W I
— CO ~ CO ~ t'’
an LIBRARIES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHilWS S3iaVHan LIBRAR ^ CO z: » ^ ^ ^ 5 ■'
! Ai .
CO
^VASt''
_ > \JVAsq>^ 2 " ^ XovosvJiX >■ 5 ■'
nON ‘^N0liniliSNI^NVIN0SHims‘^S3 I avaa n^LI BRAR lEs'^SMITHSONlAN INSTITUTION Noiinii.
— •- m — ^ 2 co_
H
/.V /- _
o
an LIBRARIES SMITHSONIAN INSTITUTION NOlidillSNI NVINOSHillNS S3iavaail LIBRAR z r- ^ ^ z: r- z
_ _
TIOn"^ NOliniliSNl-NVINOSHimS^^SB I ava a n~LI B RAR I ES^SMITHS0NIAN"|NSTITUTI0N^ NOlinil s .< ^ S ^Vx - — vx^ ' ^
> J> '/sr ^ X z 5 ^
d a LI B RAR I Es‘^SMITHSONIAN^INSTITUTlON‘^NOIiniliSN!_NVINOSHims‘^S3 iavaan_LIBRAF
Q Xi|(V_DC>' _ XiOiosf^ o ", — O
TION ^ N0lini!iSNI~*NVIN0SHiU^S^S3 lavaail^LIBRARI ES^ SMITHSONIAN^ INSTITUTION ^ NOlifUl
ian_LIBRARI ES '^SMITHSONIAN INSTITUTION'^NOIlDiliSNI ^NVINOSHiUNS^SB lavaail^LIBRAI
X ^
CO S o
TiON NoiiniiiSNi^NviNOSHims^^sa I ava a I n\i b rar i es'^smithsonian_institution ^ Noiini
$
IhOi^
lan libraries smithsonian institution NoiiniiiSNi nvinoshiii/ms saiavaan libra
rn
TION NOliniliSNI NVINOSHillMS S3iavaan U B RAR l ES^SMITHSONIAN mSTITUTlON^NOliUl
I I x^s I i ^
1S0NIAN INSTITUTION NOIiniliSNI NVINOSHiU^S S3ldVHan LIBRARIES SMITHSONIAN INSTT (/) z: co z: ,
> _
(f\ -Z 00 2 ^ ^ CO
JSHillMS S3IBVaan libraries SMITHSONIAN institution NOIiniliSNI NVINOSHimS S3IB
O XiivpitA^ I_ O '^'Vosv''>/ ?
Z _l z
HSONIAN INSTITUTION NOIiniliSNI NVINOSHillMS S3IHVaan LIBRARIES SMITHSONIAN INSTI
CO
osHims' S3 iavaan“LiBRARi es“'smithsonian“ institution ^NoiiniiiSNi nvinoshhins s3 i a
CO Z to 2 , c/5 Z to
? , , V- 5 1 y- -
i-iriPif
> ~'W' 2 > x^OiM^^ 2 2 xO\oC^ ^
Z t/5*Z CO*^ Z to z
3S0NIAN INSTITUTION NOIiniliSNI NVINOSHimS S3iavaan LIBRARIES SMITHSONIAN INSTI
- CO _
to
Q " _ '<^ua‘=^ Q x^Ai DX--^ O
DSH1IIAIS"'S3 lava a n~LI B RAR I ES^SMITHS0NIAN~’lNSTITUTI0N^N0linillSNl“'NVIN0SHlllNS_S3 I a
CO
^ x^tvAsv^ m ^ m xcfVQsv^jx ^
HS0NIAN~INSTITUTI0N^N0linillSNI~NVIN0SHilWS^S3 I ava a 11 ~LI B RAR I ES SMITHSONIAN INSTI
Z r CO Z CO Z > ^ Z
< '->9, S s s
o
S ' 1>F'’ >' 5 >■'
OSHlItMS^^Sa lavaan^LIBRARI Es'^SMITHSONIAN institution 'NOIinillSNI_NVINOSHlllA!S ’ S3 I a
~ ‘ CO 2:
HSONIAN INSTITUTION NOIiniliSNI NVINOSHIIWS S3iavaail LIBRARIES SMITHSONIAN INSTI
osHiiiNS S3iavaan libraries Smithsonian institution NoiiniiisNi nvinoshiiins S3ia
CO z CO z V, CO z ^ CO
VOLUME 15
Palaeontology
1972
PUBLISHED BY THE PALAEONTOLOGICAL ASSOCIATION LONDON
Dates of publication of parts of Volume 15
Part 1, pp. 1-185, pis. 1-38 Part 2, pp. 187-380, pis. 39-65 Part 3, pp. 379A-518, pis. 66-105 Part 4, pp. 519-693, pis. 106-132
February 1972 June 1972 September 1972 December 1972
THIS VOLUME EDITED BY N. F. HUGHES, GWYN THOMAS, ISLES STRACHAN, ROLAND GOLDRING, J. D. HUDSON, D. J. GOBBETT AND L. R. M. COCKS
Dates of publication of Special Papers in Palaeontology
Special Paper No. 1 Special Paper No. 2 Special Paper No. 3 Special Paper No. 4 Special Paper No. 5 Special Paper No. 6 Special Paper No. 7 Special Paper No. 8 Special Paper No. 9 Special Paper No. 10 Special Paper No. 11
21 June 1967 31 January 1968 7 October 1968 1 May 1969 17 October 1969 26 April 1970
6 August 1970
11 November 1971 11 December 1971 10 February 1972
7 December 1972
© The Palaeontological Association, 1972
Printed in Great Britain at the University Press, Oxford by Vivian Ridler Printer to the University
CONTENTS
Part Page
Ambrose, T., and Romano, M. New Upper Carboniferous Chelicerata (Arthropoda) from Somerset 4 569
Ash, S. R. Marcouia gen. nov., a problematical plant from the Late Triassic of the south-western U.S.A. 3 423
Late Triassic plants from the Chinie Formation in north-eastern Arizona 4 598
Baker, P. G. The development of the loop in the Jurassic brachiopod Zeilleria leckeiibyi 3 450
Banks, H. P. The stratigraphic occurrence of early land plants 2 365
Bate, R. H. Fossil and living Hemicypris (Ostracoda) from Lake Rudolf, Kenya 1 184
Phosphatized ostracods with appendages from the Lower Cretaceous of Brazil 3 379A
Bates, D. E. B. A new Devonian crinoid from Australia 2 326
Bishop, G. A. Moults of Dakoticancer overanus, an Upper Cretaceous crab from the Pierre Shale of South Dakota 4 631
Black, C. C. Review of fossil rodents from the Neogene Siwalik beds of India and Pakistan 2 238
Black, M. Crystal development in Discoasteraceae and Braarudosphaeraceae (plank- tonic algae) 3 476
Brett, D. W. Fossil wood of Platanns from the British Eocene 3 496
Brunton, C. H. C., and Mackinnon, D. L. The systematic position of the Jurassic brachiopod Cadomella 3 405
Clayton, G. Compression structures in the Lower Carboniferous miospore Dictyo-
triletes admirabilis Flayford 1 121
Cocks, L. R. M. The origin of the Silurian Clarkeia shelly fauna of South America and its extension to West Africa 4 623
Creber, G. T. Gymnospermous wood from the Kimmeridgian of East Sutherland and from the Sandringham Sands of Norfolk 4 655
Eager, R. M. C. Use of the pictograph 2 378
Elliott, G. F. Cretacicrusta gen. nov., a possible alga from the English Cretaceous 3 501
Trinocladus exoticus, a new dasycladacean alga from the Upper Cretaceous of
Borneo 4 619
Earrow, G. E. Periodicity structures in the bivalve shell: analysis of stunting in
Cerastoderma edide from the Burry inlet (South Wales) 1 61
Hancock, J. M., Kennedy, W. J., and Klaumann, H. Ammonites from the trans- gressive Cretaceous on the Rhenish Massif, Germany 3 445
Jackson, D. E., and Lenz, A. C. Monograptids from the Upper Silurian and Lower Devonian of Yukon Territory, Canada 4 579
Jago, j. B. Two new Cambrian trilobites from Tasmania 2 226
Jeletzky, j. a. Morphology and taxonomic status of the Jurassic belemnite "Rhopalo- teuthis' somaliensis Spath 1935 1 158
Kaueeman, E. G. Ptychodus predation upon a Cretaceous Inoceramus 3 439
Keen, M. C. The Sannoisian and some other Upper Palaeogene Ostracoda from north- west Europe 2 267
Kennedy, W. J. The affinities of Idiohamites ellipticoides Spath (Cretaceous ammon- oidea) 3 400
and Klinger, H. C. A Texanites-Spinaptychus association from the Upper
Cretaceous of Zululand 3 394
Hiatus concretions and hardground horizons in the Cretaceous of Zululand 4 539
CONTENTS
iv
Part Page
Kennedy W. J. See Hancock, J. M.
Klaumann, H. See Hancock, J. M.
Klinger, H. C. See Kennedy, W. J. (two papers)
Lane, P. D. New trilobites from the Silurian of north-east Greenland, with a note on trilobite faunas in pure limestones 2 336
Layman, M. See Taylor, J. D.
Lenz, a. C. See Jackson, D. E.
Lord, A. iVicherella and Gramannella, two new genera of Lower Jurassic Ostracoda from England 2 1 87
MACKINNON, D. L. See Brunton, C. H. C.
Matthews, S. C., Sadler, P. M., and Selwood, E. B. A Lower Carboniferous cono- dont fauna from Chillaton, south-west Devonshire 4 550
Mishra, V. P. See Sahni, A.
Morton, N. The Bajocian ammonite Dorsetensia in Skye, Scotland 3 504
Nichols, D. The water-vascular system in living and fossil echinoderms 4 519
Paul, C. R. C. Morphology and function of exothecal pore-structures in cystoids 1 I
Peel, J. S. Observations on some Lower Palaeozoic tremanotiform Bellerophontacea
(Gastropoda) from North America 3 412
Rigby, J. F. On Arberia White, and some related Lower Gondwana female fructifications 1 108
Romano, M. See Ambrose, T.
Sadler, P. M. See Matthews, S. C.
Sahni, A., and Mishra, V. P. A new species of Protocetus (Cetacea) from the Middle Eocene of Kutch, western India 3 490
Sellwood, B. W. Regional environmental changes across a Lower Jurassic stage- boundary in Britain 1 125
Selwood, E. B. See Matthews, S. C.
SoRAUF, J. E. Skeletal microstructure and microarchitecture in Scleractinia (Coelen- terata) 1 88
Taylor, J. D., and Layman, M. The mechanical properties of bivalve (Mollusca) shell structures 1 73
Thulborn, R. a. The post-cranial skeleton of the Triassic ornithischian dinosaur Fabrosaurus australis 1 29
Tozer, E. T. Observations on the shell structure of Triassic ammonoids 4 637
Tucker, M. E. See Van Straaten, P.
Van Straaten, P., and Tucker, M. E. The Upper Devonian Saltern Cove goniatite bed is an intraformational slump 3 430
Wade, M. Hydrozoa and Scyphozoa and other medusoids from the Precambrian Ediacara fauna. South Australia 2 197
WooTTON, R. Nymphs of Palaeodictyoptera (Insecta) from the Westphalian of England 4 662
Wright, A. D. The brachiopod Acanthocrania in the Ordovician of Wales 3 473
Notes for authors submitting papers for publication in Palaeontology 4 676
VOLUME 15 • PART 1
Palaeontology
FEBRUARY 1972
PUBLISHED BY THE PALAEONTOLOGICAL ASSOCIATION LONDON
THE PALAEONTOLOGICAL ASSOCIATION
The Association was founded in 1957 to further the study of palaeontology. It holds meetings and demonstrations, and publishes the quarterly journal Palaeontology and Special Papers in Palaeontology. Membership is open to individuals, institutions, libraries, etc., on payment of the appropriate annual subscription:
Institute membership ..... £10-00 (U.S. $26.00)
Ordinary membership £5-00 (U.S. $13.00)
Student membership £3-00 (U.S. $8.00)
There is no admission fee. Institute membership is only available by direct appli- cation, not through agents. Student members are persons receiving full-time instruc- tion at educational institutions recognized by the Council; on first applying for membership, they should obtain an application form from the Membership Treasurer. All subscriptions are due each January, and should be sent to the Membership Treasurer, Dr. A. J. Lloyd, Department of Geology, University College, Gower Street, London, W.C. 1, England.
COUNCIL 1971-2
President: Dr. W. S. McKerrow, Department of Geology, Oxford Vice-Presidents: Professor M. R. House, The University, Kingston upon Hull, Yorkshire Dr. Gwyn Thomas, Department of Geology, Imperial College, London, S.W. 7
Treasurer: Dr. J. M. Hancock, Department of Geology, King’s College, London,
W.C. 2
Membership Treasurer: Dr. A. J. Lloyd, Department of Geology, University College, Gower Street, London, W.C. 1
Secretary: Dr. W. D. I. Rolfe, Hunterian Museum, The University, Glasgow, W. 2
Editors
Mr. N. F. Hughes, Sedgwick Museum, Cambridge Dr. Isles Strachan, Department of Geology, The University, Birmingham 15 Dr. R. Goldring, Department of Geology, The University, Reading, Berks.
Dr. J. D. Hudson, Department of Geology, The University, Leicester Dr. D. J. Gobbett, Sedgwick Museum, Cambridge
Other members of Council
Dr. E. N. K. Clarkson, Edinburgh
Dr. L. R. M. Cocks, London
Dr. R. H. Cummings, Abergele
Dr. Julia Hubbard, London {co-opted)
Dr. W. J. Kennedy, Oxford
Mr. M. Mitchell, Leeds
Dr. Marjorie D. Muir, London
Dr. B. Owens, Leeds Dr A. D. Wright, Belfast Dr. W. H. C. Ramsbottom, Leeds Dr. Pamela L. Robinson, London Dr. E. P. F. Rose, London Dr. C. T. ScRUTTON, Newcastle Dr. V. G. Walmsley, Swansea
Overseas Representatives
Australia: Professor Dorothy Hill, Department of Geology, University of Queens- land, Brisbane
Canada: Dr. B. S. Norford, Institute of Sedimentary and Petroleum Geology, 3303- 33rd Street NW., Calgary, Alberta India: Professor M. R. Sahni, 98 The Mall, Lucknow (U.P.), India New Zealand: Dr. C. A. Fleming, New Zealand Geological Survey, P.O. Box 30368, Lower Hutt
West Indies and Central America: Mr. John B. Saunders, Geological Laboratory, Texaco Trinidad, Inc., Pointe-a-Pierre, Trinidad, West Indies Western U.S.A.: Professor J. Wyatt Durham, Department of Paleontology, Univer- sity of California, BerKeley 4, California
Eastern U.S. A.: Professor J. W. Wells, Department of Geology, Cornell University, Ithaca, New York
© The Palaeontological Association, 1972
MORPHOLOGY AND FUNCTION OF EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
by c. R. C. PAUL
Abstract. Humatirhombs, humatipores, and diplopores have external respiratory exchange surfaces. Their thecal canals open internally and body fluids flowed through them in life. Four types of humatirhomb are dis- tinguished on morphology and arrangement of canals. Raised and buried humatipores occur and diplopores may have had extensile podia in life.
All cystoid pore-structures were respiratory. Exothecal pore-structures were individually less efficient in exchange than endothecal (dichoporite) pore-structures. Their relative inefficiency is due to requirements of protection and is counteracted by their large number per theca. Cystoids with exothecal pore-structures attain great size. Less efficient pore-structures (humatirhombs, humatipores) have shorter stratigraphic ranges and become extinct before more efficient types (pectinirhombs, cryptorhombs, diplopores).
Recent echinoderms as a group lack a specialized circulatory system and utilize varied exchange surfaces as did cystoids. Efficient exchange surfaces must be thinner than 1-3 mm: cystoid exchange surfaces are 0 01- OT mm thick. Diplopores and humatipores may have been connected to an internal water vascular system but humatirhombs were not. Rhombifera probably had external radial water vessels but Diploporita lacked them. Some Rhombifera may have had both internal and external branches of the water vascular system. Classes Rhombifera and Diploporita are defined and cystoid classification is reviewed.
The cystoids constitute a heterogeneous grouping of primitive echinoderms which range from the basal Ordovician just into the Upper Devonian. The vital organs of cystoids (and other primitive echinoderms) were completely enclosed within a rigid cup or theca which provided them with protection. At the same time the theca restricted communication with the ambient sea water from which both food and oxygen neces- sary for life were obtained. The purpose of this paper is to show that three major types of pore-structure which occur in cystoids evolved in response to the respiratory ‘prob- lems’ created by the rigid theca. These pore-structures in cystoids, and by implication similar pore-structures in other primitive echinoderms, were effectively gills.
Traditionally cystoid pore-structures have been grouped into ‘diplopores’ and ‘pore- rhombs’ on morphological grounds. Functionally however division of all echinoderm pore-structures into endothecal and exothecal groups, where primary exchange from sea water to body fluids took place within and outside the theca respectively, is more appropriate (text-fig. 1). The morphology and function of endothecal (dichoporite) pore- structures in cystoids have been described (Paul 1968). The present paper considers exothecal pore-structures for which Hudson (1915, p. 166) originally proposed the term ‘exospires’.
Three basic types of exothecal pore-structures occur in cystoids : one type of rhomb and two types of dipore. Brief descriptions of their morphology were given in Paul (1968). In the next two sections the morphology of exothecal pore-structures in cystoids is described and they are analysed functionally as exchange structures. Since the most likely form of exchange is oxygen and carbon dioxide transfer a section on respiration
[Palaeontology, Vol. 15, Part 1, 1972, pp. 1-28, pis. 1-7.]
C 8472 B
2
PALAEONTOLOGY, VOLUME 15
in recent echinoderms follows. The last two sections deal with the water vascular system in cystoids and with the taxonomic and evolutionary implications of this study.
