, from 1964 to 1970, of a number of skulls and vertebrae of dicynodont-like animals in Ecca defined beds, reported during the 2nd Inter- national Gondwana Symposium (Barry 1972) resulted in the reappraisal of this view. Comparative osteological studies of the skulls show that they predate the Tapinocephalus zone forms in the retention of primitive features such as paired vomers, paired premaxillaries and the retention of the lateral processes of the pterygoid. These features are reminiscent of the Pelycosauria but have dis- appeared in the Lower Beaufort forms.
Eodicynodon has proved to be more primitive than any known dicynodont from the Tapinocephalus zone. On the other hand it shows so many typically dicynodont features that it cannot be excluded from the direct ancestry of at least some branches of the Tapinocephalus zone dicynodonts.
The affinities and systematic position of Eodicynodon formed the subject of a paper presented to the 3rd International Gondwana Symposium. It is scheduled to appear in 1974.
ACKNOWLEDGEMENTS
I wish to record my sincere thanks to Mr Roy Oosthuizen for bringing his finds to my attention and for allowing me to have them prepared for description. My thanks are also due to Mrs I. Rudner for the meticulous acid preparation of the skulls. Mr V. Branco is thanked for the illustrations and Mr A. Byron for photographing the fossils.
A NEW DICYNODONT ANCESTOR FROM THE UPPER ECCA 135
REFERENCES
Barry, T. H. 1965. On the epipterygoid-alisphenoid transition in Therapsida. Ann. S. Afr. Mus 48: 399-426.
Barry, T. H. 1967. The cranial morphology of the Permo-Triassic anomodont Pristerodon buffaloensis with special reference to the neural endocranium and visceral arch skeleton. Ann. S. Afr. Mus. 50: 131-161.
Barry, T. H. 1972. Terrestrial vertebrate fossils from Ecca-defined beds in South Africa. In: INTERNATIONAL UNION OF GEOLOGICAL SCIENCES. Commission on Stratigraphy. Sub-commission on Gondwana Stratigraphy and Palaeontology. Gondwana Symposium, 2nd, South Africa, 1970. Pro- ceedings and papers: 653-659. Pretoria: C.S.I.R.
Barry, T. H. (In press.) Affinities and systematic position of the South African lower Middle Permian dicynodonts (Therapsida: Dicynodontidae). Int. Gondwana Symp. 3, Canberra 1973-
Boonstra, L. D. 1969. The fauna of the Tapinocephalus zone (Beaufort Beds of the Karoo). Ann. S. Afr. Mus. 56: 1-73.
Broom, R. 1905a. Notice of some new South African fossil amphibians and reptiles. Ann. S. Afr. Mus. 7: 270-278.
Broom, R. 19050. Notice of some new fossil reptiles from the Karoo beds of South Africa. Rec. Albany Mus. 1: 331-337.
Broom, R. 1909. An attempt to determine the horizons of the fossil vertebrates of the Karroo. Ann. S. Afr. Mus. 7: 285-289.
Broom, R. 1913. On some new genera and species of dicynodont reptiles, with notes on a few others. Bull. Am. Mus. nat. Hist. 32: 441-457.
Broom, R. 1921. On some new genera and species of anomodont reptiles from the Karroo Beds of South Africa. Proc. zool. Soc. Lond. 1921 (2): 647-674.
Broom, R. 1932. The mammal-like reptiles of South Africa and the origin of mammals. London: Witherby.
Broom, R. 1935. A new genus and some new species of the mammal-like reptiles. Ann. Transv. Mus. 18: 1-12.
Broom, R. 1936. On some new genera and species of Karroo fossil reptiles. Ann. Transv. Mus. 18: 55-72.
Broom, R. 1940. On some new genera and species of fossil reptiles from the Karroo Beds of Graaff Reinet. Ann. Transv. Mus. 20: 157-192.
Broom, R. 1941. Some new Karroo reptiles, with notes on a few others. Ann. Transv. Mus. 20: 193-213.
Camp, C. L. & WELLs, S. P. 1956. Triassic dicynodont reptiles. Part 1. The North American genus Placerias. Mem. Univ. Calif. 13: 255-304.
Case, E. C. 1934. Description of a skull of Kannemeyeria erithea Haughton. Contr. Mus. Paleont. Univ. Mich. 4: 115-127.
Ciuver, M. A. 1971. The cranial morphology of the dicynodont genus Lystrosaurus. Ann. S. Afr. Mus. 56: 155-274.
Du Torr, A. L. 1948. The climatic setting of the vertebrate faunas of the Karroo System and its significance. In: ROYAL SOCIETY OF SOUTH AFRICA. Robert Broom commemorative volume: 113-125. Cape Town: The Society. (Special publication.)
Du Tort, A. L. 1954. Geology of South Africa. 3rd ed. Edinburgh: Oliver & Boyd.
Cox, C. B. 1959. On the anatomy of a new dicynodont genus with evidence of the position of the tympanum. Proc. zool. Soc. Lond. 132: 321-367.
Erremov, J. A. 1940. Preliminary description of the new Permian and Triassic Tetrapoda from USSR. Trudy paleont. Inst. 10 (2): 1-140.
Ewer, R. F. 1961. The anatomy of the anomodont Daptocephalus leoniceps (Owen). Proc. zool. Soc. Lond. 136: 375-402.
Haucuton, S. H. 1924. On some gorgonopsian skulls in the collection of the South African Museum. Ann. S. Afr. Mus. 12: 499-517.
Hux.ey, T. H. 1865. Indian pre-Tertiary Vertebrata. On a collection of vertebrate fossils from the Panchet Rocks near Ranigung, Bengal. Mem. geol. Surv. India Palaeont indica (4) (1): I-24.
Owen, R. 1876. Descriptive and illustrated catalogue of fossil Reptilia of South Africa in the collection of the British Museum. London: British Museum.
136 ANNALS OF THE SOUTH AFRICAN MUSEUM
Romer, A. S. & Price, L. W. 1940. Review of the Pelycosauria. Spec. Pap. geol. Soc. Am. 28: i-x, 1-538.
eer G. 1898. On the skull of Mochlorhinus platyceps from Bethulie, Orange Free State, preserved in the Albany Museum, Grahamstown. Ann. Mag. nat. Hist. (7) 1: 164-176.
SIGOGNEAU, D. 1970. Révision systématique des gorgonopsiens sud-Africains. Cah. Paléont C.N.R.S.: i-xil, 1-414.
TorriEn, M. J. 1953. The evolution of the palate in South African Anomodontia and its classi- ficatory significance. Palaeont. afr. 1: 49-117.
Watson, D. M.S. 1911. The skull of Diademodon with notes on some other cynodonts. Ann. Mag. nat. Hist. (8) 8: 293-330.
Watson, D. M.S. 1913. Further notes on the skull, brain and organs of special sense of Diademo- don. Ann. Mag. nat. Hist. (8) 12: 217-228.
ABBREVIATIONS JA Chait Arteria carotis interna ORB.SPH. Orbitosphenoid APPREP TL Anterior process of the ptery- PAL. Palatine goid PAR. Parietal B.E.C Base of the epipterygoid colu- P.B.SPH. Parabasisphenoid mella P.FOR. Pineal foramen B.OC. Basioccipital PL.SPH. Pleurosphenoid B.P.B.SPH. Base of the parabasisphenoid P.MX. Premaxillary B.SPH Basisphenoid PO.FR. Postfrontal CG: Canine tusk PO.O. Postorbital EC.PT. Ectopterygoid PO.PAR Postparietal ERE. Epipterygoid PO.PAR.R. Post parietal ridge EX.OCC Exoccipital PR.OT. Prootic F.EPT. Footplate of the epipterygoid PR.PAR Preparietal F.OV. Fenestra ovalis PT. Pterygoid FR. Frontal P.T.FO Posttemporal fossa I.A.M. Internal auditory meatus ale Quadratojugal I.V.I.PT.B Intervomero-interpterygoidal Q.R.EPT Quadrate ramus of the epiptery- vacuity goid LAC.FOR. Lacrymal foramen OoR Ea Quadrate ramus of the ptery- L.PR.PT Lateral process of the pterygoid goid MX. Maxillary QU. Quadrate MX.ANT. Maxillary antrum S.OC. Supra-occipital O.C. Occipital condyle SOC.C.T Socket of the canine tusk O.OT. Opisthotic S.Q. Squamosal O.OT.PR. Opisthotic process VO. Vomer
THE SKULL AND MANDIBLE OF A NEW CISTECEPHALID DICYNODONT
By MicHarEt A. CLUVER South African Museum, Cape Town (With 4 figures)
CONTENTS
PAGE Introduction : : ‘ : : : i : 137 Classification : : : é 138
Cranial morphology of Cisteephaloides ponpaee Skull ‘ : : : ; 139 Snout and skull so5 , : : : : : 140 Occipital region : : : ‘ 2 ; I4I Basicranial axis and selec : é ; : : 143 Temporal region . 2 : ‘ 3 : : 146 Lower jaw : : 147 Comparison with Cistecephalus ped oniracats d , 147 Functional considerations : - : é : : 150 Summary . : , 2 : : : : 153 soe leenens : E : : : : : 154 References : ‘ : ; : 3 : : 154 Abbreviations : 4 : é ‘ : : : 155
INTRODUCTION
The dicynodont genus Cistecephalus is one of the more well-known groups of South African therapsids, and its unusual, readily identifiable skull form and stratigraphically limited occurrence (Keyser 1965; Kitching 1970) make ita valuable marker genus in the Beaufort Series zoning system. Several species (11) have to date been described but reviews of the genus have shown that only 9 of these are in fact based on specimens of Cistecephalus. These are C. microrhinus, chelydroides, planiceps and bathygnathus (Owen 1876); angusticeps (Broom 1932); major, rubidger and platyfrons (Broom 1948) and Jaticeps (Brink 1950). Several synonymies have at various times been declared among these species (see Brink 1950; Keyser 1965); in the most recent review, Keyser (1965) recognizes only C. microrhinus Owen but at the same time has found evidence for a new species.
Cox (1972) has recently described a new East African dicynodont, Kawinga- saurus fossilis, which he states is fairly closely related to Cistecephalus but has important differences in the skull and especially the postcranial skeleton.
The specimen at present under discussion (SAM 6243) could at first glance be regarded as a species of Cistecephalus with somewhat unusual skull proportions, but removal of the lower jaw and exposure of the palate and basicranium revealed a basic skull structure which differs widely from the Cistecephalus and, indeed, the standard dicynodont condition. These differences
137
Ann. S. Afr. Mus. 64, 1974: 137-155, 4 figs.
(ON.
138 ANNALS OF THE SOUTH AFRICAN MUSEUM
appear to warrant generic distinction, and are summarized in the following section.
CLASSIFICATION
The creation of a new genus, closely allied to Cistecephalus, raises the ques- tion of the status of Cistecephalus itself. Although the recent classification of Romer (1966) places the genus in the family Dicynodontidae, earlier workers tended to emphasize the unusual skull structure of the genus by setting it apart from other dicynodonts. ‘Thus Seeley (1894) was led by the unusual skull roof (the palate was at that stage unknown) to regard the genus as being sufficiently far removed from his suborder Dicynodontia to warrant a separate suborder, Kistecephalia. Most subsequent authors have, however, been content to leave the genus in a separate family. Broom in 1903 proposed a family Cistecephalidae for the genus, distinct from the Endothiodontidae, Dicynodontidae and Lystro- sauridae, and distinguished from these by the structure of the quadrate and occipital regions. Broili & Schréder (1935: 17) found additional grounds for maintaining this family, which they diagnosed as:
‘Skull short-snouted, relatively high, intertemporal width greater than interorbital width,
no preparietal, no independent postfrontal and probably no tabulars, and also no inter-
pterygoidal opening. Toothless.’
Haughton & Brink (1954: 69) have retained the family Cistecephatidees which they define as having:
‘Small broad skull. Molarless. Very wide and flat intertemporal region.’
Investigations into the anatomy of Cistecephalus undertaken for the purposes of the present work make it clear that the genus’s strikingly unusual cranial structure demands familial distinction, on the lines proposed by Broili & Schroder (1935) and Haughton & Brink (1954). The pertinent cranial charac- ters of the family and genus, and the relationships of Cox’s Kawingasaurus and the proposed new genus and species, may be formalized as follows:
Class REPTILIA Order THERAPSIDA Suborder ANOMODONTIA (sensu Romer, 1966) Infraorder DICYNODONTIA (sensu Romer, 1966) Family Cistecephalidae Broom, 1903
Diagnosis: Small, toothless dicynodonts with very broad intertemporal region, lacking a preparietal and postfrontal. Squamosal not extended posteriorly beyond level of the supraoccipital; stapes perforated or deeply incised; inter- pterygoidal vacuity vestigial or absent.
Genus Cistecephalus Owen, 1876
Diagnosis: Cistecephalids with broad or narrow interorbital region. Lacrimal foramen absent or low down within orbit; pterygoids meeting below para- basisphenoid complex; interpterygoidal vacuity absent.
THE SKULL AND MANDIBLE OF A NEW CISTECEPHALID DICYNODONT 139
Genus Kawingasaurus Cox, 1972
Diagnosis: Cistecephalids with moderately broad interorbital region and rela- tively small orbits. Postorbital and prefrontal separated by frontal. Lacrimal foramen in normal dicynodont position, pineal foramen absent. Zygomatic arch slender. Stapedial foramen present and interpterygoidal vacuity absent.
Genus Cistecephaloides gen. nov.
Diagnosis: Cistecephalids with broad intertemporal and interorbital regions, reduced orbits and robust postorbital bars. Prefrontal meeting postorbital, lacrimal foramen in usual dicynodont position; stapes deeply incised and reduced to slender bar; pterygoids extending far backwards onto basioccipital tubera, separated in midline by parasphenoidal rostrum which runs forward in palatal view to meet vomer. Interpterygoidal vacuity represented by small fossae (possibly foramina) alongside parasphenoid. Dorsal dentary surface of jaw ramus drawn up laterally as a sharp, longitudinal blade.
Cistecephaloides boonstrai sp. nov. Diagnosis: as for genus. Holotype: SAM 6243. Partially weathered skull and mandible. Locality: Ely Cottage, near Alice, Cape Province, South Africa.
Horizon: Cistecephalus zone (Daptocephalus zone of Kitching, 1972) of Upper Permian Middle Beaufort Series.
I have pleasure in naming the new species after Dr L. D. Boonstra, who introduced me to vertebrate palaeontology.
Haughton & Brink (1954) include the genus Emydorhinus Broom under their Cistecephalidae, but, with the exception of the loss of the postfrontal, there do not appear to be any significant resemblances between this genus and Cistecephalus, Kawingasaurus and Cistecephaloides.
CRANIAL MorPHOLOGY OF CISTECEPHALOIDES BOONSTRAI SKULL
The skull lacks the left zygomatic arch, the right quadrate and otic region, and a portion of the skull roof above the left temporal fossa. In addition, the left cheek region and the adjoining palatal surface is incomplete, while in ventral view the anterior portion of the palate and basicranium is displaced dorsally at a shear just anterior to the level of the internal carotid artery canals. These imperfections are relatively easily compensated for, and the illustrations of the reconstructed skull may be regarded as being reasonably accurate.
The skull shows the usual cistecephalid proportions and is triangular in dorsal view, being widest over the rear of the squamosals. In the actual speci- men, before allowance is made for distortion, the width of the skuli over this region is 58 mm, as compared with the basioccipital-premaxilla skull length of
140 ANNALS OF THE SOUTH AFRICAN MUSEUM
61 mm. In the reconstructed skull, these measurements become respectively 63 and 62 mm.
Snout and skull roof
The edges of the palatal rim are lacking, and the ventral limits of the snout are thus difficult to determine. On the more complete right-hand side the maxilla is flared out laterally in front of the orbit to form a prominent and robust ledge, which lies as a forward prolongation of the zygomatic arch. The septomaxilla is exposed on the surface of the snout behind the external naris, and forms a small part of the floor of the anterior nasal chamber. Behind the septomaxilla the maxilla rises up high in the side of the snout to meet the nasal.
The nasal covers a relatively large area on the snout and skull roof, and forms a small boss over the external naris. Anteriorly, in the region of the external naris, the nasal is perforated by numerous nutritive foramina and channels.
The small size of the orbit is one of the striking features of the skull. The postorbital bar is a robust pillar, and the bone extends forward above the orbit to make contact with the prefrontal, thereby excluding the frontal from the orbital border. The lacrimal and prefrontal have little exposure on the lateral snout surface, but the lacrimal forms a large part of the inner wall of the orbit, and is pierced by a foramen for the duct of the lacrimal gland in the anteroventral corner of the orbit, in the normal dicynodont position. ‘The pre- frontal forms only the anterior half of the inner orbital border, and behind it the lacrimal makes contact with a forward extension of the frontal.
The maxilla does not reach far back on the zygomatic arch, and the jugal, extending out from below the lacrimal inside the orbit, forms the major part of the infraorbital bar.
The intertemporal part of the skull roof is weathered and discoloured, and sutures are consequently difficult to trace. The frontonasal suture is clear, and extends back to the level of the rear of the orbits. This level is well within the rear half of the skull, and accordingly the frontals and parietals are short, wide elements. A frontoparietal suture can be distinguished near the midline of the skull, while traces of a parietal-interparietal contact can be seen on the rear edge of the skull table. From these it appears that the frontals are con- siderably shorter than the nasals, while the parietals, shortest of all, are confined to the extreme rear of the skull roof. Laterally each parietal passes forward for a short distance, between the postorbital and frontal of its side. An inner flange of the frontal passes forward and down medially to the postorbital to form part of the inner wall of the orbit, where it meets the prefrontal and lacrimal. A similar but more vertical flange of the parietal forms the median wall of the temporal chamber and meets the dorsal tip of the epipterygoid.
No clear parietal foramen can be made out, but a circular, damaged area in the midline of the skull in the parietal region has within it portions of smooth,
THE SKULL AND MANDIBLE OF A NEW CISTECEPHALID DICYNODONT I4I
vertically standing bone, which may represent part of the pineal canal. In any event, the parietal foramen, if present, could not have been more than a small, relatively insignificant opening.
Occipital region
As in Cistecephalus, the squamosals in Cistecephaloides do not extend back beyond supraoccipital level, and the occiput (Fig. 4) is thus an open and slightly rounded surface, the occipital condyle being the most posterior structure. From the condyle the occiput slopes forward fairly steeply, so that the plane of the occiput meets that of the skull roof at an angle slightly in excess of go°.
Considerable damage to the occipital region has been caused through weathering, and some doubt exists as to the exact relationships held between supraoccipital, interparietal and squamosal, especially in the dorsal part of the occiput. In addition, the presence or absence of a tabular is open to question. The interparietal forms a low nuchal crest immediately behind its meeting with the parietal and extends down and laterally on the occiput to meet the supraoccipital and squamosal. Both supraoccipital and interparietal slope back towards the exoccipital, the condyle of which lies in the most posterior plane of the skull. Medially the exoccipital carries a short, blunt process for the proatlas on the margin of the foramen magnum, and laterally it terminates in the inner corner of the posttemporal fossa. Above the condyle it forms the posterior wall of the jugular foramen.
Because of the considerable lateral extent of the exoccipital, the opisthotic does not make any contact with the supraoccipital on the occipital surface. From the posttemporal fenestra the opisthotic-squamosal suture is deeply interdigitated and runs ventrolaterally, but the meeting lower down between the squamosal and the lateral paroccipital process is, in contrast, a fairly loose one. Ventrolaterally (Figs 1, 3, 4) the opisthotic bears a sharp crest which is drawn out posteriorly as a fairly prominent ‘opisthotic’ process, the equivalent of the tympanic process of Cox (1959). Below this is the channel for the lateral head vein, narrower than in other dicynodonts, and deeply incised into the opisthotic’s ventral border.
The dorsal extent of the squamosal is uncertain, but below its meeting with the postorbital it forms an extensive sheet-like border to the occiput; in common with species of Cistecephalus, this occipital part of the squamosal is flat and lacks the posterior flanges found in other dicynodont groups.
The occipital condyle is the most posterior part of the skull. The lower half is formed by the basioccipital, while the two exoccipitals complete the structure dorsally. The exoccipital part of the condyle does not project back farther than the basioccipital part, as is the case in some dicynodonts, and, since the basioccipital and exoccipitals are depressed to form a notochordal pit in the centre of the condyle, the actual articular surface is a flat, ring- shaped area.
The quadrate and quadratojugal hold the normal dicynodont relation-
142 ANNALS OF THE SOUTH AFRICAN MUSEUM ships with each other. Ventrally the quadratojugal is indistinguishably fused with the dorsal surface of the lateral quadrate condyle, but dorsally its suture with the squamosal is clearly visible above the quadrate foramen, where the quadratojugal and quadrate are separated by a ventral process of the squamosal.
pm.
sm xX.
lac. for
COC. soc.
Fig. 1. Cistecephaloides boonstrai gen. et sp. nov. Skull in dorsal view. Natural size.
boc.
Fig. 2. Cistecephaloides boonstrai gen. et sp. nov. Skull in ventral view. Natural size.
THE SKULL AND MANDIBLE OF A NEW CISTECEPHALID DICYNODONT 143
The articular surface of the quadrate has the form of a wide trough, dorsally curved anteriorly and posteriorly, and is not easily divided into distinct lateral and medial condyles. The major part of the articular area faces forwards and down, but at the rear this surface curves sharply upwards to meet the ventral edge of the squamosal.
pa.
po. fr
den.
art: lat. den. |.
| cm. a Ging: (b)
den. lat.den.l.
med.
Fig. 3. Cistecephaloides boonstrai gen. et sp. nov.
(a) Skull in lateral view. (b) Mandible in lateral view. (c) Man- dible in dorsal view. Natural size.
Basicranial axis and palate
The basioccipital-basisphenoidal tubera are prominent features in the ventral part of the braincase, and their structure can be fairly accurately determined in the more complete left-hand side. The rear of the tuber is widely
144 ANNALS OF THE SOUTH AFRICAN MUSEUM
excavated over the basioccipital-basisphenoid suture, and it is possible that this part, which is unfinished in periosteum, was completed in cartilage during life. Above this area, the opisthotic is flared backwards as a slightly concave, horizontal ledge, below and lateral to the opening of the jugular foramen.
In the ventral midline the highly interdigitating basioccipital-basisphenoid suture is clearly visible. The suture straightens out on the inner side of each tuber, but weathering on these ventralmost points in the skull obscures the meeting between basioccipital and basisphenoid in the region of the fenestra ovalis. Interpretation of this area is further complicated by the unfinished, porous nature of the bone surface, suggesting that the rim of the fenestra ovalis was at least partly formed in cartilage.
The stapes itself appears to have been partly cartilaginous. A fairly sub- stantial footplate is present, but the body of the element is represented only by a very thin bony splint, arising from the anterior end of the footplate and not quite reaching the quadrate. As such, the stapes is an extremely light and delicate structure, and the considerable posterior extent of the footplate, finished in periosteum, suggests that the bone was deeply notched, possibly to allow passage for the stapedial artery and chorda tympani nerve. The distal end of the stapes, which normally abuts against the inner side of the quadrate condyle, is not preserved, and during life was probably present as cartilage.
On the anterior slope of the basioccipital tuber the basisphenoid 1s overlain by a long, irregular posterior extension of the pterygoid, which reaches back almost to the level of the fenestra ovalis. In front of the meeting between the body of the pterygoid with the quadrate ramus, the basicranial girder is sheared through, and the anterior portion displaced dorsally, but in spite of this complication it is possible to make out the relations which the various bones hold with each other. Most striking is the lack of contact between the pterygoids in the ventral midline. In front of the internal carotid artery foramina
bo.
| cm. a |
Fig. 4. Cistecephaloides boonstrai gen. et sp. nov.
Skull in posterior view. Natural size.
THE SKULL AND MANDIBLE OF A NEW CISTECEPHALID DICYNODONT 145
the pterygoids approach each other, but are clearly separated by an anterior extension of the parabasisphenoid complex. Anterior to this, the pterygoids diverge to form the lateral borders of the internal choanae, and a pair of deep recesses are formed between the pterygoid and the median, parabasisphenoid bar. These recesses are apparently vestiges of the interpterygoidal vacuity which is found in other dicynodonts (except Cistecephalus). The lateral borders of the choana, formed by pterygoids, palatines and ectopterygoids, are criss- crossed by a network of small cracks, and sutures are thus difficult to determine with complete certainty. It seems, though, that each pterygoid extends forward to make a short contact with the maxilla and, together with the palatine and a posterior process of the premaxilla, closes off a large lateral palatal foramen at the level of the anterior part of the choana. In this region the pterygoid is wedged between the palatine medially and the ectopterygoid laterally.
The palatine has a limited palatal portion and the major part of the bone lies as a low outer wall of the choanal passage. Posteriorly this palatine wall makes contact with an anterior extension of the parabasisphenoid complex laterally, and with the vomer medially, where this latter bone curves outwards to form the posterior wall of the choanal passage. Anteriorly the palatine passes forward to the ventral part of the orbital wall, where an anterior lappet meets the lacrimal and jugal.
The ectopterygoid is a thin splint of bone lying entirely lateral to the anterior part of the pterygoid. It makes contact with the maxilla but does not border onto the lateral palatal foramen. The vomer, on the other hand, is an important bone in the palate and lies as a deep and stout median septum which descends to the level of the lateral, pterygoidal borders of the choana, effec- tively separating the two internal nares from each other. Dorsally the septum widens as the vomer curves out on each side to form a partial median roof for the choanal passage, the roof becoming complete at the meeting of the vomer with the palatine at the rear of the choana. Anteriorly the vomer is clasped by a pair of diverging processes of the premaxilla, while posteriorly it, in turn, contains the anterior parasphenoidal prolongation of the parabasisphenoid.
The bony secondary palate, formed by premaxilla and maxilla, is wide and bounded by a low palatal rim. On either side of its median palatal ridge (a functional continuation of the parasphenoid-vomer bar), the premaxilla slopes away to form a pair of shallow recesses, which merge in front of the median ridge to create a vaulted area behind the anterior palatal rim. Medial to the lateral palatal rim, along the maxilla-premaxilla suture, the palate is raised as a narrow ledge, which terminates anteriorly as a weak lateral palatal ridge on the premaxillary palatal rim. The palatal rim itself is damaged on both sides of the snout, but it is nevertheless apparent that the rim was thin and sharp-edged anteriorly, while posteriorly it probably became more robust near the short, blunt caniniform process. The posterior, maxillary portion of the palatal rim is medially scalloped, and two embayments lead onto the lateral palatal ledge. The posterior of these is the more pronounced, and lies
146 ANNALS OF THE SOUTH AFRICAN MUSEUM
anterior to a smooth, concave portion of the maxilla which slopes postero- medially from the caniniform process to merge with the palatal surface level with the anterior margin of the choana.
Temporal region
The left side of the skull has been completely cleared of matrix and portions of the braincase and basicranial axis have been exposed. The prootic-basi- sphenoid suture is clear, and a small depression, half-way along its length, leads to the foramen for the exit of the facial nerve. There is no deep sulcus in the prootic in front of the posttemporal fenestra, such as is found in many other dicynodonts and which appears to have stood in connection with the venous system of the head (Cox 1959; Cluver 1971), and no notch (venous notch of Agnew 1959) is found between the prootic and parietal in the sidewall of the braincase.
The braincase lies deep within the greatly enlarged temporal vault. Frontal, postorbital and parietal combine to make an extensive roof behind the orbit, and the temporal fenestra opens almost entirely laterally, the roofed- over area being best developed posteriorly, where the skull roof joins the occipital plate. The bone forming the temporal roof is uniformly thin, but above the temporal fenestra the ventral border of the postorbital is inflated to a powerful, bar-like rim, which extends from behind and medial to the postorbital bar to the posterolateral corner of the temporal fenestra (Fig. 2).
In front of the hypophyseal region the basisphenoid lies as a horizontal sheet, meeting the pterygoids laterally and the palatine anteriorly, and carrying the presphenoid dorsally. A notch visible in lateral view between the basi- sphenoid and the anteroventral edge of the presphenoid lies above the pair of deep fossae in the palate which represent the vestigial interpterygoidal vacuity. In typical dicynodonts a similar notch serves as the dorsal opening of the interpterygoidal vacuity, and although it is reduced in Cistecephaloides, its presence suggests that the interpterygoidal openings might in fact pass upwards to open out above the pterygoids, a condition which has been lost in Cistece- phalus. Verification of this possibility by further preparation was, however, prevented by the delicate nature of the bones in this region of the skull.
As stated earlier, the rear of the palate in the type specimen has been displaced upwards at the level of the internal carotid artery canals, and the presphenoid is consequently partly concealed behind a sheet of bone descending from the skull roof. This bone (os, Fig. 3), wedged between the descending laminae of the parietal and frontal, appears to be part of a damaged orbito- sphenoid complex. Few details can be made out, but it is clear that it forms a considerable part of the upper sidewall of the braincase, and is slightly curved. There is no clear division of the element into a dorsal trough-like portion and a ventral keel.
A small portion of the epipterygoid footplate, in front of the ascending shaft, is present on the dorsolateral edge of the pterygoid, but the major part
THE SKULL AND MANDIBLE OF A NEW CISTECEPHALID DICYNODONT 147
of the bone was apparently damaged by the displacement of the palate, and is only partially preserved. Dorsally a lappet of the parietal, descending to meet the upper extremity of the epipterygoid, can be distinguished.
LOWER JAW
The mandible is short and deep in its overall dimensions. The powerful dentaries are completely fused in the symphyseal region and carry a sharp, transverse cutting edge anteriorly. Behind this blade the dentaries are scooped out as a wide trough, and ne dentary tables, as found in dicynodonts such as Lystrosaurus (Crompton & Hotton 1967; Cluver 1971), are present. Behind the symphyseal region, however, the dorsal edge of the dentary widens, and a wide, shallow dentary groove is formed. The medial wall of this groove rises posteriorly as a sharp blade (med.r., Fig. 3c) forming the medial edge of the dorsal dentary surface. Below this sharp edge the inner surface of the dentary is slightly concave, so that the Meckelian sulcus in the prearticular is exposed in dorsal view. The lateral wall of the dorsal dentary groove, on the other hand, is highest anteriorly, where it rises as a sharp, delicate cutting edge (lat.r., Fig. 3c). Below and lateral to this the dentary is expanded above the mandibular fenestra to form a robust lateral dentary ledge. Below the large mandibular fenestra the dentary passes back to meet the angular.
The dorsal, surangular margin of the jaw curves down steeply, and the articular is consequently carried relatively low in the mandible; this feature may, however, be partly the result of distortion which the mandible appears to have undergone. The main articular condyle is a wide, slightly concave facet, while a smaller median facet is carried on a delicate inner ledge. The two surfaces are separated by a rounded and ventrally curved longitudinal ridge. The ventral extremity of the articular, together with any evidence of a retroarticular process, has been lost through weathering.
The medial border of the Meckelian sulcus is formed by the prearticular posteriorly and the splenial anteriorly. As stated earlier, the inner surface of the dentary above the sulcus is hollowed out, and a wide, smooth depression leads down anteroventrally into the trough.
COMPARISON WITH CISTECEPHALUS AND KAWINGASAURUS
A number of skull features, notably in the skull roof, occiput and palate, indicate that the affinities of Cistecephaloides lie more with Cistecephalus and Kawingasaurus than any other dicynodont genera. The broad intertemporal region in Cistecephaloides is (as in Cistecephalus) composed of the very wide frontals and parietals, which on each side meet the sheet-like postorbitals. In Cistecepha- loides, however, an additional feature is the enlargement of the nasals, which extend farther laterally and posteriorly than in any Cistecephalus species. Their lateral extent is reflected in the small size of the orbits, while their posterior borders are well within the rear half of the skull. The frontals and parietals are consequently confined to a relatively short and wide area of the skull roof.
148 ANNALS OF THE SOUTH AFRICAN MUSEUM
The parietals in Cistecephaloides do not pass forward between the frontals in the midline, as they do in Cistecephalus, but laterally each parietal curves forward between the frontal and postorbital of its side. In dorsal view, therefore, the parietals together cover an open, horseshoe-shaped area on the skull roof, in contrast to the triangular area common in Cistecephalus. The fronto-parietal suture is not shown in Cox’s (1972) illustrations of Kawingasaurus.
The enlargement of the nasal in Cistecephaloides and the accompanying increase in the interorbital width and reduction in orbit size have resulted in crowding together of the circumorbital bones and an unusual meeting of the prefrontal with the postorbital above the orbit. The frontal is thereby excluded from the orbital border, but, passing down medially to the prefrontal and post- orbital, still forms part of the inner wall of the orbit. Although a prefrontal- postorbital contact does not occur in Cistecephalus, these bones do approach each other in some species (C. planiceps of Broili & Schréder 1935; C. laticeps of Brink 1950), and an anterior extension of the postorbital seems to be characteristic of the genus.
The roofing-over of the intertemporal region in Cistecephaloides has been accompanied, as in Cistecephalus and Kawingasaurus, by the loss of the posterior flange of the squamosal, so that the occiput is an open, slightly convex surface and totally unlike that of other dicynodonts. Apparently related to this modifi- cation of the squamosal is the relatively low position of the zygomatic arch in the skull of both Cistecephalus and Cistecephaloides.
In the basicranium and palate similarities between Cistecephalus and Cistecephaloides are fewer, although some features appear to parallel each other functionally. Thus an apparent tendency in both genera to strengthen the basicranial girder (connecting the palate with the floor of the braincase) by the reduction or loss of the interpterygoidal vacuity has been accomplished in two different ways. In Cistecephalus the pterygoids meet in the midline in front of the internal carotid artery foramina in normal dicynodont fashion, but this contact is strengthened anteriorly by medial pterygoidal extensions which meet the rear end of the vomer, completely obliterating the interpterygoidal vacuity. In at least one species of Cistecephalus (C’. microrhinus of Keyser 1965) a small area of parabasisphenoid complex is still visible in ventral view in the midline between the pterygoids and vomer. Cox (1972) has not given a detailed illustration of this region in Kawingasaurus, but states that the interpterygoidal vacuity is absent, and from the drawings it appears that the pterygoids meet in the midline as in Cistecephalus.
In Cistecephaloides, on the other hand, the pterygoids have parted in the ventral midline but have improved their attachment to the basicranium by developing extensive posterior processes which overlap the basisphenoid to the level of the fenestra ovalis. The parabasisphenoid lies exposed in ventral view between the pterygoids, and the parasphenoid rostrum, which in dicynodonts is usually clasped by the vomers above palatal level (cf. Lystrosaurus, Cluver 1971) lies low down in the palate and is clasped by posterior diverging wings of the
THE SKULL AND MANDIBLE OF A NEW CISTECEPHALID DICYNODONT I49
vomer to form part of the median septum separating the internal nares.
The parting of the pterygoids has caused the basicranial girder of Cistece- phaloides to become unusually wide, and this may be associated with the very broad and shallow secondary palate, formed by the maxillae and premaxilla. A second result of the pterygoidal separation is the apparently incomplete closure of the interpterygoidal vacuity, since the pair of deep fossae in front of the internal carotid artery foramina, and in the position of the primitive interpterygoidal opening, may possibly lead to foramina which open out in front of the presphenoid above the level of the pterygoids.
In both Cistecephalus and Cistecephaloides the vomer lies as a low, stout septum within the choana and, meeting either the pterygoid (Crstecephalus) or the basi- sphenoid complex (Cistecephaloides) posteriorly, acts as a median strut between the palate and basicranium.
Several similarities in the palate of Cistecephalus and Cistecephaloides also exist. In both, the trough-shaped central portion of the premaxilla, which is divided by the median palatal ridge, is separated from the palatal rim by a lateral maxillary ledge, while the anterior part of the palatal rim lacks the anterior palatal ridges which are found in many dicynodonts. Also in both genera, the palatine meets the premaxilla anteriorly, so that the maxilla is excluded from the choana. However, in Cistecephaloides the ectopterygoid is reduced to a thin, lateral splint of bone, unlike the relatively large ectoptery- goid of Crstecephalus and Kawingasaurus. Associated with the small ectopterygoid in Cistecephaloides is the unusual forward extension of the pterygoid, which meets the maxilla laterally to the lateral palatal foramen.
Both Crstecephaloides and Kawingasaurus lack the two curious circular depressions in the caniniform process region of Cistecephalus, one medial to and the other lateral to the palatal rim. In Cistecephaloides, instead, the palatal rim is notched medially in two places at the caniniform process, the posterior notch being the larger and in the position of the inner circular depression found in Cistecephalus. Kawingasaurus represents a third type, as Cox (1972) states that the palate is unusually flat, and lacks sharp cutting edges.
The structure of the mandible in Cistecephalus and Cistecephaloides is basically the same. The dentary is widely excavated in the symphyseal region behind the transverse anterior cutting edge, and the dorsal dentary surface of each jaw ramus bears a pair of ridges (more pronounced and sharp in Cistecephaloides). Above the mandibular fenestra a distinct lateral dentary shelf is present. Com- parisons with Kawizngasaurus are difficult, as the symphyseal region is incom- pletely preserved, but Cox’s (1972) drawings suggest a low, blunt symphysis, and it seems unlikely that an upright, anterior cutting blade, such as in Cistece- phaloides, could have been present. All three genera, however, resemble each other in that the dorsal, dentary part of the jaw ramus is bowed out laterally so that the Meckelian sulcus is exposed in dorsal view.
From these comparisons, it appears that Cistecephaloides, Cistecephalus and Kawingasaurus all fulfilled certain common functional requirements but, by
150 ANNALS OF THE SOUTH AFRICAN MUSEUM
following slightly different, parallel pathways, arrived at their general adapta- tional level with considerable differences in their cranial structure.