Acknowledgements. Thanks are due to the following for the loan of, or access to, specimens in their care: Dr. R. L. Batten, American Museum of Natural History (AMNH); Drs. R. P. S. Jefferies and H. G. Owen, British Museum, Natural History (BMNH); Drs. R. V. Melville and A. Rushton, Insti- tute of Geological Sciences, London (GSM); Dr. H. Mutvei, Naturhistoriska Riksmuseum, Stock- holm (RM); Mr. A. G. Brighton and Dr. C. L. Forbes, Sedgwick Museum, Cambridge (SM); and Dr. P. M. Kier and Mr. T. Phelan, United States National Museum (USNM); Dr. E. S. Richardson, Field Museum of Natural History, Chicago (FMNH).
Parts of this work were undertaken during the tenure of a Harkness Scholarship and a Natural Environment Research Council research scholarship at the Sedgwick Museum, Cambridge. Both are gratefully acknowledged. Dr. Jefferies kindly read the manuscript and suggested several improvements; however, I accept responsibility for all opinions expressed in the work.
TEXT-FIG. 1. Diagrammatic representations of endothecal (a) and exothecal (b) canals. In a sea-water flows through the canal and exchange takes place within the theca; in b body fluids flow in the canal and exchange takes place outside the theca. Thecal wall shown with vertical lines. In this and following diagrams the external medium is towards the top of the figure.
TEXT-FIG. 2. Simple (a) and compound (b) thecal canals. In a a single tangential canal (tc) connects a pair of perpendicular canals (pc); in b three tangential canals connect the perpendicular canals. Thecal canals open in internal pores (ip) in all exothecal pore-structures.
MORPHOLOGY OF EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
All cystoid pore-structures are composed of U-shaped thecal canals (text-fig. 2) with one or more connections (tangential canals) between the limbs of the U (perpendicular canals). The openings (thecal pores) of exothecal canals are internal. The three types of exothecal pore-structures in cystoids are humatirhombs, humatipores, and diplopores s.s. (Paul 1968, p. 700).
1. Humatirhombs (text-figs. 3a-d, Pis. 1-4)
Humatirhombs (humare: Lat. to bury) are composed of a set of thecal canals pores, fistula, Lat. a canal), all of which arise from pores on the inner surface of one
PAUL: EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
3
plate, pass through the plate, and cross a plate suture to pores on the internal surface of an adjacent plate (text-fig. 3). The pores are always simple and circular (PI. 4, fig. 2). The tangential canals may be single (simple fistulipores, text-fig. 2>a-b) or multiple (compound fistulipores, text-fig. Zc-d) and they lie either just below the external sur-
TEXT-FiG. 3. Four types of humatirhomb. a, simple humatirhombs with simple fistulipores (sf), b, complex humatirhombs with simple fistulipores, c, simple humatirhombs with compound fistulipores (cf), d, complex humatirhombs with compound fistulipores. In humatirhombs with simple fistulipores (a, b) the tangential canals (xc) are raised in ridges; humatirhombs with compound fistulipores (c, d) have tangential canals buried beneath plate surfaces. Complex humatirhombs (b, d) have both principal fistulipores (pf) and shorter intermediate fistulipores (if), pc = perpendicular canal.
face (Pis. 3-4) or in the crests of ridges on the external surface of the theca (PI. 1). Usually only simple fistulipores are associated with ridges (text-fig. 3a-b).
Four types of humatirhomb may be distinguished on the structure and arrangement of fistulipores. In simple humatirhombs all the fistulipores run the entire length of the rhomb from margin to margin (text-fig. 3a, c). Complex humatirhombs (text-fig. 3b, d) have additional shorter fistulipores within the intra-rhomb area. Both types of humati- rhomb may be composed of either simple or compound fistulipores. Thus four types of humatirhomb may be recognized :
4 PALAEONTOLOGY, VOLUME 15
Simple humatirhombs with simple fistulipores (text-fig. 3a)
Complex humatirhombs with simple fistulipores (text-fig. 2>b)
Simple humatirhombs with compound fistulipores (text-fig. 3c)
Complex humatirhombs with compound fistulipores (text-fig. 2d)
Humatirhombs are characteristic of, and confined to, the superfamily Caryocystitida, members of which have thecae composed of an indefinite and usually large number of thecal plates which are randomly arranged. The number of thecal plates often increases
TEXT-FIG. 4. Camera lucida drawings of the humatirhombs of Caryocystites lagenalis Regnell to show principal (pf) and intermediate (if) fistulipores. A, traces of tangential canals (SM A57362). b, traces of perpendicular canals (SM A30656) to show that intermediate fistulipores define smaller rhombs within the main rhomb, s = plate suture.
during growth and it is impossible to describe the position in a theca of an individual plate or rhomb. However since all plate sutures have rhombs developed across them and all rhombs of any one theca are invariably the same, distinction of individual rhombs is unnecessary. No part of a theca is better provided with rhombs than any other part.
EXPLANATION OF PLATE 1
Stereophotos of simple humatirhombs with simple fistulipores.
Figs. 1, 2, 7. Lophotocystis granatum (Wahl.). 1, 2, Two small weathered examples to show fine granules on external surface. 1, RM Ec4353; 2, RM Ec4352; 7, SM A57330, Part of a large example with well-developed humatirhombs. All x 3.
Eigs. 3-6. Ulrichocystis eximia Bassler. 3, 4, Unweathered isolated plate; 3, oblique sutural view to show tangential canals beneath external ridges, x 6; 4, external surface, x 4. 5, 6, Another isolated plate ; 5, weathered external surface to show exposed tangential canals and positions of perpendicular canals ; 6, internal surface to show canals partially covered near plate centre, both X 4 (cf. text-fig. \2b). Specimens in author’s colln.
Eig. 8. Lophotocystis sp. nov. (Shole’shook, S. Wales). SM A53070c. Latex impression of fragmentary example. X2.
Fig. 9. Lophotocystis malaisei (Regnell). SM A 50361. Latex impression of part of theca. X2.
All figures whitened with ammonium chloride sublimate.
Palaeontology, Vol. 15
PLATE 1
PAUL, Simple humatirhombs with simple fistulipores
PAUL: EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
5
The distribution of the four types of humatirhomb within the superfamily Caryo- cystitida does not correspond to taxonomic subdivisions. The first two types of humati- rhomb (text-fig. 3>a-b) are confined to the genera Ulrichocystis Bassler and Lophotocystis nov. (= Heliocrinites of the ‘planata’ group of Regnell 1951, p. 22; Bather 1906, p. 18; see Appendix 2, p. 26). Lophotocystis has tangential canals developed in prominent ridges on the external surface of the theca (PI. 1). The humatirhombs of Ulrichocystis are simple and their tangential canals are less distinctly raised (PI. 1, figs. 3-4).
The third and most common type of humatirhomb (text-fig. 3c) occurs in all species of Heliocrinites s.s. (i.e. Regnell’s ‘plicata’ group, 1951, p. 22), all Echinosphaeritidae and Caryocystites dubia (Angelin) = C. angelini Auctt. The fourth type (text-fig. M) occurs only in C. lagenalis Regnell as far as is known.
TEXT-FIG. 5. Two possible interpretations of the structure of the canals in Stichocystis Jaekel as seen in longitudinal section. In a each pair of perpendicular canals (pc) is connected by a separate tangential canal (tc). In b a single tangential canal connects all perpendicular canals. The tangential canals were made entirely of soft tissue and are not preserved. Both the pairing of the perpendicular canals and the efficiency of currents (indicated by arrows) favour interpretation a. s = plate suture.
The genus Stichocystis Jaekel which on other morphological grounds belongs in the Caryoeystitida bears unusual rhombs with sets of perpendicular canals developed in ridges on the external surface of the plates. Not all details of the structure of these rhombs are known but they seem to be functionally and morphologically related to humatirhombs. I interpret them as having a rhomb-in-rhomb structure (text-fig. 5a) but this is not certain. These rhombs may bear the same relationship to simple humati- rhombs that multi-disjunct pectinirhombs bear to disjunet pectinirhombs.
The next two types of exothecal pore-struetures (diplopores and humatipores) are dipores which consist of a single thecal canal, not a set of canals (Paul 1968, p. 700).
2. Diplopores (text-figs. 6-8, Pis. 5-6)
Diplopores are dipores composed of a simple thecal canal, the tangential portion of which was not normally calcified and probably formed a papula or podium in life. As a result only the pair of perpendicular canals is preserved in fossils, which led Muller (1854) to propose the term ‘Doppelporen’ or diplopore (see Huxley 1854).
If a diplopore is considered as a functional rather than a morphological unit, the pores are internal and the supposed podium represents the tangential portion of the thecal canal (text-fig. 6). No podium has yet been found preserved but diplopore
6 PALAEONTOLOGY, VOLUME 15
tangential canals are sometimes calcified in the Aristocystitidae and Sphaeronitidae. Normal diplopores show as two pores which are usually paired within a shallow depres- sion {peripore) on the external surface of the theca. Peripores may have rims, or peripheral or central tubercles (text-fig. la-d). In general morphology, diplopores strongly resemble the pore-pairs of echinoids.
Only the two perpendicular canals of a diplopore are preserved in most fossils. They may pass straight through the plates, follow sinuous courses, or unite with one or more other perpendicular canals (text- fig. 8). Previously it has been assumed that when two perpendicular canals unite to form a Y-shaped canal both branches fed the same diplopore (e.g. Chauvel 1941, figs. 39c-c, p. 100). However, in the only example where I have been able to trace the course of the two branches they fed separate diplopores (text-fig. 8Z>). Functionally this is a more efficient arrangement since it allows circulation. In Codia- cyslis Jaekel and Sphaeronites Hisinger large pits, off which a number of perpendicular canals branch, occur on the internal surfaces of the plates. These pits could be centres of radiation for afferent and efferent canals but this is not certain. They show up as prominent tubercles on internal moulds (e.g. Barrande 1887, pi. 19, figs. 30, 32-33, 35-36).
Diplopores show wide variations in morphology but no clearly defined types exist. Diplopores of certain genera and families may have characteristic morphology however (e.g. Haplosphaewnis, PI. 6, figs. 1-3; Sphaeronites s.s. PI. 5, figs. 1-2). Diplopores occur in all superfamilies of the Diploporita. They are usually randomly distributed over a theca but may be more prolific on certain parts of the theca (or of individual thecal plates) than on other parts. In the Aristocystitidae certain areas of the theca may have sealed canals. Usually these areas were permanently in contact with something solid
TEXT-FIG. 6. Diagrammatic representa- tion of the structure of a diplopore. The two perpendicular canals open into a depression, the peripore, over which a podium or papula extended in life.
EXPLANATION OF PLATE 2
Complex humatirhombs with simple fistulipores.
Fig. 2. Lophotocystis araneus (Sehlotheim), RM Ec5370. Note intermediate fistulipores in rhombs.
Fig. 7. Lophotocystis sp. RM Ec25233a.
Simple humatirhombs with compound fistulipores.
Fig. 1. Heliocrinites stellatus RegneU, RM Ec25985. Note compound fistulipores with pairs of tan- gential canals.
Fig. 3. Echinosphaerites aiirantiurn aiirantium (Gyll.), SM A57365. Note fistulipores reach plate centres.
Fig. 5. Echinosphaerites aurantium suecicus Jaekel, SM A57343. Note large area without fistulipores in centres of large plates.
Figs. 4, 6. Heliocrinites ovalis Angelin. Two examples with weathered surfaces revealing tangential canals in groups of two and three, 4, RM Ec3324; 6, RM Ec3327.
Figs. 1, 2, 4, 6, 7, x3, whitened with ammonium chloride sublimate; Figs. 3, 5, x5, photographed under water.
Palaeontology, Vol. 15
PLATE 2
PAUL, Complex humatirhombs with simple fistulipores
PAUL; EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
7
TEXT-FIG. 7. Four diplopores with different arrangements of tubercles and ridges associated with their peripores. a, Sphaeronites pomum (Gyll.). Oval peripores deeply impressed into the plate surface and with spine-like tubercles on the ridges between them (PI. 5, fig. 2). b, Sphaeronites globulus (Ang.). Polygonal peripores with a large flat-topped central tubercle which produces moat-like channels within the peripore (PL 5, fig. 3). c, Archegocystis sp. nov. (Shole’shook, S. Wales). Oval peripores with simple raised rims (cf. PI. 6, fig. 4). d, Haplosphaeronis sp. nov. (Shole’shook, S. Wales). Peripore divided into pyriform and circular depressions by a subcentral ridge, peripheral ridge with two tubercles between pores (PI. 6, fig. 2).
A B
TEXT-FIG. 8. Two possible arrangements of Y-shaped perpendicular canals in diplopores. Current systems (indicated by arrows) are more efficient in b than in A. p = podia, pc = perpendicular canal.
during life (e.g. ambulacral facets, attachment areas, etc.). In the Dactylocystidae diplopores are confined to five ambulacral tracts.
3. Humatipores (text-fig. 9, PI. 7)
A humatipore is a dipore which consists of a wholly calcified compound thecal canal (text-fig. 9). In undamaged humatipores no pores show on the external surface. Two internal circular pores lead to two or more tangential canals which may lie either
8 PALAEONTOLOGY, VOLUME 15
beneath the flat external surface of the plates (buried humatipores, text-fig. 9b, PI. 7, figs. 1, 5-10), or in a prominent external tubercle (text-fig. 9a, PI. 7, figs. 2-4).
Humatipores are characteristic of, and confined to, the family Holocystitidae Miller. Buried humatipores occur in all five genera of Holocystitidae but tubercular humati- pores are confined to the genera HoJocystites s.s. and Pustulocystis Paul. Humatipores are always evenly developed over a theca.
TEXT-FIG. 9. Diagrammatic representations of the morphology of a, raised, and b, buried humatipores. In both, part of one tangential canal (tc) is cut away for clarity, pc = perpendicular canal, t = tubercle.
4. Haplopores
A number of species of the Aristocystitidae have been claimed (Bather 1900; Chauvel 1941) to bear haplopores: a type of pore-structure which consists of a single perpen- dicular canal. Since I have not examined all the relevant species I cannot state that haplopores do not occur in cystoids. However I have not seen a single specimen of a cystoid which characteristically bears haplopores and hence their functional mor- phology is not analysed. In the last fifty years or so Chauvel is the only person to have
EXPLANATION OF PLATE 3
Simple humatirhombs with compound fistulipores.
Figs. 1, 7. Heliocrinites sp. nov. (Rhiwlas, N. Wales). 1, GSM 102326, Part of an unweathered theca with tangential canals filled with dark sediment, x 5, photographed under water. 7, GSM 102325, Stereophotos of weathered theca to show tangential canals in groups of two to four, x 3.
Fig. 3. Echinosphaerites aurantium americanum Bassler. Example with tangential canals reaching plate centres (cf. PI. 4, fig. 1), x4 (author’s colb.).
Figs. 4, 6. Caryocystites dubia (Angelin). 4, SM A51333, A large weathered theca. 6, SM A57335, Stereophotos of a small example to show tangential canals in groups of two to four, both x 2.
Fig. 5. Heliocrinites guttaeformis Regnell. RM Ec4780. An example with prominent tangential canals,
x3.
Complex humatirhombs with compound fistulipores.
Fig. 2. Caryocystites lagenolis Regnell. SM A57362. Stereophotos to show details of rhombs (cf. PI. 4
fig. 6), X 6.
All figs, except fig. 1 whitened with ammonium chloride sublimate.
Palaeontology, Vol. 15
PLATE 3
PAUL, Simple humatirhombs with compound fistulipores
PAUL: EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
9
examined cystoid species which may bear haplopores. He wrote (1941, p. 60): ‘If one reserves the name haplopores (canaux haploporiques) for these sinuous ramifying canals, the canals of Arislocystis [sic], even though united in pairs, are incontestably haplopores’ (my italics). But if the perpendicular canals were paired circulation was possible and whatever modifications occurred within the plates, functionally the struc- ture is the same as a typical diplopore (text-fig. 6). Hence I regard the pore-structures of Aristocystites as diplopores. Only one true haplopore has come to my notice in an isolated plate of Eucystis sp. from Knock, Westmorland which also bears several typical diplopores. I interpret this as a damaged or incompletely developed diplopore.
FUNCTION OF EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
Untestable functional interpretations are inherently weak, since they can be neither substantiated nor disproved, and undesirable since they cast doubt on the validity of all functional interpretations. Many functional hypotheses are open to test however, by Rudwick’s paradigm method (Rudwick 1964). To test a functional hypothesis the detailed morphology of a fossil structure is compared with that of an ideal structure (paradigm) which would serve the supposed function with maximum efficiency. Close comparison indicates that the fossil structure could have served the supposed function efficiently but does not prove that in fact it did so.
The paradigm method can be used to test the mechanics, but not the physiology of the structures investigated. It depends on acceptance of ‘mechanical uniformity’ (i.e. that the ‘laws’ of mechanics applied in the past as they do now) but may be totally independent of knowledge of living organisms. All morphological structures have more or less well-defined mechanical effects. Their function is taken to be that effect which is most beneficial to the vital needs of the organism or which confers most selective advantage on the organism. By considering effects rather than function, analysis may be made more rigorous and conclusions stated more positively. For example rigidity is an undeniable mechanical property of triangles. The triangulation ‘ornament’ of ridges connecting plate centres which is so commonly found in echinoderms undoubtedly increased the strength and rigidity of the test. This is not an interpretation: it is a fact. If we try to explain this effect in terms of function or selective advantage, then we make interpretations. In doing so we may draw sound conclusions if we can demonstrate that a vital function necessary for the survival of the organism (e.g. nutrition, protection, respiration, excretion, etc.) was performed more efficiently with, than without, the structure involved. Echinoderm tests with triangulation ‘ornament’ provided better protection for the enclosed vital organs than those without this ‘ornament’, by virtue of their increased strength and rigidity. Thus we may conclude that most probably the function of triangulation ‘ornament’ was protection but un- doubtedly its effect was to increase the strength and rigidity of the test.