FUNCTIONAL CONSIDERATIONS
The unusual skull morphology of Cistecephalus has prompted a number of workers to speculate on the most likely mode of life common to the various species of the genus. The possibility of an aquatic or semi-aquatic way of life was mentioned by Broom (1948) and Brink (1950), but Brink in 1952 described the left manus of a specimen of C. microrhinus which, he suggested, indicated a burrowing existence. Brink was led to this opinion by the fusion of the distal phalanges in the 2nd and 3rd digits, and the degeneration of the 1st, 4th and 5th digits. According to Brink the hand had only two powerful, functional fingers, and was thus well adapted for fossorial or burrowing habits.
Keyser (1965) argued that Brink (1952) had misinterpreted his specimen, stating that the digital formula in additional specimens of Cistecephalus was of the standard therapsid type, although the two distal phalanges of the 3rd and Ath digits were fused, strongly developed and elongate, and were probably equipped with strong claws during life. In addition, Keyser maintained that the 4th and 5th digits could probably appose the Ist, 2nd and 3rd digits, so that the hand was of the grasping type. Keyser concluded that these features, taken together with indications of stereoscopic vision (such as the large, anteriorly directed orbits) were strongly suggestive of arboreal activities.
Cox (1972) has interpreted the unusual pectoral girdle, humerus and fore- limb of Kawingasaurus as being adapted for powerful digging, and suggests that the animal might have rooted in mudflats in search of small invertebrates. As stated by Cox, this view is supported by the absence of sharp ‘herbivore’ cutting edges which form the palatal rim in other dicynodonts.
The skull of Czstecephaloides boonstrai does not in itself offer conclusive evidence in support, or otherwise, of any of these theories, but a consideration of the three obviously related genera does to an extent reveal that several of their specializations are the consequence of broadly similar functional require- ments. Thus it could be argued that the reduction and loss in all three genera of the interpterygoidal vacuity, together with the increase in thickness and vertical extent of the vomerine septum, represent strengthening of the basi- cranial axis of the skull to withstand fore-and-aft stress. Again, the broadly roofed-over intertemporal region strengthens the overall skull structure, but this feature of the cistecephalids has an important corollary, and may not have arisen as a purely reinforcing device. Thus, the relatively lateral position of the cistecephalid temporal fenestra, a direct consequence of the broadened inter- temporal region, must have brought about considerable changes in the jaw adductor musculature, the deep part of which arose from the bones surrounding the fenestra. Crompton & Hotton (1967) have reconstructed this muscle, the external adductor muscle, in dicynodonts, and suggest the dorsal surface of the postorbital and posterior flange of the squamosal (being respectively the
THE SKULL AND MANDIBLE OF A NEW CISTECEPHALID DICYNODONT nye
medial and posterior margins of the fenestra) as the most likely areas of origin. The moulding of these bones in dicynodonts, especially of the postorbital behind the postorbital bar, makes such an arrangement very probable. Cromp- ton & Hotton (1967) have indicated the dorsal surface of the dentary and, possibly, surangular as the main insertion areas of this muscle, and suggest that a portion may have extended a short way down the medial side of the mandible.
Of interest in this connection is the fact that in Cistecephalus, Cistecephaloides and (judging by Cox’s 1972 illustrations) AKawingasaurus, the dorsum of the dentary is bowed out laterally and the inner surface, together with that of the surangular, moulded to form a wide, shallow trough leading down to the exposed Meckelian sulcus. ‘The sulcus, bounded by prearticular and splenial medially and angular and dentary laterally, is in a position equivalent to the opening described as the adductor fossa by Romer (1956) in reptiles, and which is primitively the major insertion area of the external jaw adductor muscles. The condition in cistecephalids can, it appears, be best explained in terms of a fairly substantial slip of the medial external adductor muscle inserting on the bones surrounding the Meckelian sulcus in dicynodonts, while in dicyno- donts with narrow or medium-width intertemporal regions this adductor fossa would be an easily accessible insertion area for such a muscle, in cistecephalids the inner surfaces of the dentary and surangular above the sulcus and mandibu- lar fenestra would inevitably have been influenced by the lateral migration of the muscle’s temporal area of origin, and the course of the fibres is thus more clearly shown in them than in other dicynodonts.
Thus reconstructed, three sets of external adductor muscles appear to have inserted on the mandible in cistecephalids and presumably in other dicynodonts as well. An inner division of the medial external adductor muscle inserted in the adductor fossa, while, as reconstructed in dicynodonts by Crompton & Hotton (1967) and Cluver (1970, 1971), a lateral division of the medial external adductor muscle inserted on the dorsal dentary surface. (In Cistecephaloides this surface is slightly excavated to receive the muscle’s insertion.) On the outer side of the dentary a lateral dentary shelf is developed in Cistecephalus and Cistecephaloides, and the lateral external adductor muscle would have inserted here. Cox (1972) does not mention a dentary shelf in Kawingasaurus, and the muscle would presumably have inserted on the outer side of the dentary, as suggested for Dicynodon testudirostris by Cluver (1970).
The effect of widening of the skull roof in cistecephalids would, it seems, have resulted in an overall lateral shift of the areas of origin of the external adductor muscles relative to their areas of insertion. A result of this would be increased control by these muscles over any lateral movement of the mandible across the palate. Indications that such movement was in fact possible are provided in the structure of the palate and the mandible, which differ in important respects from the condition in other dicynodonts. Among forms in which the jaw musculature and jaw movements have been reconstructed
152 ANNALS OF THE SOUTH AFRICAN MUSEUM
Emydops (Crompton & Hotton 1967) could have utilized only a straightforward anteroposterior displacement during the masticatory stroke, guided by adjacent rows of upper and lower postcanine teeth, while a step in front of the articula- tion surface of the articular condyle prevented excessive sliding over the quadrate. In Lystrosaurus (Crompton & Hotton 1967; Cluver 1971) the anterior symphyseal region of the lower jaw was guided by the tight-fitting tusks, canini- form processes and palatal rim, and the quadrate and articular surfaces are closely-matched counterparts which appose any tendency for lateral displace- ment during the jaw movements. In a third group, represented by Dicynodon testudirostris (Cluver 1970), strongly developed caniniform flanges act as lower jaw guides, and enlarged crushing areas on the dorsal surface of the dentary meet apposing surfaces on the palate. In effect, in all these forms the dentary is so controlled as to meet specific areas of the palate during each masticatory stroke.
In cistecephalids, on the other hand, no such restrictions on the mandibular movements appear to be developed. The articular surface of the quadrate has two rounded condyles, but they are not separated by a deep cleft as in Emydops or Lystrosaurus, and there is considerable freedom of movement between it and the articular bone. No bony step, such as is found in Emydops, is present in cistecephalids, and, in Cistecephalus at least, the articular surface continues from the articular itself on to the surangular (this area is not clearly preserved in Cistecephaloides or described in Kawingasaurus). Anteriorly, the palatal rim is low in Cistecephalus and Cistecephaloides and evidently almost absent in Kawinga- saurus. In none of the genera is a proper caniniform process found. The palate itself is wider than the symphyseal region of the lower jaw, and no crushing dentary tables are developed to appose raised areas of the palate. A sharp, spade-like anterior termination of the lower jaw is found in Cistecephalus and Cistecephaloides, but in Kawingasaurus the mandible appears to have been blunt anteriorly. In Cistecephalus and Cistecephaloides, and possibly in Kawingasaurus, the mandible is too short to meet the anterior part of the palate in ‘beak bite’ (Crompton & Hotton 1967), while contact between mandible and the lateral palatal rim could not have been achieved without sideways movement of the lower jaw.
In brief, then, the possibility of controlled lateral movements of the lower jaw during the masticatory cycle cannot be ruled out in the three cistecephalid genera, and there is in fact some positive evidence to show that this type of movement did take place, and was associated with the broadening of the intertemporal region.
The broad, open and slightly convex occiput of cistecephalids, which contrasts sharply with the condition in other dicynodonts, might be considered next. The loss of the posterior flange of the squamosal, which served as part of the area of origin of the medial external adductor muscle in other dicynodonts, might have been associated with the roofing-over of the temporal region, whereby the root of the zygomatic part of the squamosal lies virtually within the
eer” rOeewee—
THE SKULL AND MANDIBLE OF A NEW CISTECEPHALID DICYNODONT 153
temporal vault. Alternatively, the shape of the occiput may well have been determined by changes in the neck musculature. Here comparison with recent mammals yields suggestive results. Among insectivores, inferences may be drawn from the skull of the Cape Golden Mole (Chrysochloris asiatica). In this fossorial form the rounded occipital region, lacking pronounced occipital and sagittal crests, is reminiscent of the condition in cistecephalids, and has evi- dently been strongly influenced by development of the powerful neck and shoulder muscles (Dobson 1882) required for its digging and burrowing way of life. Gregory (1951) discussed the muscular and osteological adaptations in a number of insectivores, including the Oligocene Apternodus, where a special squamosal flange in the temporal region served to accommodate forwardly extended neck and forelimb muscles.
Direct and detailed comparisons between dicynodonts and _ fossorial mammals must inevitably be undertaken with caution, but it is tempting to link the insectivore-like occiput in cistecephalids with the digging adaptations in the forelimb of Kawingasaurus, described by Cox (1972). The manus in Kawingasaurus is unfortunately not known, but if the fused and elongated distal phalanges of two digits of the hand in Cistecephalus are taken into consideration, together with evidence of adaptations for skull strengthening and reinforcement in all three cistecephalid genera, then it seems reasonable to suggest that an at least semi-fossorial mode of life was generally selected for in the family. That this direction of specialization was accompanied by basic changes in mastication is suggested by the unusual palate—mandible relationships and the changed orientation of the medial division of the external jaw adductor muscles. However, differences between the palate and mandible of Aawingasaurus on the one hand and Cistecephalus and Cistecephaloides on the other show that more than one type of feeding activity was developed within this adaptive framework. While the varying degree of commitment to digging and perhaps burrowing activities in the three genera can be determined only when the postcranial skeleton of each group becomes better known, it does seem that the sharp, shovel-shaped jaw symphysis of Czstecephaloides, and to an extent Cistecephalus, indicates dietary preferences quite different to those of the possibly carnivorous Kawingasaurus, which (Cox 1972) may have dug for small, mudflat-dwelling invertebrates.
SUMMARY
The skull and mandible of a new dicynodont genus are described and figured. It is concluded that the new form, for which the name Cistecephalozdes boonstrai is proposed, is closely allied to Cistecephalus and the newly established Kawingasaurus (Cox 1972) and that a family, Cistecephalidae, should be retained to include these three genera. Cistecephaloides is unusual in the small size of the orbits, the prefrontal-postorbital contact, and the separation of the pterygoids in the ventral midline, with consequent incomplete closure of the interptery-
154 ANNALS OF THE SOUTH AFRICAN MUSEUM
goidal vacuity. Comparisons between the three cistecephalid genera indicate the possibility of parallel evolution, and it is suggested that digging and perhaps burrowing activities were characteristic of this family, and could account for the specializations encountered in the group.
ACKNOWLEDGEMENTS
I am grateful to Professor M. E. Malan for her helpful suggestions and critical reading of the manuscript. Mrs I. M. Chesselet of the Department of Palaeontology was responsible for the skilful preparation of much of the cistece- phalid material used in this investigation.
REFERENCES
AGNEw, J. D. 1959. Cranio-osteological studies in Dicynodon grimbeeki with special reference to the sphenethmoid region and cranial kinesis. Palaeont. afr. 6: 77-107.
Brink, A. S. 1950. On a new species of Cistecephalus Owen. Ann. Mag. nat. Nist. (12) 3: 985-997.
Brink, A. S. 1952. Studies on Karoo reptiles. III. The manus of Cistecephalus. S. Afr. F. Sci. 49: 13-15.
Brom, F. & ScHRODER, J. 1935. Beobachtungen an Wirbeltieren der Karrooformation. VI. Uber den Schadel von Cistecephalus Owen. Sher. bayer. Akad. Wiss. 1935: 1-20.
Broom, R. 1903. On the classification of the theriodonts and their allies. Rep. S. Afr. Ass. Adumt ie, $19 160),
Broom, R. 1932. The mammal-like reptiles of South Africa and the origin of mammals. London: Witherby.
Broom, R. 1948. A contribution to our knowledge of the vertebrates of the Karroo Beds of South Africa. Trans. R. Soc. Edinb. 61: 577-929.
Criuver, M. A. 1970. The palate and mandible in some specimens of Dicynodon testudirostris Broom & Haughton (Reptilia, Therapsida). Ann. S. Afr. Mus. 56: 133-153.
Cruver, M. A. 1971. The cranial morphology of the dicynodont genus Lystrosaurus. Ann. S. Afr. Mus. 56: 155-274.
Cox, C. B. 1959. On the anatomy of a new dicynodont genus with evidence of the position of the tympanum. Proc. zool. Soc. Lond. 132: 321-367.
Cox, C. B. 1972. A new digging dicynodont from the Upper Permian of Tanzania. In: JoysEy, K. A. & KEMP, T. Ss. eds. Studies in vertebrate evolution: 173-189. Edinburgh: Oliver & Boyd.
Crompton, A. W. & Hotton, N. 1967. Functional morphology of the masticatory apparatus of two dicynodonts (Reptilia, Therapsida). Postilla 109: 1-51.
Dosson, G. E. 1883. A monograph of the Insectivora, systematic and anatomical. Part I1. London: John van Voorst.
Grecory, W. K. 1951. Evolution emerging. New York: Macmillan.
Haueuton, S. H. & Brink, A. S. 1954. A bibliographic list of Reptilia from the Karoo Beds of South Africa. Palaeont. afr. 2: 1-187.
Keyser, A. W. 1965. The morphology of the anomodont genus Cistecephalus Owen Myo (M. Sc. thesis, University of Pretoria.)
Kitcuina, J. W. 1972. A short review of the Beaufort zoning in South Africa. In: INTERNATIONAL UNION OF GEOLOGICAL SCIENCES. Commission on Stratigraphy. Subcommission on Gondwana Strati- graphy and Palaeontology. Gondwana Symposium 2nd, South Africa, 1970. Proceedings and papers: 309-311. Pretoria: C.S.I.R.
Owen, R. Descriptive and illustrated catalogue of the fossil Reptilia of South Africa in the collection of the British Museum. London: British Museum.
Romer, A. S. 1966. Vertebrate paleontology. Chicago: University of Chicago Press.
SEELEY, H. G. 1894. Researches on the structure, organisation and classification of the fossil Reptilia. Part IX, section I: On the Therosuchia. Phil. Trans. R. Soc. (B) 185: 987-1018.
THE SKULL AND MANDIBLE OF A NEW CISTECEPHALID DICYNODONT
ang. art. bas. boc. den. ect. e€0c. ept. ar
1.C.a.
ip. ipt.vac. jug.
jug. for. lac. lac.for. lat.den.l. lat.pal.for. lat.r. max. med. r.
ABBREVIATIONS angular nas. articular op. basisphenoid os. basioccipital pa. dentary pal. ectopterygoid par. exoccipital pas epipterygoid pfr. frontal pm. foramen for internal carotid po. artery pro interparietal prs interpterygoidal vacuity pt. jugal pt.f. jugular foramen q. lacrimal qj. lacrimal foramen soc. lateral dentary ledge smx. lateral palatal foramen spl. lateral ridge on dentary sq. maxilla st. medial ridge on dentary vo.
nasal opisthotic orbitosphenoid parietal palatine prearticular parasphenoid rostrum prefrontal premaxilla postorbital prootic presphenoid pterygoid posttemporal fenestra quadrate quadratojugal supraoccipital septomaxilla splenial squamosal stapes
vomer
BaD
A NEW DICYNODONT (REPTILIA, THERAPSIDA) FROM CYNOGNATHUS ZONE DEPOSITS OF SOUTH AFRICA?!
By NicHotas Hotton III Smithsonian Institution, Washington, D.C. 20560
(With 2 plates, 1 figure and 1 table)
CONTENTS PAGE Introduction . F : : ‘ : 157 Taxonomy ; : : 3 : : 158 Discussion ‘ , ; : : , 161 References 5 ; : ; : ‘ 165 INTRODUCTION
A small dicynodont skull, complete with lower jaw and stapes, was collected by J. W. Kitching about 1,6 km south of Lady Frere, Cape Province, on the north side of the Cacadu River in 1961, during the writer’s first visit to South Africa. It was originally catalogued as number 22936 in the collections of the National Museum of Natural History, Smithsonian Institution, but as it must be designated the type of a new genus, it has been transferred to the collections of the Bernard Price Institute for Palaeontological Research, Johannesburg, where it is catalogued as Museum Number 430.
The specimen is clearly a member of the Cynognathus zone fauna, for it was found in the same shale layer as a good skull of the Cynognathus zone gomphodont Diademodon, about 275 m west of the latter. No. 430 lay right side up, about 0,3 m below a layer of fine gray sandstone that was about 0,6 m thick. Although the skull is markedly distorted by dorsoventral flattening, preservation of the bone is good and I do not think it possible that the specimen was reworked from older deposits. The area is part of Lady Frere Commonage, which has yielded numerous specimens characteristic of the Cynognathus zone but nothing to suggest the presence of any other therapsid faunal element, either by survival or by reworking of older sediments.
The specimen is definitely not a member of the genus Kannemeyeria, which until now was the only dicynodont present in the Cynognathus zone fauna. It is also distinguishable from other dicynodont genera, and so constitutes a second, very rare, dicynodont member of the Cynognathus fauna.
' The research upon which this paper is based was supported by a National Science Founda- tion Grant, G 14705.
15/7
Ann. S. Afr. Mus. 64, 1974: 157-165, 2 pls, 1 fig., 1 table.
158 ANNALS OF THE SOUTH AFRICAN MUSEUM
‘TAXONOMY Class REPTILIA Order THERAPSIDA Family Dicynodontidae Genus Kombuisia* gen. nov. Kombuisia frerensis® sp. nov. Genoholotype. No. 430. Skull with articulated stapes and lower jaw, strongly distorted by dorsoventral flattening, but very little cracked or broken; zygomae
lost. Sub-periosteal bone very well preserved; little evidence of surface cracks due to subaerial weathering.
Generic diagnosis. Small dicynodont (see Table I for dimensions) without post- canine teeth, much like Kingoria (Cox 1959) in the following features. Inter-
TABLE I
Measurements of Kombuisia frerensis, No. 430, TYPE, in millimetres. Greatest height and height at snout omitted because of crushing.
Overall length, occipital condyle to tip of premaxillary beak (L) . : ‘ : 76,8 Width at postorbital bar (W) . : : : : : : : . 4 52.3 Smallest interorbital distance (Wio) . ‘ : : : 16,1 Intertemporal width, anterior end of isieniompauel bar (Wit) ; 2 : : L267, Width across paroccipital processes (Wo) . : : , : Boe Bottom of occipital condyle to top of occipital sieie. deer (Ho) : ; 18,6 Horizontal distance, anterior margin of orbit to tip of premaxillary beak Raw) . } 15,9 Horizontal distance, posterior margin of orbit to tip of premaxillary beak (Rpo) . ; 35,1 Length of temporal fossa, medial side (Rt) . : ; , ‘ 27,5 Length of intertemporal bar, postorbital bar to top af paciiot (Rit) 5 ; 26,0 Posterior width of mouth, measured between posteromedial margins of eespisiNioinn processes at level of anterior maxillary margin (P,) ; ; : 20,0 Length of mouth, along midline from P, to tip of premaxillary beak (P a) 6 “ ‘ 23,0
orbital bar wider than intertemporal bar, and length of postorbital region nearly equal to distance from postorbital bar to snout (Plate 1 A). Nasal boss expressed only as a low, rugose median swelling on frontals, nasals, and pre- maxilla; no postfrontal bone; preparietal fusiform, long and narrow; post- orbitals cover only about three-quarters of medial sides of temporal fenestrae; parietals exposed dorsally as narrow ridge between postorbitals. Caniniform processes small, rounded. Secondary palate very long, choanae well behind back of caniniform processes (Plate 2 A). Secondary palate consists almost entirely of premaxilla, which is in contact with small palatal process of palatine bone and blocks maxilla from contact with choanal margin (Plate 2 A). Palatine makes very minor contribution to secoihdary palate; ectopterygoid separates maxilla from pterygoid. Insertion of temporal musculature on lateral face of dentary produced laterally to form prominent shelf or ‘wing’ (cf. Cox 1959).
* The genus is named for its discoverer, J. W. Kitching, who during his military service bore
the nickname ‘Kombuis’, Afrikaans for kitchen. * The specific name refers to the type locality of Kombuisia.
A NEW DICYNODONT FROM CYNOGNATHUS ZONE DEPOSITS T59
Dentary symphysis long. External naris large, distance from front of orbit about twice diameter of external naris, which lies very close indeed to margin of upper jaw.
Differs from Aingoria, and from other dicynodonts as well, in configuration of dentary. Symphysis shallow, broad, flat dorsally; ventral profile slopes upward and forward very gently from splenial margin to tip of beak, so that general appearance of front of beak is spatulate (Plate 2 C). Front and side margins of beak sharp. In dorsal aspect (Plate 2 B) the symphysis is not parallel-
PLATE 1. Aombuisia frerensis. No. 430.
A. Skull, dorsal aspect. B. Parietal region, enlarged to show condition of bone surface and absence of pineal opening.
160 ANNALS OF THE SOUTH AFRICAN MUSEUM
sided as in most dicynodonts, but tapers forward to a squared-off anterior margin. Median groove between dentary tables, which is deep and distinct in nearly all dicynodonts (Crompton & Hotton 1967), is very shallow in Kombuisia; in fact, dentary tables are separated from median groove only by a pair of low, sharp ridges that run straight back from the corners of the squared-off anterior margin of the beak. Lateral and parallel to each of these ridges is a low, rounded ridge
PLATE 2. Kombuisia frerensis. No. 430. Stereo pairs. A. Skull, palatal aspect. B. Dentary, dorsal aspect. C. Dentary, left lateral aspect.
A NEW DICYNODONT FROM CYNOGNATHUS ZONE DEPOSITS 161
that rises at the symphyseal margin about half-way between the tip of the beak and the splenial margin, and runs back to disappear a little behind the splenial margin.
DiIscussION
The most distinctive single feature of the type of Kombuisia is that it has no pineal opening. In the type specimen, the pineal region of the skull roof is virtually undistorted and the surface of the bone is undamaged (Plate 1 B); there is thus no evidence of pathology and the lack of a pineal opening appears to be the normal condition. A second specimen (National Museum, Bloemfon- tein, C. 3006, excellent stereophotographs of which were provided by A. R. I. Cruickshank) comes from a Cynognathus zone locality and is clearly referable to Kombuisia on the basis of palate and dentary symphysis. Unfortunately the Bloemfontein specimen consists of snout and dentaries only, and the presence or absence of a pineal opening cannot be confirmed. Because the lack of a pineal opening is so unexpected in dicynodonts, and because at present it is indicated in Kombuisia by only the one specimen, I prefer to delay incorporating this character into the formal diagnosis of the genus until more evidence comes to light.
The two identified specimens of Kombuisia are very nearly the same size, and are slightly less than half as large as the genotype of Kingoria, University of Tubingen, catalogue number K. 12. This specimen, as well as those described by Cox, appear to have been adult at the time of death, as does the type of K. galecephalus (Broom & Robinson 1948), Rubidge Collection, catalogue number 97.* The type of Kombuisia, though smaller, does not appear to have been appreciably younger at time of death than the various specimens of Kingoria, and it is therefore probably close to the adult size of the genus.
Kombuisia and Kingoria are most usefully compared with the more common Beaufort dicynodonts, animals of small to moderate size that lack postcanine teeth, including Dicynodon as it is generally accepted at the present time, and Lystrosaurus. The genus Dicynodon is under study by the writer and M. A. Cluver, and there is every indication that the name as currently used includes two distinct genera. However, differences between these groups are not critical to comparison with Kombuisia and Kingoria, and for present purposes the name Dicynodon is employed in its current usage. The informal term ‘common dicyno- donts’ as used in the remainder of this discussion includes both Dicynodon, in this inclusive sense, and Lystrosaurus.
Neither the type nor the Bloemfontein specimen of Kombuisia has any trace of tusks. ‘The caniniform processes are small and the snout is short and delicate. These facts alone are not sufficient to indicate that tusks were absent in all
4 Listed by Haughton & Brink (1954) as catalogue number 77 (in errore). Number 77 is the type of Dicynodon clarencei Broom, which Haughton & Brink list incorrectly as number 97. The
two specimens are quite distinctive, and the error seems to be simply a matter of catalogue numbers having been switched inadvertently.
162 ANNALS OF THE SOUTH AFRICAN MUSEUM
individuals of the genus for, as Broom notes (1935), tuskless specimens of Dicynodon grimbeekt have more lightly-built snouts than tusked specimens. Cox reports that only two of ten specimens of Aingoria at his disposal had tusks, but describes no differences between the snouts of tusked and tuskless individuals. However, in Kombuisia the snout is more delicate and the caniniform processes are proportionately smaller than in most other dicynodonts, including tuskless specimens of Kingoria and D. grimbeeki, and it seems unlikely that any individuals of Kombuisia had tusks. The palates of Kombuisia and Kingoria are distinctive in their lack of relief. In most dicynodonts (cf. Ewer 1961; Crompton & Hotton 1967; Cluver 1970, 1971) there is a pair of strong ridges on the back of the premaxillary beak, one on either side of the midline, and the contact between the vomer and the pre- maxilla is produced into a prominent midline ridge. In Kombuisia and Aingoria the premaxillary ridges are lacking and the midline vomerine ridge is much subdued; there is, however, a low, longitudinal ridge just medial to each maxillo-premaxillary suture (Fig. 1: LR) which is not present in other dicynodonts. The punctate and rugose sculpture on the beaks of dicynodonts is generally agreed to indicate the presence, in life, of a horny covering, because of its resemblance to comparable areas in birds and chelonians. On the rostral
Fig. 1. Kombuisia frerensis. No. 430.
Skull, palatal aspect: AR, articular; LR, longitudinal ridge at
lateral edge of premaxilla; MR, median ridge at junction of
premaxilla and vomer; PAL, palatine; PMX, maxillo-pre- maxillary suture; Q , quadrate.
A NEW DICYNODONT FROM CYNOGNATHUS ZONE DEPOSITS 163
surface of the snout and on the dentary symphysis the distribution of this covering must have been relatively uniform throughout the Dicynodontia; it is much the same in Kombuisia and Kingoria as in other dicynodonts. On the buccal surfaces of the palate, however, it shows much greater variation; Kombuisia and Azngoria represent one extreme and common dicynodonts the other, and these extremes parallel the differences in relief of the palate. In common dicynodonts the entire buccal surface of the palate is punctate, even to the hollows between the ridges, and was presumably entirely covered by horn. In Kombuisia and Kingoria, punctate sculpture, and hence a putative horn covering, is restricted to the buccal surfaces of premaxilla and maxillae, to the latero- posterior corners of the premaxilla, and to the small palatal processes of the palatines adjacent to the latter. This leaves most of the premaxillary secondary palate smooth and presumably covered by mucous membrane, much like the mesial surface of the palate in chelonians. Because punctate sculpture is not developed medial to the low, longitudinal ridges at the lateral margins of the premaxilla in Azngoria, Cox (1959) suggests that these ridges merely reflect the medial margins of horn covering. The structure of Kombuisia should probably be interpreted in the same way.
In the common dicynodonts, the horn-covered secondary palate, with its elaborate system of ridges and rugosities, is associated with a mouth of trenchant appearance: maxillary and premaxillary margins are blade-like, the dorso- lateral margins of the dentary symphysis are squared off, and the tip of the dentary beak supports a transversely oriented blade. This arrangement forms a system of blades and surfaces on the upper and lower jaws (Crompton & Hotton 1967) that comminuted food by a shearing action, the force being provided by retraction of the jaw; it probably arose co-ordinately with the retractive sliding action of the jaw that defines the Dicynodontia as a group. In various pristerodonts and Lystrosaurus (Crompton & Hotton, 1967), and in most other dicynodonts as well, the jaw joint is characteristically very unstable in full protraction, because the bearing surfaces of the joint that are in contact during full protraction, i.e. the posterior surface of the articular and the front of the quadrate, are both convex. Because of this instability, the vertical force that could be exerted on the jaw at this time was limited to the small vertical component of the very obliquely directed anterior fibres of the temporalis musculature, which served primarily to initiate elevation of the jaw. Elevation and retraction were concurrent, but the bulk of force exerted on the lower jaw in early stages was horizontal, and the motion that was most effective in shearing was retraction. The more vertically oriented posterior temporalis fibres must have been relaxed at protraction, and came fully into play only as the jaw approached full retraction.
In Kombuisia and Kingoria, in which the secondary palate lacks horn covering and marked relief, the appearance of the mouth is less trenchant than in most other dicynodonts. Although maxillary and premaxillary margins are sharp in Kombuisia, the shallow palate is occluded by a flattened, spatulate
164 ANNALS OF THE SOUTH AFRICAN MUSEUM
dentary. The lateral premaxillary ridges are not comparable in either position or prominence to the ridges of common dicynodonts. In Aingoria, both upper and. lower jaw margins appear to be rounded and smooth, and the tip of the beak tapers to a blunt point. The palate is deeper than that of Kombuisia but is equally featureless.
Reduction of trenchancy of mouth and palate suggests marked modifica- tion of the shearing function in Kombuisia and Kingoria. In Kingoria this modifica- tion does not seem to have affected the jaw mechanism, which is much the same as in other dicynodonts, but in Kombuisia it apparently extends to the jaw joint. The lower jaw of the type specimen of Kombuisia is articulated in full protraction, and the right articuloquadrate joint is quite well preserved and exposed. The retroarticular portion of the articular bone is produced backward so that the lower (and posterior) part of the bearing surface lies beneath the quadrate in full protraction (Fig. 1: AR), rather than in front of it as in most dicynodonts. The posterior bearing surface is therefore somewhat concave, so that the surfaces of quadrate and articular are more nearly congruent and the joint is correspondingly more stable when the jaw is fully protracted. Thus it is possible that in Kombuisia greater vertical force could be applied at the beak much earlier in the cycle, perhaps even during full protraction.
The mouth of Kombuisia more than that of any other dicynodont is like the mouth of many chelonians, in the characteristic flatness of the dorsum of the dentary symphysis and the restriction of horn to the margins of the palate. Furthermore, the ridges at the lateral margins of the premaxilla (Fig. 1: LR) appear to meet the dentary tables, i.e. the grooves between each of the pairs of lateral ridges on the dorsum of the dentary (Plate 2 B), when the jaw is at or close to full protraction. The dorsum of the symphysis thus appears to be more nearly congruent to the buccal surface of the shallow palate than in most dicynodonts, another characteristic reminiscent of chelonians. There is no doubt that Kombuisia retained the retractive sliding mechanisms of the dicyno- dont jaw, but the stability of the joint during protraction, coupled with morpho- logical features of the mouth, suggest that it had superimposed a more turtle- like anterior bite upon the basic dicynodont jaw mechanism.
Conclusions about the habits of Kombuisia must await the appearance of new material, for verification of the patterns described here and for information as to the postcranial structure of the animal. The possibly superficial resem- blance of the feeding mechanism to that of chelonians is not very useful in this context, for chelonians exhibit the widest range of habitats and habits of living reptiles. However, the fact that the clearest distinctions among Kombuisia, Kingoria and most other dicynodonts are associated with the feeding mechanism indicates that dicynodont radiation, like dicynodont origin, was primarily a matter of exploiting the possibilities of different modes of feeding.
—_ —_— ll
A NEW DICYNODONT FROM CYNOGNATHUS ZONE DEPOSITS 165
REFERENCES
Broom, R. 1935. A new genus and some new species of mammal-like reptiles. Ann. Transv. Mus. 18: I-12.
Broom, R. & Rosinson, J. T. 1948. Some new fossil reptiles from the Karroo beds of South Africa. Proc. zool. Soc. Lond. 118: 392-407.
Ciuver, M. A. 1970. The palate and mandible in some specimens of Dicynodon testudirostris Broom and Haughton (Reptilia, Therapsida). Ann. S. Afr. Mus. 56: 133-153.
Ciuver, M. A. 1971. The cranial morphology of the dicynodont genus Lystrosaurus. Ann. S. Afr. Mus. 56: 155-274.
Cox, C. B. 1959. On the anatomy of a new dicynodont genus, with evidence of the position of the tympanum. Proc. zool. Soc. Lond. 132: 321-367.
Crompton, A. W. & Horton, N. 1967. Functional morphology of the masticatory apparatus of two dicynodonts (Reptilia, Therapsida). Postilla 109: 1-51.
Ewer, R. F. 1961. The anatomy of the anomodont Daptocephalus leoniceps (Owen). Proc. zool. Soc. Lond. 136: 375-402.
Haucuton, S. H. & Brink, A. S. 1954. A bibliographical list of Reptilia from the Karroo beds of Africa. Palaeont. afr. 2: 1-187.
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS
By
A. J. CHARIG
British Museum (Natural History), London
&
A. W. CROMPTON
Museum of Comparative Xoology, Harvard University, Cambridge, Mass.
(With 11 figures.)
For more than half a century the only pre-Jurassic ornithischian known was Geranosaurus atavus Broom, 1911, an incomplete lower jaw (with some doubt- fully associated postcranial bones) from the Upper Triassic Cave Sandstone of South Africa. In 1962, however, Crompton & Charig gave a preliminary
description of an almost complete and extremely well-preserved skull, also
from the Cave Sandstone, which they named Heterodontosaurus tuckt. One of the most remarkable features of Heterodontosaurus is its dentition, shown in Figures 10 and 11. There are (a) simple pointed premaxillary teeth working against a toothless predentary, () one enlarged pointed caniniform tooth in each jaw, this being the last premaxillary tooth in the upper jaw and the first dentary tooth in the lower, and (c) more complex, closely packed maxillary and dentary teeth, with ridged lateral and medial surfaces and with enamel on one side only, abrading and sharpening each other like opposing chisels to produce oblique, flat, continuous occlusal surfaces. ‘These cheek teeth are set in from the side of the face. Most curious of all, there seems to be no tooth replacement whatever in the type skull of H. tuck: or in another nearly complete skull of the same species collected subsequently (S.A.M. No. K1332). In their original description Crompton & Charig noted also the verbal observation of their colleague, J. Attridge, that an incomplete dentary from the underlying Red Beds (Figs 8, 9), described by Haughton in 1924 as Lycorhinus angustidens gen. et sp. nov. and believed to belong to a cynodont, possessed a dentition very like that of Hetero- dontosaurus and probably represented an ornithischian dinosaur rather than a therapsid. Romer (1966) classified Lycorhinus as an ornithischian, placing it tentatively in the family Hypsilophodontidae of the suborder Ornithopoda. A very detailed description of the holotype skull of Heterodontosaurus tucki has now been prepared by Charig & Crompton (in press) and it is intended that a similar description of S.A.M. No. K1332 shall follow.
Since 1962 Upper Triassic ornithischians have been turning up more often, mostly in the Stormberg Series of southern Africa (Republic of South Africa and Lesotho) but elsewhere too. Until now, however, only one other Stormberg ornithischian has been named—Fabrosaurus australis Ginsburg, 1964—and once again the holotype consists of nothing more than a fragment of a dentary from
167
Ann. S. Afr. Mus. 64, 1974: 167-189, 11 figs.
168 ANNALS OF THE SOUTH AFRICAN MUSEUM
the Red Beds. This has virtually no distinguishing features other than the highly characteristic features of its teeth, which are clearly very different from those of Heterodontosaurus. In Fabrosaurus—if we assume its complete dentition to have been like that of similar specimens discovered later (Fig. 1) —the heterodonty is less marked, without development of caniniforms; the more posterior teeth are of a leaf-like shape, probably not very different from the unworn teeth of Heterodontosaurus but more widely spaced and never worn down to a flat con- tinuous occlusal surface; the cheek teeth are not set in from the side of the face, and normal alternating reptilian replacement appears to occur. We think it very likely that such teeth will prove to be common to several early ornithischian genera as well as to persistently primitive forms of later date, e.g. Echinodon Owen, 1861 from the Purbeck Beds of England (on the Jurassic—Cretaceous boundary). Indeed, this ‘fabrosaur’ type of tooth is accepted by all workers as primitive for the Ornithischia; Bakker & Galton (1974) are even claiming that the basic pattern of the dentition in primitive ornithischians (such as Fabrosaurus) was similar to that of small prosauropods. But if this characteristic tooth structure is diagnostic of a whole family (or of an even higher taxon) rather than of a single genus, then the nominal genus Fabrosaurus must be indeterminate. The names Fabrosaurus and F. australis must therefore be regarded as nomina dubia; no other material should be referred to the species, and no other species to the genus.
Thulborn has nevertheless described more ornithischian material from the Stormberg Series, including both skull (1970b, 1971a) and postcranial elements (1972), and he refers it all to Fabrosaurus australis; he does so because of the ‘highly distinctive’ nature of the teeth (Fig. 1). But for the reasons given above, and because it is evident from our own collections that the Stormbergs of southern Africa contain other, hitherto undescribed ornithischians, we cannot accept that Thulborn’s reference of this material to F. australis is justified — even though the reptile he described is again very different from Heterodonto- saurus. We should prefer to call it ‘fabrosaurid gen. et sp. indet.’.
It is therefore apparent that all the Triassic ornithischians that have been found in southern Africa up till now (even the little-known Geranosaurus) may be classified as either ‘Heterodontosaurus-like’ or ‘Fabrosaurus-like’, according to the nature of their teeth. The rest of this paper is mainly concerned with those resembling Heterodontosaurus in this respect.
Our greatest concern is with Thulborn’s contention (1970a) that Lycorhinus angustidens and Heterodontosaurus tucki are congeneric, i.e. that the nominal genus Heterodontosaurus Crompton & Charig, 1962 is a subjective junior synonym of Lycorhinus Haughton, 1924, is therefore invalid and should be discarded. (At the same time he agrees that the skull from the Cave Sandstone merits specific separation as L. tuckt.) In consequence he now uses the name Lycorhinus instead of Heterodontosaurus in his published discussions on ornithischian phylo- geny (19704, 19714, b, 1972), sometimes adding the name ‘Heterodontosaurus’ in parentheses or square brackets with or without inverted commas. He also con-
YT!
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS 169
tends (19704, 19710) that both valid genera of ornithischians from the Upper Trias of southern Africa (i.e. Lycorhinus and Fabrosaurus, according to him) should be placed in the family Hypsilophodontidae, the type-genus of which is Wealden in age; and (19710) that these Upper Triassic forms should be classified further into two subordinate ‘groups’ of that family, to be called ‘lycorhinids’ and ‘fabrosaurs’ respectively.
In this connection it should be mentioned that in Britain and the U.S.A. there are—apart from Thulborn himself—four established palaeontologists who have worked on primitive ornithischians within the last few years: Charig, Crompton, Galton and Ostrom. None of the four accepts Thulborn’s synonymy. Nor does Cluver, Curator of Fossil Vertebrates in the South African Museum, the museum where all the material of both genera is housed. Why, then, should we bother to discuss the synonymy further? We do so partly to determine our position in relation to such a declared synonymy, whether we are free to accept it or reject it as we think fit, and partly to consider, if we are free to reject it, whether we should be justified in so doing. But we are less concerned with resolving this particular case than with using it as an illustration of certain general points, especially the inadvisability of publishing a synonymy without logical consideration of all the available evidence and without due regard for its taxonomic and nomenclatural consequences.
What, then, are our opinions of the generic synonymy proposed by Thulborn ? First, we believe that such a synonymy can serve no useful purpose whatever; indeed, from a practical point of view it is highly undesirable. We shall explain this further below. Secondly, it seems to us that the synonymy is based upon alleged similarities for which the evidence, in many instances, is either incorrect or not pertinent or both; that, because of the fragmentary nature of the Lycorhinus material, the only useful evidence concerns the lower teeth; that there is as much evidence of differences between the teeth of the two genera as of similarities; and that this severe limitation of the comparison, coupled with our lack of knowledge of the variability of the tooth structure of ‘Triassic ornithischians, would probably make it difficult, if not impossible, to estimate the closeness of the relationship between the two holotypes. Did they belong to the same genus, as claimed by Thulborn, or only to the same sub- family? Perhaps only to the same family? Obviously this problem requires detailed consideration.
Before embarking on this, however, we shall deal briefly with Thulborn’s other contention, concerning his two special ‘groups’ within the family Hypsilo- phodontidae. These, being suprageneric, are manifestly of subfamilial rank and should be recognized as such; yet one of the names he gives to those sub- families is incorrect in its stem, irrespective of whether or not his synonymy be accepted, and both names are incorrect in their terminations (even in their vernacular forms). Our first objection to Thulborn’s subfamily names is based on the provisions of Article 40 in the International Code of Zoological Nomen- clature, which states: “Synonymy of the type-genus. When, after 1960, a
170 ANNALS OF THE SOUTH AFRICAN MUSEUM
nominal type-genus is rejected as a junior synonym (objective or subjective), a family-group name based on it is not to be changed, but continues to be the valid name of the family-group taxon that contains both the senior and junior synonyms.’ In other words, a family or subfamily name must be retained even though the type-genus on which it is based has been rejected as a junior syno- nym. Applying this to our particular case, the family name Heterodontosauridae Kuhn or Romer (which ?), 1966 could not be replaced by a new family name Lycorhinidae, even if Heterodontosaurus were accepted as a junior synonym of Lycorhinus; and the same is true of the corresponding subfamily names. As for our second objection, Article 29 of the Code lays down that ‘A family-group name is formed by the addition, to the stem of the name of the type-genus, of .. . -INAE in the case of a subfamily’. Thulborn’s two ‘groups’, therefore, should be called Heterodontosaurinae and Fabrosaurinae respectively. In any case, Galton (1972) has already pointed out the undesirability of placing these Triassic forms within the Hypsilophodontidae; although we do not agree with every aspect of Galton’s phylogeny and classification, we do agree with him that Thulborn’s two ‘groups’ merit independent familial status, as Hetero- dontosauridae and Fabrosauridae. (Authorship of the latter family is attribut- able to Galton 1972.)
Let us return to our central theme—the matter of the alleged generic synonymy. It sometimes happens that the Law of Priority (Article 23 of the Code) obliges one to place a well-established, familiar name in the junior synonymy of one much less familiar or altogether unknown; or, even more unfortunate, to place a taxon based on excellent type-material in the subjective junior synonymy of another based on poor type-material. The placing of Heterodontosaurus in the junior synonymy of Lycorhinus would have both these unhappy consequences without any possible advantage accruing therefrom. Heterodontosaurus is already a well-established name appearing in many articles and recent textbooks, for, as the first fairly complete skull of a Triassic (or indeed pre-Upper Jurassic) ornithischian, it attracted a great deal of attention; Lycorhinus is a name familiar to no one and, where it does appear in lists of genera (e.g. in Romer 1956), it has usually been classified as a therapsid. The Heterodontosaurus holotype has now been completely developed and the detailed description of its skull (Charig & Crompton, in press) is likely to make it the most completely described dinosaur skull in existence; moreover (as mentioned above) there is now another skull (S.A.M. No. K1332), equally good if not better, in association with good postcranial material. The Lycorhinus holotype, on the other hand, was an incomplete left dentary, with eight teeth and the impressions of four others, and is today (see p. 174) represented only by the most anterior, caniniform tooth, and the impression of the dentary with its row of pos- terior teeth (S.A.M. No. 3606). If the two holotype specimens were congeneric, beyond all reasonable doubt, and if we were given the choice of which to use as our standard of reference for the combined genus, the choice would obviously fall upon Heterodontosaurus; the use of the now very fragmentary
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS _§ I7I
Lycorhinus type would impose great practical difficulties.
But we should have no such choice. L. angustidens is the type-species of Lycorhinus, and must remain so. The true systematist is like a judge in a civil case; he must apply the law impartially, interpreting it rigidly where the wording is unequivocal even though he may think it to be unjust. To disregard the law (even in some small particular) or to permit others to do so can only bring about a general disregard for the law as a whole. If the systematist feels that a certain Rule should not be applied in a certain case, he can request the International Commission on Zoological Nomenclature to exercise its plenary powers; if he thinks that a particular Rule or part thereof is a bad rule, his remedy is to seek to change it through the next Congress. The wisest course, however, for anyone suspecting that a possible synonymy might prove incon- venient or embarrassing could well be to ignore it altogether; on the other hand, once it has been dragged into the open it must be judged on only one aspect—not on its consequences or its practicability but simply on whether or not it is possible to show, to the reasonable satisfaction of other workers in the field, that the specimens or taxa concerned are sufficiently alike to be given the same name. Charig & Reig (1970) state that: ‘In deciding synonymy or otherwise the onus of proof lies with him who wishes to establish it.’ Kermack, Mussett & Rigney (1973) are even more positive: ‘Once a genus has been described and named, the onus is entirely on those who wish to make the name a synonym to show that the two genera concerned are identical beyond any possible doubt [their italics]: the onus is in no way on those who wish to maintain the status quo to prove that the names are not synonyms.’
This is the case with Lycorhinus and Heterodontosaurus. It should have been obvious to Thulborn that, even if he could have proved the synonymy, here was a case for letting sleeping dogs lie; but once he had chosen to stir them up we found ourselves compelled to examine his arguments very carefully in order to determine the degree of acceptability of his conclusions. We have done this as objectively as possible, well aware of the fact that, as co-authors of the name Heterodontosaurus, we might be accused of some emotional attachment thereto!
Essentially Thulborn’s arguments for synonymizing Heterodontosaurus with Lycorhinus rely upon a new specimen, U.C.L. No. A.1oo (Figs 4-7); he claims that this specimen, a block of sandstone with dissociated pieces of skull, is a topotype of L. angustidens. (While he admits that the holotype of L. angus- tidens did not provide enough evidence even to investigate the possibility that Lycorhinus was an ornithischian, apparently a latex impression of that holotype now affords sufficient proof of its specific identity with A.100!) A.100, in turn, is regarded as congeneric with Heterodontosaurus tuck.
The first question is, is such use of intermediate specimens legitimate? If the holotype of one species cannot be synonymized generically with the holotype of another, is it proper to synonymize one with an assumed topotype— or even a mere referred specimen—of the other? The answer to this question,
172 ANNALS OF THE SOUTH AFRICAN MUSEUM
in general, is indubitably yes. It may well be that two well-preserved but incomplete holotypes in the same deposit consist of different parts of conspecific individuals but cannot be shown to belong to the same species until a new, more complete specimen including both parts associated proves to be identical to both. For example, most species of chelonian in the London Clay are based either on skulls or on shells; it is likely that many ‘skull species’ could be paired up with ‘shell species’, but this can be done only by the finding of well-preserved specimens with skull and shell in association. Such specimens are rare.
The case of Lycorhinus and Heterodontosaurus, however, is not like that at all. All that we have of Lycorhinus angustidens is an impression of the dentary, an element of which the greater part is present also in the holotype of Heterodonto- saurus tucki; the use of another dentary as an intermediate cannot help matters at all because, without any additional elements in common to the two specimens concerned, we are no better equipped to prove the identity of that dentary with the Lycorhinus holotype either. Indeed, even if we were provided with an absolutely perfect skull as our intermediate we should not be able to prove that its dentary and lower teeth were identical with those of Lycorhinus angustidens, although we might be able to prove the identity of that perfect skull with the type skull of Heterodontosaurus tucki. The simple fact is that the holotype of Lycorhinus angustidens seems to be too incomplete and too poorly preserved to allow anything to be certainly identified with it.
Thulborn fails to comment on the fact that specimen No. A.100 has heavily worn maxillary teeth and unworn dentary teeth; it is interesting to speculate as to how he imagines the animal wore down the former while preserving the latter in their newly erupted condition. This apparent anomaly did lead us to suspect, albeit fleetingly, that the lower jaw of A.100 might not have belonged to the same animal as did the upper jaws; we now regard that possibility, however, as no more than extremely remote because the various elements present on the A.100 block (premaxilla, maxillae, jugals, frontal, postorbital, dentary) show no duplication and are all of commensurate size. In any case, the premaxilla, maxillae and upper teeth all give clear indications of hetero- dontosaurid affinities, and the lower jaw—except for its unworn teeth—does the same. Even so, we remain puzzled by the unworn condition of those lower teeth. Thulborn, on the other hand, appears to ignore the fact that those unworn lower teeth do not bear much resemblance to any Heterodontosaurus teeth described by Crompton & Charig in the holotype of H. tucki but are in some respects more reminiscent of the teeth of Fabrosaurus australis. The details which he gives of the ‘second specimen of Heterodontosaurus’, which he refers to as being in the care of one of us (A.J.C.) in the British Museum (Natural History), are inaccurate; and, while it seems very likely that the fragment of maxilla in question belonged to a member of the family Heterodontosauridae, it differs in several particulars from the holotype of Heterodontosaurus tucki. (Had we known that he intended to mention our research material in his article we should gladly have supplied him with the necessary information.)
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS 173
Thulborn takes sufficient cognisance of the differences between the lower canini- form teeth of Lycorhinus and those of Heterodontosaurus—the former crenellated on the anterior! margin only, the latter on both margins—to admit that this signifies a specific difference between the two forms. We agree that this difference exists; for we maintain that the type of Lycorhinus angustidens, while too incomplete and too poorly preserved to permit anything to be certainly identi- fied with it, was well enough preserved to show that it was different from both A.100 and Heterodontosaurus tuckt.
Despite this minor difference there is no doubt that all three specimens concerned possess four common characters of the dentition which are typical of the ‘Heterodontosaurus-like’ forms and which are lacking altogether in Fabro- saurus and its allies, the only other ornithischians from the Upper ‘Trias of southern Africa of which we have any real knowledge. These are (a) the promi- nent caniniform tooth at the front of the dentary, (5) a trend towards the possession of closely packed postcaniniform teeth which are worn down to a continuous occlusal surface, (c) the apparent complete absence of any tooth replacement (see Charig & Crompton, in press), and (d) the mediad recession of the tooth row into the side of the face, suggesting the presence of muscular cheeks (see Galton 1973). Thulborn, oddly enough, draws attention only to the first of these. Characters (a) and (c) are unique to this particular group within the Ornithischia, but (6) is found also in some Cretaceous members of the order and (d) in all post-Triassic members.
Let us now analyse Thulborn’s comparison of Lycorhinus with Heterodonto- saurus, detail by detail. It relies entirely on the unquestioned acceptance of two points: his unproven, tacit but very evident assumption that all the material on the A.100 block represents a single individual, and his allegedly proven belief that ‘specimen A.100 does represent Lycorhinus angustidens’ (p. 242). We shall—with some hesitation—accept the former point, supposing that all the elements visible on the A.1oo block (Fig. 1) belonged to the same animal and ignoring the surprising and seemingly inexplicable fact that the maxillary teeth are heavily worn while the dentary teeth are scarcely worn at all. But we feel that his reference of A.100 to Lycorhinus angustidens should be submitted to a critical examination.
The lower jaw of A.100 is exposed only on its lateral side (Fig. 4). Thulborn (p. 241) refers it to L. angustidens because, according to him, it shares the follow- ing common features with Haughton’s holotype:
1. The mandible is exceptionally deep. 2. There is a large, caniniform first dentary tooth, with only its anterior edge crenellated.
* Thulborn uses terms employed in dentistry and human odontology when describing the teeth of reptiles; thus anterior becomes ‘mesial’, posterior ‘distal’ and so on. We disagree with Edmund (1969) that the use of these terms is convenient and that ‘The terms distal and mesial are to be preferred to anterior and posterior because of the curved shape of the dental arcade.’ ‘Distal’, in particular, is generally used by comparative anatomists to mean something entirely different (namely, opposite to proximal) and it is therefore ambiguous or misleading to apply it to reptile teeth.
174 ANNALS OF THE SOUTH AFRICAN MUSEUM
3. There is a small gap between the first (caniniform) and second dentary teeth.
Haughton, however, described and figured (1924: 343-344, fig. 8; our Fig. 8) only the medial aspect of his holotype. Later the specimen itself became sepa- rated from the piece of matrix to which it had been attached by its lateral surface and was lost; the lateral aspect of the dentary, as evidenced by its natural mould on the matrix, was described and figured by Broom (1932: 307, fig. 104, I; our Fig. g). It is of this lateral aspect that ‘Thulborn possesses a latex impression (p. 236) and we ourselves have another. But Thulborn does not mention Broom’s account of Lycorhinus (although it is evident from personal communication that he is aware of its existence, indeed we must acknowledge our gratitude to him for reminding us of it) and he tacitly implies that his latex impression is of the surface described by Haughton, without indicating at this point whether Haughton described the medial or the lateral surface. Later, however (p. 242), Thulborn states that further comparisons are difficult because only the medial side of the Lycorhinus angustidens holotype was exposed and described by Haughton, whereas, of course, the mandible and lower teeth of specimen A.100 are visible only in lateral aspect; had Thulborn used his latex impression of the lateral side of the holotype the comparison would have been easier and more meaningful.
Whichever surface of the Lycorhinus mandible be compared with the lateral surface of the A.100 mandible, it is immediately obvious to us that the two specimens are different in many respects. With regard to the three similarities cited by Thulborn (see above), the second—concerning the presence of a canini- form with only its anterior edge crenellated—is certainly correct. But the illustrations by Haughton, Broom and Thulborn, taken in conjunction with the latex mould, suggest that the base of the caniniform tooth is relatively more slender in A.100 than in Lycorhinus. Further, the tip of the Lycorhinus caniniform is bevelled off anteromedially, as shown in Haughton’s figure and described in his text; the lower caniniform of A.100 is not exposed on its medial side, but Thulborn’s illustration of it in lateral view suggests that the tooth is unworn right up to its remarkably acute tip. (Haughton commented that in Lycorhinus it was ‘bevelled off by rubbing against the canine of the upper jaw’, but, with no upper jaw preserved, this was pure speculation; in Heterodontosaurus the upper and lower caniniforms, seemingly in their natural relative positions, are nowhere near each other.) The same illustrations by Haughton, Broom and Thulborn suggest also that Thulborn’s other two similarities are not especially close: the mandible beneath the caniniform is much shallower in A.100 than in Lycorhinus, and the gap between the first (caniniform) and second dentary teeth is much wider.
There are other differences too, outweighing the similarities in number and importance. The postcaniniform teeth of Lycorhinus are inclined slightly forwards, they are so close together that the distal ends of their crowns are in contact with each other, and they are heavily worn down to a continuous,
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS 175
though rather irregular, occlusal surface. On the other hand, the teeth of A.100 show no forward inclination whatever, they do not touch each other (Thul- born’s fig. 5, our Fig. 4, suggests that the third and fourth teeth might have touched before the third was broken) and, as illustrated, they show no signs of wear; this last point is confirmed by Thulborn’s description (p. 240). In both specimens a cingulum separates the more or less parallel-sided root from the upwardly widening crown, the anterior and posterior edges of the latter diverging towards the occlusal margin; in Lycorhinus this cingulum is absent on the lateral surface and present (fide Haughton) on the medial surface, but in A.100 (where the medial surface of the lower teeth is unknown) it is well developed on the lateral surface. ‘The teeth of Lycorhinus are not each symmetrical in lateral or medial view, for on each tooth from the fifth onwards (‘4th molar’ of Haughton) there is ‘a large anterior cusp occupying two-thirds of the grind- ing surface, and a much smaller, somewhat lower posterior cusp’ (Haughton 1924: 344); the groove on the medial surface, running between the ridges which descend from the cusps to the cingulum, is therefore markedly posterior in position. But in A.100, where we can see only the lateral surface of the tooth row, each crown has the form of a symmetrical arrowhead; there are three or four small cuspules on either side (anterior and posterior) of the tip, and the well-pronounced ridges which run down from each cuspule are roughly symme- trical on either side of a stronger central ridge. (It must be admitted that the lateral surfaces of the teeth of the Lycorhinus angustidens holotype, as figured by Broom and as shown in the latex mould, do look a little more symmetrical than the medial surfaces, but they are still quite unlike the lateral surfaces of the teeth of A.100.)
Because of the difficulties alleged by Thulborn of comparing the lower teeth of Lycorhinus with those of A.100 (difficulties which, as we have seen, are actually unnecessary if use be made of Broom’s description and figure and of the latex mould), Thulborn also compares the medial aspect of the dentary teeth of the former—as described and figured by Haughton—with the medial aspect of the maxillary teeth of A.100 (Fig. 7); this comparison, of extremely dubious validity, enables him to claim (p. 242) of the teeth in the latter speci- men that their ‘worn crowns (Fig. 4) are virtually identical with those described and figured by Haughton (1924). In each case the lingual crown surface bears a broad median rib and the distal [posterior] margin is produced as a thin and erect ridge, the mesial [anterior] edge showing only a faint tendency towards elaboration into a similar ridge.’ In fact none of these features is apparent in Haughton’s figure of the Lycorhinus holotype or mentioned in his description, unless we include ‘the ridges which descend from the cusps to the cingulum’. According to Thulborn, these alleged but actually non-existent similarities between the upper teeth of one specimen and the lower teeth of another ‘leave little doubt that specimen A.100 does represent Lycorhinus angustidens’. We repeat that the medial surfaces of the dentary teeth of A.100 remain unknown and may well differ from those of the maxillary teeth; but they could not have
I 76 ANNALS OF THE SOUTH AFRICAN MUSEUM
resembled the corresponding surfaces in Lycorhinus, for, as indicated above, the dentary teeth of A.100 are unworn and their whole shape is quite different from that found in Lycorhinus.
Thulborn goes even further and states (p. 236) that: ‘Since both of these specimens are from the same locality there is a slight possibility that they might represent different parts of a single individual, though this seems unlikely in view of the great hiatus between their two dates of collection.’ He ignores the several differences that we have pointed out between Lycorhinus and A.100, evidently preferring to believe that it is possible for the teeth on the left dentary to undergo extensive wear while those on the right dentary of the same animal remain virtually unworn! He also ignores the fact that the exposure of the Red Beds at Paballong (neither Haughton nor Thulborn localized their specimens more precisely than that) is so large as to render negligible the chances of two bones found on different occasions being parts of the same individual—especi- ally when those two occasions are separated by more than 40 years. In any case, the rate of erosion at Paballong during the rainy season is so high that the time required for the destruction of a piece of bone weathered out of the rock is more likely to be, on average, a matter of days or even hours rather than of decades.
In conclusion, we are much less impressed by Thulborn’s similarities between Lycorhinus angustidens and A.1oo than by the differences between them; and we feel that we have established a prima facie case for their generic separation.
Because of his determination of A.100 as Lycorhinus augustidens Thulborn next claims that ‘Specimen A.100, imperfect though it is, considerably amplifies our knowledge of Lycorhinus’, and then proceeds, in the light of that ‘amplified’ knowledge, to compare Lycorhinus with the holotype of Heterodontosaurus tucki; in that comparison he draws his Lycorhinus characters partly from the holotype, partly from the A.100 lower jaw, but mostly from the upper jaw and skull of A.100. He concludes that: ‘Whilst there are definite differences between Lycorhinus angustidens and L. (Heterodontosaurus) tucki these serve only to distin- guish the animals at species level and do not warrant their separation into distinct genera.’ But we have rejected his determination of A.100 as Lycorhinus angustidens. In that circumstance Lycorhinus and A.100 should be compared separately with Heterodontosaurus: first the holotype dentary of Lycorhinus with the dentary of the holotype of Heterodontosaurus, then the dentary of A.100 with the dentary of the holotype of Heterodontosaurus, and finally the various elements of the upper jaw and skull of A.100 with the corresponding elements of the Heterodontosaurus upper jaw and skull. Further, since we believe that Lycorhinus and A.100 are not identical, it follows that—in our view —Heterodontosaurus cannot be the same as both of them; it may be ether the same as Lycorhinus, or the same as A.100. (If it is not the same as A.100, then the latter must represent yet another ornithischian in the Upper Trias of southern Africa.) It is even possible that Heterodontosaurus is not the same as either Lycorhinus or A.100, in
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS- 177
which case we must be dealing with three separate forms.
First let us compare the Lycorhinus holotype (Fig. 8) directly with the dentary and lower teeth of Heterodontosaurus (Fig. 10). ‘Thulborn, drawing his Lycorhinus characters from the holotype and not from A.1oo, can claim only three points of similarity: the presence of a prominent caniniform first dentary tooth, the planing-off of the lateral sides of the mandibular crowns to produce sharp chisel-like ends, and the size gradation of those mandibular crowns so that the largest are in the middle of the tooth row. (It is not clear from his paper (p. 243) whether the size gradation is supposed to pertain to the upper cheek teeth only or to the lower teeth as well, but it does in fact pertain to both.) We indeed agree that all three features appear to be present in both genera. On the other hand, Thulborn admits to one difference: the lower caniniform is crenellated only on its anterior edge in Lycorhinus, on both edges in Hetero- dontosaurus. ‘There are also other important differences, unrecognized by Thulborn. Haughton (1924) mentioned that the ‘molars’ of Lycorhinus had a pronounced cingulum, which, as pointed out by Crompton & Charig (1962), is absent in the postcaniniforms of Heterodontosaurus. ‘The crowns of Lycorhinus taper down towards that cingulum so that, although they touch their neighbours occlusally, large triangular gaps are left between their more basal portions; the latex impression shows that their lateral surfaces are convex in both directions, vertical and horizontal. In Heterodontosaurus, by contrast, the lateral surface of each lower postcaniniform crown has sub-parallel edges which touch (or almost touch) its neighbours throughout its length and is more or less flat. There is a further difference in the vertical ridging-and-grooving on the medial surface of the teeth: in Lycorhinus a narrow groove lies behind a wide anterior ridge, in Heterodontosaurus a wide shallow trough lies behind a narrow anterior ridge. Because the similarities between Lycorhinus and Heterodontosaurus are so few and so general (indeed, they could well be family characters rather-than generic) we regard the generic identity of Lycorhinus and Heterodontosaurus as unproven; because there are also certain differences between them, we consider it unlikely. As we wrote in 1962: ‘This [the possession by Lycorhinus of a distinct cingulum] and the nature of the wear of the teeth appear to indicate that Heterodontosaurus and Lycorhinus are generically distinct.’ We see no reason to change our minds; rather has our detailed analysis confirmed our opinion. The fragmentary nature and poor preservation of the Lycorhinus angustidens holotype, taken together with the fact that it is now represented only by an impression, make it improbable that the matter could ever be settled really conclusively one way or the other, and for that reason we prefer to regard the names Lycorhinus and L. angustidens as nomina dubia, names which have been founded on inadequate material and which ought not to be used except in connection with the holotype itself.
Adoption of this point of view is not only sufficient in itself to justify rejection of any synonymy involving the name Lycorhinus but logically requires such rejec- tion. Charig (in press) states that ‘... no worker can be compelled to accept
178 ANNALS OF THE SOUTH AFRICAN MUSEUM
the validity of a specific or generic name based on type-material which he considers unsatisfactory; if he rejects such a name it becomes—to him—a nomen dubium, and on those grounds he is not only entitled to reject other authors’ synonymies involving the name in question but is logically obliged to do so’.
The alleged identity of the lower jaw of A.100 (Fig. 4) with that of Hetero- dontosaurus tucki (Fig. 10) is another matter. Comparison of the postcaniniform teeth is difficult because specimen A.100 includes only anterior dentary teeth (numbers 1~7) and the H. tuck: holotype includes only the caniniform and the more posterior teeth (numbers 1 and probably 7-14); further, the lower jaw of A.100 and its teeth are exposed only on the lateral side, while the lower teeth of H. tucki, though visible in both aspects, are well exposed only on the medial side. Even so, the alleged identity is easy to disprove, for though the similarities are few the differences are many. In fact, a list of those differences is essentially the list of differences between A.100 and the Lycorhinus angustidens holotype given earlier in this article together with another couple of differences between Lycorhinus and Heterodontosaurus as given immediately above. We shall put these down in tabular form. Let us include also a column for the corresponding condition in the Lycorhinus holotype—where known—and insert it between the A.100 and Heterodontosaurus columns (the reason for this arrangement will soon become apparent). Let us include also two additional characters in which A.100 differs from Lycorhinus but which are not visible in the holotype of Heterodontosaurus tuck.
Thulborn also writes (p. 243): “The teeth of Lycorhinus [viz., A.100] bear on their labial surfaces median ribs which are neither as thin nor as sharp as those in Heterodontosaurus.’ From the context he could be referring to vertical ridges on either the lateral surface or the medial surface of either the upper or the lower teeth of Heterodontosaurus, all of which are different. On the next page (p. 244), where he is presumably referring to the same feature, he partly clarifies the situation by writing: ‘Fluting on the labial surfaces of the cheek teeth is more pronounced in the species tucki.’ In Heterodontosaurus, however, there are no thin sharp median ridges on the lateral surfaces of the lower teeth, indeed all that is known of the ornamentation of the lower teeth in the type of f1. tuckt (where they were badly damaged during development) is that a broad vertical ridge appears to have been present; yet a comparison of the lateral surface of the lower teeth of A.100 with the lateral surface of the upper teeth of Heterodontosaurus would mean nothing.
It is obvious from the comparative table that A.100 is the least specialized and Heterodontosaurus the most specialized of the three forms, with Lycorhinus angustidens fitting in as an intermediate. This accords with their stratigraphical positions: A.100 and Lycorhinus are from the same locality and therefore from approximately the same horizon within the Red Beds, Heterodontosaurus is from the overlying Cave Sandstone and is likely to be a little younger.
Heterodontosaurus tucki (Figs 10, 11) and A.100 (Figs 5~7) may also be com-
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS 179
COMPARISON OF LOWER JAWS AND TEETH
depth of mandible beneath caniniform
width of base at alveolar margin
crenellate edges
caniniform
width of gap between
A.100 Holotype of Holotype of Lycorhinus angustidens | Heterodontosaurus tucki much shallower exceptionally deep more slender broader anterior edge only both edges wider narrower ? (not seen)
caniniform and second dentary tooth
distance apart
inclination of anterior teeth
wear
appearance of each crown in lateral or medial view
postcaniniform teeth
distinct neck between root and crown
cingulum
close together well apart, no
contact crowns in contact | crowns in contact distally throughout length not inclined crowns inclined ? (not seen)
slightly forwards heavily abraded to form continuous occlusal surface virtually unworn
occlusal surface occlusal surface irregular smooth symmetrical asymmetrical (shown by cusps & (shown by ridging-and-grooving on ridging-&-grooving medial surface) on lateral surface) present ? absent, teeth essentially columnar present absent on lateral surface on medial surface (medial not seen) only
pared on the upper jaw and the rest of the skull, parts which are lacking entirely in Lycorhinus angustidens. Both specimens have (a) a large, triangular and widely open antorbital fossa, (6) a premaxilla which is remarkably deep below the external naris, (c) a wide, deep diastema between the premaxilla and the maxilla, and (d) a distinct recess, forming a ‘step’, in the lateral surface of the maxilla above the tooth row (possibly more pronounced in Heterodontosaurus than in A.100). In both specimens a wide edentulous zone at the front of the premaxilla is followed by three teeth, the first two being simple conical pegs and the third a prominent caniniform (large, but not as large as the lower caniniform) with its posterior border crenellate (in Heterodontosaurus the anterior
ANNALS OF THE SOUTH AFRICAN MUSEUM
180
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS I8I
border is not visible). The maxillary teeth number about a dozen (12 in Hetero- dontosaurus, 13 in A.100) with the largest crowns in the middle of the series; their medial sides are planed off to produce sharp, chisel-like ends. This planing-off is oblique in Heterodontosaurus ; in A.100, however, although Thulborn does mention in his text (p. 243) that the medial sides of the maxillary teeth are planed off, his Figure 4 gives no indication whatever of this character.
On the other hand, there are also a number of differences. Specimen A.100 seems to lack the suborbital boss and the jugal process which are so characteristic of the jugal of Heterodontosaurus, indeed the whole form of the jugal seems to be very different. The disparity between the lengths of the upper caniniform and of the other premaxillary teeth is less marked in A.100 than in Heterodontosaurus. In Heterodontosaurus the most anterior tooth in the maxilla lies immediately beneath the anterior end of the antorbital fossa; in A.100, however, the maxilla itself and its tooth row extended forwards far beyond the limits of that fossa. The maxillary teeth of Heterodontosaurus are close together, with only small gaps between them at the alveolar margin and touching each other occlusally, even overlapping a little in most instances; but those of A.100 are separated by wide gaps at the alveolar margin, sometimes wider than the teeth themselves. The maxillary teeth of Heterodontosaurus possess no neck and are essentially columnar in form, while those of A.100 have a pronounced neck and cingulum demarcating the root from the crown. In Heterodontosaurus the lateral surface of each maxillary tooth bears three prominent ridges—anterior, central and posterior —separated by sharply defined excavated regions, the central ridge in particular being thin and sharp; the medial surface, however, is quite different in that it has only a poorly developed central ridge. In A.100, by contrast, the lateral and medial surfaces of the maxillary teeth are essentially alike, the central ridges being weaker, thicker and blunter than on the lateral surfaces of the maxillary teeth of Heterodontosaurus.
Fig. 1. Fabrosaurid gen. et sp. indet. Described and figured Thulborn 19708, 19714 as Fabrosaurus australis. Department of Zoology, University College London, field no. B.17. Likhoele Moun- tain, near Mafeteng, Lesotho; Red Beds. Left premaxilla and maxilla. Lateral view, x 33. (After Thulborn 1971a; maxilla reversed from right side.)
Fig. 2. Pisanosaurus mertii Casamiquela, 1967. Holotype. Laboratorio de Paleontologia de Verte- brados, Instituto Miguel Lillo, Tucuman, Argentina, no. 2577. ‘Agua de las Catas’, opposite Km 461 on Ruta Nacional no. 40, La Rioja Province, Argentina; middle section of Ischigualasto Formation. Left mandible. Medial view, x 3. (After Bonaparte, in press; reversed from right
side.)
Fig. 3. Same individual. Left mandible. Lateral view, x 3. (After Bonaparte, in press; reversed from right side.)
Abbreviations: m— maxilla; pm—premaxilla; q—fragment of the quadrate.
ANNALS OF THE SOUTH AFRICAN MUSEUM
182
S
SSS
S
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS 183
All the premaxillary teeth of Heterodontosaurus (including the caniniform) appear to have more or less rounded tips; but they are seen only in lateral view, and it is not possible to be sure that they do not have wear facets on the medial side like those of A.100 which, Thulborn suggests, ‘resulted from their working against a horn-sheathed predentary at the mandibular symphysis’.
There are also certain very characteristic features of the cheek teeth of Heterodontosaurus, both upper and lower, which cannot be used in this compari- son because Thulborn does not mention the corresponding condition in the latter specimen. In Heterodontosaurus the cheek teeth are curved in the transverse plane so that, in anterior or posterior view, the lateral profile of each upper crown is convex with a distinct ‘knee’, the medial profile is more or less straight, and the planar occlusal surface faces medioventrally; the opposite holds true for the lower crowns, though their convexity (on the medial side) is weaker. There is an extremely thin layer of enamel on the occlusal half of each convex surface only, i.e. on the lateral surface of each upper tooth and the medial surface of each lower tooth. Thulborn does state, however, that the unworn cheek teeth of Lycorhinus (by which he presumably means the lower postcanini- form teeth of A.100) are completely enamelled.
In this connection it may be noted that Galton (1973) reproduces (fig. 2 K) Thulborn’s figure 3 of the right maxilla of A.100; his new caption reads: ‘Heterodontosaurus sp., right maxilla in lateral view, figured as Lycorhinus augusti- dens by Thulborn (19702) who does not and cannot demonstrate that this specimen is identical to the lost holotype of Lycorhinus angustidens Haughton (1924).’ While we agree with Galton that A.100 cannot be referred to Lycorhinus angustidens, we would certainly not place it in the genus Heterodontosaurus (see our conclusions below) ; the differences between A.1oo and the type of H. tucki are too many and too substantial.
Fig. 4. Heterodontosaurid gen. et sp. indet., described and figured Thulborn 1970a as Lycorhinus angustidens. Department of Zoology, University College London, no. A.100. Paballong, near Mount Fletcher, Herschel District, Cape Province, South Africa; Red Beds. Left dentary, with teeth. Lateral view, x 2. (After Thulborn 1970a; reversed from right side.)
Fig. 5. Same specimen. Left premaxilla, with teeth. Medial view, x 2. (After Thulborn 1970a; reversed from right side.)
Fig. 6. Same specimen. Left maxilla, with teeth. Lateral view, x 2. (After Thulborn 1970a; reversed from right side.)
Fig. 7. Same specimen. Left maxillary teeth. Medial view, x 2. (After Thulborn 19702.)
Abbreviations: aof—antorbital fenestra; can—caniniform tooth; di—diastema.
ANNALS OF THE SOUTH AFRICAN MUSEUM
184
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS 185
If it be accepted that the whole of specimen A.100 represents a single individual, then certain characters of the upper jaw and its teeth also accord with the view that A.10o0 is less specialized than Heterodontosaurus. ‘They include the simpler form of the jugal (without suborbital boss or jugal process), the slightly weaker development of the upper caniniform, the wider spacing of the maxillary teeth in the tooth row, the simpler form of those teeth with lateral and medial surfaces more or less alike, and possibly a lesser development of the recess and ‘step’ in the lateral surface of the maxilla above the tooth row.
The graded differences between A.100, Lycorhinus and Heterodontosaurus are indeed significant; but, as mentioned above, they do not obscure the essen- tial similarities of the three forms—some of them unique to the group. There are three characters which are diagnostic of the group, no matter what rank the latter may be accorded in the classification: the prominent caniniform at the front of the dentary, a trend towards the possession of closely packed post- caniniform teeth with antero-posteriorly expanded crowns worn down to a continuous occlusal surface (not evident in the lower jaw of A.100), and the concavity extending along the outer sides of the maxilla and dentary above and beneath the postcaniniform tooth row (suggesting a mediad recession of the tooth row into the side of the face). We believe that this suite of common characters justifies the association of the forms in question into a taxon of suprageneric rank—the family Heterodontosauridae.
The three specimens considered here also share a unique peculiarity in that their teeth show no signs of replacement. It seems, however, that this character is not invariably associated with the rest of the suite; we have the incomplete maxilla of what appears to be another heterodontosaurid from
Fig. 8. Lycorhinus angustidens Haughton, 1924. Holotype. South African Museum, no. 3606. Paballong, near Mount Fletcher, Herschel District, Cape Province, South Africa; Red Beds. Left dentary, with teeth. Medial view, x 14. (After Haughton 1924.)
Fig. 9. Same individual, same bone. Lateral view, x 14. (After Broom 1932.)
Fig. 10. Heterodontosaurus tucki Crompton & Charig, 1962. Holotype. South African Museum, no. K337. Mountain behind Tyindini trading store, Herschel District, Cape Province, South Africa; Cave Sandstone. Reconstruction of skull. Left lateral view, natural size. (After Charig & Crompton, in press; reversed from right side.)