The following analysis is an attempt to estimate the mechanical efficiency of cystoid pore-structures as exchange systems and they are compared with the appropriate para- digm. Any exchange system must have an exchange surface to prevent mixing of fluids. The amount of exchange is controlled by the following factors :
1. The area of the exchange surface: the larger the area the greater the amount of exchange.
2. The resistance to exchange of the exchange surface: the thinner the surface the less its resistance will be.
3. The concentration gradient across the exchange surface: the higher the gradient (i.e. the greater the difference in concentration of the exchange substance on either side of the exchange surface) the greater the potential exchange. A counter-current system (text-fig. lOu) is the most efficient method of maintaining a high concentration gradient.
10
PALAEONTOLOGY, VOLUME 15
Thus the paradigm of an exchange system will have a large area of exchange surface which is as thin as is compatible with its strength and a counter-current system. A more detailed account of the above is presented in Paul 1968, pp. 708-709.
Detailed functional analysis
In exothecal pore-structures the fluids within the canals were body fluids. A healthy animal presumably had control over both their composition and circulation. Hence devices to prevent recirculation and choking of the canals by foreign particles were unnecessary and cannot in fact be recognized. The exchange surfaces were outside the theca and therefore liable to mechanical damage. This brings into opposition two requirements of the paradigm: the thinner the exchange surface the greater the amount of exchange but the greater the chances of rupture and mixing. With the above ideas in mind the detailed morphology of exothecal pore-structures in cystoids will be compared with the paradigm of an exchange system and estimates of the efficiencies of the various types made.
1. The area of the exchange surface. Four of the five basic types of cystoid pore- structures have calcified exchange surfaces which are frequently preserved in fossils and the areas of which can be measured or at least estimated fairly accurately. Echinoderm skeletal material is a meshwork of fine calcite rods and soft tissue fibres ; exchange would have taken place through the latter. Only about half the exchange surface area (the soft tissue half) functioned actively in exchange during life.
In cystoids individual calcified exothecal pore-structures are much less efficient than endothecal (dichoporite) pore-structures in terms of the area of exchange surface. The
50
70-
90-
II
50
>30
-»I0
100 .
TEXT-FIG. 10. Idealized exchange systems, a, counter current; b, concurrent. Maximum potential exchange in b is half that of a. Figures represent percent concentration of the exchange substance. Heavy arrows indicate current directions, light arrows indicate ex- change.
EXPLANATION OF PLATE 4
Stereophotos of simple humatirhombs with compound fistulipores.
Fig. 1. Echinosphaerites aurantiiim americanum Bassler. Weathered example with fistulipores which do not reach plate centres in largest plates, x 4 (author’s coUn.).
Figs. 2, 3. Echinosphaerites aurantiiim s.l. 2, BMNH E7803, Internal surface of part of theca to show openings of perpendicular canals (cf. PI. 1, fig. 6), x2. 3, BMNH (unreg.), Weathered portion of theca, x3.
Fig. 4. Heliocrinites giittaeformis Regnell. RM Ec4763. Portion of weathered theca with large rhombs, X3.
Fig. 5. Caryocystites dubia (Angelin). SM A57332. Example with unweathered external surface show- ing outlines of rhombs, x 2.
Stereophotos of complex humatirhombs with compound fistulipores.
Fig. 6. Caryocystites lagenalis Regnell. SM A57362. Note intermediate fistuhpores, X 2.
Palaeontology, Vol. 15
PLATE 4
PAUL, Simple humatirhombs with compound fistulipores
PAUL; EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
11
ratio AJA( where A^ is the area of the exchange surface and Ai is the area of the thecal surface occupied by the pore-structure is a measure of the efficiency of an individual pore-structure. In endothecal pore-structures this ratio is always greater than one and in one measured pectinirhomb was 7-84. Table 1 shows that this ratio varies from 0-28-0-86 in humatirhombs. All three types of exothecal pore-structure in cystoids exhibit modifications of their basic design which increase the ratio AJA( but no evo- lutionary trends towards increased efficiency are apparent in contrast to pectinirhombs (Paul 1968).
TABLE I . Estimates of the ratio AJAt in humatirhombs
|
Species |
Ratio |
Humatirhomb type |
|
Lophotocystis angustiporus (Regnell) |
0-37 |
Simple rhombs with simple fistulipores |
|
L. granatum (Wahl.) |
0-30-0-42 |
„ „ „ |
|
L. sp. (Shole’shook) |
0-28 |
„ „ „ |
|
L. malaisei (Regnell) |
0-28-0-41 |
,, ,, ,, |
|
Ulrichocystis eximia Bassler |
0-59-0-67 |
,, ,, ,, |
|
L. sp. nov. (Skalberget) |
0-42 |
Complex rhombs with simple fistulipores |
|
Heliocrmites ovalis Ang. |
0-58-0-68 |
Simple rhombs with compound fistulipores |
|
FI. guttaeformis Regnell |
0-38-0-80 |
,, ,, ,, |
|
H. sp. nov. (Rhiwlas) |
0-43 |
„ „ ,, |
|
Echiuosphaerites aurantium (Gyll) |
0-29 |
„ „ „ |
|
E. a. suecicus Jaekel |
0-75 |
,, >9 99 |
|
E. a. americamm Bassler |
0-57-0-77 |
99 99 99 |
|
Caryocystites dubia (Ang.) |
0-71 |
99 99 99 |
|
C. lagenalis Regnell |
0-86 |
Complex rhombs with compound fistulipores |
For simple fistulipores raised in ridges: Ag ~ (\-5xFW), and At = {RS+RW) where RS = ridge separation, RW = ridge width, and FW = fistulipore width.
For compound fistulipores buried in plates; Ag ~ FW and At ^ FW+IFW. (IFW = separation of fistulipores.)
For complex rhombs the area covered by intermediate fistulipores is calculated and the ratio Ag/At doubled for that area, i.e.{4e (total) + 4;, (complex)}//! (.Thus if the complex rhomb area is half the total rhomb area the total exchange area = \-5Ag/At-
In an ideal simple humatirhomb without raised ridges the ratio of the width of the tangential canals to the width of the gap between them is a close approximation to AglAf. When the two widths are equal AJAi is approximately 0-5. Humatirhombs with compound fistulipores generally have more closely spaced tangential canals than those with simple fistulipores. However, the latter are usually raised in ridges and have a larger area of exchange surface than buried tangential canals (text-fig. 11). The increase in area due to the ridges is as much as 50%. Both arrangements (i.e. raised simple fistulipores and buried compound fistulipores) seem to be alternative methods of increasing the ratio AgjA,,
In terms of exchange surface area humatirhombs are individually much less efficient than dichoporite pore-structures. However, every plate suture bears a humatirhomb in all Caryocystitida and there are thus several hundred rhombs per theca. The total exchange area per theca was probably as high as in dichoporite cystoids with only 1-25 rhombs per theca. Since almost all the thecal surface is covered with humatirhombs the ratio AgjAt is only slightly greater than the ratio of the total exchange area to the total thecal surface area. Thus anything from about 25% to 75% of the thecal surface area
12 PALAEONTOLOGY, VOLUME 15
was exchange surface in humatirhomb-bearing cystoids but only about half aetively functioned in exchange.
Humatipores are very similar to humatirhombs in terms of exehange area but it is much more difficult to measure or estimate AJAi. An increase in the number of tan- gential canals increases the exchange area per humatipore (text-fig. 9b). Equally, raising the humatipore into a prominent tubercle increases exchange area (text-fig. 9a). With regard to exchange area per humatipore buried humatipores are less efficient than tubercular humatipores. However, the former are frequently more densely packed than the latter. Again this seems to reflect two alternatives: fewer, more efficient structures or a larger number of less efficient structures. In humatipore-bearing cystoids total
A
<— 2r — ►
TEXT- FIG. 11. Diagram to illustrate the exchange areas of (a) buried, and (b) raised tangential canals (tc) in fistulipores. Width of exchange area in one canal in a is 2r, in b it is nr. In rhomb width W and length / areas are 6rl in A and Inrl in b. Since 77 ~ 3 two canals in b have the same exchange area as three canals in a.
exchange area per theca probably lies within the same limits as for humatirhomb-bearing cystoids.
Some diplopores resemble echinoid pore-pairs very strongly and probably gave rise to podia analogous, if not homologous, to echinoid tube-feet. The podia are never
EXPLANATION OF PLATE 5
Stereophotos of diplopores.
Fig. 1. Sphaeronites sp. nov. (Raback, Vastergotland, Sweden). SM A35317. A small theca with densely packed diplopores over entire surface.
Fig. 2. Sphaeronites pomum (Gyll.) SM (unreg.). Note spinose tubercles between deeply sunken peri- pores (cf. text-fig. 6a).
Fig. 3. Sphaeronites globulus (Angelin). SM A57321. Note polygonal peripores with large flat-topped central tubercle (cf. text-fig. 6b).
Fig. 4. Sphaeronites sp. nov. (Skalberget, Dalarna, Sweden). SM A57407. Note irregular diplopores like those of S. globulus.
Fig. 5. Sphaeronites pyriformis (Forbes). BMNH E16340.
Fig. 6. Sphaeronites litchi (Forbes). GSM 7431. Note very prominent central tubercles (cf. fig. 8 this plate).
Stereophotos of echinoid pore-pairs.
Fig. 7. Arbacia punctulata (Lam.). Note tubercles around pore-pairs (author’s colln.).
Fig. 8. Echinocorys scutatus (Leske). SM (unreg.). Pore-pairs of buccal tube-feet. Note large central tubercle (cf. fig. 6, this plate).
Figs. 1-6 X 3, 7, 8 x4. All whitened with ammonium chloride subfimate.
Palaeontology, Vol. 15
PLATE 5
PAUL, Diplopores
PAUL: EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
13
preserved, so measurement of their surface area is impossible. Nevertheless the area of podium wall could not have been less than the area of the peripore and since it was entirely made of soft tissue all the area could have functioned in exchange. Thus in diplopores AJAirmxst have been greater than or equal to 1. Most diplopores have rims and tubercles associated with their peripores as do most echinoid pore-pairs. In the latter the rims and tubercles are attachment structures for the longitudinal muscles of the tube-feet (Nichols 1959, p. 70). As a broad generalization, the more strongly developed the rims, etc. are, the stronger the muscles and the greater the flexibility of the tube-feet. For example, many regular sea urchins have much more prominent rims and tubercles associated with pore-pairs on the oral surface up to the ambitus, and the tube-feet on this surface are the main ones used in locomotion. In some cases the com- parison between diplopores and pore-pairs is so strong (e.g. Sphaeronites and Echino- corys oral tube-feet, PI. 5, cf. figs. 3-6 with fig. 8) that the conclusion that they represent almost identical structures seems inescapable. Although impossible to prove, available evidence strongly suggests that some diplopores had extensile podia. This would increase their efficiency as exchange surfaces in two ways: it increases the area and it decreases the thickness of the exchange surface.
The density of packing of diplopores varies from genus to genus or even species to species. In Archegocystis the number of diplopores could apparently increase or decrease during life (Paul 1971). This forms a very delicate exchange mechanism that could respond to changes of the environment. Such was definitely not the case in Sphaeronites, all species of which have diplopores evenly developed all over the theca. The latter genus shows an interesting evolutionary trend towards larger diplopores throughout the Middle and Upper Ordovician.
Many representatives of the Aristocystitidae have sealed diplopores in some part of the theca. Chauvel (1966, p. 109) has interpreted this as a ‘maladie calcaire’ reminiscent of W. D. Lang’s fatalistic trends in various calcium carbonate secreting organisms (Lang 1923n, b). Calcification of diplopores decreases their efficiency by at least halving their exchange surface area but it does not necessarily render them useless. Indeed calcifica- tion is much more likely to represent protection against predators eating soft tissue podia than ill health. Cystoids lacked spines, at least as far as is known; so podia were not mechanically protected as sea urchin tube-feet are.
The following conclusions can be drawn as regards area of exchange surface :
(i) Individual exothecal pore-structures are very much less efficient than individual endothecal pore-structures but far more of them are developed on any one cystoid. The total area for exchange per theca was probably the same for both endothecal and exo- thecal pore-structures. In humatirhombs available measurements indicate that the total exchange area was between 25% and 75% of the total thecal surface area.
(ii) In humatirhombs the raising of tangential canals in ridges, and the development of compound and additional fistulipores increase exchange area. In humatipores production of many tangential canals and development of tubercular humatipores both increase exchange area. Some diplopores may have had extensile podia which also increased exchange area. Within the limits of their geometry all three types of exothecal pore- structure tend to maximize their exchange area. Diplopores were probably the most efficient of the three but were liable to predation since their exchange surfaces were made entirely of soft tissue.
14 PALAEONTOLOGY, VOLUME 15
(iii) Although individually inefficient all exothecal pore-structures probably provided adequate exchange area per theca by sheer weight of numbers.
2. The resistance to exchange of the exchange surface. For maximum efficiency the exchange surface should be as thin as possible; however, rupture and mixing of fluids must be prevented. Exothecal pore-structures are much more susceptible to mechanical damage than endothecal pore-structures since their exchange surfaces are external. It is not surprising therefore that measurable exothecal exchange surfaces are thicker (0-05-0-10 mm) than endothecal exchange surfaces (always less than 0-03 mm and reaching as little as 0-007 mm). Nevertheless the thecal wall in most cystoids with exo- thecal pore-structures is 1-3 mm thick and often much thicker in aristocystitids. Chauvel (1966, p. 27) records a maximum thickness of 26 mm in Maghrebocystis. Although exchange surfaces of humatipores and humatirhombs are thicker than those of endo- thecal pore-structures they are still very much thinner than the thecal wall. Since no measurements of thickness are possible in diplopores their efficiency in terms of re- sistance to exchange cannot be estimated. However, extensible podia would have had very thin walls in all probability.
Exothecal pore-structures do not seriously weaken the theca and no strengthening structures have been recognized. Again this contrasts with pectinirhombs (Paul 1968).
3. Maintenance of a concentration gradient. The most efficient method of maintaining a concentration gradient is a counter-current system (text-fig. lOn). The best evidence for current directions is given by protective devices and devices to prevent recirculation. Unfortunately neither type of device is necessary with exothecal pore-structures since the fluids flowing in the thecal canals were body fluids. Neither type of device has been recognized. Some indirect evidence of currents and their directions of flow is available, however.
Nearly all recent echinoderms have ciliated external epithelia and cystoids probably had too. From purely hydrodynamic considerations fluids within the canals would not have moved without cilia due to the viscous effect of the boundary layer (Paul 1968, pp. 719, 721). Almost certainly both internal and external ciliary currents were present in cystoids. The humatirhombs of Lophotocystis granatum (Wahlenberg) have fine granules developed on the ridges. In the best preserved example the granules are elongate parallel to the rhomb axes (PI. 1, figs. 1-2). If a ciliated epithelium was present
EXPLANATION OF PLATE 6
Stereophotos of diplopores.
Figs. 1, 3. Haplospliaeronis obloiiga (Angelin). 1, SM A57381, An example with oval peripores without rims. 3, SM A57356, An example with peripores with strongly raised rims.
Fig. 2. Haplospliaeronis sp. nov. (Shole’shook, S. Wales). SM A57520. Latex impression of theca showing asymmetrical diplopores (cf. text-fig. 6d).
Fig. 4. Archegocystis steUiilifera (Salter). BMNH E16200. Latex impression showing elongate and oval diplopores with simple rims (cf. text-fig. 6c).
Fig. 5. Aristocystites bohemicus Barrande. SM A49868c. Latex impression showing irregular peripores between gonopore (left) and hydropore (right). Even though the peripores are very irregular it is still possible to see that the perpendicular canals are arranged in pairs as in typical diplopores.
Eig. 6. Triamara sp. USNM 166580. Example with small oval peripores. x4.
Eigs. 1-5 X 3. All whitened with ammonium chloride sublimate.
Palaeonlology, Vol. 15
PLATE 6
PAUL, Diplopores
PAUL: EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
15
these granules would have increased the area of ciliated surface and enhanced the cur- rents parallel to the rhomb axis. Such granules are similar to those on the periplastronal areas of recent spatangoid sea urchins and are present on other species of Lophotocystis. Granular ornament is also characteristic of most species of humatipore-bearing cystoids although the granules are not elongate. Since all the thecal surface is covered with humatirhombs in caryocystitid cystoids, external currents of different rhombs would interfere with each other. Nevertheless, water in contact with the external surface of the theca would be continually changed.
Evidence for the presence and direction of currents in Diploporita is virtually non- existent. However the genera Holocystites and Haplosphaeronis regularly have asym- metrical dipores and this asymmetry may be associated with current flow. Clearly, internal fluids came up one perpendicular canal and descended down the other. Which canal was efferent and which afferent is not certain in either genus. In Holocystites, which bears humatipores, one perpendicular canal is subcentral and one peripheral (text-fig. 9a, PI. 7, figs. 2, 3). I suggest that body fluids ascended the subcentral perpen- dicular canal but the alternative direction seems equally plausible. From the point of view of exchange either direction would seem to be equally efficient.
Some species of Haplosphaeronis have asymmetrical diplopores. The peripore rim is thickened and raised and the peripore floor raised between the perpendicular canals, but closer to one than the other. Thus one canal opens in a roughly circular depression and the other in a pyriform depression (text-fig. Id, PI. 6, fig. 2). The diplopores of Haplosphaeronis are elongate and most are aligned in an oral-aboral direction. In the oral half of the theca the pyriform depression is adoral in the diplopore but the converse is true in the aboral half of the theca. If the theca was orientated with the mouth upwards, surface ciliary cleaning currents may have moved in an aboral direction on the upper half of the theca as they do in many recent sea urchins. A current in the reverse direction in the aboral half of the theca would also help to keep the theca clean (text-fig. 12).