Fig. 11. Same individual. Left maxillary teeth. Lateral view, x 4. (After Charig & Crompton, in press; reversed from right side.)
Abbreviations: a—angular; aofo—antorbital fossa; ar—articular; can—caniniform tooth; d—dentary; j—jugal; jp—jugal process; m—maxilla; pd—predentary; pm—premaxilla; q—quadrate; qj—quadratojugal; sob—suborbital boss. On all the illustrations the arrow indicates the anterior direction.
186 ANNALS OF THE SOUTH AFRICAN MUSEUM
the Stormberg Series with functional teeth possessing the typical characters of the family but also with two unerupted replacing teeth and other evidence of replacement. This is the fragment referred to by Thulborn (1970a: 243) as ‘A second specimen of Heterodontosaurus’.
The relative sizes of specimen A.100, Lycorhinus and Heterodontosaurus may be of interest. Comparisons are difficult, but the average interval at which the teeth are inserted in the middle of the dentary seems to be about the same (3,6 mm) in all three. The depth of the dentary beneath those teeth, however, is a little greater in specimen A.100 than in either Heterodontosaurus or Lycorhinus, and the maxilla is approximately one-third longer in A.100 than in Hetero- dontosaurus.
Also relevant to our discussion is the genus Pisanosaurus Casamiquela, 1967 (only species P. mertzi Casamiquela, 1967 from the Ischigualasto Formation of Argentina), which is of Carnian or perhaps even of Ladinian age; it is there- fore older than the African forms with which this paper is mainly concerned and which are generally considered to be Norian or Rhaetian. Indeed, Prsanosaurus is the earliest ornithischian known. The unique holotype comprises a fragment of maxilla and an almost complete mandible (Figs 2, 3), both with teeth; cervi- cal, dorsal and (as natural moulds) sacral vertebrae; a little rib material; an incomplete scapula and an impression of part of the pelvis; the impressions of three metacarpals; and a good part of the hind limb. Pisanosaurus has been variously placed in the families Pisanosauridae (Casamiquela 1967), Hypsilo- phodontidae (Thulborn 197167, 1972; Galton 1972, 1973) and Heterodonto- sauridae (Bonaparte, in press). The last work cited, a redescription and reassess- ment of Prsanosaurus, shows that it possesses all the heterodontosaurid tooth characters mentioned above and several others too, with the notable exception of the caniniform teeth; and even the absence of those teeth is by no means certain because Pisanosaurus lacks both the premaxilla and the front end of the mandible —a sufficient length thereof to have borne a caniniform. Galton (1972, fig. 1) indicates his belief that Pisanosaurus occupies a position in ornithischian phylogeny at the base of the ornithopod line of ancestry, just after the diverging of the heterodontosaurid, ankylosaur and stegosaur lines; indeed, he places it on or close to the line of ancestry of all other ornithischians except Fabrosaurus and Echinodon. He therefore regards it as the earliest and most primitive hypsilo- phodontid, for he writes (p. 466): ‘All hypsilophodonts were probably derived from a form similar to Pisanosaurus.’
It seems to us, however, that Bonaparte’s revelation of the specialized, truly heterodontosaurid character of the Pisanosaurus dentition makes Galton’s beliefs completely untenable; we therefore support Bonaparte’s reference of the genus to the Heterodontosauridae, a family known otherwise only from southern Africa. But the higher tooth count (maxillary teeth estimated at 16-18 by Bonaparte, 15 dentary teeth actually preserved) suggests that Pisanosaurus is
e Thulborn (19715) includes Pisanosaurus among the ‘fabrosaur group’ of the family Hypsilo- phodontidae. The separate family Fabrosauridae was formally proposed by Galton in 1972.
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS 187
more primitive than its African relatives; this suggestion, which will probably be confirmed by other characters of the genus when they become known, accords with its lower stratigraphical position.
Pe
Our conclusions may be summarized as follows:
All three specimens considered here—holotype of Lycorhinus angustidens, holotype of Heterodontosaurus tucki and specimen A.100—are sufficiently alike to be placed in the same family, for which the correct name is Hetero- dontosauridae. The diagnostic characters of that family are the prominent caniniform tooth at the front of the dentary (and in the premaxilla too where known), a trend towards the possession of closely packed postcanini- form teeth which are worn down to a continuous occlusal surface, and the mediad recession of the postcaniniform tooth row into the outer side of the jaws. The three specimens concerned are also remarkable in their apparent total lack of tooth replacement, but this cannot be regarded as an absolutely diagnostic character because another heterodontosaurid jaw fragment from the Stormbergs shows unequivocal evidence of replacement.
Whether the lower jaw of A.10o did in fact belong to the same individual as the other skull elements, or even to the same species, might be considered a little doubtful because of the unworn nature of the lower teeth and the heavily worn nature of the uppers. On the other hand, the various skull elements of A.100 show no duplication, they are all of commensurate size, and both upper and _ lower jaws give clear indications of heterodontosaurid affinities.
The lower jaw of A.100, however, is sufficiently distinct from Lycorhinus and Heterodontosaurus, and the upper jaw (unknown in the former genus) is sufficiently distinct from Heterodontosaurus, to rule out any possibility that A.100 might be congeneric with either. It must represent a new genus of heterodontosaurid from the Upper Trias of southern Africa.
Lycorhinus angustidens, as far as can be seen in so imperfect a specimen, bears a strong resemblance to Heterodontosaurus tucki but differs sufficiently in certain details of tooth structure to rule out any possibility of conspecificity. The fragmentary nature, poor preservation and subsequent loss of most of the Lycorhinus angustidens holotype make it impossible to determine whether or not the two forms were congeneric, but without positive evidence of congenericity we are obliged to regard them as separate genera. In any case, we consider the names Lycorhinus and L. angustidens to be nomina dubua.
If all three lower jaws be compared together, Heterodontosaurus and A.100 seem to be very different; Lycorhinus is intermediate between them, resem- bling Heterodontosaurus rather than A.100 in most of the contrasting charac- ters but more like A.100 in a few. A.100 is the most primitive of the three and Heterodontosaurus the most specialized. This is borne out by a comparison of the upper jaws and teeth (also the jugal) of A.100 with the corresponding elements of Heterodontosaurus.
188 ANNALS OF THE SOUTH AFRICAN MUSEUM
6. The slightly earlier genus from Argentina, Pisanosaurus, is also a heterodonto- saurid, probably more primitive than the forms from southern Africa but certainly too specialized to be ancestral to any post-Triassic ornithischians known.
7. As will be shown elsewhere, the heterodontosaurids possessed highly specialized masticatory and locomotor adaptations which indicated that they had moved into ecological niches quite distinct from those occupied by other contemporary dinosaurs, both ornithischian and saurischian. Diversity at the generic level might therefore be expected.
ACKNOWLEDGMENTS
We are grateful to Dr T. H. Barry, Director of the South African Museum, Cape Town, for allowing us access to the holotype of Heterodontosaurus tuck1; likewise to Dr J. F. Bonaparte, of the Fundacion Lillo of the Universidad Nacional de ‘Tucuman, Argentina, for permitting us to refer to his unpublished manuscript on Pisanosaurus and to reproduce two text-figures therefrom.
REFERENCES
Bakker, R. T. & Gatton, P. M. 1974. Dinosaur monophyly and a new class of vertebrates. Nature, Lond. 248: 168-172.
BONAPARTE, J. F. In press. Pisanosaurus mertii Casamiquela and the origin of the Ornithischia. J. Paleont.
Broom, R. 1911. On the dinosaurs of the Stormberg, South Africa. Ann. S. Afr. Mus. 7: 291-308.
Broom, R. 1932. The mammal-like reptiles of South Africa and the origin of mammals. London: Witherby.
CASAMIQUELA, R. M. 1967. Un nuevo dinosaurio ornitisquio tridsico (Pisanosaurus mertit ; Ornitho- poda) de la Formacién Ischigualasto, Argentina. Ameghiniana 5: 47-64.
Cuarie, A. J. In press. Thoughts on synonymy. Palaeontolog y.
Cuaric, A. J. & Crompton, A. W. In press. The Triassic ornithischian Heterodontosaurus tucki: skull, dentition and systematic relationships. Ann. S. Afr. Mus.
Cuaric, A. J. & Reic, O. A. 1970. The classification of the Proterosuchia. Biol. 7. Linn. Soc. 2: 125-171.
Crompton, A. W. & Cuaric, A. J. 1962. A new ornithischian from the Upper Triassic of South Africa. Nature, Lond. 196: 1074-1077.
Epmunp, A. G. 1969. Dentition. In: GANS, C., BELLAIRS, A. D’A. & PARSONS, T. 8., eds. Biology of the Reptilia. 1: 117-200. London, New York: Academic Press.
Gatton, P. M. 1972. Classification and evolution of ornithopod dinosaurs. Nature, Lond. 239: 464-466.
Gatton, P. M. 1973. The cheeks of ornithischian dinosaurs. Lethaia 6: 67-89.
GinspurG, L. 1964. Découverte d’un scélidosaurien (dinosaure ornithischien) dans le Trias supérieur du Basutoland. C. r. hebd. Séanc. Acad. Sci., Paris 258: 2366-2368.
Haucuton, S. H. 1924. The fauna and stratigraphy of the Stormberg Series, Ann. S. Afr. Mus. I2: 323-4097.
Kermack, K. A., Mussert, F. & Ricney, H. W. 1973. The lower jaw of Morganucodon. Zool. 7. Linn. Soc. 53: 87-175.
Kuun, O. 1966. Die Reptilien, System und Stammesgeschichte. Krailling bei Miinchen: Oeben.
Owen, R. 1861. On the fossil Reptilia of the Wealden and Purbeck Formations: Part 5 (Lacer- tilia). Palaeontogr. Soc. [Monogr.]: 31-39.
THE ALLEGED SYNONYMY OF LYCORHINUS AND HETERODONTOSAURUS 189
Romer, A. S. 1956. Osteology of the reptiles. Chicago: University of Chicago Press.
Romer, A. S. 1966. Vertebrate paleontology. 3rd ed. Chicago, London: University of Chicago Press.
THULBORN, R. A. 1970a. The systematic position of the Triassic ornithischian dinosaur Lycorhinus angustidens. Zool. F. Linn. Soc. 49: 235-245.
Tuuporn, R. A. 1970). The skull of Fabrosaurus australis, a Triassic ornithischian dinosaur. Palaeontology 13: 414-432.
TuuLporn, R. A. 19714. Tooth wear and jaw action in the Triassic ornithischian dinosaur Fabrosaurus. F. Kool., Lond. 264: 165-179.
THuLBoRN, R. A. 19715. Origins and evolution of ornithischian dinosaurs. JVature, Lond. 234: 75-78.
TuHuLporn, R. A. 1972. The post-cranial skeleton of the Triassic ornithischian dinosaur Fabro- saurus australis. Palaeontology 15: 29-60.
THE TARSUS OF THEROPOD DINOSAURS By SAMUEL P. WELLES & ROBERT A. LONG University of California Museum of Paleontology, Berkeley (With 13 figures)
CONTENTS PAGE Introduction . : : ‘ : 19] General evolution of the reptilian tarsus . 1QI The pivot tarsus of the crocodiloids . . 195 The hinge tarsus of the dinosaurs . : ‘ 196 The ceratosauroid tarsus : » OY The allosauroid tarsus 4 : : 203 The albertosauroid tarsus. : 5 ROT The ornithomimoid tarsus. 5 AM The tyrannosauroid tarsus - : ° 213 Summary : : : : 5 BIG Acknowledgements. 4 : : S25 Postscript k : ‘ : . 5 RUG References : : : : ; 2) INTRODUCTION
The theropod tarsus is a very highly evolved structure, and the investiga- tion of its origin and development is a most intriguing subject. As with most such studies, each new discovery reveals new unsolved problems, most of which cannot be pursued because they diverge too far from the main topic of research. Our interest in this rather specialized subject was aroused by the tarsi of two individuals of Dzlophosaurus, both in a fine state of preservation. We have long searched for theropods that might show relationship to Dilophosaurus. This led us to a comparative study of theropod tarsi and an inquiry into the origin of this peculiar mechanism. We have tried to limit our thoughts to the tarsus, but we must caution the reader that, consciously or unconsciously, our conclusions have probably been influenced by our knowledge of other parts of the skeleton.
GENERAL EVOLUTION OF THE REPTILIAN T'ARSUS
The theropod hind limb was derived from that of a quadrupedal archosaur or protoarchosaur, probably by or in the early Triassic. In the ancestral forms the femur is essentially horizontal, projecting laterally and swinging forward and backward about the knee. It has been customary to think of the leg as rotating in the acetabulum, and this is indeed the condition in the recovery stroke. However, once the foot becomes planted in its forward position the whole action is reversed and the foot becomes the pivot about which the epipodials and propodial revolve. We therefore think of a fixed pes with the limb lever
1QI
Ann. S. Afr. Mus. 64, 1974: 191-218, 13 figs.
ANNALS OF THE SOUTH AFRICAN MUSEUM
192
é a Is1oquieq sninesorydelyé OY SNINVSOUIL]L") ¢ Ja]20 sndoysinue
IOJCUSAOINIAY snyoXAuoutsq
snj[e sjenbout SNUITUIOTYINIIG snainesoyoAuoua}s
XO[OA SNUITWIOYITUIC
sm8unjmbe snanesojdAiq snjesense yseID [PH
(} / a / lmMeq ; sisusisapoys / sishydoja00 | snsiejudg 1
! u 1UJ9}SUITII]
/ sninesoor[epy / i snyjesesse 3 ee | mropom Ipuepyong sninesoydoyiq uomo]doj1990g
ee
SISUOSOATP SISUITUOXO
snjApuodsoidaysnq snjApuodsoidansny
/ Tuds[oO / sninesolj9[V
| |
sn} e1qi]
(DV
|! (sInsunjo1e —) snsevydooies
snunesozoq Vy
SI[Iseay sI[Isu SIUIOOISVU snineso][y snanja05, sninesojeiar) ¢ / f / / / / / sniesese vjowe'T]
| | | | | { |
xol sninesouuess J,
sie} podosoy) jo sdrysuoneay
THE TARSUS OF THEROPOD DINOSAURS 193
Table of indices and measurements of theropod tarsi
4 Cet = en se ee 25 pac icamne ee Boe ats ne Se Ge Hae £o Ws 88 aa Tyrannosaurus rex : i a ts 286 — 10 372 —_— Hell Creek LACM a ds # oo — — 196 — Ornithomimus velox a = a a EBA 123 60 52 65 Albertosaurus sarcophagus Se ae incrassatus, cfd.) .. ae Q7 — 52 248 — Albertosaurus sarcophagus, cfd. (A. eee 123 — — 216 265 Struthiomimus altus Lamb 1902 ee Bae 154 — 129 146 Albertosaurus libratus (Gorgosaurus 1.) ah LAG 128 67 Q12 246 Alectrosaurus olseni be ae aH SSE 5O 129 55 120 I4I Deinonychus antirrhopus | _ vit de bao 129 72 59 65 Microvenator celer ae ae oe = / 2H0 — 75 22,3 — Calamosaurus foxiestimated .. Se a pn — — 25 — Coelurus agilis.. ue oe a = 7S _ — 74 — Allosaurus fragilis cast .. is vf see ST 77 50 185 208 A. f. juv. cast - i = tes se 104: 03 — 98 110 Ceratosaurus nasicornis, type .. ae oo 46 35 — 130 Gon. type. leit, YPM. - 4 ihe oo 45 — — 118 C. ?n. UUVP 5682 oe - a oo 49 — — 147 C. ?n. UUVP 5681 a ae ie ee) ei 42 — a 175 Elaphrosaurus bambergi ay ae oo 100 — 88 = ?Eustreptospondylus Dives. Cuvier’s .. ae 88 — 38 120 — Eustreptospondylus oxoniensis = Xo al — 32 87 = Poecilopleuron bucklandi a Ce see 75 a 60 148 = Dilophosaurus wetherilli UCMP 37302 mee ey 61 Ba. gI = D. w., cfd. UCMP 37303 estimated... Go 48 20 — 114 Lufeng astragalus Pes ue ee sey ig 53 18 — 157 Halticosaurus liliensterni oe oe eke SAGE 53 II 65 79 Syntarsus rhodesiensis_.. i of 5) cas 49 12 = 20 “Coelurosaur’ Bonaparte est. .. Ee set OO 60 — 15 15 Coelophysis bauri Ra of Be oo 65 20 = 23
acting from this base. It functions by lifting the lower leg and foot, swinging them forward in a nearly horizontal arc centred at the acetabulum, or even farther inward if there is a concomitant sigmoid flexion of the vertebrae. This plants the foot in a forward position. Next there is a transfer of part of the
194 ANNALS OF THE SOUTH AFRICAN MUSEUM
body weight to this new base and its epipodial pillar. Then follows a horizontal rotation of the femur about the knee, atop the more or less fixed epipodials, which moves the acetabulum forward. At the same time the body weight is held above the ground, or at least lifted enough to be slid forward. The mecha- nics are first to lift and swing the epipodium and foot forward, then to reverse the forces and lift the body on this new fulcrum and propel the body forward.
As described by Schaeffer (1941: 412) in the salamanders, the propulsion phase begins immediately on the contact of the plantar surface of the foot with the ground. The recovery phase starts with the dorsal surface of the foot turning laterally into a vertical position. The crus is carried forward as an extension of the femur until just in front of a position at right angles to the body. The last movements involve a flexion of the knee, the extension of the foot and its rotation into a horizontal plane. This kind of limb movement, Schaeffer noted, was probably used by the smaller primitive amphibians. He went on to point out that Gregory & Camp (1918) had described the Permian reptile knee as permanently bent, and he thought that the heavier labyrinthodonts must have had similarly permanently flexed knees.
Broom (1921) presented the first general review of the reptilian tarsus, tracing it from the early Permian labyrinthodont Trematops miller: up into the therapsid reptiles. While he did include some other reptiles, he did not discuss the thecodonts or the dinosaurs. He noted Gegenbaur’s pioneer work and used his terminology.
Schaeffer (1941) also began with the tarsus of Trematops. From this he derived that represented by the early Permian cotylosaur Labidosaurus, and this became the ancestral pattern for later tetrapods. He found these later forms to have three divergent kinds of tarsi. His type I is a retention of the primitive pattern or its modification into an aquatic paddle as seen in the marine reptiles. The movement is originally infratarsal. Type II developed a mesotarsal joint with various specializations. Here the astragalus and calcaneum are functionally part of the epipodium, except in the crocodiles and most thecodonts. This joint developed in the Chelonia and, independently, in the lines leading from the Eosuchia, although not present in the Eosuchia themselves. In this type the distal tarsals become reduced. Type ITI is that leading to the epitarsal joint of the pelycosaurs, therapsids, and mammals.
Fig. 1. Reptilian tarsi.
a, type I, Mesosaurus; b, type II, Plateosaurus; c, type III, Youngina. Modified from Schaeffer 1941, fig. 11, not to scale.
THE TARSUS OF THEROPOD DINOSAURS 195
Schaeffer’s type II is, of course, the one that interests us. He wrote (p. 446), ‘the final stage in the perfection of the mesotarsal joint is found in the Saurischia, Ornithischia, Aves, and Pterosauria’. Here the astragalus and calcaneum have become depressed and the distal tarsalia are reduced to two or three. The astragalus and calcaneum may become fused, and are attached to the tibia and fibula. The distal surfaces of the astragalus and calcaneum become rounded into a transverse hemicylinder. At the same time the femur develops a medial head and the tibia becomes the prominent weight-bearer, with the fibula much reduced. In the theropods the tibia develops a distal lateral flange that extends behind the fibula. It also develops a sulcus on the anterolateral distal face, into which fits the ascending process of the astragalus.
Since the publication of Schaeffer’s work evidence has accumulated to indicate that there are at least two basic kinds of tarsi in the archosaurs. The first is the crocodiloid tarsus with its intertarsal movement, having the astra- galus locked to the tibia and the calcaneum locked to the pes. Movement between the astragalus and calcaneum is rotation about a horizontal axis. The second kind has both astragalus and calcaneum locked to the epipodials, or at least moving with them, and the movement is a hinge below these proximal tarsals. We can therefore expand Schaeffer’s classification to include II A, the crocodiloid tarsus, and II B, the dinosauroid tarsus.
Tue Pivot Tarsus OF THE CROCODILOIDS
The crocodiloid tarsus probably had as its ancestral stages those of the amphibians, cotylosaurs, and eosuchians. The movement was originally one of flexion, but later, in the thecodonts developed into a pivot, a rotation of the crus on or through the tarsus as the body moved forward. Some of this rotation was absorbed in the knee but much or most of it involved the crurotarsal joint. This pivot joint is found in such crawling forms as aetosaurs and crocodilians where the tibia can rotate upon the astragalus. Especially in the aetosaurs, this resulted in spiral articular surfaces between the two bones. This prohibited any hinge motion between tibia and astragalus and caused them to act as a unit so that the hinge developed between the astragalus and the distal tarsals on the inside, and the fibula and the calcaneum on the outside.
The astragalus in this foot was so named as early as Cuvier (1836, 9:207). Gegenbaur (1864: 88) first demonstrated that the bone was a composite of tibiale, intermedium and centrale, based on his embryological studies. Rabl (1910: 242) also found it to originate in three embryonic centres and so called it the ‘tritibiale’ This was later beautifully demonstrated in the cotylosaur Captorhinus by Peabody (1951: 340). Steiner, however, (1934, fide Krebs, 1963: 92) derived the astragalus from four embryonic centres representing the intermedium, proximal centrale, distal tibial centrale, and tibiale. Gegenbaur (1864: 92) noted the unique character of the crocodile foot and wrote ‘Der Calcaneous bewegt sich mit dem Unterschenkel fester verbundene Astragalo-
196 ANNALS OF THE SOUTH AFRICAN MUSEUM
Scaphoideum nur mit dem Unterschenkel sich bewegt’. He named this kind of joint, with the movement between the astragalus and calcaneum ‘intertarsal’ (p. 93). Schaeffer (1941: 442) considered the crocodiloid tarsus unique, yet noted the similarity between that of Protosuchus and Aetosaurus. He analysed the motion as a 45° rotation of calcaneum on astragalus, the latter immobile and functionally united with the crus, while the calcaneum moved as a part of the foot.
This crocodiloid intertarsal joint is widespread among the thecodonts. Ewer (1965: 426) described an ideal pre-crocodiloid ankle in the Olenekian Euparkeria and demonstrated that its astragalus functions as a unit with the tibia, while the calcaneum is attached to and works with the foot. This is the same crocodiloid structure that was described by Gegenbaur (1864: 92) in the crocodile but is more primitive in lacking the peg-and-socket joint between astragalus and calcaneum that is found in the crocodiles. Although Ewer con- sidered Euparkeria to be the earliest example of such a tarsus, a calcaneum was referred to Wangisuchus tzey Young (1964: 179, fig. 60 A) which has a well- defined astragalar sulcus, head, and tuber. This is from the Ehrmaying Series (Olenekian), the same age as Euparkeria, but has a more advanced, truly crocodiloid structure.
A possible mechanical explanation for the retention of this peculiar tarsus in the living crocodiles is that these reptiles have two distinct gaits. In the slow gait the animal crawls with the body close to the ground. The femur is horizon- tal and a certain amount of rotation of tibia on astragalus is required. Also, the fibula must be free to rotate on the calcaneum, which explains the ball-on- ball articulation between the two bones. In the fast gait the body is held high off the ground, the femur moves in a plane that dips about 60° laterally, and the astragalus is locked to the tibia. The movement then becomes entirely hinge, or dorsoflexion, between the metatarsals and the unit composed of tarsalia, astragalus and tibia on the inside, and the calcaneum and fibula on the outside.
Tue Hince Tarsus OF THE DINOSAURS
The dinosaur tarsus with its mesotarsal hinge is very different from the tarsus discussed above. Krebs (1963: 91), Ewer (1965: 426), and Bonaparte (1969: 474) each maintained, and we agree, that the mesotarsal joint cannot be derived from the intertarsal joint. Furthermore, as noted by Bonaparte, the twisted tibia of the Triassic dinosaurs is very different from the straight tibia of the pseudosuchians. Attridge (in Bonaparte) pointed out the association of the twisted tibia with this mesotarsal dinosaur articulation. Present evidence indicates an entirely separate origin for the dinosaur mesotarsal hinge.
The mesotarsal joint is also correlated with an inturned head of the femur and with the femur moving in a nearly vertical plane. The mesotarsal joint limits the movement to a parasagittal plane and is an adaptation to bipedal locomotion. However, bipedal locomotion may also develop in the crocodiloid
THE TARSUS OF THEROPOD DINOSAURS 197
tarsus, as in Saltopus. Furthermore, Dr Alan Leviton of the California Academy of Sciences informs us that several of the desert lizards, including Callisaurus and Crotaphytus can run bipedally with the femur in a vertical plane and the feet under the body. This is to say that although the mesotarsal joint is a bipedal adaptation, bipedalism may also develop in reptiles with other kinds of tarsi.
The earliest of the mesotarsal ankles are in the two ?Anisic genera Lagerpe- ton and Lagosuchus described by Romer (1971, 1972). Lagerpeton has a short astra- galus, fused to the calcaneum, and with a dorsal process up the posterior side of the tibia. This is a unique arrangement among dinosaurs, for in all other known forms the dorsal process develops up the anterior face of the tibia. The question naturally arises as to whether Lagerpeton can truly be included in the dinosaurs. Lagosuchus has a quite different tarsus in that the astragalocalcaneum is much longer, there is no dorsal process, and there is a small posterior spur on the calcaneum. These two reptiles are quite different from each other and neither shows any direct relationship to later archosaurs.
Our studies indicate that the theropod tarsi can be grouped into five rather distinct kinds:
1. Ceratosauroid—the astragalus with a low dorsal process that is sunk into the tibia, and a large calcaneum that is closely applied to or fused with the astragalus. Late Triassic to late Cretaceous. Ceratosaurus, Coelophysis, Dilophosaurus, Halticosaurus, Lameta astragalus, Lufeng astragalus, Syntarsus.
2. Allosauroid—the astragalus with a moderately high dorsal process that is set into the anterior face of the tibia and yet has a free horizontal medial component. The medial condyle is large and the medial end rounded. A deep upper horizontal groove develops at the anterior base of the dorsal process and a deep lower horizontal groove across the face of the condyles. The calcaneum is free and large, and in Allosaurus develops a medial tongue and two pits to fit the astragalus. Middle Jurassic to late Jurassic. Allosaurus, Coelurus, Eustreptospondylus, ‘Megalosaurus bucklandi’, Phillips, Poecilopleuron.
3. Albertosauroid—the astragalus with a high dorsal process that is applied to the entire anterior and anterolateral faces of the tibia. A deep upper horizontal groove develops above the condyles. This is interrupted laterally by a ventrolateral buttress to the lateral condyle. The condylar surface does not develop the lower horizontal groove. The calcaneum is reduced and free from the astragalus. Late Cretaceous. Albertosaurus, Alectrosaurus, Dryptosaurus, Hell Creek astragalus.
4. Ornithomimoid — the astragalus with an extremely high dorsal process that is applied to the entire anterior face of the tibia. The calcaneum is reduced or absent. Late Jurassic to late Cretaceous. ?Calamosaurus, Elaphrosaurus, Microvenator, Ornithomimus, Stenonychosaurus, Struthiomimus.
5. Tyrannosauroid—the astragalus with a high, wide dorsal process. The upper horizontal groove shallow and restricted to the medial half. No lower horizontal groove, but a large oval foramen in the centre of the anteroventral condylar face. The medial condyle high, its dorsal border sloping steeply ventromedially; the lateral condyle weakly developed.
These five kinds of tarsi will be described in the following paragraphs.
THE CERATOSAUROID TARSUS
Coelophysis baurt from the upper Chinle Formation of New Mexico, was previously thought to have an astragalocalcaneum without a dorsal process, but preparation of the tarsus at the American Museum of Natural History (block 5) has revealed a well-developed dorsal process on the astragalus. This
198 ANNALS OF THE SOUTH AFRICAN MUSEUM
was not known to Dr E. H. Colbert and so was not indicated in the figure published by Raath (1969, fig. 6d) on information from Colbert. Dr Colbert is preparing a monograph on Coelophysis but he has kindly consented to our advance publication of this information. This astragalocalcaneum is skewed anterointernally, has a maximum breadth of 23 mm anteriorly and 22 mm posteriorly. It is 11 mm long internally and 12 mm externally. The dorsal process is 15 mm high above the plane of the distal condyles, yielding an H:B
O ZS)
Fig. 2. Coelophysis bauri. a, internal view, dorsal process not shown; b, anterior view, of left astragalocalcaneum and distal end of fibula. AMNH block 5. x 1.
index of 65. The dorsal process has a bearing surface 3 mm wide that underlies a flange of the tibia, and dips 20° medially. There is a shallow upper horizontal groove across the anterior face separating the condyles from the dorsal process. This groove ends laterally in a notch in the calcaneum. Just above this groove, and 5 mm medial to the fibular edge, a 2 mm foramen marks the opening of a canal which penetrates laterodistaliy into the bone. The body of the astragalus
Bigenge Syntarsus rhodesiensis, left tarsus in a, anterior; b, section through centre, anterior to right. From Raath 1969, fig. 6, approximately x 1,75.
THE TARSUS OF THEROPOD DINOSAURS 199
is but 4 mm thick laterally below the tibia. Its medial edge slopes 45° postero- laterally. The posterior edge is sharp and rounds upward about 4 mm so that the astragalocalcaneum forms the entire bearing condyle. The calcaneum is fused, is 8 mm deep below the fibula and forms the lateral part of the condyle.
Syntarsus rhodesiensis Raath, 1969, from near the top of the Forest Sandstone of Rhodesia (equivalent to the upper Red Beds of the Stormberg Series, of Norian age), is a small theropod with an astragalocalcaneum but 20 mm wide and 9,7 mm high, an H:B index of 49. This is a very low, almost incipient dorsal process, the slope of its medial surface but 12°. However, the cross section of the right tarsus given by Raath shows that the ascending process continues up into the tibia and is covered anteriorly by an overlap of the tibia. The true H:B index is therefore somewhat greater than 49. The general structure of this tarsus, with its low, deeply set dorsal process seems to us to be ceratosauroid, although it should be noted that Galton (1971: 788) considers the ankle to be ‘typically sauropodomorph’.
Young (1951: 75, fig. 15) describes a theropod astragalocalcaneum (no. V 73) from the Dark Red Beds of Haungchiatien, the uppermost beds of the Lower Lufeng Series, generally considered to be upper Norian to Rhaetic. This is a gigantic bone, 157 mm broad, with a total height of 84 mm, an H:B index of 53. The slope of the dorsal process is 18° and it begins at the lateral
Fig. 4. Ceratosauroid from Lufeng Beds.
Left astragalocalcaneum in a, anterior; b, posterior; c, dorsal; d, medial views. From Young 1951, fig. 15. x t.
200 ANNALS OF THE SOUTH AFRICAN MUSEUM
border and is almost straight. The fusion of the two bones and the medial extent of the dorsal process are not allosauroid but are similar to Dilophosaurus and Ceratosaurus.
Halticosaurus liliensterni (Knollenmergel, Norian, of Thuringia) has a primi- tive astragalus, with the calcaneum solidly fused into an astragalocalcaneum. The two bones are 79 mm wide anteriorly, the astragalus 65 mm. The dorsal process is 42 mm high and dips medially 11°, the H:B index is 65. The dorsal
| cm) O 5
Fig. 5. Halticosaurus liliensterni, type.
Anterior view of left astragalus and distal end of tibia. x $.
process is thick (16 mm) and in lateral view the entire element, including the dorsal process, is almost circular, measuring 46 mm anteroposteriorly.
The type of Dilophosaurus wetherilli (UCMP 37302) has an astragalus 9g! mm broad anteriorly and but 76 mm broad posteriorly. In anterior view the dorsal process is 70 mm high above a straight line across the base (H:B index 76), 14 mm thick at mid-height, and shows little if any bearing surface. It begins at the median border and rises in a uniform slope of 16° to about 20 mm from the summit where the slope increases to give a total slope of 27°. The anterior condylar surface is concave transversely, and rounds below into a medial condyle and a lateral condyle that continues onto the calcaneum. There is a slight transverse groove in the centre of the rounded medial edge. The anterodorsal border of this surface projects forward and is separated from the dorsal process by a broad upper horizontal groove that deepens laterally and terminates at the fibular socket. The rounded anterolateral projection of the astragalus forms the anterior border of the fibular socket. About half of this socket is in the astragalus. A rounded buttress runs down the anterolateral corner of the dorsal process and extends onto the anterolateral projection of the lateral condyle. In dorsal view the sulcus for the tibia runs posterolaterally,
THE TARSUS OF THEROPOD DINOSAURS 201
and a small boss in the posterior border, towards the medial edge, fits into a socket in the tibia. The tibial sulcus does not extend behind that for the fibula, as it does in Allosaurus, indicating that the distal end of the tibia lacks the great
Fig. 6. Dilophosaurus wetherilli, type. Left astragalus in a, anterior; b, posterior; c, dorsal; d, ventral views. UGMP 37302. x 3.
lateral expansion of the latter, as is shown by the tibia itself. The socket for the fibula is 17 mm broad in front and wedges out posteriorly, its bearing surface increasingly taken over by the calcaneum. In medial view the bone 1s very thick beneath the tibia, contrasting strongly with the thin bone of Allosaurus. In posterior view the lateral edge is sharp, but it thickens medially to form a posteromedial knob. A sharp ridge runs down posterolaterally from the dorsal process to project above and behind the calcaneum, and to divide the tibial sulcus from that for the fibula. There is a pit near the base of the dorsal process, probably for ligaments binding the astragalus to the tibia. The medial surface of the dorsal tip of the dorsal process is a flattened surface 14. mm wide that faces toward the anteromedial corner. The ventral surface of the astragalus is saddle-shaped, convex anteroposteriorly and concave laterally. It forms a lateral and a medial condyle, constricted between. Its lateral and medial edges converge posteriorly, the former sharp and forming a straight suture with the calcaneum. The medial edge is rounded. In lateral view the socket for the fibula is 30 mm above the distal end, 10 mm higher than that for the tibia. The surface for the calcaneum is subcrescentic, slightly concave, with a triangular projection overhanging the calcaneum near the posterior edge.
The isolated right tarsus of another specimen (UCMP 37303) also from the base of Kayenta Formation of Arizona (late Triassic or early Jurassic) has the calcaneum solidly in place and the total breadth is estimated to be 114 mm. The dorsal process has been crushed down somewhat by the anterolateral shoulder of the tibia and it is but 55 mm high, an H:B index of 48.
In the right tarsus of the type of Ceratosaurus nasicornis, from the late Jurassic
202 ANNALS OF THE SOUTH AFRICAN MUSEUM
Morrison Formation of Colorado USNM 4735 fide Gilmore (1920: 111, fig. 65) the two proximal tarsalia are fused into an astragalocalcaneum and the dorsal process of the astragalus is quite restricted. In anterior view the dorsomedial border slopes upward rather regularly from the inner base instead of beginning part way laterally as in Allosaurus—a difference noted by Gil- more. There is a triangular area above the lateral border of the astragalus and below the tibia and fibula which does not seem to have been filled by the dorsal process. This triangle is evident on the right and left type tarsi and also on two others at the University of Utah. This area seems to be ossified differently from the rest of the bone and could possibly have been occupied by calcified cartilage. The lateral end of the tightly fused calcaneum is deeply concave. The left tibia and astragalocalcaneum of this same individual,
YPM 4681, has a breadth of 118 mm and an H:B index of 45, and is similar to
O /0
Fig. 7. Ceratosaurus nasicornis, type. Anterior view of left astragalocalcaneum and distal end of tibia, YPM 4681. The rest of this skeleton is at the United States National Museum. x }.
THE TARSUS OF THEROPOD DINOSAURS 203
that of the left side. On both right and left the suture between astragalus and calcaneum is visible on the anterior face.
A fine right tibia and tarsus at the University of Utah (UUVP 5682), referable to the genus Ceratosaurus, if not the type species, has a breadth of 147 mm and an H:B index of 49. A second associated specimen (UUVP 5681), the largest known, has a breadth of 175 mm and an H:B index of 42.
Huene (1933: 56, pl. 19, fig. 1) describes an astragalus that he identifies as allosaurid, from the Lameta beds (Turonian) of Jubbulpore. ‘This astragalus (K27/684) is incomplete medially and is described as thick laterally and very thin, almost platelike in all other parts. It is 50 mm anteroposteriorly, with the dorsal process 20 mm high and 30 mm broad. It is ‘twisted in the direction of the transverse axis’. Huene gives no reasons for identifying this as allosaurid and his illustrations do not support his view. His ‘dorsal view’, fig. 1a, is, with the tibia vertical, evidently posterodorsal with the dorsal process down. The ligamentary pit is near the base of the dorsal process. Figure 1b, an anterior view, shows a low dorsal process with an external vertical ridge curving ventro- medially from the summit. The lateral face of the tibial sulcus rises into a high wall behind the sulcus. The lateral view shows a smooth calcaneal face. All of these features contrast with those of the allosauroids and are much more ceratosauroid. An isolated calcaneum is also ceratosauroid in being broad and parallel-sided.
THE ALLOSAUROID TARSUS
Phillips (1871: 214) describes an astragalus as that of Megalosaurus buck- land. It rises upward in a broad flat plate which fits in a well-defined hollow in the flattened anterior face of the tibia. Its lateral extension toward the fibula is cut off by a nearly vertical surface for the calcaneum. Phillips makes no mention of an interdigitation of calcaneus and astragalus, but the dorsal process and bulbous medial condyle are of the general allosauroid pattern. We therefore consider this a possible early allosauroid pending further investigation. It is presumably from the Oxford Clay.
Deslongchamps (1838, pl. 6, fig. 12) described and illustrated the astragalus of his Poecilopleuron bucklandi. This specimen was formerly at Caen but Dr Donald Russell informs us that it was destroyed during the war. It came from Vesulian or Lower Bathonian rocks, approximately equivalent to the Stones- field deposits. The astragalus is large, its breadth 148 mm and the H:B index is 75. The medial edge of the dorsal process begins rather far laterally, and slopes upward at about 60°. The medial half is low, the condyle gently convex transversely. The lateral face for the calcaneum is concave dorsoventrally and is much the thickest part of the body. The astragalus is advanced in the height of the dorsal process and its lateral restriction. The shape and position of the dorsal process, and the free medial condyle, are allosauroid but there does not seem to have been any interdigitation of astragalus and calcaneum. It could be a primitive representative of this group.
204 ANNALS OF THE SOUTH AFRICAN MUSEUM
Eustreptospondylus oxoniensis Walker, 1964 (UMOJ 13558), is from the Middle Oxford Clay (Upper Callovian) of Wolverton, southern England. The astragalus is separate from the calcaneum and measures 87 mm broad and 62 mm high, an H:B index of 71. The right astragalus
Fig. 8. Eustreptospondylus oxoniensis, type.
Right astragalus (reversed to left) in a, anterior; b, medial; c, lateral views. x 3. was sketched by Huene (1926, fig. 36) and he also illustrated the separate calcaneum, since lost. A detailed drawing sent us by Mr H. P. Powell allows us to make a more complete description. It has a breadth of 87 mm and a height of 62 mm, an H:B index of 71. The anterior view shows a 33° slope of the medial edge of the dorsal process. This edge is concave and slopes but 26° at the base. It has a long, low, rounded medial condyle and a fairly sharp ventro- lateral border of the lateral condyle. The dorsal edge of the calcaneal border is set forward of the dorsal process and has a rounded dorsolateral projection. The dorsal process is separated from the anterior surface by an upper horizontal groove that fades out medially. There is no deep pit for a process of the cal- caneum as in Allosaurus, but there is a concave edge for the calcaneum. A slight lower horizontal groove runs about a quarter of the way inward from the centre of the medial condyle. The lateral view shows a high calcaneal facet with the upper horizontal groove at the base of the dorsal process separating the rounded anterodorsal projection. The posterior foot below the tibia is bluntly rounded behind and becomes longer anteroposteriorly on the inner side. The striking features of this view are the strong posterodorsal projection above the lateral condyle and the upper horizontal groove above and behind it.
THE TARSUS OF THEROPOD DINOSAURS 205
The medial view shows a rather thick (high) medial condyle that is 43 mm anteroposteriorly on the right astragalus and 48 mm on the left. The foot is rounded posteriorly. The general shape of the dorsal process and its upper horizontal groove, and the bulbous medial condyle are allosauroid, but the deep interdigitation of the astragalus and calcaneum found in Allosaurus has not developed.
The famous Honfleur specimen described by Cuvier (1836, 10:204, pl. 249, figs 34-36) in his 4th edition, and probably earlier, was stated to have come from one of the two blue marls above the Caen stone. Huene (1926: 38) places the horizon as Upper Callovian. It consists of the right astragalus and the distal end of the tibia. The taxonomic problems arising out of this and the other Honfleur material are very difficult and need not concern us here, but Walker (1964: 122) reviewed them and referred this specimen to his Eustrepto- spondylus divesensis. Cuvier’s original description was inadequate for our com- parisons, and he used a different orientation. The specimen was well illustrated by Piveteau (1923: 9, pl. 4, figs 4, 5). One of us (Welles) was privileged to examine this specimen at Paris and noted that the tibia had a sharp bend in the supra-astragalar shelf and no real bearing surface. The astragalus had a breadth of 115 mm and a height of 94 mm, a H:B index of 82. In anterior view the dorsal process is evidently nearly entire, its tip of 2 mm finished in plaster, and is broad and rounded above. Its inner edge is concave, the slope 30°. The upper horizontal groove is very weak, present laterally but fading out medially, and there is a strong dorsolateral process above the lateral condyle, widely separate from the dorsal process. The lower horizontal groove is very poorly developed. In dorsal view the sulcus for the tibia is very narrow laterally and there is no development of a projection to insert in the tibia. In lateral view the calcaneal facet is a vertical oval with a fibular sulcus 41 mm above the base. In posterior view the posterior edge is sharp laterally but thickens medially. In general features this astragalus is allosauroid, but much more primitive than Allosaurus. In addition to the astragalus, Cuvier (pl. 249, fig. 38) illustrates a bone that he identifies as a tarsal bone. This is evidently an allosauroid calcaneum in that it has a central projection below the fibular facet. There is no proof that the two bones are from the same individual, but if they are, this is the earliest known occurrence of the pattern found in Allosaurus, consisting of separate though interlocked astragalus and calcaneum.
An astragalus at Yale (YPM 1252) from the Morrison Formation at Como Bluff is possibly referable to Coelurus agilis. In anterior view it is 74 mm broad and 58 mm high, an H:B index of 78. It has a vertical lateral base to the dorsal process and an inner border that slopes about 68°. The dorsal process is confined to the lateral half and there is a broad upper horizontal groove at the lateral base of the dorsal process. This groove separates a posterodorsal projection of the lateral condyle from the dorsal process. The posterior face shows a ligamen- tary pit at the centre of the base of the dorsal process. The body is deepest medially, about the same depth below the ligamentary pit, and thinnest
206 ANNALS OF THE SOUTH AFRICAN MUSEUM
laterally. The sulcus for the distal end of the tibia is divided into halves, the inner half sloping slightly medially (3°) and the lateral half rather steeply (10°) laterally. All of these features are of a general allosauroid nature.
A cast of a large specimen of Allosaurus fragilis from the Upper Morrison Formation of central Utah sent us by Professor James H. Madsen, Jr., of the University of Utah, has a maximum breadth of 185 mm along the anteroventral border, a posterior breadth of 160 mm, and a height of 160 mm. The H:B index is 87. (The specimen illustrated by Gilmore (1920, fig. 50, USNM 7336) has a breadth of 200 mm and a height of 170 mm, an H:B index of 85.) In anterior view the dorsal process begins 67 mm lateral to the inner edge and rises at a steep slope (53°) to the summit. This slope is interrupted by a dorsal bulge near the top. A strip along the lateral border, its widest part
Fig. 9. Allosaurus fragilis, cast of right astragalus and calcaneum from University of Utah. a, internal view of calcaneum; b, lateral view of astragalus; c, anterior view of both. x 4.
30 mm, is set at an angle of 35° to the anterior face, and forms a plane surface for the fibula. The upper horizontal groove, along the base of the dorsal process, is deep and rounds posterolaterally into the sulcus for the fibula. This groove is deepest laterally and ends medially at the base of the dorsal process. The dorsal process is broad, flat, and thin, extending laterally and not excavated by the fibular socket. The lower horizontal groove is pronounced and extends laterally for half the width of the astragalus. The lateral edge, against the calcaneum, has a deep notch that received a point of the calcaneum. In posterior view the ligament pit at the base of the dorsal process is large and deep. The lateral half forms a fairly thin (20 mm) floor below the tibial sulcus and has a sharp posterior edge. The medial half thickens, both upward and downward, to form a bulbous medial condyle rising 70 mm above the base. Behind this
THE TARSUS OF THEROPOD DINOSAURS 207
condyle the floor of the tibial sulcus is 42 mm thick. This floor is obscured by the upturned posterior edge, but the medial 70 mm makes an angle of 17° with the outer 110 mm. In dorsal view the sulcus for the distal end of the tibia is 83 mm broad medially and but 32 mm broad laterally. In lateral view the socket for the distal end of the fibula is set into the astragalus between the dorsal process and the anterodorsal knob above the lateral condyle. This inset is a continuation of the upper horizontal groove. Below and behind the fibular sulcus is a complex notch for the calcaneum, and although not fused, the two bones are deeply interlocked. In addition to the deep notch in the centre of the astragalus below the fibular socket there are two rounded processes which project into sockets in the calcaneum. One of these processes is anterodorsal, the other ventral, and this complex interdigitation of the notch and two processes prevents any movement between astragalus and calcaneum. The socket for the fibula is high above that for the tibia. (A specimen illustrated by Gilmore (1920, fig. 48, USNM 4734) is pathologic in having a crushed, and evidently healed, distal end of the fibula that lowered the calcaneum about half its height. This same specimen was illustrated by Marsh (1806, pl. 11) with the distortion corrected.) In ventral view the astragalus is distinctive in that it narrows laterally to about half of its medial anteroposterior dimension. The medial border is at nearly right angles to the posterior border but makes only a 55° angle with the anterior border. The lateral border is rounded, but the notch for the process of the calcaneum is evident.
The calcaneum of Allosaurus has its medial face modified to match the astragalus. There is a central projection that carries the floor of the fibular socket into the astragalus. Anteroventral to this is a socket that receives the anterodorsal projection of the astragalus, and posteroventral to this is another socket facing ventromedially, below and anteromedial to the tibial sulcus. This socket receives the ventral projection of the astragalus. The posteromedial face of the calcaneum is a concave semilunar sulcus for the lateral edge of the tibia. The lateral face is also semilunar with its upper concavity forming the outer rim of the fibular socket. The calcaneum is larger than in ornithomimoids or albertosauroids, but smaller than in ceratosauroids.
Another Allosaurus astragalar cast, representing a small, immature indivi- dual, was also sent to us by Professor James J. Madsen, Jr. This specimen is particularly interesting in that it has an H:B index of 104, much greater than the 87 of the large cast. The dorsal process also differs from the adult in that the inner edge begins to slope up gently at 22° from the inner border, then, about half-way out the dip increases gradually to about 72°.
THE ALBERTOSAUROID TARSUS
Gilmore (1933: 38, fig. 10) describes the astragalus and calcaneum of Alectrosaurus olseni from the Iren Dabasu Formation (Cenomanian) of Mongolia. This is his second specimen, AMNH 6554. Gilmore designated this and the fore limb from 30,5 m away (AMNH 6368) cotypes yet treated them as repre-
208 ANNALS OF THE SOUTH AFRICAN MUSEUM
senting two individuals. To avoid confusion we here designate this specimen, AMNH 6554, the type of the species. The fore limb therefore becomes a referred specimen. The tibia and tarsus were seen by Gilmore to be closest to Albertosaurus libratus (which he called Gorgosaurus sternbergi). He also noted the difference in the proportion of the dorsal process of the astragalus in the two species, rising 4 of the height of the tibia plus astragalus in A. libratus while it is 4 in Alectrosaurus olsen. ‘The height of the astragalus is given as 180 mm and measurements taken from the illustration give an estimated breadth of 120 mm, an H:B index of 150. The medial edge of this specimen is missing but enough is present to show the deep upper horizontal groove. The lateral part shows a slight ventrolateral buttress against the lateral condyle, weaker than in Albertosaurus arctunguis, but stronger than in A. sarcophagus. The calcaneum is more strongly developed and has a higher anteromedial process than in A.
arctunguis.
O £0
Fig. 10. Albertosaurus libratus, type.
Anterior view of left tarsus and distal end of tibia, from Lambe 1917, fig. 42. x 4d.
ce ee ee ee ee ee ee re eer
THE TARSUS OF THEROPOD DINOSAURS 209
Lambe (1917: 69, fig. 42) described and figured the tibia and tarsus of Gorgosaurus libratus from the Belly River Formation (Judith River, Campanian) of Alberta. This was placed in the genus Albertosaurus by Russell (1970: 4). The dorsal process of the astragalus is nearly as broad as the base and the inner face of the base is ‘flat or only slightly concave’. ‘The dorsal process is set back from the body and separated by a deep upper horizontal groove which is very deep at the centre. The height of the astragalus, taken from the figure, is 314 mm, the breadth 212 mm, an H:B index of 148. The laterodistal buttress against the top of the lateral condyle is rather strongly developed, not so pronounced as in A. sarcophagus, referred (= A. arctunguis), but more so than in either A. sarco- phagus, referred (= Dryptosaurus incrassatus, referred, Lambe 1904) or Alectro- saurus olsem. ‘The calcaneum is also intermediate in its stage of reduction between A. olsent and Albertosaurus ‘arctunguis’.
Parks (1928: 20, fig. 12) described the astragalus of Albertosaurus arctunguis from the Horseshoe Canyon Formation (Campanian-Maastrichtian) of Alberta. This was placed in synonymy with A. sarcophagus by Russell (1970: 4). It has the calcaneum in place, although probably not fused because it is free in the specimen described as Dryptosaurus incrassatus (Cope) by Lambe (1904: 20, pl. 6, fig. 6), later referred to A. sarcophagus by Osborn (1905: 263). The breadth of the combined astragalus and calcaneum is 265 mm, that of the astragalus 216 mm (from the illustration). Parks notes ‘no significant difference between’ this astragalus and that ascribed by Lambe to A. sarcophagus, except that the dorsal process is a little wider and the body a little thicker. However, a com- parison of the two illustrations shows a considerable difference in the shallowness of the body, the deep narrow anterior concavity, and especially the development of the large rounded ventrolateral buttress onto the lateral condyle in A. arctun- guis. Although the two species have astragali developed on the same basic principles, that of A. arctunguis is more specialized.
Lambe (1904: 20, pl. 6, figs 6-8) illustrated and described an astragalus and the distal end of a tibia as Dryptosaurus incrassatus (Cope). This was referred to Albertosaurus sarcophagus by Osborn (1905: 265), and is from the Horseshoe Canyon Formation (Campanian-Maastrichtian) of Alberta. Russell (1970: 10) identified this as NMC 5601. The breadth is given as 248 mm and an estimate of the height is 241 mm, an H:B index of 97. The dorsal process is set laterally about 30 mm, its lateral edge rising steeply at first, then sloping more gently at about 52°. In anterior view this astragalus has a large medial condyle that is 86 mm high. The central constriction is roughened by vertical grooves and ridges. The condylar body is separated from the dorsal process by a deep upper horizontal groove that is deepest in the centre. The surface for the cal- caneum is notched anteroventrally and this surface is described as a concave area. The medial and lateral ends of the bone are oblique, a common feature that makes the posterior border narrower than the anterior. The calcaneum is said to have contributed only to the support of the fibula.
Cope (1866: 316; 1870: 105, pl. 9, fig. 3) described the astragalus of his
210 ANNALS OF THE SOUTH AFRICAN MUSEUM
Laelaps aquilunguis from the Barnsboro marl pits of Gloucester Co., New Jersey. The generic name was preoccupied and Marsh (1877: 88) proposed Dryptosaurus to replace it. The horizon is the Greensand Navesink Formation, of Maastrich- tian age (Miller 1962: 195). Cope’s figures show an astragalus that underlies only the anterior half of the tibia and has a very peculiar dorsal process. This is probably a badly eroded specimen and should be restored along the lines of Albertosaurus. This would make the dorsal process at least 50°% higher than the breadth of the astragalus. It would make the peculiar opening near the lateral base into the lateral, eroded edge of the upper horizontal groove. The ventral posterior projection below the tibia should be somewhat increased. So restored, the astragalus has a breadth of 184 mm. The ventrolateral buttress does not reach the calcanear surface and its development is between that of Alberto- saurus libratus and A. arctunguis. The facet for the calcaneum indicates that the calcaneum was nearly as reduced as in the latter species. This, therefore, seems to be an albertosauroid astragalus.
An astragalus from an incomplete skeleton collected in the Hell Creek Formation (Maastrichtian) of Garfield Co., Montana in the Los Angeles County Museum collections (LACM 23849) kindly lent us by Dr Ralph Molnar is badly broken, but the lateral and medial ends are preserved along with the base of the dorsal process. It can be restored with confidence, except for the dorsal process, with a breadth of 196 mm. In anterior view the dorsal process arises 17 mm late-
Fig. 11. Albertosauroid astragalus (right, reversed) from the Hell Creek Beds.
a, lateral; b, medial; c, anterior views. LACM 23849. x 4.
THE TARSUS OF THEROPOD DINOSAURS 211
rally from the bulbous medial condyle. The upper horizontal groove at the base of the process is shallow medially and laterally, so was probably not very deep centrally. A 5 mm perforation enters the body of the bone at the lateral end of the groove. The tunnel from this perforation leads ventromedially. The ventro- lateral buttress is weak and rounds gently into the lateral condyle. In ventral view the condyles are smoothly rounded and not marked by the lower horizontal groove that in Allosaurus separates a distal face from an anterior face. The medial condyle is very large, bulbous, and smoothly rounded, somewhat like that of Dryptosaurus, except that the lower posterior edge slants more steeply antero- laterally. The lateral condyle is also round, yet has a flattening of the antero- ventral face. The lateral condyle has a narrow cleft near the base, in the same position as that of Allosaurus, but very much smaller. No such cleft is evident in Dryptosaurus.
THE ORNITHOMIMOID TARSUS
The astragalus of the partial skeleton from the late Jurassic Tendaguru beds referred to FElaphrosaurus bambergi by Janensch (1925: 41, pl. 6, fig. 1) is 88 mm broad and 56 mm anteroposteriorly. The condyles are nearly equal and are separated by a gentle constriction. The condylar surface ends posteriorly in a sharp, straight edge. In anterior view the bone is pinched in between the condyles. There is a dorsolateral projection that is separated from the dorsal process by the upper horizontal groove, which runs across the front of the bone. Janensch describes a dorso- medial projection in front of this groove which is a triangular process about 13 mm high. The dorsal process is described as weakly developed and near the medial edge, yet its height is indicated by an 85 mm high sulcus on the anterior face of the tibia. From this we estimate an H:B index of 100. In dorsal view the tibial sulcus is divided by a central rise. Its lateral half narrows and also has a concavity for the fibula. We have not examined this specimen but Janensch’s description indicates that it is the only theropod astragalus with a dorsomedial process in front of the upper horizontal groove. Janensch considers the entire bone to be astragalus, yet he describes a suture near the lateral edge which sets off a small dorsally-narrowed surface that he thought must be the calcaneum. If the astragalocalcaneum is shifted to its proper position on the ventral face of the tibia, we find that the lateral part is in the proper position for a calca- neum, and that the resulting calcaneum is rather large. This is considerably advanced over the coelophysoid type and the height of the dorsal process excludes it from the ceratosauroids. It lacks the ventrolateral buttress of the albertosauroids and seems to represent a primitive ornithomimoid.
Ostrom (1970: 75, pl. 13, figs D, E) described his Microvenator celer from the Cloverly Formation of Montana (Aptian-Albian) as having the astragalus extending the full breadth of the tibia with a small recess for a tiny, chiplike calcaneum. The breadth is 22,3 mm, the height 42,3 mm so the H:B index is 190, the highest known. The dip of the inner margin of the high dorsal process is
212 ANNALS OF THE SOUTH AFRICAN MUSEUM
75°. This is certainly an advanced tarsus of the ornithomimoid pattern, to be expected in a form of Aptian-Albian age.
Stenonychosaurus inequalis Sternberg, from the Judith River Formation was figured by Russell (1969 fig. 11). The tibia and astragalus of the type specimen is broken some 65 mm up the shaft so the dorsal extent of the astragalus is not known. It shows an astragalus very much like that of Microvenator but the inner condyle slants more medially and the lateral condyle is more nearly vertical.
Lydekker (1891, pl. 5, fig. 2) described the tibia of a small theropod from the Wealden of the Isle of Wight, BMNH R186. This he referred to Calamo- saurus foxi. There are no tarsal bones associated with this tibia but the area for articulation of the dorsal process of the astragalus is an acute triangle, indicating a high dorsal process with an estimated H:B index of 132.
Ostrom (1969: 120, fig. 68) described the astragalus and calcaneum of his Deinonychus antirrhopus from the Cloverly Formation of south-central Montana. The two bones were not fused, nor were they fused to the tibia. The astragalus is described as similar to that of Ornithomimus (YPM 542) but larger and more perfectly rounded distally. This distal rounding is extreme, given by Ostrom as 155° to 180° of arc, and evidently continued by the tibia to approximately 200°, providing an ‘unusual extension of the foot’. The breadth is 5g mm and the H:B index at least 142. The dorsal process is very thin and at least 84 mm high. It rises about 5 mm from the medial edge at a steep angle of 72° and is very broad (52 mm) at the base, extending laterally behind the calcaneum. The tibial sulcus is carried laterally onto the posterior face of the calcaneum behind the high, small fibular sulcus. This tarsus is ornithomimoid in its extremely high dorsal process, but is aberrant in retaining so large a calcaneum.
Gilmore (1920: 138) describes the astragalus that he considers to be a cotype of Ornithomimus affinis, from the Arundel Formation, Lower Cretaceous of Maryland. This is 78 mm broad, the dorsal process much eroded to a height of 56 mm, and is narrow anteroposteriorly. Russell (1972: 379) transfers this species to the genus Archaeornithomimus.
Lambe (1902: 50, fig. 11) described the right hind limb of Ornithomimus altus from the Judith River Formation, (Russell 1972: 383), Red Deer River, Alberta. This was referred to the genus Struthiomimus by Osborn (1917: 744). From the illustration the astragalus is 129 mm wide and 224 mm high, an H:B index of 174. The dorsal process lies along the fibular side of the tibia, the lower border of each concave, with the tibia expanding rapidly distally. Lambe does not describe this bone in detail, but later (1904: 20) notes that the medial and lateral ends are less oblique than in Dryptosaurus incrassatus (referred specimen). He also states that the astragalus does not coalesce with the tibia. The calcaneum has an upper sulcus for the fibula and ‘one behind to aid the astragalus in supporting the tibia’.
Ormthomimus velox was figured by Marsh (1890, pl. 1, fig. 1), as from the Ceratops beds of Colorado, the Denver Formation, Maastrichtian (Russell 1972: 379). The astragalus is 52 mm broad and 80 mm high, an H:B index of
THE TARSUS OF THEROPOD DINOSAURS 213
154. The calcaneum is very thin and Marsh did not state whether it was fused or not. If fused, the astragalocalcaneum had a breadth of 65 mm. This is an
Fig. 12. Ornithomimus velox, type.
Anterior view of distal end of left tibia and tarsus, from Marsh 1890, pl. 1, OmgTs DG
extremely specialized tarsus in that the calcaneum is so reduced that there was little or no support for the fibula, and this, in turn, must have been greatly reduced. These features were noted by Marsh, who likened it to the ostrich tarsus. The dorsal process is extremely high, its lateral edge nearly vertical and its medial edge dipping 67° near the base, the total angle of dip being 60°. Another specialization of this tarsus is seen in ventral view where the astragalus does not completely cover the lower end of the tibia, but the tibia forms a narrow rim of bearing surface along the posterior edge and wraps around the medial edge of the astragalus. This narrow tibial rim also extends laterally behind the calcaneum.
THE TYRANNOSAUROID TARSUS
The astragalus of Tyrannosaurus rex, from the Hell Creek beds of Montana, was illustrated by Osborn (1917, pl. 25) as about 425 mm broad and 530 mm
214 ANNALS OF THE SOUTH AFRICAN MUSEUM
high, an H:B index of 125. However, in the archives of the American Museum of Natural History we found a detailed unpublished illustration by Erwin S. Christman, which was probably done under the supervision of Osborn or Barnum Brown. This differs drastically from Osborn’s illustration and it evi- dently represents an entirely different kind of astragalus. It therefore seems likely to us that Osborn’s illustration was based on Allosaurus. We made a thorough search for the specimen, and although we failed to locate it we feel that Christman’s drawing is authentic. It bears the same number (AMNH 5027) as the beautiful skull from the Hell Creek Formation of Montana collected by Brown in 1907. The locality is the east side of Big Dry, 13 km from Willis Ranch, 40 km south of Lisman, Mont., top of basal ss., 24 m above the Fox
O AO)
Fig. 13. Tyrannosaurus rex. Left astragalus in a, lateral view; b, anterior view. Redrawn from unpublished illustration by Erwin S. Christman. AMNH 5027. x 4.
Hills. From Christman’s illustration the anterior view shows a breadth of 372 mm and a height of 320 mm, yielding an H:B index of 86. The dorsal process is broad and is distinctive in that the medial edge slants dorsomedially and is concave. The lateral edge slopes ventrolaterally and ends above the pit for the calcaneum. The upper horizontal groove is but weakly developed medially and is entirely absent laterally where the dorsal process rounds gently into the lateral condyle. The medial condyle is 160 mm high, protrudes dorso- medially, and its upper edge slopes ventrolaterally to form a shelf above a groove (? the lower horizontal groove) which ends by rounding back and down into the lateral condyle 60 mm internal to the calcaneal border. At about the centre of the astragalus, just below the shelf is a large (68 x 28) horizontal foramen that opens into the tibial sulcus. In lateral view the edge of the dorsal process is rounded and does not seem to have formed the usual flat surface for the fibula. The lateral condyle is low, rounding gently into the dorsal process without the separation of an upper horizontal groove. An open vertical groove
a —
THE TARSUS OF THEROPOD DINOSAURS 2 LU
could be for either the fibula or the calcaneum, probably the latter. The posterior process is very thin, with its edge high (172 mm) above the condyle. The trough for the tibia is very narrow laterally where it descends to within 20 mm of the base. It broadens medially, as with most theropod astragali.
SUMMARY
In reviewing this study of the theropod astragali, the ceratosauroids develop a low, thick dorsal process which is sunk into the tibia. ‘The calcaneum is large and may be free or fused. This tarsus undergoes little change from late Triassic to late Cretaceous and does not seem to have given rise to any other line. The allosauroids have a moderately high dorsal process which is confined to the lateral half of the tibial surface. The medial condyle is large and rounded. The calcaneum is large and free, and in Allosaurus is deeply interdigitated with the astragalus. The time range is Middle Jurassic through late Jurassic. The albertosauroid astragalus has a high, broad dorsal process with a deep upper horizontal groove at its anterior base. This groove is interrupted laterally by a ventrolateral buttress to the lateral condyle. There is no lower horizontal groove and the calcaneum becomes very small and is loosely attached to the astragalus. This kind of tarsus is found throughout the late Cretaceous. The ornithomimoid astragalus develops an extremely high and very thin dorsal process that overlies the face of the tibia without being deeply inset. Distally it is equal to the tibia in breadth. The upper horizontal groove at the anterior base of the dorsal process is strongly arched ventrally and there is no lower horizontal groove. The calcaneum becomes extremely reduced and may be lost. The time range is late Jurassic to late Cretaceous. The tyrannosauroid tarsus only appears in the latest Cretaceous in a single known form. The astra- galus bears a high, wide dorsal process as in other late Cretaceous theropods, but it differs in numerous other characters and appears to indicate a separate origin from known Cretaceous forms.
It is obvious that there has been a general tendency among theropods to increase the height of the dorsal process of the astragalus. The ceratosauroids seem to have minimized this tendency, the ornithomimoids to have maximized it. Whether or not this tendency can be used as an indicator of geologic age, as was attempted by Welles (1954: 597), can only be determined by additional specimens and further study.
This survey suggests some tentative relationships based on a very few specimens. Many undescribed astragali are known, and when they are studied we will have a much more solid basis for our conclusions.
ACKNOWLEDGEMENTS
We should like to thank Professor James H. Madsen, Jr., Department of Geology and Geophysics, University of Utah; Professor John H. Ostrom, Yale Peabody Museum; Dr Rudolph Daber, Museum fiir Palaontologie,
216 ANNALS OF THE SOUTH AFRICAN MUSEUM
Museum fiir Naturkunde; Dr E. H. Colbert, Museum of Northern Arizona; and Dr P. Taquet, Institut de Paléontologie, Muséum National d’Histoire Naturelle for permission to study and photograph specimens. Dr Ralph Molnar, Los Angeles County Museum, sent us the Hell Creek astragalus. We are especially indebted to Mr H. P. Powell of the University Museum, Oxford, for his drawings of Eustreptospondylus. Dr Eugene Gaffney of the American Museum of Natural History allowed us to use the unpublished illustration of Tyrannosaurus. The illustrations are by Mr David Cook, staff artist of the University of California Museum of Paleontology, except Figures 2 and 13 which are by Welles.
POSTSCRIPT
Upon the receipt of page proof we note the subsequent papers that include descriptions of theropod tarsi which have not been included in our survey:
Bier, A., Demay, L. & THomeEL, G. 1972. Compsognathus corallestris, nouvelle espece de
dinosaurien théropode du Portlandien de Canjuers (Sud-Est de la France). Annis Mus. Hist. nat. Nice t (1): 9-40.
Osmotska, H., Rontewicz, E. & BARsBoLp, R. 1972. A new dinosaur, Gallimimus bullatus n. gen., n. sp. (Ornithomimidae) from the Upper Cretaceous of Mongolia. In: KIELAN- jAworowsKkA, Z., ed. Results of the Polish-Mongolian Palaeontological Expeditions. IV. Palaeont. pol. 27: 103-143.
The tarsi of Compsognathus corallestris and Gallimimus bullatus display numerous characters which undoubtedly place them in our ornithomimoid group.
It should also be noted that Cruickshank has recently described the tarsus of the Lystrosaurus zone thecodont Proterosuchus which is ‘almost identical to that of an eosuchian’. The calcaneum has a well-developed tuber.
(CruicksHANK, A. R. I. 1972. The proterosuchian thecodonts. In: JoysEy, k.A. & KEMP,
T. S., eds. Studies in vertebrate evolution: 897-119. Edinburgh: Oliver & Boyd.) Also Bonaparte has described the astragalus of an indeterminate saurischian from the upper Los Colorados Formation (Norian), of La Rioja, Argentina. This is a very primitive astragalus with only an incipient dorsal process and its relationship to theropod tarsi is uncertain.
(BONAPARTE, J. 1971. Los tetrapodos del sector superior de la Formacion Lés Colorados,
La Rioja, Argentina. (Tridsico Superior.) I Parte. Op. lilloana 22: 1-185.
REFERENCES
Broom, R. 1921. On the structure of the reptilian tarsus. Proc. zool. Soc. Lond. 1921: 143-155.
Corer, E. D. 1866. [Remarks on Laelaps.| Proc. Acad. nat. Sci. Philad. 1866: 316-317.
Core, E. D. 1870. Synopsis of the extinct Batrachia, Reptilia and Aves of North America. Trans. Am. phil. Soc. (n.s.) 14: i-viii, 252. [This is a corrected reprint of Cope 1869, and is a second preprint of the whole volume, which was issued in 1871 without further changes]
Cuvier, G. 1836. Recherches sur les ossemens fossiles, ou Von rétablit les caractéres de plusieurs animaux dont les révolutions du globe ont détruit les espéces. 4th ed. g, 10. Paris: D’Ocagne.
DeEsLonccuamps, J. 1838. Mémoire sur le Poekilopleuron bucklandi, grand saurien fossile, inter- médiare entre les crocodiles et les lézards. Mém. Soc. linn. Normandie 6: 37-146.
Ewer, R. T. 1965. The anatomy of the thecodont reptile Euparkeria capensis Broom. Phil. Trans. R. Soc. (B) 248: 379-435.
Gatton, P. M. 1971. The prosauropod dinosaur Ammosaurus, the crocodile Protoschus, and ae bearing on the age of the Navajo Sandstone of northeastern Arizona. 7. Paleont. 45: 791—795-
a
THE TARSUS OF THEROPOD DINOSAURS 207,
GEGENBAUR, C. 1864. Untersuchungen zur vergleichenden Anatomie der Wirbelthiere. Erstes Heft. Carpus und Tarsus. Leipzig: Engelmann.
GitmoreE, C. W. 1920. Osteology of the carniverous Dinosauria in the United States National Museum, with special reference to the genera Antrodemus (Allosaurus) and Ceratosaurus. Bull. U.S. natn. Mus. 110: 1-154.
Gitmor_E, C. W. 1933. On the dinosaurian fauna of the Iren Dabasu formation. Bull. Am. Mus. nat. Hist. 67: 23-78.
Grecory, W. K. & Camp, C. L. 1918. Studies in comparative myology and osteology. III. Bull. Am. Mus. nat. Hist. 38: 447-563.
HueEng, F. von. 1926. The carnivorous Saurischia in the Jura and Cretaceous formations princi- pally in Europe. Revta Mus. La Plata 29: 35-167.
HueEne, F. von. 1933. Sub-Order: Carnosauria. In: HUENE, F. VON & MATLEY, c.A. The Creta- ceous Saurischia and Ornithischia of the Central Provinces of India. Mem. geol. Surv. India Palaeont. indica (n.s.) 24: 41-59.
JANENscH, W. 1925. Wissenschaftliche Ergebnisse der Tendaguru-Expedition 1909-1912. Die Coelurosaurier und Theropoden der Tendaguruschichten Deutsch-Ostafrikas. Palaeonto- graphica Suppl. 7 (1. Reihe, Teil 1, Lief. 2): 1-99.
Kress, B. 1963. Bau und Function des Tarsus eines Pseudosuchiers aus der Trias des Monte San Giorgio (Kanton Tessin, Schweiz). Palaeont. <. 37: 88-95.
LamBe, L. M. 1902. New genera and species from the Belly River series (Mid-Cretaceous). In: OSBORN, H. F. & LAMBE, L. M. On Vertebrata of the Mid-Cretaceous of the North West Territory. Contr. Can. Palaeont. 3 (2): 23-81.
Lampe, L. M. 1904. On Dryptosaurus incrassatus (Cope) from the Edmonton series of the North- West Territory. Contr. Can. Palaeont. 3 (3): 1-27.
LamBe, L. M. 1917. The Cretaceous theropodous dinosaur Gorgosaurus. Mem. geol. Surv. Brch Can. too: 1-84.
LypDEKKER, R. 1891. On certain ornithosaurian and dinosaurian remains. Q. Jl geol. Soc. Lond. 47° 41-44.
Marsu, O. C. 1877. Notice of a new and gigantic dinosaur. Am. 7. Sci. (3) 14: 87-88.
Marsu, O. C. 1890. Description of a new dinosaurian reptile. Am. 7. Sci. (3) 39: 81-86.
Marsu, O. C. 1896. The dinosaurs of North America. Rep. U.S. geol. Surv. 16 (1): 133-414.
MatTTHew W. D. & Brown, B. 1923. Preliminary notices of skeletons and skulls of Deinodontidae from the Cretaceous of Alberta. Am. Mus. Novit. 89: 1-10.
Ossporn, H. F. 1917. Skeletal adaptations of Ornitholestes, Struthiomimus, Tyrannosaurus. Bull. Am. Mus. nat. Hist. 35: 733-771.
Ostrom, J. H. 1969. Osteology of Deinonychus antirrhopus, an unusual theropod from the Lower Cretaceous of Montana. Bull. Peabody Mus. nat. Hist. 30: 1-165.
Ostrom, J. H. 1970. Stratigraphy and paleontology of the Cloverly Formation (Lower Creta- ceous) of the Bighorn Basin area, Wyoming and Montana. Bull. Peabody Mus. nat. Hist. 35: I-vll, 1-234.
Parks, W. A. 1928. Albertosaurus arctunguis, a new species of theropodous dinosaur from the Edmonton Formation of Alberta. Univ. Toronto Stud. geol. Ser. 25: 1-42.
Peasopy, F. E. 1951. The origin of the astragalus of reptiles. Evolution 5: 339-344.
Puituirs, J. 1871. Geology of Oxford and the valley of the Thames. Oxford: Clarendon Press.
PIVETEAU, J. 1923. L’arriére-crane d’un dinosaurien carnivore de |’Oxfordien de Dives. Annls Paléont. 12: 1-11.
Raatu, M. A. 1969. A new coelurosaurian dinosaur from the Forest Sandstone of Rhodesia. Aynoldia (Rhodesia) 4 (28): 1-25.
Ras, C. 1910. Bausteine zu einer Theorie der Extremitdten der Wirbeltiere. 1 Teil. Leipzig: Engelmann.
Romer, A. S. 1971. The Chafiares (Argentina) Triassic reptile fauna. X. Two new but incom- pletely known long-limbed pseudosuchians. Breviora 378: 1-10.
Romer, A. S. 1972. The Chafiares (Argentina) Triassic reptile fauna. XV. Further remains of the thecodonts Lagerpeton and Lagosuchus. Breviora 394: 1-7.
RussELt, D. A. 1969. A new specimen of Stenonychosaurus from the Oldman Formation (Creta- ceous) of Alberta. Can. J. Earth Sci. 6: 595-612.
RussELL, D. A. 1970. Tyrannosaurs from the Late Cretaceous of western Canada. Publs Palacont. nain. Mus. nat. Sci. (Can.) 1: v—viii, 1-34.
SCHAEFFER, B. 1941. The morphological and functional evolution of the tarsus in amphibians and reptiles. Bull. Am. Mus. nat. Hist. 78: 395-472.
218 ANNALS OF THE SOUTH AFRICAN MUSEUM
Wa ker, A. D. 1964. Triassic reptiles from the Elgin area: Ornithosuchus and the origin of carnosaurs. Phil. Trans. R. Soc. (B) 248: 53-135.
WELLES, S. P. 1954. New Jurassic dinosaur from the Kayenta Formation of Arizona. Bull. geol. Soc. Am. 65: 591-598.
Youne, C. C. 1951. The Lufeng,saurischian fauna in China. Palaeont. sin. (n.s.): 13: 1-96.
Youne, C. C. 1964. The pseudosuchians in China. Palaeont. sin. (C) 19: 105-205.