If such external currents were present, alignment of the diplopores parallel to them would allow internal counter-currents to operate. Recent sea urchins with specialized respiratory tube-feet have opposed exter- nal and internal currents. Again the opposite current directions would seem equally plausible and equally efficient in exchange.
The pattern of external currents proposed for Haplosphaeronis involves flow from the oral and aboral extremities towards the ambitus. Precisely similar current patterns were proposed for the external currents of dichoporite rhombiferans (Paul 1968). Since the two groups are not closely related, similar external cleaning currents may have preceded the evolution of internal respiratory currents. The latter probably developed in a fixed relationship to the former, namely counter to them.
In summary, there is little direct evidence for the presence or direction of currents in exothecal pore-structures. However, in at least one example of each major type there
TEXT-FIG. 12. Possible surface current directions in Haplosphaeronis Jaekel. Such currents would help to clean the thecal surface.
16
PALAEONTOLOGY, VOLUME 15
is some indirect evidence for currents. Indeed the basic morphology of thecal canals is ideal for current flow since one perpendicular canal could act as the afferent canal and the other as the efferent. In all but two genera the two perpendicular canals are identical and it is generally impossible to say which canal was which.
Conclusions
All three exothecal pore-structures (humatirhombs, humatipores, and diplopores) differ from the paradigm of an exchange system to some degree and they were indi- vidually less efficient than endothecal pore-structures in terms of the area and thickness of the exchange surface and possibly of current systems too. This relative inefficiency can be explained in terms of the need to prevent rupture of the exchange surface. Although individually less efficient than endothecal pore-structures, exothecal pore- structures still allowed exchange to take place. Indeed if currents were present this would inevitably have been their effect.
Large numbers of exothecal pore-structures are developed in any one theca which compensates for their individual inefficiency. For example, the ratio AJAi is an estimate of individual efficiency in terms of exchange surface area. An average value for humati- rhombs is probably about 0-5, for pectinirhombs about 10. In equal-sized thecae with equal-sized rhombs there should be 20 times as many humatirhombs as pectinirhombs to achieve the same amount of exchange. This ratio of humatirhombs to pectinirhombs is easily exceeded in practice since humatirhombs are developed in all available space on a theca. The exchange surfaces are thicker in humatirhombs than in pectinirhombs and hence the ratio should be higher than 20 to 1 .
There can be little doubt that the mechanical effect of exothecal pore-structures in cystoids was to allow exchange between sea water and body fluids. Since oxygen and carbon dioxide transfer constitute the most likely form of this exchange, exothecal pore-structures were respiratory structures. It is now pertinent to consider respiration in more detail.
EXPLANATION OF PLATE 7
Humatipores.
Figs. 1, 5, 7. Trematocystis globosus (Miller). 1, USNM S3058b. Note tangential canals exposed by weathering. 5, 7, FMNH 8766a; 5, General view of plate, X 6 approx. 7, Detail of humatipores to show plate meshwork in weathered tangential canals, x 25 approx.
Figs. 2-3. Stereophotos of Holocystites alteniatiis (Hall). BMNH E7629. 2, Detail of single humati- pore, X 25 approx. 3, General view of plate showing tubercular humatipores with radiating tan- gential canals, x 6 approx.
Fig. 4. Holocystites scutellatiis Hall. Detail of some weathered tubercular humatipores, X 10 (author’s colln.).
Fig. 6. Brightonicystis gregarius Paul. SM A32814a. Detail of humatipores with 6-8 tangential canals, X5.
Fig. 8. Stereophotos of Pentacystis sphaeroidalis (Miller and Gurley). FMNH 6000. Detail of weathered humatipores, Xl3.
Fig. 9. Stereophotos of Pentacystis simplex Paul. AMNH 20271a. Detail of pit bored into cystoid by parasite which shows tangential canals of three humatipores parallel to the sides of the pit. These canals were formed after the pit was bored, X 10.
Fig. 10. Stereophotos of Trematocystis rotundas (Miller).
All figures whitened with ammonium chloride sublimate.
Palaeontology, Vol. 15
PLATE 7
PAUL, Humatipores
PAUL: EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
17
RESPIRATION IN RECENT AND EOSSIL ECHINODERMS
A specialized respiratory system consists of three distinct parts. The respiratory system in vertebrates, for example, includes : (i) an external exchange surface (lungs or gills) whereby oxygen is gained from (and carbon dioxide lost to) the surrounding medium, (ii) a circulation system (blood stream) to distribute oxygen internally, and (iii) internal exchange surfaces (the capillaries) whereby oxygen is transferred from the circulation system for use in cellular metabolism. Unless sites of metabolism are very close to the external exchange surfaces some system of oxygen transport is vital for efficient respiration. Most triploblastic metazoa have a specialized circulation system but recent echinoderms do not.
The preceding analysis of exothecal pore-structures in cystoids considered only the first part, i.e. the external exchange surfaces, and depended on paradigmatic methods. This section considers the internal portions of the respiratory system, and depends more on biological uniformitarianism. It is pertinent to consider what is known of respiration in recent echinoderms.
Recent echinoderms
Relatively little systematic information is available on respiration in recent echinoderms. Farman- farmaian (1966, p. 245) in his summary uses Harvey’s (1928) equation for the dilTusion of oxygen into a spherical organism to prove conclusively that no echinoderm could rely on diffusion alone to gain oxygen from sea-water. Harvey’s equation is as follows:
Co = Ar^-I6D
where €„ is the concentration in atmospheres of oxygen in sea water; A is the rate of oxygen con- sumption by the organism in ml Oa/grm/minute ; r is the radius of the sphere in cm; and D is the diffusion coefficient in atmospheres/cm/cm^. Using the following values, the equation can be solved for r (which is the depth to which oxygen will penetrate into an echinoderm by diffusion alone).
Co = 0-21 (Farmanfarmaian 1966)
D = 0000011 (Krogh 1941)
A = 0 001233 maximum and 0 000166 minimum (Farmanfarmaian 1966)
___6D.Cq ' A
_ 6x0000011 X 0-21 6x000001 lx 0-21
” 0001233 “ 0000166
- 0 0112408 or 0 0834939
r =106 mm or 2-89 mm
Farmanfarmaian argues that the success of the echinoderms is to a large extent dependent upon the development of specialized respiratory surfaces (i.e. external exchange surfaces) since they are clearly essential to survival. For efficient respiration these surfaces must be significantly less than 1-3 mm thick. Recent echinoderm respiratory surfaces include respiratory trees and tentacles (holothurians), podia (all classes), papulae (asteroids), peristomial gills (regular echinoids), genital bursae (ophiuroids), and the general external surface (all classes). All but the last fall well within the thickness limits estabhshed above.
An efficient external exchange surface requires currents to replenish depleted fluids. Such currents are either oscillatory, involving current reversals, or circulatory. Oscillatory currents occur in blind structures, for instance in the respiratory trees of holothurians and the peristomial gills of regular echinoids. Circulatory currents occur in closed ring-shaped structures such as the tube-foot/ampulla system of echinoids. Recent echinoderms possess no specialized circulation system to distribute
C
C 8472
18 PALAEONTOLOGY, VOLUME 15
oxygen internally. Oxygen must be transferred into the fluids of the major eoelomic pouches via the external exchange surfaces and all organs involved in metabolism are either bathed in these fluids or directly in sea water.
The role of the water vascular system in respiration needs some clarification. Farmanfarmaian (1966, p. 250) has conclusively shown that in sea urchins the tube-foot/ampulla systems transfer oxygen from sea water to eoelomic fluids. But the radial water vessels are much less involved in respiration than the tube-foot/ampulla system because they are not directly in contact with sea water and are blind struc- tures without circulatory currents. The water vascular system can only transfer oxygen into eoelomic fluids efficiently where there is a tube-foot/ampulla system in which the tube-foot is external and the ampulla is internal (text-fig. 13). Hence the respiratory contribution of the water vascular system in crinoids, which have an almost totally external water vascular system, is negligible. This would be equally true of external branches of the water vascular system in cystoids if these were present.
In crinoids with large calyces and no calycal pore-structures, migration of coelomocytes would seem the only plausible way to oxygenate organs within the calyx. However, the role of coelomocytes in respiration (summarized in Endean, 1966) among recent echinoderms is not understood. Systematic movement of coelomocytes could explain the absence of a distinct circulation system in echinoderms. However, it is almost impossible to observe whether such movements of coelomocytes actually occur. One type of coelomocyte found in recent holothurians contains haemoglobin (i.e. the haemocytes). Wandering haemocytes would inevitably carry oxygen and carbon dioxide with them but again the actual course of wandering is impossible to observe. Haemocytes are unknown in recent crinoids and were presumably absent in fossil groups such as cystoids.
It may be noted that Farmanfarmaian (1966, p. 246) rejects the suggestion that the digestive traet can be involved in respiration.
In summary: recent echinoderms respire through a variety of specialized external exchange surfaces but lack any internal circulation system to distribute the oxygen so gained. No recent echinoderm relies exclusively on one type of external exchange surface and all types are made entirely of soft tissue. Efficient external exchange surfaces in recent echinoderms must be significantly less than 1-3 mm thick.
Comparison of recent and fossil echinoderms
Comparison of primary exchange surfaces in recent echinoderms and cystoids demonstrates both similarities and differences. Perhaps the most obvious difference is total lack of calcified exchange surfaces in recent echinoderms whereas these are com- monly present in four of the five basic types of cystoid pore-structures and occur rarely in the fifth. A second important difference is the apparent lack of oscillatory currents in cystoids. These two features may be correlated. It is difficult to develop an oscillatory current in a rigid calcified structure whereas soft tissue is ideally suited to produce the expansions and contractions necessary for oscillation. Perhaps the most important similarity is the wide variety of respiratory exchange surfaces which may be internal (e.g. holothurians) or external (e.g. asteroids and echinoids) among recent echinoderms just as in cystoids.
The over-all efficiency of a respiratory system depends not only on the external exchange surfaces but also on the internal distribution of oxygen. It is relevant therefore to speculate on the nature of the internal connections of cystoid pore-structures. If a cystoid relied on diffusion alone to distribute oxygen internally it would be advan- tageous to have pore-structures developed evenly over the entire surface. All internal organs would then be approximately equidistant from an oxygen source. The dichopores of pectinirhombs and cryptorhombs extended into internal eoelomic spaces. In the Hemicosmitida, cryptorhombs are more or less evenly developed over the theca and probably no internal circulation was developed. The supposed internal ciliary counter- currents in the cryptorhombs and movement of organs may have provided adequate
PAUL: EXOTHECAL PORE-STRUCTURES IN CYSTOIDS 19
circulation of coelomic fluids within any one coelomic pouch. In the Glyptocystitida, however, there is a progressive reduction in the number of pectinirhombs per theca and all Silurian and Devonian species have four or fewer pectinirhombs. This is believed to be correlated with the development of a specialized internal circulation system. Grooves on internal moulds of Callocystitidae from the Middle Silurian of North America pro- vide some evidence as to the path of this circulation system (Paul \961a). Since the grooves are apparently connected to the hydropore the supposed circulation system is believed to have been associated with the water vascular system.
The thecal canals of exothecal pore-structures could represent simple evaginations of internal coelomic pouches in which case all the canals could have been connected to one coelomic pouch. Alternatively each coelomic pouch could have had a number of such evaginations.
Both humatirhombs and humatipores are always developed over the entire thecal surface and hence the latter alternative would seem more likely.
However, Nichols (1962, p. 135, figs. 18<f-/) has proposed that the fistulipores of humatirhombs form discrete closed systems in themselves (text-fig.
\Aa). The tangential canals of fistulipores correspond to Nichols’s ‘external bulb’. Distinct evidence of the corresponding ‘internal bulb’ is found in the huma- tipores of Ulrichocyslis. Traces of canals are found on the internal surfaces of the plates (PI. 1, fig. 6) and the central portions of the plates show partially calcified canals internally. A section through a fistuli- pore of Ub’ichocystis resembles text-fig. \Ab. Such a system is less efficient than simple evaginations of the coelom since there are two exchange surfaces not one. If such internal-canals were generally present in humati- rhombs this might explain why humatirhombs are confined to the Ordovician whereas
TEXT-FIG. 13. O2 and CO2 exchange in a tube-foot/ampulla system, v = valve. Arrows indicate current and exchange directions.
TEXT-FIG. 14. Interpretation of the structure of fistulipores. A, Nichols’s (1962) interpretation with a soft tissue internal bulb (ib). b, diagrammatic section through fistulipore of Ulrichocystis which tends to confirm Nichols’s interpre- tation. PC ^ perpendicular canal, tc = tangential canal, s = suture.
other rhombs survived well into the Devonian. However, the evidence for such internal canals is preserved only in Ulrichocystis which is unique among the Caryocystitida in having recumbent arms. Clearly, Ulrichocystis is not a typical caryocystitid cystoid and it is unwise to regard it as such.
The situation with diplopores is equally complex and ambiguous. Many diplopores
20
PALAEONTOLOGY, VOLUME 15
have raised rims and tubercles associated with them (text-fig. 7) which strongly resemble similar features associated with echinoid pore-pairs. In echinoids these ridges are the points of insertion for longitudinal muscles of the tube-feet. If the rims and tubercles of diplopores represent similar muscle attachments then diplopore ‘podia’ were ex- tensible. Mechanical and hydrostatic considerations require compensating reservoirs for extensile podia. Nichols (1962, p. 135, figs. 18n-c) again proposed internal and external ‘bulbs’ for diplopores but he did not envisage extensile structures. Extensible podia imply a hydraulic system which is bound to leak just as tube-feet leak. If lost fluid is to be replenished some connection to a large reservoir is needed. The most obvious suggestion is that diplopores were connected to an internal development of the water vascular system. It may be significant that diplopores are not always evenly developed over the entire thecal surface: in the Dactylocystidae diplopores are confined to five ambulacral tracts which quite strongly resemble the ambulacra of echinoids. Now it is necessary to consider the nature of the water vascular system in cystoids.
WATER VASCULAR SYSTEM OF CYSTOIDS
The nature of the water vascular system in cystoids is pertinent to the function of the pore-structures and significant to echinoderm evolution in general. In recent crinoids the water vascular system consists of a circumoesophageal ring canal off which five radial canals branch and pass up the arms beneath the floors of the ciliated food grooves. Since cystoids are probably most closely related to crinoids, among living echinoderms, it has generally been assumed that their water vascular system was essentially similar to that of extant crinoids.
Examination of the food-gathering system of cystoids should provide evidence about the nature of the water vascular system. Food grooves have been described as hypo- thecal (within the theca), epithecal (on the external surface of the theca), and exothecal (free of the thecal surface in arms and brachioles). It is more accurate to say that food- gathering systems contain some hypothecal, epithecal, and exothecal portions. The first usually provides no evidence about the water vascular system; the latter two may do so but are developed to differing degrees in different groups of cystoids. The epithecal portion of the food grooves is very limited in all Rhombifera but fortunately in the Glyptocystitida and Hemicosmitida the exothecal portions are frequently preserved and hence their morphology is known. In both groups the main food grooves are rela- tively deep and wide and are provided with a permanent roof of cover plates. Should external extensions of the water vascular system have been present, they would have been adequately protected from damage. The spacious main food groove could have housed extensions of the water vascular, haemal, perihaemal, and oral nervous systems in the same manner as these are housed in the arms of recent crinoids.
Exothecal ambulacral appendages of the Caryocystitida are almost never preserved. Barrande (1887, pis. 23, 25) figures some examples of Arachnocystis wfaustus with three long brachioles preserved and Ulrichocystis eximia has three recumbent arms. Speci- mens of the latter species collected by the writer show a wide main food groove with recessed ledges on either side for the insertion of cover plates. Again if external branches of the tubular coelomic systems were present they would have been adequately housed and protected.
PAUL: EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
21
In diploporites virtually nothing is known about the exothecal ambulacral appendages since there are only two reports of preserved structures in the literature. Fortunately epithecal portions of the food grooves are frequently extensive. At least two distinct types of epithecal food grooves occur among Diploporita. The first type consists of very narrow and shallow grooves with no evidence of any covering structures. Possibly the grooves were provided with soft tissue lappets but this is of course unknown. If these shallow grooves housed extensions of the water vascular system, etc. these exten- sions would have been very delicate and yet apparently they were inadequately pro- tected from damage. It seems quite plausible that diploporites with shallow grooves lacked external extensions of the water vascular system in which case one would imagine that internal extensions were present.
The second type of diploporite food groove is found in the Aristocystitidae. Here the food groove runs along the base of a massive ambulacral tract, 5-10 mm wide which is lined with ambulacral pores in Aristocystites and Triamara (Paul 1971) and probably in other genera too. In all genera the ambulacral tracts are covered with immovable cover plates arranged in four series. In the absence of diplopores the ambulacral pores of Triamara and Aristocystites would be taken as clear evidence of internal ampullae. These ambulacral pores are however only specialized diplopores which happen to open within the ambulacral tracts. The ambulacral podia could have been connected to external radial water vessels lying in the floors of the ambulacral tracts but no such external connection is available for the supposed podia of the diplopores on the general thecal surface. If the ambulacral podia were connected to the water vascular system so were all diplopores and these connections must have been internal.
The principal functions of the water vascular system in echinoderms seem to be respiration and food gathering. If, in diploporites, the water vascular system did not contribute to food gathering it was probably respiratory. Hence I suggest that diplo- pores were connected to the water vascular system internally rather than to a wholly unique internal organ system although this latter idea is quite plausible. Humatipore- bearing cystoids evolved from diplopore -bearing cystoids so these too probably had an internal water vascular system. Oral pores, which correspond to the ambulacral pores of the Aristocystitidae, occur in all genera of Holocystitidae.
In summary, study of the food gathering organs and pore-structures in cystoids suggests that (i) Diploporita probably had an internal water vascular system connected to the dipores, (ii) most rhombiferans probably had external radial water vessels in the ambulacra, and (iii) the Callocystitidae (rhombiferans) may well have had both internal and external extensions of the water vascular system.