ALFRED SHERWOOD ROMER
Professor Alfred S. Romer was fond of comparing vertebrate palaeontolo- gists with the conies (Hyracoidea) of Scripture: ‘a feeble folk, which dwelleth among the rocks.’ Through the sudden death of Professor Romer in November 1973; this ‘feeble folk’ has lost an able and courageous leader, and one of its best-known and loved members.
Born in 1894, Alfred Sherwood Romer completed a doctoral thesis at Columbia University in 1921 before joining the faculty of the University of Chicago in 1923 as Assistant Professor, becoming Professor of Vertebrate Paleontology in 1931. In 1934 he moved to Harvard as Professor of Zoology and Curator of Vertebrate Paleontology. In 1947 he received the Alexander Agassiz Professorship in Zoology, and was director of the Museum of Com- parative Zoology from 1946 to 1961. He retired from the Harvard staff in 1965, becoming Professor Emeritus, but was actively involved in research up to the time of his death.
Among the medals and awards received by Professor Romer during his long career were the Thompson Medal of the National Academy of Sciences for 1956, the Academy’s Elliot Medal for 1960, the Hayden Memorial Geological Award of the Philadelphia Academy of Natural Sciences for 1962, the Penrose Medal of the Geological Society of America in 1962, the Medal of the Linnean Society of London 1972, and the Wollaston Medal of the Geological Society of London in 1973. He held honorary doctorates from the universities of Harvard, Amherst, Dartmouth, Buffalo and Lehigh.
219
Ann. S. Afr. Mus. 64, 1974: 219-220.
220 ANNALS OF THE SOUTH AFRICAN MUSEUM
Besides regular collecting in the Permian red-beds of ‘Texas, and (most recently) pioneering field work in two rich Triassic deposits in western Argen- tina, Romer became the first American to make an important collection of South African Karroo reptiles when in 1929, accompanied by the veteran collector Paul C. Miller of Chicago’s Walker Museum, he spent seven months collecting in most of the classic Karroo localities.
The wideness of the field that Professor Romer covered during his career makes a full appreciation of his contribution to science difficult. Always the biologist, he applied biological principles to the extinct organisms which he studied. To him a fossil vertebrate was a once-living animal, to be recreated in its body form, its functional activities, and its way of life within its contemporary environment, and to him this may have represented the ultimate achievement of palaeontology. His standard textbooks The vertebrate body (4th ed. 1970), Vertebrate paleontology (3rd ed. 1966) and Osteology of the reptiles (1956) need no introduction, nor do his more general works Man and the vertebrates (1933) (retitled The vertebrate story in 1959) and The procession of life (1968). His mono- graphs include ‘Review of the Pelycosauria’ (1940), “Review of the Labyrin- thodontia’ (1947) and the series (up to 1973) “The Chafiares (Argentina) Triassic reptile fauna’. These works, and his numerous other publications, all bear a very typical Romerian stamp: a combination of clear, logical reasoning, a total morphological insight, and a presentation couched in the clear, uncluttered literary style that set Romer apart from nearly all his contemporaries.
Romer the scientist, Romer the writer, and Romer the man will be sorely missed in the forums of comparative anatomy and vertebrate palaeontology, where he was for so long a dominant figure.
AQUATIC ADAPTATION IN REPTILES—PRIMARY OR SECONDARY?
By + ALFRED SHERWOOD ROMER
Museum of Comparative Zoology, Harvard University, Cambridge, Mass.
A fair number of modern reptiles, such as the crocodilians, most chelonians and a number of ophidians, are amphibious to aquatic in habits and, in the Mesozoic, aquatic reptiles were abundant—ichthyosaurs, plesiosaurs and mosasaurs being the most prominent forms. Are such forms secondarily aquatic, descended from purely terrestrial ancestors, or may certain of them have been derived from primitive reptiles which had never completely abandoned the aquatic habitat of their lower vertebrate ancestors? Since the most positive definitive characteristic of the Reptilia is their amniote mode of development that freed them from the water, and since, farther back, limbs capable of terrestrial locomotion were developed in even the oldest known amphibians of the late Devonian, it seems reasonable, at first sight, to believe that early tetra- pods rapidly emerged from the water onto land, and that water-dwelling was a secondary condition.
This is not necessarily the case. Despite the development of limbs in early amphibians, few members of this class ever achieved a truly terrestrial mode of life. And although the evidence is none too clear, it seems far from certain that all aquatic reptiles have reverted from land to water.
THE AMPHIBIAN STAGE
Let us broadly review the early history of tetrapods. The oldest known amphibians—the late Devonian ichthyostegids (Jarvik 1955) —were crossopte- rygian descendants which, despite the retention of certain primitive features, had developed their paired fins into short but sturdy tetrapod limbs, enabling them to leave the water and travel overland. Offhand, one would assume that with the development of paired limbs the ancestral amphibians immediately took up a terrestrial existence. But further reflection brings one to conclude that nothing could be further from the truth. For what purpose would these early amphibians come forth on to the land? It has been suggested that they were taking refuge from enemies in the water; but that can hardly be the case, for the amphibians are an offshoot of the rhipidistian crossopterygians who were the most predaceous of fishes in Devonian fresh waters. More reasonable, at first sight, is the belief that these fish descendants were setting forth to explore the terrestrial environment where new opportunities might await them. Here, however, the fatal objection is that there was precious little on land for them to eat. The rhipidistians were predaceous, eaters of smaller fishes and, pre- sumably, freshwater invertebrates. It seems certain that all early amphibians
22
Ann. S. Afr. Mus. 64, 1974: 221-230.
222 ANNALS OF THE SOUTH AFRICAN MUSEUM
were likewise eaters of animal food; universally, the teeth, like those of crosso- pterygians, were sharp-pointed conical structures quite incapable of dealing with vegetable materials. It is possible that there were already present on land in the Devonian some soft-bodied invertebrates which have escaped detection as fossils, but the only land animals known in the Devonian upon which a hope- fully terrestrial amphibian could feed were tiny Collembola (known from the Rhynie chert of Scotland) and a few myriapods. Most notably, not a single true insect is known from the Devonian.
Amphibians attempting to live on land in the Devonian might admire the scenery; but they would surely starve to death. And in the early Carboni- ferous the terrestrial food supply was little better. One might expect primitive insects to appear, but none have as yet been found. It appears possible that in the Carboniferous scorpions were in process of migrating from water to land. But even the possible addition of scorpions to a diet of centipedes and micro- scopic soil-dwelling springtails does not appear to make life on land too attrac- tive for eaters of animal food. It was not until the later Carboniferous that land life really became possible with the appearance of a wealth of insects of primitive types. It is only toward the end of the Carboniferous that we find among early reptiles (and perhaps in one family of amphibians) forms which at long last were adapted to feed upon the readily available vegetation.
Why, then, the development of land limbs in amphibians? The answer, apparently, is a paradoxical one—that these limbs, which enabled them to walk on land, were an adaptation for water-dwelling under the conditions in which early amphibian evolution took place. Barrell (1916) long ago pointed out that the late Paleozoic, when tetrapod evolution was taking place, is charac- terized by an abundance of ‘redbeds’ formations. Beds of this sort are frequently indicative of their having been laid down under climatic conditions of seasonal drought such as are found in limited regions of the tropics today. For part of the year, rainfall is abundant, and life for water-dwellers is easy. When a drought comes, streams dry up, waters become stagnant, and oxygen is at a premium. However, a good fraction of Paleozoic freshwater fishes appear to have had lungs (which very few fish have today), and atmospheric oxygen was available to them if they came to the surface. But if the stream or pond dried up completely in the drought, what then? An ordinary fish would be, literally, stuck in the mud, and if the water did not return quickly would soon perish. But if rudimentary limbs of tetrapod type were present, the animal might be able to leave his dried-up pool, crawl up or down the streambed or even travel some distance overland, find a pool with water still in it, plunge in, and resume his normal aquatic life. Under seasonal drought conditions, tetrapod limbs, even if little developed, would be of immediate selective advantage to a freshwater dweller.
Watson years ago (1926) suggested this use of legs in the Carboniferous
* Originally described as insects were two specimens from the Devonian of the USSR;; these, however, were subsequently recognized as crustacean.
AQUATIC ADAPTATION IN REPTILES— PRIMARY OR SECONDARY ? 223
and I have since pointed out (Romer 1957, 1958, etc.) that this situation may have been true of early amphibians generally. Had early limb development been for the ‘purpose’ of exploring land life, one would expect that the post- Devonian history of early amphibians would have exhibited the retention and even improvement of the limb development seen in the Devonian ichthyoste- gids. But the opposite is almost always the case. In the Carboniferous, with the frequent development of coal-swamp conditions, the importance of limbs as a means of survival under drought conditions would decrease; and if, in any standard text (such as Romer 1966: 87-98), we follow through the history of amphibians in the post-Devonian periods, we see, in accordance, a strong trend for limb reduction, rather than limb improvement, in most amphibian groups.
Pre-Jurassic amphibians are currently customarily divided into two major groups: the Lepospondyli (very probably an artificial assemblage) and the Labyrinthodontia. The Lepospondyli include a considerable variety of small Carboniferous and early Permian forms, apparently mainly coal-swamp dwellers, in nearly all of which the paired limbs are greatly reduced and may be lost completely. Exceptional are a number of late Carboniferous and early Permian members of the Microsauria in which rather well-developed limbs are present.
The much more important ancient group is that of the Labyrinthodontia. Here, as in the lepospondyls, there is in general a strong trend toward limb reduction, although not as marked or as rapid as was the case in most lepospon- dyls. We are not sure that the primitive ichthyostegids were directly ancestral to any later labyrinthodonts; however, one restricted series of Carboniferous forms, the Colosteidae (which I hope to discuss in the near future) may be ichthyostegid descendants; however, they have limbs more feebly developed than in the Devonian forms.
Ichthyostegids apart, however, nearly all labyrinthodonts may be arrayed in two main series, the Temnospondyli, in which a key character is the domi- nance in the vertebral column of the intercentrum, and the Anthracosauria, in which the pleurocentrum develops (to become eventually the true centrum of reptiles). The temnospondyls are by far the more abundant of the two, and are the only amphibians (apart from a ‘pre-frog’) present in the later Permian and ‘Triassic. In some Carboniferous and early Permian forms the legs are reasonably well developed, as in the familiar Eryops of the early Permian which could quite surely waddle comfortably about its native swamps, and in one group, the dissorophoids, there was apparently, in the Permian, a considerable trend toward true terrestrial existence.
But while limbs were maintained in proper fashion in certain temnospon- dyls, the trend toward reduction was strong within the group. Even among
2 See, for example, the dissorophids Cacops and Broiliellus (Williston 1914, etc.), in which there were not only well-developed limbs but also dorsal armor, suggesting a terrestrial life where there was danger from reptile predators. These forms lived at a time in the Permian when terrestrial food was becoming more abundant.
224 ANNALS OF THE SOUTH AFRICAN MUSEUM
primitive temnospondyls, such as the trimerorhachids, the limbs were much reduced, and from the later Permian on through the Triassic there was con- tinued reduction, so that in typical Triassic members of the group, such as Capitosaurus and Metoposaurus, we find amphibians with a broad flat head and body with tiny limbs quite incompetent to support them out of the water. In the later Paleozoic and Triassic the dangers of seasonal drought were much less than in earlier days, and in correlation with this, there was a strong trend to reduction of the limbs. But if drought did occur, these late temnospondyls were helpless. I have described a locality in the Triassic of New Mexico (Romer 1939) in which a mass death of a myriad of metoposaurs (“Buettneria’) occurred, obviously because of the drying up of the swamps and pools in which they lived.
The other main division of the labyrinthodonts is that of the Anthra- cosauria,? much less abundant but of interest as the group from which the Reptilia arose. Here, as in the temnospondyls, we see considerable variation in limb development. Most prominent of anthracosaurs in the Carboniferous (and persisting into the early Permian) were the Embolomeri, good-sized amphibians, essentially water-dwellers, quite surely with paired limbs of rela- tively small size which, nevertheless, could enable them to walk on land to some degree. Quite rare, known only from a few genera, were Carboniferous forms in which sturdy limbs persisted. One is Gephyrostegus, known only from two specimens from the European Upper Carboniferous; long confused with other genera, its nature and structure has been recently clarified by Carroll (19706: 268-286). An apparently similar form, Mauchchunkia, has now been discovered in the Lower Carboniferous of Greer, West Virginia. So far there has been published (Hotton 1970) only an account of a fragmentary skull and incomplete skeletal materials. Now available to me, however, are two excellent postcranial skeletons, which I hope to describe shortly. The structure is quite similar to that of Gephyrostegus; we appear to have here a fortunate persistence of anthracosaurs with well-developed limbs through the ‘hard times’ (for potential terrestrial life) of the early Carboniferous. From forms of this sort the earliest reptiles presumably arose in the latter half of the Carboniferous. But also there may have come from this stock the Seymouriamorpha of the Permian, best known from Seymouria of the early Permian of Texas, but also represented by a number of European genera. Derived from the same general stock as the ancestral reptiles, the seymouriamorphs show many reptile-like characteristics, but although taking advantage of the relatively favourable conditions of Permian
* O. Kuhn, in his Handbuch der Palaeoherpetologie (Panchen 1970), has, regrettably, tended to bring about confusion in amphibian terminology by substituting for Anthracosauria the Efremov term Batrachosauria (Efremov 1946) and reducing Anthracosauria to cover merely the embolo- meres. But Efremov intended Batrachosauria (as the name implies) to apply only to reptile-like amphibians, notably the Seymouriamorpha, and not to such unreptile-like forms as the embolo- meres. And for the latter Cope nearly a century ago coined the term Embolomeri, of which Anthracosauria, as used by Kuhn, becomes merely a junior synonym. I trust that Kuhn’s nomen- clatorial aberration here will be generally neglected and presently forgotten.
AQUATIC ADAPTATION IN REPTILES— PRIMARY OR SECONDARY ? 225
times to become essentially terrestrial in habits, they remained, it appears, in an amphibious stage as regards reproduction.
Above, we have reviewed the history of the older amphibious groups as regards their locomotor potentialities and general mode of life. Tetrapod limbs were early developed and in a few cases, such as Permian dissorophoids and seymouriamorphs, were not merely retained but used for terrestrial locomotion. But in general there appears to have been a strong trend for limb reduction. The story agrees well with the assumption that the Amphibia were, and gene- rally remained, aquatic or at the most amphibious in habits; that limb develop- ment at the beginning was not related to any trend toward terrestrial life but as a useful aid to successful aquatic life under seasonal drought conditions; and that when such climatic conditions were lessened, limb reduction very generally took place.
EARLY REPTILES
What was the mode of life of the earliest reptiles, descended from water- dwelling amphibian ancestors? How rapidly did they take on the terrestrial mode of life, facilitated by the ‘invention’ of the amniote type of development? In recent years it has become increasingly clear that the basic stock of the Reptilia lies in the cotylosaur group termed the Captorhinomorpha. Members of this group have long been known from the early Permian, but in recent years it has become clear, mainly through an excellent series of papers by Carroll (1969, 19708, etc.), that they were already flourishing in the late Carboniferous (the Pennsylvanian of American usage). We are here at a stage when, in con- trast to earlier times, true land life had at long last become possible; with a known wealth of primitive insects present, there was now available a basic source of food upon which a terrestrial fauna might feed. Central in this picture are members of the family Romeriidae. These were small reptiles whose appearance in life might have been rather like that of modern lizards, but which were very different from any modern reptiles in that their structure was of a very generalized and primitive type, of a sort from which, one may believe, later reptiles of almost any group might have been derived. Still more primitive are members of the family Limnoscelidae. The type form, Limnoscelis palustris (Williston 1912; Romer 1946), is of early Permian age, but representatives are now known from the Carboniferous, and one form, Romeriscus (Baird & Carroll 1967) from the Westphalian A of Cape Breton, is the oldest known reptile. It seems certain (as the scientific name suggests) that the limnoscelids were persistently primitive in retaining amphibious habits. But it is generally agreed that the limnoscelids were already technically reptiles, strongly suggesting that the amniote type of development had already been invented before the reptiles had abandoned the amphibious mode of life of the older tetrapods.
Carroll believes, however, that the limnoscelids were too archaic in struc- ture to have been a group from which the radiation of later reptilian groups
226 ANNALS OF THE SOUTH AFRICAN MUSEUM
could have taken place, and that the basic stock from which higher reptiles evolved was that of the more advanced family Romeriidae. What was the mode of life pursued by the romeriids—was it, at this level, completely terrestrial, or were these forms still somewhat aquatic in habits? Carroll gives a good argument for the terrestrial side of the question—namely, that the skeleton in romeriids is well ossified, whereas in amphibious reptiles, generally, ossification is less complete. Had we good continental beds in the Upper Carboniferous, with a true land fauna, the case for the purely terrestrial nature of the romeriids would be greatly strengthened. Unfortunately, our knowledge of Upper Carboniferous vertebrate faunas is almost entirely confined to those of coal swamps. In these, as at Linton and Mazon Creek, romeriids are relatively rare, but are present. Only in the fossil tree stumps of the Nova Scotia Joggins does Carroll (1969: 36) believe that he has a truly ‘terrestrial’ fauna in which romeriids are abundant. |
I have a very high regard for Carroll’s work on Carboniferous tetrapods, but I think that here he has (like the contained fauna) fallen into a trap. The Joggins region, well known since the days of Dawson (and recently redescribed briefly by Carroll and others in 1972) consists of a series of Carboniferous exposures along the bluffs at the head of the Bay of Fundy in western Nova Scotia. Here there are present thousands of feet of Upper Carboniferous deposits which include some dozens of coal seams separated by intervening shales. We are dealing with a long-persisting coal basin in which, time after time, a coal swamp developed. Between successive swamp deposits the area was invaded by mud flows which hardened into shales. The lycopod trees of the coal swamps tended to die, leaving hollow stumps that (as well shown by Carroll 19702, fig. 3) acted as traps for unwary amphibians and reptiles which wandered about the mud flats of the region in the periods between times of coal seam formation. The contents of the traps consist mainly of romeriids, microsaurs with well- developed limbs, and a rhachitome (Dendrerpeton) with rather good legs; absent are nectrideans, aistopods and other coal-swamp dwellers in which limbs were lost or poorly developed. Is this a ‘terrestrial’ fauna? Hardly. We are still present in a coal-swamp basin, and there is no guarantee that we are not within a short distance of the water. It simply means that we have a selective screening of a coal-swamp fauna. As Rayner (1971) notes: ‘The pure water-dwellers, such as the aistopod and nectridean lepospondyls, are absent. Those found are more ‘terrestrial’ than in other Carboniferous faunas; they walked, ambled or waddled across the muddy or sandy flats among the rotting tree trunks. The sediments they traversed, however, and which finally buried animals and trees, were purely aqueous—the normal type laid down in the open stretches of the coal forests and swamps.’
It is highly probable that some Carboniferous romeriids were completely terrestrial; but those which we do know were persistent coal-swamp dwellers. It would appear, as far as present knowledge goes, that despite the advantage given them by their amniote mode of life, and despite the fact that the abundant
AQUATIC ADAPTATION IN REPTILES— PRIMARY OR SECONDARY ? 227,
insect fauna of the later Carboniferous afforded a terrestrial food supply for eaters of animal food, early reptiles were slow to abandon the amphibious mode of life of their ancestors.
By the close of the Carboniferous and the beginning of the Permian, many reptiles were already fully ‘ashore’ and even (as shown for example by the presence of Edaphosaurus in the late Carboniferous) some were shifting from animal to vegetable food supplies. As noted later, in the case of many aquatic and amphibious reptiles of the Mesozoic and Cenozoic, evidence is lacking as to whether or not the forms leading to them had become fully terrestrial and only later returned to the water. Is there, however, any evidence in early reptile history suggesting a primitive retention of an amphibious to aquatic habit? Apart from the primitive limnoscelids, there are two positive examples of water-dwelling reptiles which are of such an early age that it seems unlikely that their pedigree included forms which had become completely terrestrial in habits.
One example is Mesosaurus of South Africa and southern Brazil (Rayner 1971: 472-476). The phylogenetic position of this small amphibious reptile is uncertain (it has been variously referred to the diapsids and synapsids, and some recent evidence suggests that it may be anapsid—an early cotylosaur offshoot). Highly specialized for feeding on minute crustaceans, Mesosaurus occurs at a horizon close to the Carboniferous-Permian boundary in the ‘white band’ of the South African Dwyka and comparable beds in Brazil. Occurring at this early stage in reptilian history, it is difficult to believe that its ancestors could have become completely terrestrial and then rapidly shifted back to an amphibious mode of life, with the speedy acquisition of its unique type of cranial and dental structures.
More striking is the story of the ophiacodont pelycosaurs. The subclass Synapsida, of which the pelycosaurs are the primitive members, is in its more advanced stages an almost purely terrestrial group. But while among the pely- cosaurs the sphenacodontoid and edaphosauroid suborders are terrestrial in nature, Ophiacodon and its relatives, most primitive of pelycosaurs, are quite surely still amphibious in habits (Romer & Price 1940). The two other pely- cosaur groups flourished greatly in the early Permian, but only appear in the fossil record at the very close of the Carboniferous. Quite different is the story of the ophiacodonts. One ophiacodont, Clepsydrops, has long been known from the Upper Carboniferous of Danville, Illinois, and it has been demonstrated that ophiacodonts were present nearly as far down in the sequence of Car- boniferous coal-swamp deposits (Romer 1961; Carroll 1964; Reisz 1972) as reptiles of any sort.
The ophiacodonts differ little from primitive captorhinomorphs, apart from the development of the synapsid temporal opening, and it seems com- pletely reasonable to believe that this basal group of synapsids branched off from the basal reptile stock while this was still persistently amphibious in habitat.
228 ANNALS OF THE SOUTH AFRICAN MUSEUM
LATER AMPHIBIOUS AND AQUATIC REPTILES
Notable in the known paleontological history of reptiles is our almost complete lack of knowledge of groups other than cotylosaurs and synapsids until late Permian and Triassic days, and the early stages in the evolutionary story of most reptile groups is purely guesswork. For example, the Chelonia is an order which includes among the tortoises a few purely terrestrial forms, and on the other hand gave rise to a number of high seas types, but in general they are typically amphibious in habits.
The first chelonians appear in the late Triassic, and while a bit more primi- tive in some regards, already show the basic structural pattern of the order and, it would seem, lived the typical amphibious life characteristic of most of their descendants. It is quite possible that in the long stretch of time between the Carboniferous and the late Triassic the chelonians had become purely terrestrial and then reverted to an aquatic mode of life. But it is equally possible that we have in the typical members of this order a retention of a truly primitive mode of amphibious life.
Unknown, too, is the pedigree of the great group of aquatic reptiles currently grouped in the subclass Euryapsida—the placodonts, nothosaurs and plesiosaurs. The mollusc-eating placodonts and the nothosaurs appear in the Middle Triassic oceans; in the Jurassic these forms disappear, to be replaced by the purely aquatic plesiosaurs, which had a spectacularly successful career in the middle and later parts of the Mesozoic. Where did these euryapsids come from? We have no clues as to their history. Araeoscelis of the early Permian has an euryapsid type of temporal opening, but Vaughn (1955) has pointed out that this genus shows no resemblance to later euryapsids in other regards.
Similarly blank is the history of the ichthyosaurs, most completely adapted of all reptiles to a marine existence. The ichthyosaurs first appear in the Middle Triassic, and although the members of the order were at that time somewhat less advanced than their abundant Jurassic (and less abundant Cretaceous) descendants, they were already definitely ichthyosaurs, quite distinct from mem- bers of any other reptijian group. I have recently (Romer 1968) demonstrated that, in contrast to earlier beliefs, their temporal structure was similar to that of the typical euryapsids; but this only adds further puzzlement to the problem of their ancestry. I have pointed out (Romer 1948) that except for the temporal opening, their skull structure is basically similar to that of ophiacodont pely- cosaurs, and like these pelycosaurs, could reasonably have originated from captorhinomorph cotylosaurs. But there is absolutely no trace of transitional forms. The ichthyosaurs could, like the ophiacodonts, have come directly from early reptiles still in an amphibious ecological condition. But the gap in time between the Carboniferous and Middle Triassic is great enough so that it is possible to argue here, as in the case of euryapsids and chelonians, that their ancestors might have become purely terrestrial and secondarily reverted to aquatic life.
The archosaurs, which include, as well as ancestral thecodonts and croco-
AQUATIC ADAPTATION IN REPTILES— PRIMARY OR SECONDARY ? 229
dilians, the great dinosaurian orders, the flying pterosaurs and bird ancestors, appear at first sight to have been ab initio a terrestrial group, in which there was a strong tendency toward a bipedal stance and, in pterosaurs and birds, on toward aerial life. But Charig (1966) has recently pointed out that there are suggestions of amphibious habits in early archosaurs. The strong hind legs and tail which were highly useful in the development of bipedal habits, may have been initially developed in relation to an amphibious mode of life (Meso- saurus is comparable). Further, many early archosaurs show aquatic tendencies (Charig & Reig 1970). The oldest and most primitive of archosaurs, the Proterosuchidae (such as “Chasmatosaurus’) were quite certainly amphibious in habits; crocodilians developed early from the thecodont base; the most abundant of later Triassic thecodonts were the amphibious crocodile-like phytosaurs; and the Middle Triassic thecodonts include a further family, Proterochampsidae, which were certainly phytosaur-like in habits (Romer 1971). All in all, it is not impossible that the ancestral archosaurs were derived from primitive reptiles which were still amphibious in habits. But the first known archosaur is late Permian in age, and it can be argued, as in the groups already discussed, that their ancestors had become completely terrestrial in Permian times before amphibious trends developed.
Of all reptile groups it is the Lepidosauria—the lizards, snakes, and rhynchocephalians—for which the best argument for a firm terrestrial ancestry can be made. The ancestors of the subclass, the Eosuchia, appear before the end of the Permian, and there is no positive evidence of any aquatic trend in the group. There are, to be sure, sea snakes and some water-living lizards in the recent fauna, but derivation of them from terrestrial ancestors can be reasonably made. Champsosaurus of the late Cretaceous and earliest Tertiary is amphibious, and certain Jurassic forms look suspiciously amphibious in nature but descent from terrestrial ancestors can be reasonably argued. Most prominent of aquatic lepidosaurs were the mosasaurs, which flourished greatly in the late Cretaceous; but these can be derived from amphibious ancestors in the early Cretaceous and, hence, mosasaur descent from Jurassic terrestrial forms can be reasonably argued. .
If we attempt to sum up this story of terrestrial versus aquatic trends in reptile history, it seems that the Reptilia were not too speedy in the late Paleo- zoic in leaving their ancestral home in fresh waters; and that some, at least, of the aquatic and amphibious members of the group may have persistently retained a primitive mode of life rather than having returned to the water from a purely terrestrial ancestry. But the question in many cases must remain an open one until new fossil evidence appears to fill in the major gaps still present in the Permian and early Triassic history of reptiles.
230 ANNALS OF THE SOUTH AFRICAN MUSEUM
REFERENCES
Bairp, D. & Carro.t, R. L. 1967. Romeriscus, the oldest known reptile. Science 1577: 56-59.
BARRELL, J. 1916. Influence of Silurian-Devonian climates on the rise of air-breathing verte- brates. Bull. geol. Soc. Am. 27: 387-436.
CarROL., R. L. 1964. The earliest reptiles. 7. Linn. Soc. (Zool.) 45: 61-83.
CarrRo.., R. L. 1969. Origin of reptiles. In: GANs, c. & PARSONS, T. S., eds. Biology of the Reptilia. 1: 1-44. London; New York: Academic Press.
Carrot, R. L. 1970a. The earliest known reptiles. Yale scient. Mag. 1970 (Oct.): 16-23.
Carro.., R. L. 1970). The ancestry of reptiles. Phil. Trans R. Soc. (B) 257: 267-308.
CarroLL, R. L. & OTHERS. 1972. Vertebrate paleontology of eastern Canada. Guidebook, Field excursion A 59. Int. geol. Congr. 24 (Montreal) P.Q. 1972: 1-113.
Cuaric, A. J. 1966. Stance and gait in the archosaur reptiles. Adumt Sci., Lond. 22 (103): 537. [ Abstract. ]
Cuaric, A. J. & Rec, O. A. 1970. The classification of the Proterosuchia. Biol. 7. Linn. Soc. 2: 125-171.
Pen A. 1946. On the subclass Batrachosauria—an intermediary group between amphi- bians and reptiles. Bull. Acad. Sci. URSS (Cl. Math. Nat. Ser. Biol.) 1946: 615-638.
Horton. N. 1970. Mauchchunkia bassa, gen. et sp. nov., an anthracosaur (Amphibia, Labyrintho- dontia) from the Upper Mississippian. Kirtlandia 12: 1-38.
Jarvik, E. 1955. The oldest tetrapods and their forerunners. Scient. Mon., N.Y. 80: 141-154.
PaNncHEN, A. L. 1970. Anthracosauria. In: KUHN, 0., ed. Handbuch der Paldoherpetologie/Encyclopedia of paleoherpetology. Teil 5A: 1-84. Stuttgart: Fischer.
Rayner, D. H. 1971. Data on the environment and preservation of late Palaeozoic tetrapods. Proc. Yorks. geol. Soc. 38: 437-495.
Reisz, R. 1972. Pelycosaurian reptiles from the Middle Pennsylvanian of North America. Bull. Mus. comp. ool. Harv. 144: 27-62.
Romer, A. S. 1939. An amphibian graveyard. Scient. Mon., N.Y. 49: 337-339.
Romer, A. S. 1946. The primitive reptile Limnoscelis restudied. Am. 7. Sci. 244: 149-188. Romer, A. S. 1948. Ichthyosaur ancestors. Am. 7. Sci. 246: 109-121.
Romer, A. S. 1956. The early evolution of land vertebrates. Proc. Am. phil. Soc. 100: 157-167. Romer, A. S. 1957. Origin of the amniote egg. Scient. Mon., N.Y. 85: 57-63.
Romer, A. S. 1958. Tetrapod limbs and early tetrapod life. Evolution 12: 365-369.
Romer, A. S. 1961. A large ophiacodont pelycosaur from the Pennsylvanian of the Pittsburgh
region. Breviora 144: 1-7.
Romer, A. S. 1966. Vertebrate paleontology. 3rd ed. Chicago: University of Chicago Press.
Romer, A. S. 1968. An ichthyosaur skull from the Cretaceous of Wyoming. Contr. Geol. Univ. Wyo. 7: 27-41.
Romer, A. S. 1971. The Chafiares (Argentina) Triassic reptile fauna. XI. Two new long- snouted thecodonts, Chanaresuchus and Gualosuchus. Breviora 379: 1-22.
Romer, A. S. & Price, L. I. 1940. Review of the Pelycosauria. Spec. Pap. geol. Soc. Am. 28:
i-x, 1-538.
Vaucun, P. P. 1955. The Permian reptile Araeoscelis restudied. Bull. Mus. comp. Zool. Harv. 113: 305-467.
Watson, D. M.S. 1926. The evolution and origin of the Amphibia. Phil. Trans. R. Soc. (B) 214: 189-257.
Wi.uston, S. W. 1912. Restoration of Limnoscelis, a cotylosaur reptile from New Mexico. Am. Ff. Sci. (4) 34: 457-468.
Wiuiston, S. W. 1914. Broiliellus, a new genus of amphibians from the Permian of Texas. J. Geol. 22: 49-56.
MIGRATIONS OF THE MULTITUBERCULATA AND THE LATE CRETACEOUS CONNECTIONS BETWEEN ASIA AND NORTH AMERICA
By ZOFIA KIELAN-JAWOROWSKA Polska Akademia Nauk, Xaklad Paleo zoologu, 02-089 Warszawa, Poland (With 3 figures)
CONTENTS
PAGE Fossil records . : ; : ‘ i : 231 Discussion : ‘ : : : : , 236 Conclusions . ; : F : : : 239 Summary ‘ : 4 ; : : ‘ 241 Acknowledgements ; : d : : 241 Addendum . : , : : ‘ : 241 References ‘ : : 3 : : ‘ 242
Fossiz RECORDS
The Multituberculata are the longest lived order of mammals. If we take into account the records generally accepted as belonging to the Multitubercu- lata, they first made their appearance in the Kimmeridgian and became extinct at the end of the Eocene, which means that they evolved for about 118 million years. It was tentatively claimed by some students of early mammals (Parring- ton 1967; Hahn 1969; Simpson 1971) that the Rhaetic Haramiyidae are related to the Multituberculata. If we regard the Haramiyidae as the earliest Multi- tuberculata (Fig. 1), the vertical range of this group would be extended to 160 million years.
The time distribution of the multituberculates is usually illustrated in diagrams as a continuous line, extending from Kimmeridgian through Eocene. However, if one combines all known data about the distribution of this group, the line would be broken in several places (Fig. 1). The first gap in the fossil records (if we accept the Haramiyidae as belonging to the Multituberculata) embraces the whole early Jurassic through Oxfordian, which is about 42 million years. The oldest known true multituberculates were found in the Kimmerid- gian of Portugal (Kuehne 1961; Hahn 1969, 1971); they embrace three genera, two of which, Paulchoffatia and Kuehneodon, are assigned to the Paulchoffatidae, suborder Plagiaulacoidea, while Guimarotodon is attributed to an undefined family, probably within the Plagiaulacoidea. After Kimmeridgian there is a second gap in the multituberculate time records, which embraces the Port- landian (possibly also latest Kimmeridgian and earliest Purbeckian) ; its dura- tion was at least 7 million years. The next records are from the Purbeckian of
231
Ann. S. Afr. Mus. 64, 1974: 231-243, 3 figs.
ANNALS OF THE SOUTH AFRICAN MUSEUM
232
“‘pyoffoynvg 10% vynffoyonvg peor :wIMyeIIy
*poqIWO sI epeuery Jo UOT}VULIO, IOAN FTP 1eddq woz stuopru14 snuas umouy A[100g
*(g961) usyeA
uv A 29 ULO]S UO paseq o1v vIOUDS ULdTIOUIY YWON jo sosuvs [eorydeisneng ‘souly UOTASIU poyYsi]qe}so-[Jam A[qeuosear dy} SMOJIe pIpOS ‘souTT UOT -eISIu pue s1aUaSoTAYd afqeqord 94} o}edIpuUT SMOIIY SNONUT}UOOSICT “sor[IUTe} 94} JO AYISIOAIP OTWIOUOXe} 9Y} 0} UOTAJOdo1d Ur Jou Iv saoRFINS poyo}eYy jo sodeys oy], “Suryo}ey yuorayIp ve Aq poxseur st AjIurey YOY “wsioues o}¥[MOIAqnIN|NUI sy} Jo UOTINGIYASIp [eorydeisnesys pue Teorydessoay “1 “SI
20pl4OpOoqQuDNs|Ny-) WW
2Dpl,UOpoWsoony f-
TIT HT 20pliALuojowWI> ey
SH2OUY
oisspan[ AAD
udioolpg
uDiuOULDg
UDIAO]/O)
= —
UDILUDDOBZK|
|
UD!P10}xQ)
UDI -bptuauiuny
UDIPUD|HOg uery>eqing|
UDILUOIO Bh)
unydy SEN 4 DLO] NIaDQnypyjneg = ee = -- 4 oe A AoY Sf = ound}f aL | NdIZqQNLIW~NUU4 t UDIg|V ISHN a IL SOx aL f I : SS a =. / ub! ubl = | =e | . - = —udwoU2) ie Y, \ Slew 1 upuioue2e4) upiuoin) oe , 1 uDiuoiny UDIDDIUO7 PR) iba SS; XQ n UDIDDIUOD ees UpluojuDg UDIUO,UDS ! “1D}O0G}O2W2h) « SRA BEEN 5 wraiuiselutag| I NRRERRRRIRRDN) |= S| |e aa ; SS oe gs / pelueseie) —_~ Terr aero wee S % . a aay Q U UDI 4 SO & | OG -—l|USIJSDD SEUSIMISPPW) SRO B es 899 x Das & 449!4;sDoW\ bers SOO ae ie (ap a= GP => ==> é z / SRS wes tOH Weqetecctereees aah 2U2902)0, 2U2IO2|0q Ree BIR oa: I"d SLRS 0 KR eee SNUWIOIO! SERBy T Koes Oo ah aeaaed Beri See yous) KEXX 0 © Sees ceyoy 2uU220 2U2905 amare Lax, Sl YsqgiodieavYIotinN s vl Yall Gl OLNOdGdoOTilid saOVIS VISV VOIVSWV HLAON jdong s30vls
MIGRATIONS OF THE MULTITUBERCULATA 233
Great Britain and from the Morrison Formation of North America, which are probably of the same age. Five genera are known from these beds—the Paul- choffatidae: Bolodon and Plioprion from Purbeck of Great Britain; Plagiaulaci- dae: Plagiaulax from Great Britain, Ctenacodon from Great Britain and North America, and Psalodon from North America (Simpson 1928, 1929; Hahn 1969). Of the latest Jurassic genera, uncertain remains of Plagiaulax are known from the Wealden of Great Britain, where, in addition, one more plagiaulacid genus Loxaulax was found (Woodward 1891; Clemens 1963a). In contra- distinction to the Jurassic multituberculates, which are represented by lower jaws with dentition, and in one case also a part of the braincase, the Wealden records are extremely scanty, represented only by a few isolated teeth. After the Wealden there is a third gap in multituberculate records, which extends for the second half of the Neocomian, through the Aptian, and embraces about 15 million years.
In 1969 the Soviet-Mongolian Palaeontological Expedition discovered in the Gobi Desert, at the locality of Khovboor, in Guchin Us somon (county), situated at 102°34 eastern longitude and 45°21 northern longitude, a fauna of early Cretaceous mammals. According to Barsbold et al. (1971) the assemblage contains the Multituberculata in addition to the Triconodonta, Symmetrodonta and Insectivora, while according to Kalandadze & Reshetov (1971), in addition to the above mentioned orders, there are also representatives of the Tritubercu- lata. The fauna has not been described as yet and the family assignment of the multituberculates is not known. Barsbold e¢ al. (1971) claimed Aptian- Albian as the presumed age of this fauna.
Another early Cretaceous fauna, which is known, but has not been adequately described, is that from the Albian of Texas. Patterson (1956: 9) when describing the Albian fauna from the locality of Forestburg stated: ‘Multituberculates are the commonest fossils at Forestburg. . . . All specimens found may be referred to the Plagiaulacidae.’ He gave a short description of the Albian multituberculate lower and upper molars, but as the specimens have not been figured, it is difficult to get an idea about their affini- ties. Slaughter (1965) mentioned some isolated multituberculate teeth in samples from Butler Farm in Texas, which are of the same age as those of Forestburg, described by Patterson. These multituberculates were neither figured nor described by Slaughter.
The fourth gap in the multituberculate record is after the Albian and embraces Cenomian, Turonian and perhaps also Coniacian, or a part of it, depending on whether the Djadokhta Formation of Asia is of Coniacian or Santonian age. In this discussion let us assume that the Djadokhta Formation is of Coniacian age (Fig. 1). This would mean that the fourth gap in the multi- tuberculate records extends until the end of the Turonian and embraces about 19 million years.
Beginning with the Djadokhta Formation, the multituberculate records are more or less continuous through the Eocene. In the Djadokhta Formation
234. ANNALS OF THE SOUTH AFRICAN MUSEUM
the representatives of modern multituberculate orders made their appearance: the Ptilodontoidea and ‘Taeniolabidoidea.*
The only representative of the order Ptilodontoidea in Asia is the Djadokhta Formation Gobibaatar, assigned to the Neoplagiaulacidae. Four other multi- tuberculate genera recorded from this formation belong to the Taeniola- bidoidea; these are Sloanbaatar, assigned to the monotypic family Sloan- baatariidae, Aryptobaatar and Bulganbaatar, assigned to the Eucosmodontidae and Kamptobaatar and Djadochtatherium, assigned to the ‘Taeniolabididae (Simp- son 1925; Kielan-Jaworowska 1969, 1971, 1974). In the younger Asiatic Barun Goyot Formation, which is of Campanian age, the multituberculate fauna so far described consists of three genera: Chulsanbaatar, assigned to the monotypic family Chulsanbaataridae, the taeniolabidid Dyadochtatherium, represented by a more advanced species than its ancestor from the Djadokhta Formation, and the eucosmodontid Nemegtbaatar, derived from the Dyjadokhta genus Bulganbaatar. In addition there are in the Barun Goyot Formation three or more, as yet not described, multituberculate genera, none of which appears to belong to the Ptilodontoidea (Kielan-Jaworowska 1974). ‘The post-Campanian multituberculate records are very scanty in Asia. A single genus Buginbaatar is known from Bugeen Tsav locality in the Gobi Desert. This is from the beds which are of the late Maastrichtian or early Paleocene age (Kielan-Jaworowska & Sochava 1969). Buginbaatar has been tentatively assigned to the ?Cimolo- myidae, but in the discussion the authors pointed out (p. 363) that: “Buginbaatar n. gen. shares the characters of both the Cimolomyidae and the Eucosmodon- tidae.’ Now it seems more probable to me that it is an eucosmodontid.
The Asiatic records of Paleocene multituberculates are extremely small. Only two taeniolabidoid genera Prionessus and Sphenopsalis are known from Khashaat (Gashato) and Naran Bulak localities, from beds belonging to either the Paleocene or Eocene (Matthew ef al. 1928; Szalay & McKenna 1971).
The late Cretaceous multituberculate records in North America begin with the early Campanian Upper Milk River Formation (Fox 1971, 1972). Younger than the Upper Milk River Formation is the Middle Campanian Judith River Formation (Sahni 1972). Beginning with the Judith River Forma- tion there is a nearly continuous multituberculate record in North America through the late Eocene (Granger & Simpson 1929; Jepsen 1930, 1940; Clemens 19635; Lillegraven 1969 and others). The multituberculate assem- blages in the Upper Milk River and Judith River Formations are very different from that in the Barun Goyot Formation and in the younger Djadokhta Formation in Asia. In the Campanian of North America, in contradistinction to the conditions in Asia, the Ptilodontoidea prevail over the Taeniolabidoidea.
* In the present paper Sloan & Van Valen’s (1965) division of Cretaceous and Tertiary Multituberculata into the Ptilodontoidea and Taeniolabidoidea is tentatively accepted, although I believe that these suborders do not form natural units. The North American taeniolabidoid family Cimolodontidae probably arose independently from the Taeniolabididae and Eucosmo- dontidae which are of Asian origin. Formal modification of this classification requires, however, . a detailed comparative study of North American and Asian multituberculates.
a ~~
MIGRATIONS OF THE MULTITUBERCULATA 235
The only taeniolabidoid family represented in these formations are the Cimolo- myidae (unknown outside of North America), represented by two genera: Meniscoessus and Cimolomys. ‘The familial and ordinal assignment of the genus Viridomys Fox from the Upper Milk River Formation is uncertain, and that is why Viridomys is not shown in Figure 1. In addition to these three genera, in the Campanian of North America three ptilodontoid families are present: the Cimolodontidae, represented by Czmolodon, the Ptilodontidae, represented by Kimbetohia and the Neoplagiaulacidae (former Ectypodontidae), represented by Czmexomys and Mesodma. All the multituberculate genera recorded in the Judith River Formation continue through the Upper Edmonton and Lance Formations (Lillegraven 1969; Clemens 19635) to the latest Cretaceous Hell Creek Formation (Sloan & Van Valen 1965; Van Valen & Sloan 1966). In addition, in Lance Formation the ?ptilodontoid Essonodon, which continues through Hell Creek Formation, made its appearance and in Hell Creek the eucosmodontid Stygimys and taeniolabidid Catopsalis appear.
The above data show that there are no pre-Campanian late Cretaceous multituberculate records in North America. In the Campanian only one taeniolabidoid family, Cimolomyidae, endemic for North America, is present on this continent, accompanied by the representatives of three ptilodontoid families: the Neoplagiaulacidae, Cimolodontidae and Ptilodontidae. By the time of the latest Cretaceous two other taeniolabidoid families, known from older beds of Asia, reach North America: the Taeniolabididae and Eucosmo- dontidae; while two taeniolabidoid families, the Sloanbaataridae and Chulsan- baataridae, remain so far monotypic and endemic for Central Asia.
The Paleocene and Eocene multituberculate records of North America bring the continuation of the Cretaceous genera and in addition the appearance of new genera in each of the families known in latest Cretaceous on this conti- nent. Within the Neoplagiaulacidae, in addition to Cimexomys, which continues in early Puercan, and Mesodma, which continues through Clarkforkian (Krishtalka 1973), the four new genera made their appearance: Mimetodon, Ectypodus, Neoplagiaulax and Parectypodus, the three latter continuing in the Eocene. One new representative of the Cimolodontidae is recorded: Anconodon, while within the Ptilodontidae, in addition to Kimbetohia, which continues through Puercan time, two new genera, Ptilodus and Prochetodon, made their appearance, the latter continuing into the Eocene (Graybullian).
Within the Eucosmodontidae, in addition to Stygimys, which continues through the Torrejonian, Eucosmodon appears in the Puercan and continues through Torrejonian time, while Microcosmodon, Pentacosmodon and Neoliotomus appear in the Tiffanian, the Veoliotomus continuing through the Graybullian.
In the Paleocene the multituberculates appear again in Europe, in France (Russell 1964) and are represented there by the North American genus JVeo- plagiaulax (representative of the Neoplagiaulacidae) and the endemic genus Liotomus, assigned by Van Valen & Sloan (1966) to the Cimolodontidae.
236 ANNALS OF THE SOUTH AFRICAN MUSEUM
Discussion
The data presented above show that during the 160 million years of pre- sumable multituberculate evolution, there are gaps in time records that embrace at least 73 million years. This shows that our information on the development of this group, especially in the first half of its evolution, is extremely scanty.
The oldest Kimmeridgian multituberculates (Hahn 1969, 1971) already show all the essential features of this group. ‘The subsequent evolution concerned only the changes of the dentition, the reduction of the first incisors, the gradual reduction of the premolars and the change of the premolars from chewing into the cutting type. But the essential features of the multituberculate skull and lower jaw structure are already developed by the Kimmeridgian. In the dentary structure there are no great changes between the Kimmeridgian and Tertiary forms. It should be also added that the multituberculate dentary as early as in Kimmeridgian has an entirely mammalian appearance, which is not the case with other groups of mammals from the same beds (Krebs 1971). This shows that the multituberculates are a group which very early branched off from the main evolutionary line of mammals.
All records of multituberculates are confined only to the three continents of the Northern hemisphere: Europe, Asia and North America. ‘The Harmiyidae are known from the Rhaetic of Europe—Switzerland and Great Britain (Peyer 1956; Simpson 1928). The Kimmeridgian multituberculates are known also from Europe only, from the single locality in Portugal (Hahn 1969, 1971). In Purbeck time the multituberculate record has been spread from Europe (Great Britain) to North America (Morrison Formation). One genus (Ctenaco- don) is common for latest Jurassic of North America and Europe, the others are fairly closely related. The earliest Cretaceous fauna (Wealden) is very poorly known and again only from Europe. The above data may lead to the con- clusion that the multituberculates originated in Europe.
More detailed data on the evolution of this group have been gathered from the Cretaceous period, particularly so from the late Cretaceous. In the second half of the late Cretaceous the multituberculates were found in two continents: in North America (Albian of Texas) and in Asia (Khovboor locality), the age of which it is either Aptian or Albian. Unfortunately, there are no data about the evolutionary level of the multituberculates at that time, except for the statement of Patterson (1956) that the multituberculates from the Albian of Texas belong to the Plagiaulacidae. It appears from the studies of Hahn (1969, 1971) and others, that of the two plagiaulacoid families, the Plagiaulacidae rather than the Paulchoffatidae are the group which gave rise to the Ptilodon- toidea.
The question arises whether in the early part of the late Cretaceous the earliest Ptilodontoidea developed in one part of the globe from the Plagiaulaci- dae and then migrated to other continents, or whether the different families of the Ptilodontoidea developed independently in various areas. The hitherto recorded data speak rather for the second hypothesis. The oldest known
MIGRATIONS OF THE MULTITUBERCULATA 237
ptilodontoid is the Djadokhta Formation Gobibaatar, belonging to the Neopla- giaulacidae. Unfortunately, the mammals of the same age as those from the Djadokhta Formation are not known from Europe or North America. In the Campanian Upper Milk River Formation, which is the youngest late Cretaceous formation yielding the multituberculates in North America, the Ptilodontoidea are already represented by three families, among the representatives of which only one genus, Czmexomys, may be regarded as related to the Asian Gobzbaatar. All other representatives of the Campanian Ptilodontoidea are much more advanced in structure than the possible Campanian descendants of Gobibaatar could be at that time. This indicates that most probably the main stock of ptilodontoid families originated in the first half of the late Cretaceous in North America. Whether Cimexomys is a descendant of Asian Gobibaatar cannot be asserted with certainty, as the skull of Cimexomys is not known and it 1s difficult to base conclusions on the evolution of multituberculate genera on teeth only. Such possibility is, however, very probable. It is thus an open question whether the Neoplagiaulacidae are a group of monophyletic (Asiatic) or diphyletic (Asiatic and North American) origin.
Nothing can be said about the origin of the North American taeniolabidoid family, the Cimolomyidae. This family embraces two genera: Cimolomys and Moeniscoessus, which in my opinion are not very closely related to each other. Both are poorly known, we do not know their skull structure and the develop- ment of the palate. Therefore it is extremely difficult to discuss their origin, and this question must remain open until more material from this family is collected. As far as the origin of the Taeniolabididae and Eucosmodontidae is concerned, the oldest representatives of both families were found in the Dja- dokhta Formation of the Gobi Desert. It is very interesting that the Djadokhta eucosmodontid A7yptobaatar is extremely similar in the general shape of the skull and in the structure of the cheek teeth to the neoplagiaulacid Gobzbaatar, which occurs in the same formation (Kielan-Jaworowska 1970, 1974). The differences between the two genera in question concern only the shape of the lower jaw and the structure of the incisors, which in Gobibaatar are entirely covered by the enamel, while in A7ryptobaatar the enamel is restricted to the antero-ventral band. It seems that restriction of the enamel to a band must be considered as the specialized character. It follows that the derivation of ‘Tae- niolabidoidea from Ptilodontoidea is theoretically the most probable hypothesis, especially considering the similarity of Gobibaatar and Kryptobaatar. Even though these two are contemporaries, the ptilodontoid Gobibaatar appears to have remained very similar to the forms that gave rise to the eucosmodontid A7rypto- baatar. In the Djadokhta Formation another eucosmodontid, Bulganbaatar also occurs, as well as the oldest representatives of the Taeniolabididae: Djadoch- tatherium and Kamptobaatar (the latter genus is very different from Djadochtathe- rium).
The recent findings of a rich multituberculate fauna in the Campanian of the Gobi Desert cast new light on the origin of North American representatives
238 ANNALS OF THE SOUTH AFRICAN MUSEUM
of the Taeniolabididae and Eucosmodontidae (Kielan-Jaworowska 1974). The eucosmodontid Bulganbaatar nemegtbaataroides from the Djadokhta Forma- tion is now regarded as an ancestor of the Campanian (Barun Goyot Forma- tion) Nemegtbaatar gobiensis, which seems to be close to the ancestors of the earliest Paleocene North American eucosmodontid Eucosmodon (Fig. 2). The
A ®
Fig. 2. Comparison of lower jaws in outer view of: A. Nemegibaatar gobiensis Kielan-Jaworowska, from the Campanian of Asia. B. Eucosmodon gratus Jepsen from the early Paleocene of North America. (A.—After Kielan-Jaworowska 1974. B. —After Jepsen 1940.)
Djadokhta Formation taeniolabidid Dyjadochtatherium matthewi is regarded as an ancestor of the Campanian (Barun Goyot Formation) Dyjadochtatherium catopsaloides, which again is an ancestor of the North American species Catopsalis joynert, known from the latest Cretaceous Hell Creek Formation (Fig. 3).
This indicates that two taeniolabidoid families, the Eucosmodontidae and and ‘Taeniolabididae, first developed in the early part of the late Cretaceous in Asia and reached North America in the latest Cretaceous.
Nothing is known on the development of the multituberculates in the
MIGRATIONS OF THE MULTITUBERCULATA 239
5mm
Ds
Fig. 3. Comparison of P*—-M? in occlusal view and P, in labial view of: A.—Djadochtatherium catopsaloides Kielan-Jaworowska from the Cam- panian of Asia, B.—Catopsalis joynert Sloan & Van Valen, from the latest Maastrichtian of North America. (A.— After Kielan-Jaworowska | | 1974; B.— After Sloan & Van Valen 1965.)
| late Cretaceous of Europe. It is probable that the descendants of European Jurassic multituberculates will be found in Cretaceous and Tertiary strata of this continent. The only known so far early Tertiary representatives of the
| multituberculates found there, appear to be closely related to the North American forms.
CONCLUSIONS
ee Es =e _—
It appears from the data presented above that in the late Cretaceous three multituberculate lines invaded North America from Asia. One of them (not very certain), the Gobibaatar — Cimexomys line, belongs to the pre-Campanian
240 ANNALS OF THE SOUTH AFRICAN MUSEUM
migration, while two others, the Dyjadochtatherium —> Catopsalis line and the Nemegtbaatar — Eucosmodon line, belong to the post-Campanian migration. ‘There are no data which indicate that any of the multituberculate groups that flourished in the Cretaceous of North America reached Asia in the late Creta- ceous. It follows that in the late Cretaceous the connections between these continents allowed the migration from Asia to North America, but not vice versa.
This conclusion, based on the evolution and distribution of the multi- tuberculates, is partly supported by the data on the distribution of the marsupial and eutherian mammals.
A major component of the North American late Cretaceous mammalian fauna consists of marsupials. As far as the occurrence of marsupials in Asia is concerned, it has been generally accepted until recently that this group never reached Asia. Butler & Kielan-Jaworowska (1973) have shown that two late Cretaceous Asian genera, Deltatheroides and Deltatheridium, assigned to the Deltatheridiidae, have a marsupial dental formula; however, they classified the Deltatheridiidae as Theria of metatherian-eutherian grade. There is no evidence that any of the late Cretaceous North American marsupials migrated to Asia.
With regard to Cretaceous Theria from North America, Slaughter (1971) believes that among the therians from the Albian of Texas true eutherians occur.
Present knowledge on the early evolution of both eutherian and marsupial mammals is too limited to draw a conclusion concerning the place of origin of both groups. However, even now it can be stated that some of the North American latest Cretaceous eutherian mammals are of Asiatic origin, which has been suggested earlier by Lillegraven (1969).
Simpson (1962) recognized three main paths of faunal interchange for land animals: corridors, filters, and sweepstakes routes. The faunistic differences between Asia and North America in the late Cretaceous are so great that one can reasonably conclude the absence of a corridor. If there was a land con- nection between these continents one should assume the existence of a filter which prevented the marsupials and certain North American dinosaurs from invading Asia. Such a filter could be the cold climatic zone in north-eastern Siberia, which would be in agreement with the position of the north pole in this region in the late Cretaceous. However, the filter (in Simpson’s sense) acts in both directions while the evidence presented in the present paper shows that with respect to mammals this filter acted in one direction only. While the Asiatic multituberculates and eutherians invaded North America, neither the marsupials nor the North American multituberculates reached Asia in the late Cretaceous. The sweepstakes route, in contradistinction to the filter, usually permits rare migrations in one direction only.
An analysis of the late Cretaceous mammalian fauna of Asia and North America leads to the conclusion that these continents were probably separated by a barrier, possibly by marine straits, and the faunal interchange was of the
MIGRATIONS OF THE MULTITUBERCULATA 241
sweepstakes route type. The existing currents allowed occasional drifts of some invaders from Asia to North America but not vice versa.
The limitation of the present hypothesis is that it is based on the study of the mammals only. Detailed comparisons of other land vertebrates that occur in the Cretaceous of both continents are needed to support or contradict the above hypothesis.
SUMMARY
The stratigraphical and geographical distribution of all described multi- tuberculate genera is discussed. It is shown that the Multituberculata originated in the early Mesozoic in Europe and then migrated to North America and Asia. In the Cretaceous of Asia the ‘Taeniolabidoidea prevail, while in North America it is the Ptilodontoidea. The two taeniolabidoid families: Eucosmodontidae and Taeniolabididae originated in Asia and reached North America in the latest Cretaceous. In addition to the multituberculates, the eutherian mammals migrated from Asia to North America during the late Cretaceous, while the groups of North American origin (e.g. late Cretaceous marsupials and certain multituberculates) never reached Asia. The possibility of a land connection between Asia and North America with a cold zone filter preventing the migration of certain vertebrates is discussed. But inasmuch as the existing barrier, with respect to mammals, allowed one-way migrations only, it is con- cluded that the faunal interchange between Asia and North America in the late Cretaceous was not of the filter type, but rather of the sweepstakes route type. These continents were probably separated by marine straits and the currents allowed occasional drifts from Asia to North America, but not vice versa.
ACKNOWLEDGEMENTS
pial thank Dr Pamela Robinson (University College, London), Dr Malcolm C. McKenna and Dr Giles MacIntyre (American Museum of Natural History, New York) for reading the manuscript and for discussion.
ADDENDUM
After this paper was submitted for publication, two papers relevant to the problems discussed herein appeared.
(1) Lillegraven (1972) gives a preliminary description of mammals from the El Gallo Formation (late Campanian) of Mexico. In this paper he described M! of ?Stygimys sp. This tooth is very similar to that of the Asian Nemegtbaatar and probably belongs to the Eucosmo- dontidae. Be this the case, the Eucosmodontidae should have migrated from Asia to North America at the middle—late Campanian boundary, and not in Maastrichtian as may appear from the data discussed herein and from Figure 1.
(2) Hahn (1973) erects for the Haramiyidae the new suborder Haramiyoidea, within the Multituberculata. This suborder could not be introduced into Figure 1.
242 ANNALS OF THE SOUTH AFRICAN MUSEUM
REFERENCES
BARSBOLD, R., Voronin, Ju. I. & ZEGALLO, V. 1971. O rabote Sovetsko-Mongol’skoj Paleontolo- giceskoj Ekspedicii v 1969-1970 gg. Paleont. Zh. 2: 139-143.
Butter, P. M. & KiELAN-JAWorRowskKA, Z. 1973. Is Deltatheridium a marsupial? Nature, Lond. 245: 105-106.
CLEMENS, W. A., JR. 1963a. Wealden mammalian fossils. Palaeontology 6: 55-69.
CiemMEnNS, W.A., JR. 19635. Fossil mammals of the type Lance Formation, Wyoming. Part I. Introduction and Multituberculata. Univ. Calif. Publs geol. Sci. 48: 1-105.
Fox, R. C. 1970. Eutherian mammal from the early Campanian, (late Cretaceous) of Alberta Canada. Nature, Lond. 227: 630-631.
Fox, R. C. 1971. Early Campanian multituberculates (Mammalia: Allotheria) from the Upper Milk River Formation, Alberta. Can. 7. Earth Sci. 8: 916-938.
Fox, R. C. 1972. A primitive therian mammal from the Upper Cretaceous of Alberta. Can. 7. Earth Sci. 9: 1479-1494.
GRANGER, W. D. & Simpson, G. G. 1929. A revision of the Tertiary Multituberculata. Bull. Am. Mus. nat. Hist. 56: 601-676.
Haun, G. 1969. Beitrage zur Fauna der Grube Guimarota. Nr. 3. Die Multituberculata. Palaeontographica 133: 1-100.
Hahn, G. 1971. The dentition of the Paulchoffatiidae (Multituberculata, Upper Jurassic). Mems Servs geol. Port. (n.s.) 17: 1-39.
Haun, G. 1973. Neue Zahne von Haramiyiden aus den Duetschen Oberen-Trias und ihre Beziehungen zu den Multituberculaten. Palaeontographica (Abt. A) 142 (Lfg 1-3): 1-15.
JepseN, G. L. 1930. Stratigraphy and paleontology of the Paleocene of northeastern Park County, Wyoming. Proc. Am. phil. Soc. 69: 463-528.
Jepsen, G. L. 1940. Paleocene faunas of the Palecat Bench Formation, Park County, Wyoming. Proc. Am. phil. Soc. 83: 217-340.
KALANDADZE, N. N. & ReEsHETov, V. Ju. 1971. Interesnye paleontologi¢eskie nachodki v Mongolu. Priroda 5: 83-84.
KIELAN-JAWOROWSKA, Z. 1969. Discovery of a multituberculate marsupial bone. Nature, Lond. 222: 1091-1002.
KIELAN-JAWoROwSKA, Z. 1970. New Upper Cretaceous multituberculate genera from Bayn Dzak, Gobi Desert. Results of the Polish-Mongolian Palaeontological Expeditions. II. Palaeont. pol. 21: 35-49.
KreLAn-JAworowska, Z. 1971. Skull structure and affinities of the Multituberculata. Results of the Polish-Mongolian Palaeontological Expeditions. III. Palaeont. pol. 25: 5-41.
KiELAN-JAworowskA, Z. 1974. Multituberculate succession in the late Cretaceous of the Gobi Desert (Mongolia). Results of the Polish-Mongolian Palaeontological Expeditions. V. Palaeont. pol. 30: 23-44
KiIELAN-JAWOROwSKA, Z. & SocHaAva, A. V. 1969. The first multituberculate from the upper- most Cretaceous of the Gobi Desert, Mongolia. Acta palaeont. pol. 14: 355-371.
Kress, B. 1971. Evolution of the mandible and lower dentition in dryolestids (Pantotheria, Mammalia). Zool. F. Linn Soc. 50: (Suppl. 1: Early mammals) : 89-102.
KRIsHTALKA, J. 1973. Late Paleocene mammals from the Cypress Hills, Alberta. Spec. Publs Mus. Texas Tech Univ. 2: 1-77.
Kuenne, W. G. 1961. Eine Mammalia Fauna aus dem Kimmeridge Portugals. Neues jb. Geol. Paléont. 7: 374-381.
LiLLEGRAVEN, J. A. 1969. Latest Cretaceous mammals of upper part of Edmonton Formation of Alberta, Canada, and review of marsupial-placental dichotomy in mammalian evolu- tion. Palaeont. Contr. Univ. Kans. (Art.) 50: (Vert. 12): 1-122.
LILLEGRAVEN, J. A. 1972. Preliminary report on late Cretaceous mammals from the El Gallo Formation, Baja California del Norte, Mexico. Contr. Sci. 232: 1-11.
Mattuew, W. D., Grancer, W. & Simpson, G. G. 1928. Paleocene multituberculates from Mongolia. Am. Mus. Novit. 331: 1-4.
ParRINGTON, F. R. 1967. The origins of mammals. Advmt Sci., Lond. 24: 165-173.
PATTERSON, B. 1956. Early Cretaceous mammals and the evolution of mammalian molar teeth. Fieldiana, Geol. 13: 1-105.
Preyer, B. 1956. Uber Zahne von Haramiyiden, von Triconodonten und von wahrscheinlich synapsiden Reptilien aus dem Rhat von Hallau. Schweiz. palaeont. Abh. 72: 1-71.
MIGRATIONS OF THE MULTITUBERCULATA 243
RussELL, D. E. 1964. Les Mammifeéres paléocénes d’Europe. Mém. Mus. natn. Hist. nat. Paris (n.s. CG) 43: 1-324.
Saunt, A. 1972. The vertebrate fauna of the Judith River Formation, Montana. Bull. Am. Mus. nat. Hist. 1477: 325-412.
SHIKAMA, T. 1947. Teilhardosaurus and Endotherium, new Jurassic Reptilia and Mammalia from the Husin coal-field, South Manchuria. Proc. Japan Acad. 23: 76-84.
Smpson, G. G. 1925. A Mesozoic mammal skull from Mongolia. Am. Mus. Novit. 201: 1-11.
Smpson, G. G. 1928. A catalogue of the Mesozoic Mammalia in the Geological Department of the British Museum. London: British Museum.
Smpson, G. G. 1929. American Mesozoic Mammalia. Mem. Peabody Mus. Yale 3: 1-235.
Smupson, G. G. 1953. Evolution and geography; an essay on historical biogeography with special reference to mammals. Condon Lectures, Oregon State System of Higher Education: 1-64. Eugene.
Smpson, G. G. 1971. Concluding remarks: Mesozoic mammals revisited. Zool. 7. Linn. Soc. 50 (Suppl. 1: Early mammals): 181-108.
SLAUGHTER, B. H. 1965. A therian from the Lower Cretaceous (Albian) of Texas. Postilla 93: 1-18.
SLUAGHTER, B. H. 1971. Mid-Cretaceous (Albian) therians of the Butler Farm local fauna, Texas. <ool. F. Linn. Soc. 50 (Suppl. 1: Early mammals): 131-142.
Stoan, R. E. & Van Valen, L. 1965. Cretaceous mammals from Montana. Science 148: 220-227.
SzaLay, F. S. & McKenna, M. C. 1971. Beginning of the age of mammals in Asia: the late Paleocene Gashato Fauna, Mongolia. Bull. Am. Mus. nat. Hist. 144: 273-317.
Van Yalen, L. & Sloan, R. E. 1966. The extinction of the multituberculates. Syst. Zool. 15: 261-278.
Woopwarp, A. S. 1911. On some mammalian teeth of the Wealden of Hastings. Q.7l geol. Soc. Lond. 67: 278.
- y i — =r,
mm | Me iors et ate ek ae
i
sa
“ ee a si Ne rar) we 4 7
1 n ao aL bh i i oa aay uo . 4 a 7 " ) i ‘ a « : \ 1 . j . aay | 7
DERIVE DES CONTINENTS ET PALEONTOLOGIE: QUELQUES REFLEXIONS
Par J.-P. Leaman Muséum National d’ Histoire Naturelle, Paris
(Avec 3 figures)
La théorie de la dérive des continents a été proposée avec netteté pour la premiere fois par A. Wegener en 1915. On insiste souvent sur le fait que Wegener avait eu des précurseurs mais ceux-ci tels que Bacon ou le pére Placet étaient des philosophes et leur intuition n’avait pas de fondement vraiment scientifique. Il est indéniable au contraire que les idées de Suess concernant l’existence d’un continent de Gondwana regroupant |’Amérique du Sud, l’Afrique du Sud, ?Inde et Madagascar en une seule unité, impli- quaient une séparation d’une masse continentale unique en blocs distincts et que lexistence du Gondwana devrait étre un important argument en faveur de la théorie de Wegener. La théorie de Wegener toutefois ne permettait pas d’expliquer pourquoi les masses continentales s’étaient séparées. Les travaux récents ont abouti a élaborer la théorie dite de la tectonique des plaques qui complete, mais sans l’infirmer en général, la théorie de Wegener.
Les critiques formulées a Pencontre de la théorie de Wegener avant guerre étaient surtout d’ordre géophysique: on n’avait pas observé de facon incon- testable de variations de latitudes, d’écartements manifestes de continents (depuis lors le paléomagnétisme a permis de calculer des paléolatitudes). Les paléontologistes dans ensemble n’étaient pas hostiles a la théorie de Wegener, mais ils pensaient que les connexions continentales, auxquelles ils devaient faire manifestement appel, devaient étre expliquées par les géologues et les géophysiciens, ces problémes ne pouvant étre résolus par la Paléontologie; Phypothése évidemment curieuse des ponts continentaux ne faisait que tra- duire lVaffirmation de l’existence de connexions intercontinentales. Elle ne semble plus avoir été prise bien au sérieux dans les années 1940 ‘Chose curieuse, cette théorie commode, voire nécessaire, ne peut étre prouvée par aucun argu- ment d’ordre géologique’ (R. Furon).
Actuellement le faisceau de preuves géophysiques assemblées en faveur de la théorie de la dérive continentale est réellement impressionnant (voir par exemple Tarling 1971, Coulomb 1969, etc.) et il est probablement temps d’essayer de voir comment cette théorie s’accorde avec les résultats de la Paléontologie. Il y a eu d’ailleurs 4 ce propos de nombreux colloques ou congrés ot ce probléme a été débattu (Congrés du Gondwana de Buenos Aires 1967 et du Cap 1970; Congrés du Nato Advanced Study Institute a Newcastle 1972; Congrés géologique international de Montréal 1972; Congrés zoologique international de Monaco 1972, etc). On doit constater qu’en particulier au
245
Ann. S. Afr. Mus. 64, 1974: 245-253, 3 figs.
246 ANNALS OF THE SOUTH AFRICAN MUSEUM
Congrés de Montréal les discussions entre “mobilistes’, partisans de la dérive, et ‘fixistes’ adversaires de celle-ci, ont été souvent trés vives, voire passionnées. Notons briévement que si le terme de mobiliste peut-étre maintenu, celui de fixiste doit étre résolument proscrit, puisque ce terme est déja employé en Géologie dans le sens d’adversaire de |’Evolution (‘Cuvier était fixiste’) et que son introduction en tectonique ne peut étre que source de confusion. Je ne chercherai pas ci-dessous a trouver des preuves paléontologiques en faveur de la théorie de Wegener ou contraires a celle-ci; je chercherai simplement a voir comment on peut utiliser les fossiles en tant que témoins de la paléogéo- graphie continentale.
L’idée méme du continent de Gondwana a été fondée sur des ressemblances de composition floristique. Ainsi cette flore se caractérise depuis le Stéphanien jusqu’au début du Trias par la présence des genres Glossopteris et Gangamopteris, sans Sigillaires ni Lépidodendrons, ni Calamites mais a Lycophytes et Coni- féres distincts de ceux de ’hémisphére nord. Les travaux de Zalessky (1932) affrmant la présence d’une flore a éléments gondwaniens dans le bassin du Kouznetsk semblent reposer sur des erreurs de déterminations (Emberger). L’individualité de la flore gondwanienne parait donc bien certaine; s’en suit-il que la répartition des Plantes est un bon critére pour affirmer l’existence d’un continent disparu? Nous ne le pensons pas: en effet les spores des plantes terrestres (ou les ‘graines’ des Ptéridospermées et celles des Coniféres) sont facilement transportables soit par le vent, soit par les animaux volants, soit par les courants. D’allieurs la carte des empires floraux actuels (voir Emberger 1968: 640, fig. 740) ne correspond nullement aux continents: par exemple (Fig. 1), ? Afrique du Nord a une flore holarctique et l’Afrique est traversée par
F259 AS. Ay
Empire floral haleranae
Empire ee africano- SS
if Empire floral antarctique - australien
Fig. 1. Les empires floristiques du monde actuel (d’apres Emberger).
la limite entre empire holarctique et l’empire africano-malgache; de méme Amérique du Nord est séparée en deux par la frontiére de l’empire holarctique et de ’empire américain, etc. De sorte que, s’il existe bien une flore a Ganga-
—
DERIVE DES CONTINENTS ET PALEONTOLOGIE 247
mopteris, celle-ci ne prouve pas pour autant l’existence d’un continent de Gon- dwana; cependant le fait que cette flore se retrouve en Amérique du Sud, en Afrique du Sud, a Madagascar, aux Indes et en Australie et pas ailleurs est évidemment une forte présomption en faveur du continent de Gondwana. L’étude des Invertébrés fossiles peut évidemment permettre des com- paraisons intéressantes s’il s’agit d’Invertébrés exclusivement continentaux. Qu’en est-il au contraire des Invertébrés marins? Les Invertébrés marins a larves planctoniques—et ils sont trés nombreux—ont évidemment des possi- bilités d’expansion d’autant plus grandes que la durée de vie larvaire est plus longue, comme l’a montré Mortensen (voir Ekman 1953); ces Invertébrés doivent logiquement avoir une répartition soumise a des conditions de tempé- rature, mais elle doit étre trés vaste. Il en est probablement de méme des Algues marines. I] faut aussi se souvenir que dans une telle discussion les argu- ments négatifs ont moins de valeur que les arguments positifs: ainsi Meyerhoff & Teichert notent la présence en Australie occidentale de séquences per- miennes trés épaisses a fossiles a affinités téthysiennes; ces auteurs en déduisent que les reconstitutions paléogéographiques (Ahmad, Veevers) qui raccordent l’Australie a Inde sont fausses; mais ces schémas pourraient étre partiellement incorrects sans que l’existence d’un continent de Gondwana soit mise en ques- tion. L’argument positif favorable est la ressemblance des formes de Stégocé- phales entre l’Inde, l’Australie, Afrique du Sud (présence de Brachyopideés en Australie [Blinasaurus] et aux Indes) et la présence de Procolophonideés en Australie (Bartholomai & Howie 1970; voir Romer 1972: 8) et en Afrique du Sud. Du point de vue paléobiogéographique, les arguments admettant le rat- tachement de Australie a l’Inde nous paraissent donc pour l’instant assez peu nombreux. Se fondant d’autre part sur la répartition des Invertébrés fossiles dans la nature actuelle, beaucoup d’auteurs font appel a la notion de gradient de
Fig. 2. Isothermes d’hiver (dans les deux hémisphéres) (d’aprés Stehli).
248 ANNALS OF THE SOUTH AFRICAN MUSEUM
diversité pour déterminer des paléolatitudes. Le principe de cette méthode est simple: on constate que sur la terre actuelle, dans l'ensemble, le nombre des différentes espéces décroit réguliérement de |’équateur au pole. Or les tem- pératures de la mer sont représentées par des isothermes qui, pour des raisons particuliéres non liées au rayonnement solaire (courants, vents locaux, etc.), ne sont pas paralléles a ’équateur (Fig. 2). On peut alors par des méthodes mathématiques éliminer en quelque sorte ces phénomenes ‘de bruit’ et con- struire des isothermes (Fig. 3) qui apparaissent étre essentiellement fonction de
Fig. 3. Diagramme obtenu a partir du précédent (Fig. 2) en corrigeant les effets de bruit (d’aprés Stehli).
la latitude (Stehli 1968); on constate alors que, dans la nature actuelle, les gradients de diversité de tous les groupes majeurs d’organismes a vaste distribu- tion sont fonction de la latitude et en général (il y a évidemment des exceptions telles que les Pingouins) atteignent leur maximum a l’équateur. Appliquée aux fossiles, cette méthode devrait permettre de localiser péles et équateur a une époque géologique donnée et donc de savoir par rapport a l’actuel sil y a eu ou non deérive.
I] n’est pas str que, méme si on pouvait connaitre de facon précise et complete l’ensemble des fossiles ayant vécu pendant un étage donné, une telle méthode serait logiquement fondée. Il est évident comme I’a déja noté Mamet (1972) que le gradient de diversité est d’abord fonction de nos connaissances et que celui-ci sera plus élevée dans les régions de Ja Terre les mieux connues. De plus, il faut étre str que les fossiles d’un méme étage sont bien contemporains ; or un étage, surtout au Primaire, peut étre trés long; il faudrait restreindre la méthode a une seule assise mais la stratigraphie n’est pas en mesure actuelle- ment de préciser a distance les synchronismes a cette échelle (c’est la restriction temporelle de Mamet [1972]). De plus il est bien clair que le témoignage des
DERIVE DES CONTINENTS ET PALEONTOLOGIE 249
fossiles n’est que partiel et que rien ne permet de supposer que les gradients de diversité des fossiles d’une méme assise sont bien proportionnels a ceux des organismes qui vivaient effectivement lors de la formation de l’assise. Mais surtout la méthode suppose qu’il y ait toujours eu sur la terre une zonation climatique du type actuel, c’est-a-dire liée a des calottes glaciaires sur les poles: or les glaciations ne sont pas connues pendant le Secondaire et le ‘Tertiaire par exemple et la distribution des étres vivants en fonction de la latitude n’est peut-étre caractéristique que du Quaternaire; il n’est pas str que le principe des causes actuelles s’applique effectivement ici (Olson 1972, Schwarzbach 1963). Enfin comme le souligne justement Mamet: ‘Les variations de tem- perature de l’eau de mer en un point donné étant bien plus influencées par la bathymetrie que par la latitude, nous devons nous borner a comparer des biota de méme bathyméetrie.’