EVOLUTIONARY AND TAXONOMIC IMPLICATIONS
Evolutionary
The evolutionary significance of the above suggestions about the water vascular system is considerable. Recent echinoderms have either internal or external branches (radial canals) of the water vascular system. This has generally been assumed to have been the case with all fossil echinoderms too, and, until the recent discovery of heli- coplacoids (Durham and Caster 1963) external branches were considered to be the primitive ancestral condition. However, internal branches may have developed as early
22 PALAEONTOLOGY, VOLUME 15
as Lower Cambrian in the helicoplacoid Waucobella (Durham 1967). In the more advanced Glyptocystitida, among cystoids, there is evidence of both internal and external branches. Bather imagined that external branches evolved first and internal branches developed later by the radial vessels ‘sinking’ through the thecal plates. Both types of branch could have evolved quite independently by growing out radially from the ring canal either beneath or above the thecal plates. If such were the case both internal and external branches could have developed within the same animal, an idea which needs further investigation. The most important implication of the idea is that internal and external branches of the water vascular system need not necessarily be homologous.
TEXT-FIG. 15. Distribution in time of the five major types of cystoid pore-structures. Pr, pectinirhombs; Cr, cryptorhombs; Hr, humatirhombs ; Dp, diplopores; Hp, humatipores. Width of stippled area proportional to number of species with each type of pore-structure.
Exothecal pore-structures were individually less efficient than endothecal pore- structures as exchange systems. Among exothecal pore-structures diplopores were prob- ably the most efficient in terms of area and thickness of the exchange surface but less so in terms of preventing rupture and mixing. The distribution in time of all five types of cystoid pore-structures is summarized in text-fig. 15. All five types first appear in the Ordovician: humatipores in the Ashgill (Upper Ordovician), the others in the Tre- madoc or Arenig (Lower Ordovician). Humatirhombs are confined to the Ordovician while humatipores appear late in the Ordovician and survived to the Middle Silurian when they flourished briefly in North America. Diplopores are common in the Ordo- vician, rather rare in the Silurian and have a second peak in the Lower and Middle Devonian. The endothecal pore-structures, pectinirhombs and cryptorhombs, are common in the Ordovician and dominate cystoid faunas in the Silurian and Devonian. Thus it seems that within the Rhombifera the least efficient type survived for the briefest period and became extinct first. In the Diploporita the situation is more complex. To a large extent humatipores and diplopores are mutually exclusive in time. Diplopores
PAUL: EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
23
are replaced by humatipores as the dominant type in the middle Silurian but reassert themselves again in the Devonian. The less efficient type again survived for the shorter period and became extinct first. A possible explanation for these reversals in relative abundance may lie in the susceptibility to rupture of diplopores. Although humatipores are less efficient than diplopores they are much less likely to be ruptured and certainly would not be subject to predation. Later Devonian diplopores may have evolved the ability to secrete noxious substances to inhibit predation. It is also possible that the reversals of dominance are artifacts of preservation or collection failure. Two points may be made: (i) If it were not for the Holocystites fauna of North America, which contains almost all known Silurian diploporites, there would be no dominance of humatipores in the Silurian, (ii) The Sphaeronitidae is a large and varied family in the Ordovician and to a lesser extent in the Devonian too, but to date not one single undoubted specimen of Sphaeronitidae has been collected from Silurian strata. {AUo- cystites Miller is not a diploporite cystoid let alone a sphaeronitid and Austrocystites Brown is not sufficiently well known to be certain of its affinities.)
From an evolutionary and functional point of view exothecal pore-structures were relatively unsuccessful. However, they can hardly be regarded as failures. All three types were able to support very much larger thecae than either endothecal type. Thecae 100-150 mm in major diameter are not at all unusual in the Caryocystitidae, Aristo- cystitidae, or Holocystitidae. Even Echinosphaerites exceeds 100 mm in diameter and Calyx may reach a length of 450 mm (Chauvel 1941, pi. 1, fig. 1). Thecae of dichoporite cystoids rarely exceed 35 mm in diameter and 60 mm is about the limit with the excep- tion of the spindle-shaped genus Rhombifera. In the Ordovician the Caryocystitidae and Diploporita were very successful both in terms of individuals and species.;This is also true of the Holocystitidae in the Middle Silurian. Diplopore-bearing cystoids survived into the Middle Devonian. Thus although less successful than cystoids with endothecal pore- structures, those with exothecal pore-structures formed an important experiment in echinoderm evolution.
Taxonomic
The principal characters which are claimed to unite the cystoids as a group are: (i) a theca which is not readily divisible into a ventral tegmen and a dorsal calyx, (ii) biserial, unbranched ambulacral appendages called brachioles, (iii) pore-structures developed in the thecal wall. The first character is generally present but Caryocrinites is an excep- tion with a distinct tegmen. Evidence for the second character is incomplete. In the Diplorita only three genera have exothecal ambulacral appendages preserved: two are undoubtedly uniserial and the third may be biserial (Chauvel 1966). Apparently all three superfamilies of Rhombifera have biserial structures, however branched and pin- nate recumbent arms are present in the Glyptocystitida, pinnate free arms in the Hemicosmitida and pinnate recumbent arms in the Caryocystitida. Cystoid pore- structures belong to five distinct types some of which are so distinct from others as to preclude close relationship. Furthermore, similar pore-structures are developed in some crinoids, paracrinoids, and eocrinoids and in all blastoids and parablastoids. Clearly, the presence of pore-structures is not unique to cystoids nor are cystoid pore-structures of a unique type. Evidence for external branches of the water vascular system is tenuous to say the least in the Diploporita but quite strong in the Rhombifera. To summarize:
24
PALAEONTOLOGY, VOLUME 15
the characters which are supposed to unify the cystoids are not found throughout the group nor are they confined to the group. The Diplorita and Rhombifera share no com- mon characteristic that they do not also share with at least one other class of primitive echinoderms.
There are two alternative solutions: either to resurrect the old class Cystoidea for cystoids, paracrinoids, eocrinoids, blastoids, and parablastoids or to recognize the Rhombifera and Diploporita as separate classes. I prefer the second to emphasize the distinctiveness of the latter two groups but in either case the Rhombifera and Diplo- porita have the same taxonomic rank as the Eocrinoidea, Paracrinoidea, etc. Paul (1967Z>, 1968) suggested that the Diploporita and Rhombifera be recognized as distinct classes but without presenting all the evidence or defining the classes. Now that detailed evidence, at least as regards the pore-structures, has been presented, formal definitions of the major taxa of cystoids may be given as follows:
CLASS DIPLOPORITA Mullcr 1854, nom. transl.
Definition. Crinozoa with exothecal pore-structures (dipores) which consist of a single thecal canal, globular or pyriform theca generally composed of a large number of randomly arranged plates, which are usually all pierced by pore-structures. With or without a true stem, ambulacral appendages uniserial but very rarely preserved, water vascular system probably internal.
The lower divisions within the class are as in Kesling (1963, 1968) except that the Holocystitidae is separated from the Aristocystitidae and the latter is elevated to Superfamily rank (Paul 1971).
CLASS RHOMBIFERA Zittcl 1880 nom. transl.
Definition. Crinozoa with exothecal or endothecal pore-structures (rhombs) which consist of rhombic sets of thecal canals, globular pyriform or oval theca, with true stem at least early in development, ambulacral appendages biserial (arms or brachioles), water vascular system probably with external radial branches.
ORDER DiCHOPORiTA Jackel 1899 emend. Paul 1968 Definition. Rhombifera with endothecal pore-structures (pectinirhombs and cryptorhombs) composed of dichopores, with theca composed of a small number of plates arranged in three to five circlets, pore-structures only developed across certain sutures, true stem throughout life.
This order contains two superfamilies, the Glyptocystitida and the Hemicosmitida. The family Polycosmitidae is assigned to the Hemicosmitida, otherwise the classification is as given in Kesling 1963, 1968.
ORDER EisTULiPORiTA Paul 1968
Definition. Rhombifera with exothecal pore-structures (humatirhombs) composed of fistulipores, with theca composed of a large number of plates which may be added during growth and are randomly arranged, pore-structures developed across all possible sutures, true stem lost in adult or possibly totally absent in rare examples.
This order contains one superfamily, the Caryocystitida. Detailed classification as in KesUng 1963, 1968 except that the family Stichocystidae is added. Kesling’s superfamily Polycosmitida thus becomes defunct.
Even this classification, which is more complex than previous classifications, may oversimplify cystoid evolution. In particular the relationship between the two rhombi- feran orders is more assumed than real. The only character unique to the two orders is the presence of rhombs. However fistulipores could not have evolved from dichopores nor vice versa. Rhombic structures or ornament may develop in any animal or plant group with a tesselated pavement of polygonal units. Rhombs can be recognized in
PAUL: EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
25
most classes of echinoderms, in fish with heavily armoured bodies, in tortoiseshell scutes, even in some calcareous algae. The rhombic outline is a geometrical result of the mode of growth of closely fitting polygons and has no other significance. The presence of respiratory rhombs in the Dichoporita and Fistuliporita is due to parallel or convergent evolution and, in the absence of other shared characteristics, implies no particularly close relationship.
APPENDIX 1
List of species with exothecal pore-structures examined in this study.
Class RHOMBiFERA Muller Type of pore-structure
Order fistuliporita Paul Superfamily caryocystitida Family caryocystitidae
Lophotocystis granatum (Wahl.)
L. malaisei (Regnell)
L. augustiporus (Regnell)
L. sp. nov. (Haverfordwest, Wales) L. araueus (Schlotheim)
L. sp. (Skalberget, Sweden) Heliocrinites ovalis (Angelin)
H. guttaeformis Regnell H. stellatus Regnell H. balticus Eichwald H. sp. nov. (Rhiwlas, Wales) Coryocystites dubia Angelin (= C. angelini Auctt.)
C. lageualis Regnell
Humatirhombs Fistulipores Rhombs
simple simple
„ complex
compound simple
complex
Family Echinosphaeritidae
Echiuosphaerites aurautium (Gyll.) compound simple E. aurautium suecicus Jaekel ,, „
E. aurautium americauum Bassler „ „
E. arachuoides Forbes „ „
Family Ulrichocystidae
Ulrichocystis eximia Bassler simple
Class Diploporita Muller Superfamily Sphaeronitida
Family Sphaeronitidae Pore-structures
Sphaerouites pomum (Gyll.) Diplopores s.s.
S. sp. nov. (Raback, Sweden) „
S. globulus (Angelin) „
S. sp. nov. (Skalberget, Sweden) „
S. litcbi (Forbes) „
S. pyriformis (Forbes) „
S. sp. (Glyn Ceiriog, Wales) „
Haplosphaerouis oblouga (Angelin) ,,
H. kiaeri Jaekel „
//. sp. nov. (Haverfordwest, Wales) „
Eucystis barreudeua Haeckel „
F. (Barrande) ,,
26
PALAEONTOLOGY, VOLUME 15
E. munitus (Forbes) Diplopores s.s.
E. quadrangularis Regnell „
E. angelini Regnell „
E. raripimctata Angelin „
Archegocystis stellulifem (Salter) „
''Sphaeronis' dalecarliciis Angelin „
'' Sphaeronis' punctatus Forbes „
Family Holocystitidae
Holocystites cylindricus (Hall) Humatipores
H. alteniatiis (Hall) „
H. obnormis Hall „
H. scutellatus Hall „
Pentacystis simplex Paul „
P. wykqffi (Miller) „
P. sphaeroidalis (Miller & Gurley) „
Trematocystis globosiis (Miller) „
T. rotimdus (Miller) „
Pustidocystis pentax Paul „
P. omatissimus (Miller) „
Brightonicystis gregariiis Paul „
Superfamily Aristocystitida Family Aristocystitidae
Aristocyslites bohemicus Barrande Diplopores
A. siibcyliiidriciis Barrande „
A. sp. (Knock, England) „
Trianwra titmida (Miller) „
T. ventricosa (Miller) „
T. multiporata Paul „
T. laevis Paul „
T. sp. (Big Creek, Indiana, U.S.A.) „
Sinocystis loczyi Reed „
'’Sphaeronis' shihtienensis Reed „
Superfamily Glyptosphaeritida Family Glyptosphaeritidae Glyptosphaerites leuchtenbergi (Volborth)
Family Dactylocystidae
IRevalocystis kearsargensis (Stauffer) „
Family Gomphocystitidae
Gomphocystites indianensis Miller „
Family Protocrinitidae
IRegnellicystis sp. (Rye Cove,
Virginia, U.S.A.)
APPENDIX 2
Formal definitions of Lophotocystis nov. and Heliocrinites s.s. : Genus Lophotocystis nov. (Lophotos, Gr. crested).
Type species. Echinosphoerites granatum Wahlenberg.
PAUL; EXOTHECAL PORE-STRUCTURES IN CYSTOIDS
27
Definition. A genus of Caryocystitidae with globular theca, fifty to several hundred thecal plates not folded about rhomb axes; humatirhombs with simple fistulipores raised in distinct ridges on external surface of plates.
Genus Heliocrinites Eichwald 1840.
Type species. Echinosphaerites balticum Eichwald.
Definition. A genus of Caryocystitidae with oval to globular theca, fifty to several hundred plates usually folded about rhomb axes to produce a surface ornament of triangular depressions; humati- rhombs with compound fistulipores almost completely or totally buried within plates.
REFERENCES
BARRANDE, J. 1887. Systeme Silurien du centre de la Bolieme. E partie: Recherches paleontologiqiies.
7. Classe des Echinodennes. Ordre des Cystides. xviH-233 pp., 39 pis. Leipzig and Prague.
BATHER, F. A. 1900, The echinoderms. In lankester, e. r. A treatise on zoology, 3, vi + 344 pp., London.
1906. Echinodermata, In reed, f. r. c. Lower Palaeozoic fossils of the Northern Shan States,
Burma. Mem. geol. Surv. India. Palaeont. indica, n.s. 2 (3), 6-40, pis. 1-2.
CHAUVEL, J. 1941. Recherches sur les Cystoides et les Carpoides Armoricaines. Mem. Soc. geol. miner. Bretagne, 5, 286 pp., 7 pis.
1966. Echinodermes de TOrdovicien du Maroc. Cahiers Paleont. 120 pp., 16 pis.
DURHAM, J. w. 1967. Notes on the Helicoplacoidea and early echinoderms. J. Paleont. 41, 97-102, pi. 14.
and CASTER, k. e. 1963. Helicoplacoidea: a new class of echinoderms. Science, 140, 820-822.
EICHWALD, E. 1840. Siir le Systeme Silurien de TEsthonie. 222 pp. St. Petersburg.
ENDEAN, R. 1966. The coelomocytes and coelomic fluids. In boolootian, r. a. (ed.). Physiology of Echinodermata. Ch. 13, 301-328.
FARMANFARMAIAN, A. 1966. The respiratory physiology of echinoderms. Ibid. Ch. 10, 245-265. HARVEY, E. N. 1928. The oxygen consumption of luminous bacteria. J. gen. Physiol. 11, 469-475. HUDSON, G. H. 1915. Some fundamental types of hydrospires with notes on Porocrinus smithi Grant. Bull. N.Y. St. Mus. Ill, 163-173, 2 pis.
HUXLEY, T. H. (Translator) 1854. On the structure of the echinoderms, by Prof. J. Muller. Ann. Mag. nat. Hist. (2 ser.) 13, 1-24, 112-123, 241-256.
HYMAN, L. H. 1955. The Invertebrates. 4. Echinodermata. The coelomate Bilateria, 763 pp., 280 figs. New York, London, and Toronto.
JAEKEL, o. 1899. Stammesgeschichte der Pelmatozoen. I. Thecoidea and Cystoidea, 442 pp., 18 pis., 88 figs. Berlin.
1918. Phylogenie und System der Pelmatozoen. Paldont. Z. 3, 1-128, 114 figs.
KESLiNG, R. V. 1963. Key for the classification of cystoids. Contr. Mus. Paleont. Univ. Mich. 18, 101- 116.
1968. Cystoids, In moore, r. c. (Ed.) Treatise on invertebrate Paleontology. S. Echinodermata, 1.
S85-S267.
KROGH, A. 1941. The comparative physiology of respiratory systems. 172 pp., 84 figs. Philadelphia. LANG, E. D. 1923o. Evolution: a resultant. Proc. geol. Ass. Lond. 34, 7-20.
19236. Trends in British Carboniferous corals. Ibid. 34, 120-136.
MULLER, J. 1854. liber den Bau der Echinodermen. Abh. preuss. Akad. IL/s'S'. (1854), 123-219, pis. 1-9 (see Huxley, T. H. above).
NICHOLS, D. 1959. Mode of life and taxonomy in irregular sea-urchins. Pubis. Syst. Ass. 3, 61-80. 1962. Echinoderms. 200 pp., 26 figs. London.
PAUL, c. R. c. 1967fl. Hallicystis attenuata, a new callocystitid cystoid from the Racine Dolomite of Wisconsin. Contr. Mus. Paleont. Univ. Mich. 21, 231-253, 4 pis., 8 figs.
19676. Cyclocystoidea, Eocrinoidea, Rhombifera, Diploporita and Paracrinoidea, In harland,
w. B. et al. (Eds.) The Fossil Record. London (Geol. Soc.), pp. 566-570.
1968. The morphology and function of dichoporite pore-structures in cystoids. Palaeontology,
11, 697-730, pis. 134-140.
28 PALAEONTOLOGY, VOLUME 15
PAUL, c. R. c. 1971. Revision of the North American Holocystites fauna (Diploporita). Fieldiaua, Geology, 24, 166 pp.
REGNELL, G. 1951. Revision of the Caradocian-Ashgillian cystoid fauna of Belgium with notes on isolated pelmatozoan stem fragments. Mem. Inst. r. Sci. not. Belg. 120, 1-47, 6 pis.