Dans une communication intéressante au Congres de Montréal, M. Mamet a montreé que les microflores algales du Viséen (couches de passage du Viséen moyen au Viséen supérieur) se groupaient d’une facon similaire a la distribu- tion actuelle des températures des eaux de l’hémisphere nord; cette distri- bution semble donc en désaccord avec la plupart des reconstitutions paléo- géographiques admettant des mouvements importants des continents par rap- port a leur latitude actuelle. Cette conclusion (d’ailleurs beaucoup plus nuancée dans la conclusion de cet article que dans le résumé) est une interprétation d’une distribution microfloristique remarquable mais cependant trop partielle pour permettre d’en tirer des déductions sires; ce n’est que si on avait affaire a des répartitions variables suivant les latitudes sur toute la terre et pour des groupes différents qu’une conclusion ‘fixiste’ pourrait étre considérée comme valable. Rappelons 4 ce propos que le Carbonifére inférieur est considéré classiquement par les géologues comme une période de climat chaud et unt- forme (Schwarzbach 1963), ce qui n’est pas conforme 4 l’observation de B. Mamet et semblerait plutét indiquer que cette distribution de la microflore nest pas la conséquence des latitudes, mais d’autres causes telles que par exemple l’existence d’un courant froid local, de variations de salinité (?), etc. De plus, il est évident que, dans un tel exemple, l’objection du pourcentage de fossilisation joue et que cette raréfaction peut n’étre due qu’a une insuffisance de témoins fossiles. Il nous parait, dans ces conditions, nettement abusif de conclure avec Teichert que les gradients de diversité spécifique sont symétriques par rapport a |’Equateur actuel; nous ne croyons pas d’ailleurs l’espéce sufh- samment bien définie en Paléontologie (ce n’est souvent guére plus qu’un numéro de catalogue de collection) pour qu’une telle assertion soit admissible. On doit noter aussi que des fossiles identiques surtout sensibles aux variations de température peuvent vivre 4 l’Equateur, en profondeur, et, en surface, dans les zones actuellement tempérées, et ainsi que certains genres se retrouvent dans les eaux superficielles au Nord et au Sud de l’Equateur: c’est le phénoméne de la bipolarité; la comparaison des fossiles ne peut donc avoir de sens que dans des conditions de faciés comparables, mais celles-ci doivent étre souvent difficiles
250 ANNALS OF THE SOUTH AFRICAN MUSEUM
a rétablir. Pour cette raison, comme par suite, des autres arguments cités ci-dessus, la méthode des gradients de diversité nous parait inapplicable aux organismes fossiles.
Les Vertébrés terrestres nous paraissent seuls des fossiles de choix pour juger de la valeur de V’hypothése de la dérive des continents; les Vertébrés marins, au contraire, sont en général peu caractéristiques d’une région: certains genres de Poissons sont pratiquement répartis dans toutes ies mers a une méme époque: les genres Boreosomus, Perleidus, Australosomus sont au ‘Trias inferieur connus au Groenland, au Spitsberg, en Angola, a Madagascar etc. ‘Toutefois les Vertébrés marins des mers €picontinentales peuvent étre trés localisés: c’est le cas des Placodontes. De méme il est probable que la répartition géogra- phique des Reptiles volants, des Oiseaux et des Cheiroptéres ne peut guére fournir, au sujet du probleme de la dérive, des renseignements faciles a inter- préter. Mais en raison de |’évolution qui est beaucoup plus rapide en général chez les Vertébrés que chez les Invertébrés, les Vertébrés terrestres nous permettent d’affirmer l’existence de phénomenes de dérive. Ainsi, si nous observons dans deux continents aujourd’ hui séparés deux faunes de Vertébrés ayant méme composition globale dans deux formations correspondantes, puis si, dans les formations plus élevées, nous trouvons des faunes de Vertebrés de plus en plus distinctes dans les assises plus jeunes, on doit en conclure que les deux masses continentales ont été continues et se sont ultérieurement séparees, probablement par dérive des continents puisque c’est le seul processus de séparation en faveur duquel nous ayons des arguments géologiques et géo- physiques. Un des exemples les plus classiques est la parenté des faunes entre Afrique du Sud et Amérique du Sud. Rappelons par exemple, et sans entrer dans les détails, que le Mésosaure n’est connu qu’en Afrique du Sud (formation de Dwyka) et au Brésil. De méme, les fouilles récentes de Bonaparte et de Romer ont montré que l’on retrouvait au Trias inférieur non seulement les mémes familles de Reptiles mais les mémes genres. Les genres sud-africains Cynognathus et Kannemeyeria sont connus a la fois en Argentine et dans le Karroo; un Cynodonte en Argentine (Colbertosaurus) est, comme le genre Diademodon d’Afrique méridionale, un Gomphodonte. Au Trias moyen, les ressemblances sont encore trés marquées (présence de Gomphodontes évolués et de Rhyncho- céphales des deux cétés de |’Atlantique).
Les travaux récents de Colbert concernant |’Antarctide ont aussi mis en évidence l’existence de relations fauniques entre l’Afrique du Sud et l’Antarc- tide: on sait que ce savant a montré la présence dans ce continent d’un Lystro- saure (qui appartiendrait méme, selon Colbert, a l’espéce L. murray, comme en Afrique du Sud), d’autres Thérapsidés, de petits Thériodontes et d’un Labyrinthodonte; c’est une faune qui rappelle celle du Karroo et le fait est d’autant plus remarquable qu’aujourd’hui |’Antarctide est trés isolée de tous les autres continents.
De toute fagon, Piveteau a montré depuis longtemps déja, quand il a fait connaitre la présence d’un Dicynodonte en Indochine, que le continent de
DERIVE DES CONTINENTS ET PALEONTOLOGIE 251
Gondwana ne devait pas en tout cas étre considéré comme isolé; des migrations ont du étre possibles a partir de Gondwana et c’est la méme explication que lon peut donner de la présence de Lystrosaurus accompagné de restes de grands Thériodontes au Turkestan chinois (Sinkiang; formation de Tunghungshan). I] est certain que, comme groupes, les Thériodontes et les Dicynodontes ont eu une repartition quasi-mondiale; on en connait classiquement en Europe et un Dicynodonte, proche de Kannemeyeria, vient d’étre découvert en Afrique du Nord. Les Thérapsidés sont surtout abondants en Russie, en dehors de l'Afrique du Sud. Ce n’est donc que la présence de genres ou d’espéces, identiques ou voisins, qui peut étre interprétée comme conséquence de la dérive et encore peut-il y avoir des exceptions. : De méme en ce qui concerne l’hémisphére nord, la présence des Pélyco- | saures exclusivement aux Etats-Unis, en Europe occidentale et en URSS, est ) un argument fort en faveur de l’existence de la Laurasie au Permien inférieur. L’identité spécifique de deux Vertébrés fossiles d’un méme genre dans des continents différents mais autrefois réunis doit cependant étre trés rare; les | Vertébrés terrestres semblent en effet plus dépendants morphologiquement du l milieu que les Vertébrés marins; ainsi il est rare de trouver des Stegocéphales de mémes genres dans des régions différentes du monde, comme si ces animaux ! étaient particuliérement sensibles aux conditions locales et subissaient facile- | ment la ségrégation géographique. Enfin il serait absurde d’essayer d’utiliser des gradients de diversité pour les Vertébrés terrestres, car les gisements sont si rares et si localisés que de tels gradients n’ont certainement aucun sens. Certains Vertébrés terrestres de petite taille comme les Rongeurs, certains | Singes etc... . peuvent avoir été transportés sur des radeaux naturels par des courants (‘islands hoppers’); ces Vertébrés sont évidemment sans intérét pour mettre en évidence des phénoménes de derive.
L’étude du Karroo malgache n’apparait pas favorable a l’existence d’un | continent Afrique-Madagascar—Indes; en effet il n’y a pas d’especes communes de Stégocéphales a Madagascar et en Afrique du Sud (du moins pas d’espéces certaines) et pour l’instant aucun Reptile mammalien n’est connu a Mada- | gascar ; les faunes de Poissons d’eau douce sont aussi différentes; le rapproche- | ment avec l’Angola est plus net (Lehman ¢ét al. 1959). D’aprés les schémas | paléogéographiques des mobilistes, Madagascar était au Permo-Trias plus au [ N qu’aujourd’hui, étant coincée entre |’Afrique et Inde environ au niveau | |
de la Somalie ex-italienne; mais ces différences de faune avec |’Afrique du Sud peuvent, dans le cas de Madagascar, étre la conséquence de circonstances fortuites: les formations sédimentaires d’origine continentale sont relativement \ peu développées a Madagascar; les Stégocéphales continentaux (Benthosu- chides, Rhinésuchides) y sont connus dans des formations marines (Lehman 1961, 1966). Quant aux Thérapsides il est possible que l’on ne les ait pas encore découverts. Cet exemple illustre le fait que seules les présences de Vertébrés fossiles peuvent servir d’argument valable, mais que les absences sont d’interpré- tation dangereuse. Les découvertes toute récentes d’un Paréiasaure et d’un
252 ANNALS OF THE SOUTH AFRICAN MUSEUM
Cotylosaure au Niger, d’un Dicynodonte au Maroc montrent que l’inventaire paléontologique mondial est loin d’étre achevé.
Enfin il est clair que la distribution des Vertébrés terrestres, si elle dépend de conditions géographiques, est aussi fonction de nombreux autres facteurs; la mer n’est pas la seule barriére écologique possible. Il ne faut donc pas s’attendre a ce que les données concernant les paléofaunes terrestres de Verte- brés traduisent de facon fidele la disposition et les rapports des anciens continents.
REFERENCES
Barrett, P. J., BAILLIE, R. J. & CoLBert, E. H. 1968. Triassic amphibian from Antarctica. Science 161: 460-462.
BarTHOLOMAI, A. & Howis, A. 1970. Vertebrate fauna from the Lower Trias of Australia. Nature, Lond. 225: 1063.
BonaAPARTE, J. F. 1972. Annotated list of South American Triassic tetrapods. Jn: INTERNATIONAL UNION OF GEOLOGICAL SCIENCES. Commission on Stratigraphy. Subcommission on Gondwana Strati- graphy and Palaeontology. Gondwana Symposium, 2nd, South Africa, 1970. Proceedings and papers: 665-682. Pretoria: C.S.I.R.
CoLBerT, E. H. 1972. Antarctic Gondwana tetrapods. In: INTERNATIONAL UNION OF GEOLOGICAL SCIENCES. Commission on Stratigraphy. Subcommission on Gondwana Stratigraphy and Palaeontolog y. Gondwana Symposium, 2nd, South Africa, 1970. Proceedings and papers: 659-664. Pretoria: C.S.1.R.
Cou.Lomp, J. 1969. L’ expansion des fonds océaniques et la dérive des continents. Paris: Presses Universi- taires de France.
EKMAN, 8S. 1953. <oogeography of the sea. London: Sidgwick & Jackson.
Ex.uiott, D. H., Coisert, E. H., BREED, W. J., JENSEN, J. A. & PoweE.t, J. S. 1970. Triassic tetrapods from Antarctica: evidence for continental drift. Science 169: 1197-1201.
EMBERGER, L. 1968. Plantes fossiles dans leurs rapports avec les végétaux vivants. Paris: Masson.
Furon, R. 1941. La paléogéographie.. Essai sur V évolution des continents et des océans. Paris: Payot.
LEHMAN, J.-P. 1961. Les Stégocéphales du Trias de Madagascar. Annis Paléont. 477: 109-154.
LEHMAN, J.-P. 1966. Nouveaux Stégocéphales de Madagascar. Annis Paléont. (Vert.) 52: 115-139.
LEHMAN, J.-P., CHATEAU, C., Laurain, M. & NaucueE, M. 1959. Paléontologie de Madagascar. XXVIII. Les poissons de la Sakamena moyenne. Annls Paléont. 45: 175-219.
Mamet, B. 1972. Un essai de reconstitution paléoclimatique basé sur les microflores algaires du Viséen. Int. geol. Congr. 24 (Montreal) (sect. 7): 282-297.
Otson, E. C. 1972. The habitat: climatic change and its influence on life and habitat Jn: INTER- NATIONAL ENCYCLOPAEDIA OF FOOD AND NUTRITION [1.E.F.N.]. 18. Biology of nutrition; ed. by R. N. Fiennes: 267-305. Oxford: Pergamon Press.
PIVETEAU, J. 1938. Un Thérapsidé d’Indochine. Remarques sur la notion de continent de Gondwana. Annls Paléont. 27: 139-152.
Romer, A. S. 1968. Fossils and Gondwanaland. Proc. Am. phil. Soc. 112: 335-343-
Romer, A. S. 1972. Tetrapod vertebrates and Gondwanaland. Jn: INTERNATIONAL UNION OF GEOLOGICAL SCIENCES. Commission on Stratigraphy. Subcommission on Gondwana Stratigraphy and Palaeontology. Gondwana Symposium, 2nd, South Africa, 1970. Proceedings and papers: 111-124. Pretoria: C.S.I.R.
SCHWARZBACH, M. 1963. Climates of the past. New York: Van Nostrand.
Taruinc, D. H. & Runcorn, S. K. 1973. Implications of continental drift to the earth sciences. London, New York: Academic Press. ([Congress proceedings] NATO Advanced Study Institute, April 1972, University of Newcastle upon Tyne.)
Taruinc, D. H. & Tartine, M. P. 1971. Continental drift. London: Bell.
TEICHERT, C. & MeEyeruorF, A. A. 1972. Continental drift and the marine environment. Int. geol. Congr. 24. (Montreal): 339-349.
—
DERIVE DES CONTINENTS ET PALEONTOLOGIE 253
STEHLI, F. G. 1968. Taxonomic diversity gradients in pole location: the recent model. Jn: DRAKE, E. T., ed. Evolution and environment; a symposium ...: 163-227. New Haven, London: Yale University Press.
Youne, C. C. 1935. On two skeletons of Dicynodontia from Sinkiang. Bull. geol. Soc. China
14: 483-517.
if = - = St =
|
oh
irs a ot ad bs a =
aria ys:
i]
eS
ei
SSS
= -
ee ey Wid
INSTRUCTIONS TO AUTHORS
Based on
CONFERENCE OF BIOLOGICAL EDITORS, COMMITTEE ON FORM AND STYLE. 1960. Style manual for biological journals. Washington: American Institute of Biological Sciences.
MANUSCRIPT
To be typewritten, double spaced, with good margins, arranged in the following order: (1) Heading, consisting of informative but brief title, name(s) of author(s), address(es) of author(s), number of illustrations (figures, enumerated maps and tables) in the article. (2) Contents. (3) The main text, divided into principal divisions with major headings; sub- headings to be used sparingly and enumeration of headings to be avoided. (4) Summary. (5) Acknowledgements. (6) References, as below.
Figure captions and tables to be on separate sheets.
ILLUSTRATIONS
To be reducible to 12 cm x 18 cm (19 cm including caption). A metric scale to appear with all photographs.
All illustrations to be termed figures (plates are not printed; half-tones will appear in their proper place in the text), with arabic numbering; items of composite figures to be designated by capital letters (A, B, C etc.).
REFERENCES
Harvard system (name and year) to be used: author’s name and year of publication given in text; full references at the end of the article, arranged alphabetically by names, chronologi- cally within each name, with suffixes a, b, etc. to the year for more than one paper by the same author in that year.
For books give title in italics, edition, volume number, place of publication, publisher.
For journal articles give title of article, title of journal in italics (abbreviated according to the World list of scientific periodicals. 4th ed. London: Butterworths, 1963), series in parentheses, volume number, part number (only if independently paged) in parentheses, pagination.
Examples (note capitalization and punctuation)
BULLOUGH, W. S. 1960. Practical invertebrate anatomy. 2nd ed. London: Macmillan.
FIscHER, P.-H. 1948. Données sur la résistance et de le vitalité des mollusques. J. Conch., Paris 88: 100-140.
FISCHER, P.-H., DuvAL, M. & Rarry, A. 1933. Etudes sur les échanges respiratoires des littorines. Archs Zool. exp. gén. 74: 627-634.
Konn, A. J. 1960a. Ecological notes on Conus (Mollusca: Gastropoda) in the Trincomalee region of Ceylon. Ann. Mag. nat. Hist. (13) 2: 309-320.
Konan, A. J. 19605. Spawning behaviour, egg masses and larval development in Conus from the Indian Ocean. Bull. Bingham oceanogr. Coll. 17 (4): 1-51.
THIELE, J. 1910. Mollusca: B. Polyplacophora, Gastropoda marina, Bivalvia. In: SCHULTZE, L. Zoologische und anthropologische Ergebnisse einer Forschungsreise im westlichen und zentralen Siid-Afrika 4: 269-270. Jena: Fischer. Denkschr. med.-naturw. Ges. Jena 16: 269-270.
ZOOLOGICAL NOMENCLATURE
To be governed by the rulings of the latest International code of zoological nomenclature issued by the International Trust for Zoological Nomenclature (particularly articles 22 and 51). The Harvard system of reference to be used in the synonymy lists, with the full references incorporated in the list at the end of the article, and not given in contracted form in the synonymy list.
Example Scalaria coronata Lamarck, 1816: pl. 451, figs 5 a, b; Liste: 11. Turton, 1932: 80.
L. D. BOONSTRA COMMEMORATIVE VOLUME
| ‘ee, ; ' : a Y). oh : JIYVvysgi 47 LIBRARI ES SMITHSONIAN _INSTITUTION NOILALILSNI_NVINOSHLIWS _Saluye ” = me = ” \S . — (Sete Fi a. ay jas = roNY oO | n vA 5 eA aa S = S Baca = beh |e cal —- igi acy NOILALIESNI NVINOSHLIWS _ My LIBRARI ES_ SMITHSONIAN INSTITU z en t vie, ; O w ee w oe o , : : 0 . . F ose re << Wr _ es * — LI B RARI ES SMITHSONIAN INSTITUTION NOILONLILSNI. NVINOSHLIWS S43 lYVe hy = ae = < ff Z = z 5 z 4 V7, © Wer n” oO ag Hy oO < le EB OWN 8 t me "2 8 we... ” Zz ” ml NOILALILSNI NVINGSHLUNS 33 lyYVvuya I1'1 LIBRARIES SMITHSONIAN INSTITU) _ 4 4 es ty oc a _ oc — | Uinoents “2 oO . ss oO ca. 5 / a Fel tere ves AF Ss : 1YVvYA 17 LIBRARIES SMITHSONIAN INSTITUTION NOLLNLILSNI NVINOSHLINS S3IYVWY a ae i dupe = w 2 aes ° is SE ° > 0] Le Ww -~ xy RS 3 = Ae = ; =| > bk , > | ce > % ~~ ro a “ty m ” m ” on i | ” = n me oN) NOILALILSNI NVINOSHLINS S31uYVYaIT LIBRARIES SMITHSONIAN INSTITU ” & n z ee 2) Zz ised = < = at ~~ = < ay = ra 4 ze SSN =i Zz . op) D Dy 2 WS F o Ws : : oy ES iG? > = > = 4 ae Ss Pad Ww) = Y) = Ps Tp LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSNI S3I1Y¥VvY Zz et Z if = ” : i : : ie : = a = ‘. = . — co = _ co eo) a Ye: oO _ oO at z aul ere val Zz *) aie NOILNLILSNI_NVINOSHLINS S3!1YvVYgIT LIBRARIES SMITHSONIAN z iN z ro z ae & —e s — ies) — w — E 2 : : E 2 om > - > cs “<, > a E a ris ae, wn n re wn * wo = a z ue ie ee hie Lavdeet OB RAR HES SMITHSONIAN INSTITUTION NOILOLILSNI_ NVINOSHLINS SaIlyvy & x z Z % Z < = | S \ = S = “ep a eu ” NK , L v) . @B oD wn - Ws © ae a ote Oo NAN \ SE ee Z = = TITUTION NOME SN NVINOSHDUAS LIBRARIES SMITHSONIAN ng z > ” Sey 5D) a GTO a . GTP 4 Ket Do) ~ SM) = GEA 5 = Be 2) 5 GAMA =
> Be < =—~ =~ — — * aig ~— es =< ie 4 Ww Zz 9) ee ” 1 MBRARIES SMITHSONIAN INSTITUTION NOILALILSNI ROSH LS — Ww > = ane fr Ss = .o “ oa 8 a < eS < a <x ip . fea) = 28) = a8) Pa) ‘OK i) Ge oO a oO . — 2 tt = feed been a al as | ALILSN! NVINOSHLINS S3IY¥VYSIT LIBRARIES SMITHSONIAN zi ze. z a z Boks 6 WKastituy . CO rae ; i > wo = w — /? 18 9) ce 2 Yy = 2 5 Ae ce — & Po] 2" YS > AJ Ae “4 A se 4 ~ a ata NY am a ae ee Ly (~ BR wase m S\ SS sl m pl m 4 2 nL ec eS = n = w . RARIES SMITHSONIAN INSTITUTION NOILALILSNI NYINOSHLIWS z i = raja w a 7) se = Se = < = INQ 2 é 3 , 8 2 Nd) S ” _ Wn sa = So iE = = =, re Z > =_ewe-. 4, eae br 2 a 2 a 2 Ml _NVINOSHLIWS SMITHSONIAN | Z = “ eat 22) aa uJ Vg \ = af = Yin, 4 ) con = o hyp ior some : = S < <A < hg ae | / — w e& “a g Se €. S 4 = af wasn oO is = a ¥ = £ = - 2 —_ z RARIES SMITHSONIAN INSTITUTION NOILNILILSNI SN = = E z aAHSON = 2 Y L D <a : — 4 £3 = as a = 7 = : = Ey TO > = Pani re 2 * Oo = i . 2 INLILSN!I NVINOSHLINS S3SIYVYGIT LIBRARIES SMITHSONIAN Ee ia w Zz oats wW = BAS fate = +e ye = < a: ; 2a 2 : A 4 a Tree D WY™ GD a d \ fe) jp * Z ES Z : Yj a D> eg a 2 nae: ee i RARIES SMITHSONIAN INSTITUTION NOILALILSNI NVINOSHLIWS c2 Sito 2 ts 3 ‘ es SN a ia ” NS m ; ya = a = XS oc % <x < c = - = = mw Ca fae) = ea) = fe: O ~ @) =a at z a a ee NVINOSHLINS S3/IY¥VYgIT LIBRARIES SMITHSONIAN to = ig = | E: Se = ow = Yip OD 2 \ NG a © GYtly > ——_ ES ‘iw = - bE > 2 WAV = 2 = G? <i = = Ww ” = uw * 2a ie) i a 7= ap) a Ww BRARIES SMITHSONIAN INSTITUTION NOILNLILSNI NVINOSHLIWS 4 WY ae n z ” | = SO :. = < = : : : Wi: : So 7) a : > o Z = = ‘ = = Pas = - >" = > siti — : “” Gi a wm 2 amVINOSHLINS S3IYWVYE!IT LIBRARIES SMITHSONIAN SR, ) w > w~ > /\} im “ 4 20 No? os es) Aa 4 fs ie at <4 = wey : ae Kain Sa a pe ee
INSTITUTION NOILN
a’
LIBRARIE
NV
saiuvugin
NOILNLILSNI
INSTITUTION NOILNLILS?
INSTITUTION
S3JIYVYsIT
SMITHSONIAN
Saluvudil LIBRARIE
SatYvVYdIT LIBRARIES
INSTITUTION
NOILALILS it:
Ol “G “iy
LIBRARII
NVINOSHLIWS
NOILNLILSNI
INSTITUTION NOILNLILS
INSTITUTION
SMITHSONIAN
INSTITUTION NOILOLILS
ARIES
i it oe i bya 1 “mer 8 Dh 6 es # ‘ « ‘ é ee ee ee GAG nae ‘ ( j etek yak ef de ve fie mbt MAO ? , wea i ' “on ‘ Peg rah bain werd, whe ge Ps od pavhope Holey ie. Aig a eis adler a Teme petted of gh gs A P r ding tg: ok ve 4 Prati ae Ch BOS Me A Ped ‘ iggy cere ot 1 i at. re % Wet Pada ted td re onl 8 ge ty rok? ig 4 P & Uh Rha ‘ ( e ment 4 " 4d eh eh ae he’ sf , Beh Oe gain a noe i O'why je ' wh if Ot wht AE jst era ee y gh e& P Ahn Dialed ey rik | aia abe ‘ 4 ‘ te hee oer eat t4 Ae 1 Gal di ‘ wo ; At” Bim hed hh eo oye of wit “ teat ee age coh & Ga 4 a 8 i ‘ l Vie iat phd Yee y mae ca ‘ re aera eat ‘ ‘ 0 @ Gee ie te pee a a ag A eas eh puny pf aca th CR the % od 1 wee ‘ ¢ Ee a ¢ ’ a4 hy wy a% "Aa Te ee 1 ‘ ‘ , ' | A pi gate vant¢ hohe godt “a 4 A hsin t fou Hong we ‘ of othe oe A ve rece | ’ Sietey sa , P P car) ‘ Hine ! ' ' As he oon 1 wna ets? UA AL ot Mot ke Ge 1 Bike Maw CF ne rihue F irae oti fil ‘ COMM Ge who 7 ‘ 440 bay 4 Sona Lg HEM TE «4a gad id ‘ % 4@ dik tee aad ge MRR vt oY shee a rhc , f Cia. dutied ‘ ‘ 4 ‘ ‘ ‘ 4 wed ag a i , 4 Poe ne ‘ wh AP Ooh wef 4 easy »F 5 y yy : 44 Y at nf ¢ a Hone ad ‘ vay. , 7 4 /( ae K'$ hed wat ie » “ ie ‘ee eo F A bing . rary, a 3 ‘ ' erie i RAE BS ne a qh (od Ahn al eet a 4 ‘ ‘ ; ‘ wy oun ara ae for a ¢ fie Bly Re 7 ae og Fie! ’ ‘ D , § i" Garry ii vat et ae aNd 3 Auth: 96h le 63, 22 Fed a tyea ined Ce A j na py A ¢ vf % gts + he F ae it WSF oy) ok Ed A Bi ua. Poe ae A dit ites lati 5 4 ai $ Kelly é hip os O38 ek on ee ald tains fh od vith y of wie be ‘ 2 A ae teh ret ae ¢ oer AT, Hind a? ve Pay: t Harn’ ee wer ‘ ' ne pai bs 6 oth on fom bys HR 6 Be oe a it os aes / shed ny ‘ ’ ‘ , ‘ V7 ae 4 Mee ge on Brey ery, Gite otek ar F ORR Fae ed ree wy ave gf ' ‘ ‘ ‘ , ie Mas dit Ae A eA ‘ ti Aged Mae Aw mi oa wie ah ‘ Wot ’ sheave | t ’ a 4 ‘ A aw . dole go Kilt ye ¢ Ass 3 ar ’ ri a bet ae ra | ' , t 4 “at i tv ne iF fd ott oy ar a, Adi i'n ‘ t tas 1 r et it p BULA se i 1. et pone Pa: Lh 4-4 ria noe 5 Whee F er ae ‘ rt bike £0 ow a wo Hie) fon U f y F ‘ 5 . : eA he jeg OL Big ey Fi W-eue : ar an ‘ Ae eon wrt ‘ i i¢ “os J hn 4 th 4 9 A 5% , apes ONG Swf 4 i ws * , ‘ eka ‘F ea hk “ os 4 ra shige - ‘ “ i ‘ “ we De ite: i ’ é ne ‘ on 1A yee ae wt of Stic he of, oF % grid % A » a hay ‘ ¥ a ‘ oath oa ‘ 106 ‘ , fy! nis Cry Oey a es a ee 1 ake I Mb. a Bigs aed i eek wee / 1 ed ane Aen ye ot, ol ae ae gh “ie F et x2 Va ects Oa Pe ' F Tard Ep 4 ‘ at . “'t “ ve act Uy wad \ Pt hi aBt lek é fac y Sf s4 ae ae na | Work | in Pike , ; ou 4 het ’ «hae ene eb FAL eat ’ ; eid 4 BE REY ' 4 , alts he ved 4 * . re seh 1 toa Be t sot ' tts < 4 Cr £ i fs ' : e y , ae 1 , * wear ad i, MM ‘ i e y i oie ay a i Rat OF ‘ ft “ ft ft gf grt se ot é a fi ; . eben phd al At Aah arg n tar 4 ‘ ‘ ‘ ‘ ‘ ‘ , alep ee “ Gon 7 Piyry 3 F 5 4 , re) Siete i dia Rs soo « + fat ie , me 4 i ' 4 , ! X rae) pv : , “yi en poe 1 P A wir J bs ‘ v . t : vo ‘ J F , re. cet Bee Wie 50 / * é fer s x 4 ja , 1 he es ie do wrbed ; She Z 7 ‘ ! 4 ed A i ‘ .) hs ae ‘ Z é ’ aiden deg, Zep ue nn t {3 } is i neo r yar ’ ? , t Fs ve F we a? Np Gad a wy on ‘ a ‘> ; ‘ / ‘ : : f rT Le Pra SAP AR he PT Ien eca tig a Z its Ait ie Mig NR i PugtS dc zi ¢ ‘ , ‘ y a iy ‘ Oi po ih idl fed it en oe AWS htt ! ' t : é Ps ‘ Z j bal ee oe gd iho , 7 # | ‘ Hy , Go i Pane ! ‘ y ne rod , i . ’ en hive rh ' P Oo # ; Fi aya ¢ A ont ‘i ra “ne nt ne a wit F. tha ; ’ F “ oad a tet ra oat ae Ait r ts ant at aa] J J ‘ ‘ / i i ¢ “ a tf ‘ t FAS He he Re &§ ’ . % , , “ ’ t é doa rine ‘ pra bR Re GAl ep hl eet be ‘ ‘ rr ae A ny tan Oe ti ’ n dos. nena rn f, bo 1 thet ' me tent Crd i 4 ns ; i a f« fae 4) suits tw ie 2 (4 ee a Caen / us F i ‘ci a F ‘ ‘ det * bd rae ea) ‘ rt, a Le ed Th ’ 4 y ier 5 ‘ s My 1 gk ‘ $Y ae Kd) re er % 5 ar nei in an a ” H ’ f ‘ r rahe WF be NOR gh ¢ ‘ ‘ 4 i . t ‘ ¢ ed a4 Aa be ft a Bt Loti Oo oy a ' te os hon ' ' f us 7 ay ee Ft Neate i Sat TEE a wor i eer r 4 t j A y Ria 632 Joust Peg la yA vb alee fe whet ‘ F , g wan He ‘ fp 8 se > ad? v4 e Me a 4 YU 4 Th ‘ gies q 4 ¢ rare ‘ Ayton nh ah a c wantin , : ra } j ¥ . od , 4 noe n . 9 7 . 7 wal wer Pah Avabas et 7 , ah Ay : b ; i a inl & Pa i Pe 4 ” Fy ) b 4 § i 5 Mei at hf } Se ood Me wh, Wet oe . ., A . . Wa of wart b Yh vee 7 tei uis ie . a ae f + ’ : re ” , _ . af 4 r t { , ee b 4 4 ; yy ‘ . ret i ae . , r . F t ¥ 5 ’ » > : i ‘ wal hte : : 2th , i ‘ ‘ meee gong gh uy : en 4.3 . ‘ ‘ B fa s r 3 : , Pind, ¢ R : . : . * 4 x ‘ , 4 rh ’ 4 ne sy : 4 ; > i " # t pay ‘ . f Vea ¥ B® Bet WSbag? be, " \ & ; L 6 ' 4 ha ‘ sie yin ok fee eo ‘ a . ‘ A : ; es Ye ay Pn crea a8 Veni tip) qsGey “ Mak ie 3 ‘ : : i ff top Wis yeas we ‘ ae ts 2 4 A iH ae F ‘ 4 Br oat aioe \ ie | Wi aa a ‘ is © BS og Aas i Pir ‘ ! ‘ dy ai bt t aes she, et £43) 4 ‘ t Pea ae et Ye A ' “ae rere a tA MM 2 SAE tN, vt ete, , ‘ aah MS : 3 : ‘ We ‘ i Mee yy oh He test kd f : ref 1.5 \ H \ : ‘ Peas ante | ‘ : ; ‘ yt ‘ ue ‘ , : " . ON v if a : if (fit ‘ Pero ey : \ He A z {i ‘ Bt ‘ a r : : hous p " F ‘ ROR TON : f 2 te tz ue Os cL eo! at i 4 482) BLY \ ; . X tee Sy : ‘ : ‘ nha y Bra eth aa * ‘ Wes | bowed a gs . : D . . i i 1pkl eta eH fae . Vg A ‘ me pac! i en vy a ‘ V5 bore Res SUP ae ae | . ya it » 1 | Linn A BR 7 4 4 « ‘ , * 7 ‘) in fo Xs ys I K as vig + ey 1 ‘ wo sg hh % ited iui Aas or bt & v. a ee 4 ea 4 f . Se fen ' ‘ ‘ Lert " % 4 . c < ‘ t 4 w bow Y & : ee fe te 1d oe Ok wd ws oy TS ag ‘ ‘ : ‘ F : ‘4 +) o an TS74 Seaver) eee a “34 RR Sy 6 oT oth 9 CS: ‘eon Sato & ‘ 4 t¥ ' 5 he pas Ces Siw ‘ % 8.8 4 Ay xi: 4 yi rt os iy wht a ‘ ay Doyres 4 t ot ou S ‘ A SEV Awe RED uy’ POa hel, \ Abe 8 Pn it en ea Oe vat * ” , xs v : 7 AY 4 "1 ‘ r H ‘ AU Uy ey at ae Vly ‘ ews : : eer eae el ee OL ee ee Be Ae ag Hn Te R Ae yy rin . . 5 o ‘ vt nn . | a ND 4 4 er ret EN | 4 i A ran £ vy . \ No An ‘ i gt | oy » 4 “wf f ‘ ah 4 ‘ % a ( o, pst tem, sh ania gt ee ‘ so 1 yess ‘ h de ‘ ra ean) \ y meek te owe ‘ VAL ye ' : a 4 ak 1 | a Uae ‘ ' ; ae ae wo . ¥ 0 wey Myr ot “ak eas v4 " 1 . a \ \, ‘ ire U es ks a) ee > i , fan galeq gost PW esas \ yi Van Ye nt woes ota hy ve e34 eR EA ‘ Sen al ae ee OTT) a ear ek (at Pac Ok ra p 7 aioe ‘ ye t Ae : i. 5 if Gy VOT ae MEE pert aD inate ae ear Oe : \ owe "fy 44s 1 WM bait eS 1 GY Ps 1 op Bp 4 oe a \ ean} hu ek i ein gh WD glint Von teat Pathe Ga hae ha Mey 6 ; sl : SY i ek ton PS ie ree LAPS alte a fb ry ‘ gat Ca £4 ay tt ie en Oe pay waira ht un ny F ih oy 2 ot Py anes wi DP Ag yA | 6 ; ‘ a POE TET Y eae ey Te a Oe ine 4 ww |e 4 aN iu ¢ Na ‘ s ¢ ‘ ‘ i. pe Oy ‘ ene PTLeae AY Shr m vee ' : 5 . uae : 4 wit) & Vip “ f » A at A muy n PAN OP of 58 sloth f wee ty, tui Oo NRW Y i tf 1 og ha A { HPlae pk bts 4 bi af yO Ny a wey Pe t 28 EA pak ony aes dot CRA ay G80 ANS "the pA ey , : sae er int vey ? ti op ie ay fees ine eA ‘ ‘ iy v i ke 4 a iy a, tat § ve a bine \ Pay yew wre im, (meat ; sae # Ara Ces Te) SUED Ha RO ape ye he \ Pr i a | i r ivotg A \ ives 1 haw ‘ Hi, 44 i, Ws The whey yy re xP kM HS) Ni fol i ’ Oyy BN a a ae ee vee c ' * ow
SMITHSONIAN INSTITUTION L|
LN,
3 9088 01206 6239
: t ae s tig 22 : i pare ade s 1 ‘ ‘ 1 , ‘ iY net yo a baa : ‘ , ' (x ' ; fey : ‘ i ‘ . . ‘ athe : yer ve elie we yous f sd Ay oy ‘ oe af 5.5 iy ; P hk ey ¢. oe + % : ree ‘ x4 : a ; Mian ke i ‘ wid GLE cine : ,
oh bat or Am Ge Wk we °F ;
: a Bins es ‘ *
3 i ee : ; ‘ . 5 , ; F ho Le Ga be a ot 3 Pa ok cf va yt es : ee Sl ha wa ‘ . ee: / : ,
4 ; apy ah ‘ ai i : . keg ‘ a i rite oy a4 ae ' poms W weil's ‘ .
, hb 009 fal : $ ae. » , ' vis :
ie yt! ; ; 4 wit a ;
ite ‘
VENER cea ht sey ; ana i fas igs
bad why Nyt
rf i i Oey ah
0) :