RUDWiCK, M. J. s. 1964. The inference of function from structure in fossils. Br. J. Phil. Sci. 15 (57), 27-40.
ziTTEL, K. A. VON, 1880. Hondbucli der Palaeontologie. 1. Protozoa, Coelenterata, Echinodermata and Molluscoidea. 765 pp., 558 figs. Miinchen and Leipzig.
C. R. C. PAUL
Dept, of Geology University of Reading
Revised typescript received 6 May 1971 Reading, RG6 2AB
THE POST-CRANIAL SKELETON OF THE TRIASSIC ORNITHISCHIAN DINOSAUR FABROSA URUS AUSTRALIS
by R. A. THULBORN
Abstract. The post -cranial skeleton of the ornithischian dinosaur Fabrosaums australis, Ginsburg 1964, is described for the first time from material from the Upper Triassic Red Beds of Lesotho. Certain skeletal features (e.g. the tibio-femoral ratio) indicate that Fabrosawus should be assigned to the family Hypsilophodontidae of the suborder Ornithopoda. Fabrosawus is envisaged as a small, unarmoured and habitually bipedal dinosaur with distinct cursorial potential. Muscle scars on the femora and pelvic girdle bones point to a system of pelvic musculature not unlike that proposed by Romer (1927) for Thescelosaurus and by Gallon (1969) for Hypsilo- phodon.
The problem of ornithischian origins is briefly examined. Fabrosawus presents few primitive characters and is of little assistance in any attempt to locate the possible ancestors of the Ornithischia. It is concluded that Fabrosaiiriis represents the earliest known portion of a hypsilophodont stock which persisted through the greater part of the Mesozoic era and which gave rise, even if indirectly, to such varied ornithischian groups as the iguanodonts, hadrosaurs, and ceratopsians. Triassic relatives of Fabrosaurus may be discerned as far afield as China (Tatisaurus) and Argentina (Pisanosaiirus). Lycorhimts [Fleterodoutosaiirus], also from the Upper Trias of southern Africa, appears to represent an extremely early, rather specialized, and short-lived hypsilo- phodont divergence.
Knowledge of the earliest recorded (Upper Triassic) ornithischian dinosaurs is based upon rare and fragmentary fossils. In only two cases, Lycorhimts [Heterodonto- saurus] and Fabrosaurus, are the skulls at all well known (Crompton and Charig 1962; Thulborn 1970n, 1970Z?). Post-cranial bones have been described only in the South American Pisanosaurus (Casamiquela 1967), and these are far from complete.
The genus Fabrosaurus was established by Ginsburg (1964) on the basis of a jaw fragment from the Upper Triassic Red Beds of Basutoland (now Lesotho). Subsequent discoveries have permitted description of the Fabrosaurus skull in near-entirety (Thul- born 1970n). This paper concerns the previously unknown post-cranial skeleton of Fabrosaurus. Hence Fabrosaurus becomes perhaps the best known pre-Jurassic orni- thischian.
Material
The material described below is preserved in the collection of the Zoology Department at University College, London. It was collected by Dr. K. A. Kermack and Mrs. F. Mussett during the 1963-1964 expedition from University College to Basutoland. The material was obtained from the Upper Triassic Red Beds of the Stormberg Series on the northern flank of Likhoele Mountain, near the settlement of Mafeteng (see map, text-fig. 1). Greater stratigraphic precision is not possible for two reasons: firstly, because the classic subdivisions of the Trias, established upon marine faunas, cannot be extended into continental deposits such as the Red Beds, and secondly, because of the lack of suitable zone fossils within the Upper Trias of southern Africa. Such zonation as has been achieved in the late Trias of this area is not at all detailed and cannot be extended successfully over large areas.
The bones described below are all from ‘assemblage B. 17’ mentioned in separate accounts of the Fabrosaurus skull and dentition (Thulborn 1970u, 1971). This assemblage (text-fig. 2) contains at least [Palaeontology, Vol. 15, Part 1, 1972, pp. 29-60.]
30
PALAEONTOLOGY, VOLUME 15
two individuals of Fabrosaurus, the smaller one being the better represented. Assemblage B. 17 com- prises: skull fragments (both individuals), numerous isolated teeth, 44 vertebrae or parts of vertebrae (? both individuals), rib fragments and ossified tendons (? both individuals), left and right scapulae, left scapula (larger individual), left humerus, left humerus (larger individual), right radius and ulna, left radius (larger individual), parts of right carpus and manus, paired ilia, ischia, and pubes, paired femora, tibiae, and fibulae, 2 left tarsal bones, left metatarsus and parts of right metatarsus, phalanges of left and right feet.
TEXT-FIG. 1. Maps showing provenance of assemblage B. 17 {Fabrosaurus australis). The material was collected at locality a, on the northern flank of Likhoele Mountain. Shaded areas (larger map) repre- sent outcrops of Drakensberg volcanics overlying the Red Beds.
Preservation and preparation of material
The matrix is a tough medium-grained sandstone of bright red colour. The bones are preserved in grey or white calcareous material which is usually stained black, brown, or red. Nearly every bone is traversed by numerous fine cracks — the ‘checkering’ noted by Simmons (1965) in his account of reptiles from the Chinese Trias. These fissures doubtless represent sun-cracking acquired by the bones prior to burial. Similar effects may be observed at present in southern Africa, where even the stoutest bones (e.g. those of horses and oxen) are completely shattered after a few weeks’ exposure. Each bone is enclosed within a coat of reddish-black ferruginous material. This coating is usually one or two millimetres thick and tends, where it is weathered, to part from the underlying bone very easily (a feature which is of considerable use in preparation). When freshly exposed, however, this encrustation adheres very firmly to the bone by virtue of innumerable intrusive veinlets. This is especially noticeable at the ends of the long bones and elsewhere at points of incomplete ossification (e.g. the dorsal margin of the scapula). It is also in these regions that the ferruginous cortex is thickest.
The material was prepared by both mechanical and chemical means. Soft or weathered matrix was
THULBORN: POST-CRANIAL SKELETON OF FABROSAURUS
31
removed with a mounted needle. Tougher matrix was removed rapidly by the use of a light hammer with small cold chisels and straight dental probes. A variable-speed vibro-tool was used to the same effect; this lent itself to much finer control and was safely employed very close to the bone. Prior to chemical treatment all exposed bone was coated in a thin (1 part to 4) solution of polybutyl meth- acrylate in ethyl acetate. This protective coating, which also served to consolidate friable bone surfaces, may be removed at any time by washing in ethyl acetate. Subsequently the material was very thoroughly dried and then immersed in cold dilute (10-15%) solutions of either acetic or formic acids in water. The period of immersion varied between 30 minutes and 3 hours. After thorough washing and drying the material was further prepared mechanically. These processes were repeated until no more matrix could safely be removed (ideally until individual bones were freed from the matrix).
TEXT-FIG. 2. Fabrosaurus australis. Assemblage B. 17. X0 38. Several bones (mainly fragments of vertebrae) have been omitted for clarity. Two individuals are present, the larger one being represented by the humerus {h I) and the scapula (s I) at left.
DESCRIPTION
Explanation of abbreviations used in text-figures
|
ac |
acetabular margin |
cleft between proximal trochanters of |
|
|
acet |
acetabulum |
femur |
|
|
a e |
anterior embayment of ilium |
fc |
facet for coracoid |
|
a pr |
anterior process |
fib |
fibula |
|
art |
proximal articular surface |
fib 1 |
left fibula |
|
c |
cranial fragments |
fib r |
right fibula |
|
c2 |
second distal carpal |
fl |
left femur |
|
c3 |
third distal carpal |
fr |
right femur |
|
cap |
capitulum |
g |
glenoid cavity |
|
cf |
facet for chevron bone |
gd |
grooved dorsal margin of ischium |
|
eg |
claw groove |
g t |
greater trochanter |
|
cn c |
cnemial crest |
h |
head |
|
d |
diapophysis |
hdr |
bones of right hand |
|
d c |
distal condyle |
hi |
left humerus |
|
dp c |
delto-pectoral crest |
ig |
intermalleolar groove |
|
32 |
PALAEONTOLOGY, VOLUME 15 |
||
|
ill |
left ilium |
P if |
posterior intercondylar fossa |
|
Up |
iliac process of ischium |
p U |
posterior iliac shelf |
|
Ur |
right ilium |
P 1 |
left pubis |
|
i m |
medial malleolus |
po |
postzygapophysis |
|
is 1 |
left ischium |
pop |
postpubis |
|
is p |
ischiadic peduncle of ilium |
PP |
prepubis |
|
is r |
right ischium |
ppd |
pubic peduncle of ilium |
|
1 c |
lateral condyle |
ppr |
posterior process |
|
It |
lesser trochanter |
P r |
right pubis |
|
m |
insertion of flexor tibialis (ischium) |
P t |
insertion of coccygeo-femoralis longus |
|
m c |
medial condyle |
(femur) |
|
|
mcl to mc4 |
metacarpals i to iv |
pub p |
pubic process of ischium |
|
mtl to mt4 |
metatarsals i to iv |
pz |
prezygapophysis |
|
mt 1 |
left metatarsus |
r c |
radial condyle of humerus |
|
mt r |
right metatarsus |
rd |
radius |
|
n |
notch in scapular margin |
rd r |
right radius |
|
n c |
neural canal |
s 1 |
left scapula |
|
n s |
neuro-central suture |
s r |
right scapula |
|
n sp |
neural spine |
tf |
fourth trochanter |
|
ob |
obturator process of ischium |
t 1 |
left tibia |
|
ob f |
obturator foramen |
t P |
transverse process |
|
o m |
lateral malleolus |
t r |
right tibia |
|
0 t |
ossified tendons |
tub |
tuberculum |
|
pa |
parapophysis |
u |
ulna |
|
p e |
insertion of ilio-femoralis externus |
u c |
ulnar condyle of humerus |
|
(femur) |
u r |
right ulna |
Vertebral column (text-figs. 3, 4, and 5)
The material includes vertebrae from most regions of the column. It is impossible to estimate the vertebral formula with any accuracy since the material, which is rather fragmentary, may have been derived from more than one anhoal.
The best-preserved cervical vertebrae (text-fig. 3a) come from the middle of the neck. Their long and narrow centra have deeply excavated flanks and give the impression, in ventral view, of having been ‘pinched in’. These excavations probably represent areas of origin for the rectus capitis muscula- ture (running forwards to insert on the occiput) and serve to distinguish the neck vertebrae from others in the column. Each centrum bears a prominent median keel on its ventral surface. The terminal articular faces of the centra are in most cases obscured by thick crusts of haematite; these faces are shield-shaped, wider than high. The foremost centrum shown in text-fig. 3 tends slightly to the opistho- coelous condition; the succeeding cervical centra are distinctly amphicoelous. The parapophysis is a poorly defined rugosity near the antero-dorsal corner of the centrum; it occurs at successively higher levels as it is traced back through the neck vertebrae. The other area of rib attachment, the diapophysis, is a small rounded eminence situated on, or slightly above, the mid-point of the persistent neuro-central suture. In the hindmost neck vertebrae the diapophysis is at a somewhat higher level and is extended into a short ventro-lateral process. The neural arch is about as high as the centrum whilst the neural spine is merely an insignificant median ridge. The rounded and tongue-like prezygapophyses overhang the front of the centrum; the postzygapophyses are shorter and are rather angular in outline. The articular faces of the zygapophyses are inclined at about 15° from vertical.
The dorsal vertebrae (text-figs. 3 and 5) are distinguished from the neck vertebrae by virtue of their more robust construction, principally through their broad centra and stout transverse processes. Each spool-shaped centrum has smoothly rounded flanks and bears a very faint median keel on its ventral surface. At their extreme anterior and posterior ends the lateral and ventral faces of the centra bear traces of wrinkling and weak longitudinal fluting. This ornament probably marks the former attachment of hypaxial trunk muscles. All of the dorsal centra are amphicoelous and have terminal articular faces of sub-circular outline. The very distinctive transverse processes are remarkably thick and massively constructed where they merge with the neural arch. Each process extends horizontally and terminates in an elliptical and convex facet (the diapophysis) for the attachment of the tuberculum from the associated rib. The facet for the capitulum of the rib (i.e. the parapophysis) is located on the
THULBORN: POST-CRANIAL SKELETON OF FABROSAURUS 33
TEXT-HG. 3. Fabrosaurus australis. Cervical and dorsal vertebrae, xl-5. a, parts of three cervical vertebrae in right lateral view, b, two dorsal vertebrae in left ventro-lateral view, c-f, reconstructed cervical vertebrae in dorsal, right lateral, anterior, and ventral views, g-k, reconstructed dorsal vertebra in left lateral, anterior, ventral, and dorsal views.
C 8472
D
34
PALAEONTOLOGY, VOLUME 15
anterior edge of the transverse process. In the hindmost dorsal vertebrae the parapophysis is found much closer to the diapophysis. The neural spines are best illustrated by three examples together with a number of ossified tendons (text-fig. 5F). Each blade-hke spine arises from the entire length of the neural arch and is of consistent antero-posterior width for its entire height. Faint vertical striae on the flanks of these spines indicate the former attachment of epaxial trunk muscles. The thickened dorsal
TEXT-FIG. 4. Fabrosaiirus australis. Sacral and caudal vertebrae, X 1-5. a-d, sacral centrum in ventral, left lateral, dorsal and anterior views, e, partial neural arch, from sacral region, in dorsal view, f, sagittal section through a sacral centrum, g, partial neural arch, from anterior caudal region, in right lateral view, h, anterior caudal vertebra in dorsal view. J, neural arch, from middle caudal region, in left lateral view, k, posterior caudal vertebra in right lateral view.
margins of the neural spines were probably embedded, during life, in the fibrous tissues of the dermis. The postzygapophyses are stout spatulate processes which overhang the rear end of the centrum ; the prezygapophyses are similar in outline, but shorter. In the anterior and middle dorsal vertebrae the articular faces of the zygapophyses are inclined at about 20° from horizontal; in the posterior dorsal region these articular faces are practically horizontal.
The sacral vertebrae (text-fig. 4) are represented by five centra and parts of two neural arches. Attachment scars on the ilium (text-fig. 8c) suggest that the Fabrosaurus sacrum incorporated five vertebrae. The sacral centra are distinguished by their remarkable width. Each one is constricted in the middle and bears a distinct median keel on its ventral surface. The terminal articular faces of the
THULBORN: POST-CRANIAL SKELETON OF FABROSAURUS
35
centra are quite flat, crescentic in outline, and wider than high. There are no indications of any fusion between the centra. The dorsal view of a centrum (text-fig. 4c) shows four large facets for neural arch attachment and the deep basm-Uke excavation which represents the floor of the neural canal. A for- tuitous sagittal section through one centrum (text-fig. 4f) reveals the full extent of the ventrally inflated neural canal. The neural arches of the sacral series are represented by fragments found at some distance from the nearest centrum. Depressed areas flanking the extremely thin neural spine merge imperceptibly with the flat dorsal surfaces of the transverse processes. Each stout and horizontal transverse process is some 7 mm long and is of uniform width to its free end. The prezygapophyses are situated close to the midfine, indicating that the postzygapophyses (not preserved) must have been very close together. Each prezygapophysis has a sub-rectangular profile and bears an articular face which is ahnost vertical.
The caudal vertebrae (text-fig. 4) are rather poorly represented. The anterior tail vertebrae are dis- tinguished by their long and extremely thin transverse processes. These processes are directed laterally and slightly to the front. The spool-shaped centra have flat terminal faces of sub-circular outline. At its postero-ventral margin each centrum is bevelled to form a crescentic facet for the attachment of a chevron bone (haemal arch); a similar, though smaller, facet is present at the antero-ventral margin. Each tall and blade-like neural spine is inclined to the rear and arises only from the posterior half of the neural arch. The bluntly rounded prezygapophyses overhang the front of the centrum ; the postzy- gapophyses are represented merely by raised areas flanking the postero-ventral part of the neural spine. The articular faces of the zygapophyses are inclined at about 20° from vertical.
The middle caudal vertebrae are represented by a single neural arch (text-fig. 4j). This differs from the neural arches of the anterior tail vertebrae in lacking any trace of the transverse processes.
The distal parts of the Fabrosaurus tail are represented by a single vertebra (text-fig. 4k). Its slender and constricted centrum terminates in fiat circular surfaces and bears distinct facets for the attachment of chevron bones. The neural arch consists of little more than a small pyramidal eminence. The slender and finger-like prezygapophyses are situated close to the mid-fine; the postzygapophyses appear to have been very weakly developed or absent.
The ribs (text-fig. 5) are represented by numerous fragments. One rib appears to have come from the posterior cervical region (text-fig. 5e). The proximal part of this rib is flattened and rather plate-like. Capitulum and tuberculum are both well defined (the latter being distinctly the longer) and enclose an angle approaching 90°. More distally the antero-lateral edge of the rib tends to a definite sharp- ness whilst the postero-medial margin remains thicker and well rounded. The dorsal ribs are also two-headed (dichocephalous), though the hindmost ones show a tendency to the single-headed (holo- cephalous) state. The posterior dorsal region is, however, rather poorly known and there is no incon- trovertible evidence that any of the ribs ever fully attained the holocephalous condition. In the larger ribs from the front of the thorax (text-figs. 5a-c) capitulum and tuberculum are both well developed and diverge at somewhat less than a right angle. The expanded proximal part of each rib passes distally into a long, slender, and rod-like portion. The entire rib is arched to the exterior (the greater part of this flexure occurring in the proximal one-third of the bone) and the antero-lateral edge is noticeably thinner and sharper than the postero-medial edge. In the posterior dorsal ribs (text-fig. 5d) the capitu- lum and tuberculum are situated closer together. These ribs are distinguished from those at the front of the thorax by being shorter, thicker, and more obviously arched to the exterior. The delicate trans- verse processes of the anterior tail vertebrae (text-fig. 4h) probably represent caudal ribs which are fused on to the vertebrae.
Eragments of ossified tendons (text-fig. 5f-h) are preserved alongside the neural spines of the dorsal and caudal regions. It is probable that the tendons originally extended to cover much of the tail in Fabrosaurus. Each slender, compressed, and rod-like tendon is applied to the flanks of up to five successive neural spines. At one end (anterior or posterior) the tendon tapers to a point; towards the other end it gradually widens and splays out into several narrow rays. Each tendon is about a milli- metre wide and is marked with fine longitudinal striae. The tendons are grouped in definite bundles and in lateral view (text-fig. 5f) there is a slight suggestion of these bundles being disposed in a lattice- like pattern with diamond-shaped interstices.
No chevron bones are preserved in assemblage B. 17. But this is not surprising in view of the paucity of tail bones in general. Chevron bones were clearly present since the caudal centra bear prominent facets for their attachment.
36
PALAEONTOLOGY, VOLUME 15 -tub
Hralis. Ribs and ossified tendons, X 1-2 (except figure H). a-b, anterior thoracic rib in anterior and medial views, c, middle thoracic rib in anterior view, d, posterior thoracic rib in medial view, e, posterior cervical rib in posterior view, f, ossified tendons associated with neural spines from three dorsal vertebrae, g, anterior view of a neural spme (from dorsal region) to show arrangement of the ossified tendons, h, detail of ossified tendons, x 2.
Pectoral girdle (text-fig. 6)
The scapula (text-fig. 6) is a tall, blade-like bone which is roughly triangular in lateral profile. The widely expanded dorsal margin is less solidly constructed than the rest of the bone and has a distinctly porous texture. This porous zone (text-fig. 6e) represents a region of transition between the scapula proper and a cartilaginous supra-scapula. At its postero-dorsal corner the bone is extended into a short tongue-like process; the antero-dorsal corner is obtusely angular. Vertical striae on the dorsal and central parts of the lateral scapular surface doubtless mark the origin of a broad sheet of muscle (the scapular deltoid) running down to insert at the proximal end of the humerus. The ventral part of the scapula is rather ‘foot-like’ in profile due to the presence of a salient and forwardly projecting ‘acro- mial’ process. In this region the depressed lateral face of the scapula probably bore the origin of a second shoulder muscle (scapulo-humeraUs anterior) which inserted, like the deltoid, on the proximal part of the humerus. The posterior view (text-fig. 6b) demonstrates the strong latero-medial compres- sion affecting the dorsal half of the scapula and also shows that the bone is elegantly curved so as to follow, in life, the convexity of the underlying rib cage. The glenoid cavity is roughly oval in plan, higher than wide, and opens postero-ventrally ; it is confluent antero-ventraUy with a shallow trough which received the dorsal margin of the coracoid. This trough is not as broad as the glenoid but is con- siderably longer owing to its anterior prolongation beneath the ‘acromial’ process.
No coracoid has been recovered from the material. Since the scapula bears a salient ‘acromial’ process (which lengthens the region for coracoid attachment) it is reasonable to infer that the coracoid was of considerable size. A distinct notch in the ventro-medial margin of the scapula (text-fig. 6c) indicates that the coracoid was perforate. In related genera (such as Hypsilophodon) this scapular notch is continuous with a coracoidal foramen which served to transmit the supracoracoid nerve and small blood vessels.
Fore limb (text-fig. 7)
The slender and columnar humerus (text-fig. 7a-e) is expanded at each end and is almost imper- ceptibly arched to the front. In anterior view (text-fig. 7a) it may be seen that the proximal end is
THULBORN: POST-CRANIAL SKELETON OF FABROSAURUS
37
a little wider than the distal end and that the humeral shaft is narrowest at a point slightly distal to its centre. Both proximal and distal expansions are directed principally to the medial side. The blunt and slightly projecting proximo-medial corner constitutes the head of the humerus; the proximo-lateral corner is obtusely angular in profile. The lateral edge of the bone is fairly straight and the erect
TEXT-FIG. 6. Fabrosawus australis. Scapula, a-b, left scapula (from larger animal in assemblage B. 17) in lateral and posterior views, X 1 . c-d, ventral part of same scapula in medial and ventral views, X 1 - 5. E, dorsal part of left scapula (smaller animal) showing incompletely ossified marginal zone, x 2.
F, ventral part of right scapula (smaller animal) in medial view, X 2.
delto-pectoral crest, which afforded insertions for some of the principal shoulder muscles, accounts for some 27% of the total humeral length. The crest is situated very high up on the humerus and appears, in fact, to be more restricted in distal extent than in any other ornithischian dinosaur. The medial distal condyle is fractionally larger than its lateral counterpart. Though the condyles are slightly expanded from back to front they are not appreciably attenuated or up-curved behind.
The radius (text-fig. 7f-k) is a slender rod-like bone, distinctly shorter than the humerus, which is expanded at both proximal and distal ends. It is quite straight and untwisted. Transverse sections near the middle of the shaft are of elliptical outline, widest antero-posteriorly. Similarly the convex and crescentic proximal surface is widest from front to back. Near the proximal end the anterior edge of
TEXT-FIG. 7. Fabrosaunis australis. Fore limb bones, x 1 (except figures Lt o r). a-e, left humerus (from larger animal) in anterior, posterior, medial, proximal, and distal views, f-h, left radius (larger animal) in lateral, proximal, and anterior views, j-k, right radius and ulna (smaller animal) in antero-lateral and distal views, l, bones of right hand, as preserved, x2. m-o, right distal carpal (the second?) in proximal, palmar, and medial views, X 5. p-q, right distal carpal (the third?) in distal and proximal views, X 5. R, reconstruction of left hand in dorsal view (reconstructed portions shaded), x2.
THULBORN; POST-CRANIAL SKELETON OF FABROSAURUS
39
the radius tends to a definite sharpness. In lateral view (text-fig. 7f) the almost straight anterior margin contrasts with the concave rear edge. The distal articular surface is oval in plan and rather flat, though a shallow depression at the postero-medial margin lends it a ‘saddle-shaped’ appearance.
The sole example of the ulna (text-fig. 7j-k) is somewhat crushed and lacks the proximal end. At the middle of the bone, which is a little stouter than the radius, the shaft is elliptical in cross-section (widest from front to back). Anterior and posterior margins of the shaft are both quite thin, though the latter is distinctly the sharper. The distal articular surface has the outline of a narrow triangle with a width (latero-medial) of barely 2-5 mm. This incomplete ulna seems to be in natural relationship with the associated radius (text-fig. 7j); the ulna is situated lateral to the radius and slightly behind it, its concave medial face accommodating the rounded flank of the radius.
The carpus is represented by two distal carpal elements, both from the right fore limb of the smaller individual in assemblage B. 17. One distal carpal (text-fig. 7m-o) was found at the proxunal end of the second metacarpal. This carpal bone is a flattened irregular quadrilateral measuring some 3 mm (latero- medial) by 4-5 mm. It is nearly 2 mm thick. All of its edges are thickened and rounded and its proximal surface is shghtly depressed. The other distal carpal (text-fig. 7p-q) was located at the proximal end of metacarpal III. This small and rectangular carpal bone measures some 4 mm (latero-medial) by 2-5 mm and is compressed to a thickness of about 1-5 mm.
The bones of the manus (text-fig. 7l) are described from the right fore limb of the smaller individual in assemblage B. 17. This partial right hand comprises metacarpals I to IV together with five phalanges. Metacarpal I is a slender rod-like bone nearly 6-5 mm long. It terminates proximally in a wide and slightly inflated articular surface of elliptical outline. Distally the metacarpal is expanded into a pair of small condyles. These condyles are about equally developed and each bears a faint circular pit on its flank. Metacarpal II is decidedly narrower than metacarpal 1 and is much longer (11 mm). Meta- carpal 111 is even larger, attaining a length (estimated) of 12-5 mm. Metacarpal IV is narrower than any of the others and has a length of 8-5 mm. The fifth metacarpal has not been recovered.
The other bones of the hand are identified as the first phalanges in digits 1 to IV and the second phalanx in digit II (text-fig. 7r). The proximal phalanx of digit I is represented by a fragment close to the distal end of metacarpal I. The first phalanx in digit II hes at the distal end of metacarpal 11; this short (5-5 mm) and stout phalanx has a maximum width of almost 4 mm and bears two sub-equal condyles at the distal end. The proximal phalanx of digit 111 is fractionally shorter than that of digit II. The first phalanx in digit IV is the smallest of the proximal phalanges ; it is 3-5 mm long and has a maxi- mum width of barely 2 mm. The remaining phalanx is, to judge from its position (text-fig. 7l), the second in digit 11. This bone resembles the other hand phalanges and is 4-5 mm long. The Fabrosaurus hand is shown reconstructed with a phalangeal formula of 2 : 3 : 4 : 3 : 0 (text-fig. 7r) ; this reconstruction is based upon the hand of the related Hypsilopbodon.
Pelvic girdle (text-figs. 8 and 9)
The ilium (text-fig. 8) is a blade-hke bone roughly twice as long as it is high. The anterior iliac pro- cess is long, slender, slightly deflexed, and acutely pointed; the posterior process seems to have been considerably shorter and broader. The pubic peduncle extends antero-ventrally to articulate with the acetabular part of the pubis (text-fig. 9f-g). Striations on the pubic peduncle (sp) mark the former presence of cartilaginous tissues serving to bind the ilium to the pubis. The ischiadic peduncle is directed straight downwards ; its swollen ventral tip met the iliac process from the ischium (text-fig. 9a-c) so as to define the rear margin of the open acetabulum. Immediately above the acetabulum the lateral surface of the ihum is strongly inflated (principally to the exterior, but also shghtly in a dorsal direction). This supra-acetabular swelling isaf), which serves to deepen the acetabulum, contrasts with the generally flat remainder of the lateral surface. The rear margin of the ischiadic peduncle is extended into a thin and sharp-edged plate of bone. Behind the peduncle this bony sheet merges with the posterior iliac process and assumes the form of a horizontal shelf, its free edge being directed medially (text-fig. 8d). The ventral surface of this shelf carried the origin of the coccygeo-femoralis brevis, an important thigh muscle which inserted on the fourth trochanter of the femur (text-fig. 10) and functioned in drawing back the hind limb during locomotion. The lateral surface of the ilium bears a number of easily dis- cerned markings (text-fig. 8b). A narrow striated zone at the dorsal margin id fib) probably defines the origin of the posterior ilio-tibialis muscle; this muscle inserted on the front of the tibia and served to extend the knee joint. Directly beneath this striated zone, and immediately over the acetabulum, there
40
PALAEONTOLOGY, VOLUME 15
TEXT-FIG. 8. Fabrosaurus australis. Right ilium, X 1 . A, lateral view, b, lateral view with surface markings shown diagrammatically. c, medial view, d, dorsal view with transverse sections at the points indicated.
THULBORN: POST-CRANIAL SKELETON OF FABROSAURUS
41
lies a long, shallow, and roughened groove {il tr)\ this doubtless represents the origin of the ilio- trochantericus muscle (which inserted on the greater trochanter of the femur and functioned in femoral protraction). The striated lateral face of the anterior process {sar) presumably bore the origin of the anterior ilio-tibial is muscle (or ‘sartorius’) ; this would have assisted the posterior ilio-tibialis in extending the knee joint.
The medial surface of the ilium (text-fig. 8c) resembles its lateral surface in that the ventral parts are somewhat inflated. The dorsal half of the medial surface is ornamented with vertical striae. These markings indicate the attachment of dorsal axial muscles which would, in life, have run antero- posteriorly between the ilia. These striae (// c and lev) are most deeply impressed on the anterior half of the iUum — suggesting that the dorsal axial muscles may have been divided into anterior (ilio-costalis) and posterior (levator coccygis) groups. The medial surface of the anterior process bears two prominent and nearly horizontal ridges. At the top of the pubic peduncle there is a shallow and ill-defined depres- sion in the medial iliac surface. Behind this, and at a slightly higher level, there extends a series of four similar depressions. The first of these four lies mid-way over the acetabulum, the second over the ischiadic peduncle, the third (and largest) at the base of the posterior process, the fourth (and faintest) on the posterior iliac shelf. These five excavations (srl-srS), representing attachment areas for the sacral vertebrae, are separated by smooth and convex surfaces.
The dorsal view of the ilium (text-fig. 8d) not only illustrates the extent of the supra-acetabular swelling but also demonstrates that the thick and flat dorsal margin is not everted above the acetabulum (i.e. there is no ‘antitrochanter’). The dorsal margin is thickest over the acetabulum and has a sinuous course; it curves slightly outwards over the rear part of the acetabulum, slightly inwards over the anterior part. The anterior process is deflected to the exterior.
The ischium (text-fig. 9a-c) comprises a blade-like distal portion separated from the expanded proximal end by a long and weakly constricted ‘neck’. Pronounced torsion of this ‘neck’ region causes the proximo-lateral face to turn forwards as it is traced distally. The proximal end is composed of two stout processes separated by a shallow embayment representing the ventral border of the acetabulum. The dorsal (iliac) process tapers slightly to end in a convex surface for articulation with the ischiadic peduncle from the ilium. The broader and longer anterior (pubic) process terminates in a flat face which meets the pubis {is /, text-fig. 9f-g). The rear margin of the ischium forms a sweeping curve (concave posteriorly); the anterior margin forms a corresponding curve, though this is interrupted by the pro- jecting obturator process. This process, a thin and sheet-like extension of the anterior margin, curves forwards and outwards to form a distinct hollow on the lateral face of the bone. Below the twisted ‘neck’ the posterior margin is very thick, being composed of two roughly parallel ridges separated by a deep and narrow groove. This groove seems to have borne the origin of the ischio-trochantericus muscle (dinosaurian equivalent of the avian ischio-femoralis). Such a muscle would have extended up and forwards to insert near the head of the femur; it doubtless served to prevent femoral dislocation during locomotion. The grooved rear margin of the ischium bears a shallow pit, a few millimetres long, just below the level of the obturator process. This pit, which is marked with a feather-like pattern of divergent striae, probably accommodated the origin of the flexor tibialis muscle (equivalent of the ischio-flexorius in birds). The small and relatively weak flexor tibialis ran forwards and down to insert on the rear face of the tibia; it would have functioned in flexing the knee and in drawing back the hind limb. At the distal end of the ischium much of the lateral surface is ornamented with fine longitudinal striae. When ischium and pubis are placed together in natural articulation it is evident that this striated face lies directly opposite the similarly marked dorsal surface of the postpubis. The groove between the two bones was probably floored in life with ligamentous tissues serving to bind the two bones together. It is from both walls of this pubo-ischiadic groove that a muscle termed the pubo- ischio-femoralis externus is presumed to have originated. Such a muscle (equivalent of the avian obturator) inserted close to the head of the femur and assisted in femoral retraction. Striations on the medial surface of the ischium define the origin of some of the ventral axial musculature (probably the ischio-caudalis, which ran back to insert upon the centra of the foremost tail vertebrae). The blade- like distal part of the ischium is slightly arched in a medial direction, presumably to allow clearance for the femur during locomotion. The pubis (text-fig. 9d-e) exhibits comparable flexure, apparently with the same functional basis.
The pubis (text-fig. 9d-g) may be divided, for convenient description, into an expanded acetabular portion and a rod-like distal portion (postpubis). The proximal part of the better preserved (left) pubis
ac
ilp
THULBORN; POST-CRANIAL SKELETON OF FABROSAURUS
43
is damaged anteriorly (i.e. the prepubis is ineomplete). The prepubis is twisted so that its lateral face is directed somewhat ventrally. A stout posterior process from the acetabular part of the pubis curves down towards the dorsal surface of the postpubis and defines the uppermost limit of the large obturator foramen (text-fig. 9f-g). The flattened dorsal surface of this process represents the pubic portion of the acetabular margin; anteriorly this same surface is modified into a facet (// /) to receive the pubic peduncle from the ilium. The deflexed tip of this sub-acetabular process lies some 2 mm away from the dorsal surface of the postpubis and defines the incomplete posterior margin of the elliptical obturator foramen. The postpubis is a long and slender bony rod, directed postero-ventrally away from the acetabulum. Its dorsal surface is slightly flattened, so that cross-sections are elliptical or oval. In lateral view (text-fig. 9d) the postpubis shows very slight ventral ‘bowing’ which is interrupted by a distinct dorsal kink about 25 mm behind the obturator foramen. Immediately behind the obturator foramen the dorsal surface of the postpubis bears a short and deep groove {g is, text-fig. 9f) which accommodated the ventral edge of the pubic process from the ischium. In its central and distal regions the postpubis is drawn out medially into a thin and sharp-edged flange nearly 2 nmi wide. The striated dorsal face of the postpubis forms one side of the pubo-ischiadic groove and carried the origin (together with the adjacent part of the ischium) of the previously considered pubo-ischio-femoralis externus muscle. Striations on the ventral face of the postpubis indicate the former attachment of part of the ventral axial musculature (running forwards to the trunk region internal to the thigh).
Hind limb (text-figs. 10, 11, and 12)
The femur (text-figs. 10, 11h-m) is a stout columnar bone which is expanded both proximally (to form the head and the proximal trochanters) and distally (to form the condylar region). The femoral shaft is not perfectly straight but is slightly arched to the front. In its central regions the latero-medially compressed shaft is elliptical in cross-section. A faint vertical ridge down the front of the shaft probably marks the line of division of the femoro-tibialis musculature into lateral and medial portions. These muscles, extending over the front of the knee to insert upon the tibia, served to open the knee.
The femoral head is not demarcated by any constriction or ‘neck’ and is a simple bulb-hke process which is directed medially away from the shaft at an angle approaching 90°. The depressed rear face of the head probably accommodated the ischiadic peduncle from the ilium when the femur was drawn up and back during locomotion. Fine vertical striae on the posterior face of the head doubtless mark the attachment of ligaments which held the head in place within the acetabulum. The greater trochanter is a blade-like process, directed dorsally, which is applied to the lateral face of the head. This trochanter is somewhat shorter than the head but is noticeably taller than the adjacent lesser trochanter. The striated lateral face of the greater trochanter (text-fig. 11h) probably represents the insertion area of the ilio-femoralis internus muscle. This muscle originated from the posterior thoracic region, and possibly from the prepubis (Galton 1969), and served to extend the femur. The roughened upper surface of the greater trochanter represents the insertion area of the ilio-trochantericus muscle (see description of ilium for details). The lesser trochanter is an erect finger-like process which diverges antero-laterally from the region where the head and the greater trochanter meet anteriorly (text-fig. 11h-j). The lesser trochanter lies antero-medial to the greater trochanter and is separated from it by a deep vertical cleft. Just below this cleft, on the lateral surface of the femoral shaft, there lies a small, rounded, and conspieuously pitted eminence (text-fig. 10b). This pitted area, together with the striated lateral face of the lesser trochanter, seems to have borne the insertion of the ilio-femoralis externus muscle (see deseription of ilium for details).
The fourth trochanter arises from the rear face of the shaft and is entirely confined to the proximal half of the femur. This trochanter is a triangular and blade-like structure with an acute and declined tip (i.e. it is of ‘pendent’ type). The surface of the fourth trochanter bore the insertion of the large eoccygeo-femoralis brevis muscle (see account of ilium for details). Medial to the fourth trochanter there lies a prominent roughened depression (text-fig. IOd). This excavation doubtless marks the
TEXT-FIG. 9. Fabrosaunis australis. Ischium and pubis, X 1 (except figures f and g). a-c, left ischium in antero-lateral, postero-medial, and posterior views, d-e, left pubis in lateral and ventral views. F-G, acetabular portion of left pubis (reconstructed) in lateral and medial views, x3. The arrow indicates the probable course of the obturator nerve and associated blood vessels.
44 PALAEONTOLOGY, VOLUME 15
insertion of the occygeo-femoralis longus (which originated from the anterior tail vertebrae and served to retract the femur).
The shaft is widest just above the distal end. The lateral distal condyle is fractionally larger than its medial counterpart; these condyles are not appreciably attenuated behind and are separated by the wide and deep posterior intercondylar fossa. There is no trace of any anterior intercondylar fossa.
TEXT-FIG. 10. Fabrosaurus australis. Right femur, xl. Posterior (a), lateral (b), anterior (c), and
medial (d) views.
The tibia (text-fig. 11a-e) is considerably longer than the femur. Its proximal end is widest from front to back whilst its distal end is expanded latero-medially. These differing directions of expansion lend the tibia a decidedly twisted appearance, the proximal view of the bone (text-fig. 1 1e) showing that this torsion ranges through some 70°. The anterior view (text-fig. 11b) shows that the bone is also affected by a weak sinuous flexure, the proximal half being arched medially whilst the distal half is arched to the exterior. The convex and roughened proximal surface is crescentic in outline. At the proximal end of the tibia the antero-medial surface is transversely convex whilst the postero-lateral
TEXT-FIG. 11. Fabrosaurus australis. Tibia, fibula, and femur, X 1. a-b, left tibia in lateral and anterior views, c, distal part of same tibia in posterior view, d, distal parts of right tibia and fibula in natural juxtaposition (anterior view), e, proximal outline of left tibia (thick line) superimposed upon distal outline (shaded) to illustrate torsion affecting the bone, f-g, right fibula in lateral and anterior views. H-J, proximal part of left femur in lateral and anterior views, k, proximal view of right femur, l, plan of head and proximal trochanters in the left femur, m, distal view of right femur.
46
PALAEONTOLOGY, VOLUME 15
face is generally depressed. This postero-lateral concavity is interrupted near the middle by a blunt triangular projection (the ‘lateral condyle’) which extends down the tibial shaft as a thick and well rounded rib. Anterior to this there Ues a similar, though rather smaller, ‘accessory condyle' (text-fig.
1 1a). The ‘inner condyle’ is a blunt projection forming the posterior corner of the tibia at its proximal end. In consequence of the torsion affecting the tibia the cnemial crest (the thickly rounded anterior margin of the bone) shifts over to the medial side as it is traced distally. The lateral malleolus is a little broader and longer than its medial counterpart. The sharpened margins of the malleoli extend for a short distance up the tibial shaft as weak ridges. The anterior faces of the malleoU are almost flat. There are few obvious surface markings on the tibia. Indistinct striae on the cnemial crest probably mark the insertion of part of the extensor musculature (the femoro-tibialis and the ilio-tibiaUs or parts thereof).
The slender fibula (text-fig. 11f-g) resembles the tibia in displaying pronounced torsion. The proximal tip is latero-medially compressed and has its posterior corner extended into a short process. From the depressed proximo-medial surface a shallow groove runs down the fibular shaft for about a quarter of its length. Below this the medial surface is nearly fiat. In its central regions the lateral face of the bone bears a prominent vertical ridge (which accounts for the almost triangular cross- sections of the shaft). As it is traced proximally this lateral ridge shifts over to the posterior margin. Distal parts of the right tibia and fibula are preserved in natural relationship (text-fig. 1 Id); the tip of the fibula lies on the flat anterior face of the outer tibial malleolus.
The tarsus (text-fig. 12c-e) is represented by two poorly preserved distal tarsals, both from the left side. One of these is a small, flattened, and disc-like bone (text-fig. 12c-d); its depressed distal surface accommodates the raised proximal end of metatarsal 111. When these bones are articulated in this fashion the overhanging medial edge of the tarsal meets the lateral half of the proximal surface of metatarsal II. The other tarsal bone (text-fig. 12e) is less well preserved; this is slightly thicker than the other tarsal and seems to have articulated with the proximal end of metatarsal IV.
The pes (text-fig. 12a-b, f-e) is represented by numerous scattered phalanges (including three unguals), a well preserved left metatarsus (text-fig. 12a-b) and fragments of the right metatarsus.
Metatarsal I is a thin, sharp-edged, and splint-like bone about half as long as the adjacent second metatarsal. Its swollen distal tip is widest transversely and bears a shallow median furrow. Metatarsal I differs from the other metatarsals in its orientation; instead of running straight downwards it is directed down and back so that its distal tip lies well behind the other bones of the foot (text-fig. 12b). Metatarsal II is a thick and rod-like bone, 58 mm long, which is slightly arched to the front. Its flat proximal surface is roughly triangular in outline owing to the narrowness of the posterior margin. Its convex and sub-rectangular distal surface is developed into two small condyles; the slightly smaller medial condyle is extended up the rear face of the bone as a sharp ridge. Metatarsal 111 is the longest of the hind limb metapodials, attaining a length of 67 mm (i.e. more than half the length of the tibia). It is basically similar to, but a little stouter than, metatarsal II. Metatarsal IV is fractionally shorter than metatarsal II (56 mm as opposed to 58 mm) and is perfectly straight. A thin and sharp ridge per- sists along the entire lateral margin of this metatarsal. Metatarsal V has not been recovered.
Fifteen phalanges of the foot are preserved in assemblage B. 17. Eight of these are assigned to the right foot, seven to the left. In the left foot the first phalanx in digit I was found in natural articulation with metatarsal I. This slender phalanx is 17 mm long (text-fig. 12h). Cross-sections of the bone are triangular in consequence of its compressed medial edge. The proximal surface is nearly flat whilst
TEXT-FIG. 12. Fabrosaiirus australis. Ankle and foot bones, X 1 (except figures c, d, e, and q). a-b, left metatarsus in anterior and medial views, c, distal tarsal bone in distal view, x 2. d, cross-section of same tarsal bone to show surfaces for articulation with metatarsals II and III, X 2. e, proximal view of a second distal tarsal bone, X 2. E, first phalanx of digit 3 (right foot) in dorsal, medial, and ventral views. G, diagrams to show structure of same phalanx, h, first phalanx of digit 1 (left foot) in dorsal and lateral views. J, first phalanx of digit 4 (left foot) in dorsal and medial views, k, first phalanx of digit 2 (right foot) in dorsal and lateral views, l, second phalanx of digit 2 (right foot) in dorsal and medial views, m, third phalanx of digit 4 (left foot) in dorsal and lateral views, n, second phalanx of digit 3 (left foot) in dorsal and lateral views, o, third phalanx of digit 3 (right foot) in dorsal and medial views, v, dorsal and medial views of an ungual phalanx (probably from digit 1 of the right foot). Q, diagrams to show structure of same ungual phalanx, X 2. r, reconstruction of left foot in dorsal view (reconstructed portions shaded).
THULBORN: POST-CRANIAL SKELETON OF FABROSAURUS
47
48
PALAEONTOLOGY, VOLUME 15
the distal end is elaborated into a pair of small condyles. Each condyle bears a shallow circular pit on its flank; similar pits, marking the attachment of digital extensor muscles, are evident in the other foot phalanges. The proximal phalanges are distinguished from the more distal phalanges of the foot through the structure of the proximal surface; in the proximal phalanges the proximal face is almost flat whilst in the distal phalanges this surface is deeply excavated (compare text-figs. 12f and 12l). This criterion serves to distinguish five more proximal phalanges within the material. Two of these are merely fragments; the remaining three all differ in their proportions. The stoutest of these phalanges (text-fig. 12f) would seem, from its size, to be the first one in digit III. A second, rather narrower and longer phalanx (text-fig. 12k) is regarded as the proximal one in digit II. The remaining proximal phalanx (text-fig. 12j) is shorter than the other two but is intermediate in stoutness. This represents, by elimination (and assuming that metatarsal V bore no phalanges), the first phalanx in digit IV.
The distal phalanges, forming the foot digits between the proximal row of phalanges and the unguals, are represented by six examples. A pair of these are left and right counterparts and there are, in effect, only five examples to be considered. The stoutest of the five (text-fig. 12n) is assumed to be the second phalanx in digit III since it articulates quite agreeably with the first phalanx in this digit. Its proximal surface is divided by a vertical ridge into two concavities which receive the distal condyles of the pre- ceding phalanx. A somewhat narrower and longer phalanx (text-fig. 1 2l) is presumed to be the second in digit II. A third example (text-fig. 12o) is tentatively identified as the third in digit III. A rather short (11 mm) phalanx is probably the second in digit IV. The smallest example (text-fig. 12m) is doubtless the third phalanx in digit IV.
Of the three ungual phalanges recovered from assemblage B. 17 only one, the smallest, is at aU well preserved (text-fig. 12p-q). This phalanx has the form of a curved cone with a transversely flattened ventral (palmar) surface. The slightly excavated proximal surface is nearly circular in outline. Half way up each side of the phalanx there lies a narrow and deeply incised claw groove; these grooves are about equally developed on each side of the phalanx and extend nearly half way along the bone from its bluntly rounded distal apex. It is assumed that the largest ungual (13-5 mm long) comes from digit III. The smallest (8-5 mm long) is assigned to digit I. The third example (10 mm in length) is assigned to digit II, though it might possibly have come from digit IV. The foot of Fabrosaurus is shown reconstructed (text-fig. 12r) with a phalangeal formula of 2: 3:4: 5:0.
Measurements in mm. Main measurements of the two individuals of Fabrosaurus australis in assemblage B. 17. *Indicates estimated figure.
|
Larger Individual |
Smaller Individual (cont.) |
||
|
scapula height |
75 |
postpubis length |
*100 |
|
humerus length |
68 |
femur length |
104 |
|
radius length |
43 |
tibia length |
129 |
|
fibula length |
*123 |
||
|
Smaller Individual |
metatarsal I length |
*30 |
|
|
scapula height |
66 |
metatarsal 11 length |
58 |
|
humerus length |
58 |
metatarsal III length |
67 |
|
ulna length |
*40 |
metatarsal IV length |
56 |
|
radius length |
37 |
lengths of foot phalanges: |
|
|
metacarpal I length |
6-5 |
proximal phalanx, digit I |
17 |
|
metacarpal II length |
11 |
proximal phalanx, digit II |
22 |
|
metacarpal III length |
*12-5 |
proximal phalanx, digit III |
20 |
|
metacarpal IV length |
8-5 |
proximal phalanx, digit IV |
14 |
|
lengths of hand phalanges : |
second phalanx, digit II |
18-5 |
|
|
proximal phalanx, digit I |
*4 |
second phalanx, digit III |
15-5 |
|
proximal phalanx, digit II |
5-5 |
second phalanx, digit IV |
11 |
|
proximal phalanx, digit III |
5 |
third phalanx, digit III |
14 |
|
proximal phalanx, digit IV |
*3-5 |
third phalanx, digit IV |
10-5 |
|
second phalanx, digit II |
4-5 |
ungual phalanx, digit I |
8-5 |
|
ilium length |
*85 |
ungual phalanx, digit II (IV ?) |
10 |
|
ilium height |
31 |
ungual phalanx, digit III |
13-5 |
|
ischium length |
*95 |
THULBORN: POST-CRANIAL SKELETON OF FABROSAURUS
49
DISCUSSION
Systematic position of Fabrosaurus
In his original account of the genus, based upon a jaw fragment, Ginsburg (1964) recognized the ornithischian status of Fabrosaurus and suggested that it might be closely related to the Liassic Scelidosaurus. In a subsequent description of the skull (Thulborn 1970a) this alliance with the armoured and somewhat problematical Scelido- saurus has been refuted and Fabrosaurus has been referred to the family Hypsilopho- dontidae of the suborder Ornithopoda. The structure of the post-cranial skeleton endorses this concept of Fabrosaurus as a hypsilophodont. The Fabrosaurus skeleton is very like that of Hypsilophodon itself, though it is distinguished through its noticeably more delicate construction. The vertebral column of Fabrosaurus, though it remains poorly known, matches that of Hypsilophodon in most essentials. The main point of difference concerns the number of sacral vertebrae; Fabrosaurus has 5 whilst Hypsilo- phodon typically has 6 (Galton in press). But since the number of sacral vertebrae is slightly variable within the ornithopods {Hypsilophodon, for example, sometimes has 5) this distinction does not seem excessively important.
The limb girdles of the two animals are remarkably similar. The Fabrosaurus scapula is distinguished from that of Hypsilophodon through its projecting postero-dorsal corner and through its prominent ‘acromial’ process. In possessing this salient ‘acromial’ pro- cess Fabrosaurus is somewhat unusual amongst ornithischians (with the exception of the stegosaurs) and resembles some of the theropod saurischians (e.g. Gorgosaurus). The most obvious difference in iliac structure concerns the immediate supra-acetabular region; in Hypsilophodon this part of the ilium is almost flat whilst in Fabrosaurus it is strongly inflated and flared out to the exterior. The Fabrosaurus ischium is not quite as straight as that of Hypsilophodon and has the obturator process situated rather nearer the acetabulum. The Fabrosaurus pubis is very similar to that of Hypsilophodon ; the postpubis demonstrates none of the shortening seen in later and larger ornithopods.
The humerus is remarkably like that figured for Hypsilophodon by Swinton (1936); in both cases the free edge of the delto-pectoral crest accounts for roughly one third of the total humeral length. The fore arm bones and the diminutive manus of Fabrosaurus are quite comparable with those of Hypsilophodon. The femora of the two animals are both characterized by a large fourth trochanter of ‘pendent’ type which is located unusually high up on the shaft. In both cases there is no anterior intercondylar fossa and the proximal trochanters are divided by a deep cleft. Within the Fabrosaurus hind limb, the tibia is considerably longer than the femur; this unusual tibio-femoral ratio is quite typical of members of the family Hypsilophodontidae. The Fabrosaurus meta- tarsus is equally as long and as narrow as that of Hypsilophodon. In both animals the phalangeal formula for the foot appears to be 2: 3:4: 5:0 and the digits terminate in slender claws. Finally Fabrosaurus resembles Hypsilophodon in possessing a system of ossified tendons along the rear parts of the vertebral column and in having hollow and thin-walled limb bones.
There can be no doubt, in view of this evidence, that Fabrosaurus is a genuine orni- thischian dinosaur of Triassic age. This is amply demonstrated by its possession of a predentary bone at the mandibular symphysis (Thulborn 1970a) and by the tetra- radiate plan of its pelvic girdle. Certain structural peculiarities, such as the toothed
C 8472 E
50
PALAEONTOLOGY, VOLUME 15
premaxilla (Thulborn op. cit.) and the unusual tibio-femoral ratio, indicate that this genus should be included within the family Hypsilophodontidae. It is proposed, in view of these findings, to remove Fabrosaurus australis from the Scelidosaurinae (Gins- berg’s assignment, 1964) and to place it within the family Hypsilophodontidae of the suborder Ornithopoda. Ginsburg’s original diagnosis is greatly amplified.
TEXT-FIG. 13. Fabrosaurus australis. Restoration of skeleton. Approximately J natural size.
Class REPTILIA Order ornithischia Suborder ornithopoda Family hypsilophodontidae Genus fabrosaurus Ginsburg 1964 Monotypic species F. australis Ginsburg 1964
Diagnosis (for genus and sole known species): unarmoured ornithischian, about 1 metre long, with slender and hollow limb bones. Skull* about 10 cm long, triangular, diapsid, with extensive circular orbits at sides. Antorbital vacuity triangular, widely open. Premaxilla extended behind naris but does not reach lacrimal. Maxilla flat above tooth row; jugal slender, without ventral flange. Parietals separate, forming broad and flat zone between upper temporal openings. Frontal with transverse crescentic depres- sion marking front limit of upper temporal opening. Quadrate tall, extended anteriorly, with front edge overlain by slender descending process from squamosal. Mandible slender, with salient finger-like retroarticular process and weak coronoid apophysis. Small median edentulous predentary at mandibular symphysis. Dentition heterodont; implantation thecodont; teeth in simple marginal row. Premaxilla with up to 6 acute.
* Details of skull construction from Thulborn 1970a.
THULBORN: POST-CRANIAL SKELETON OF FABROSAURUS 51
smooth, recurved teeth, last 2 bearing minute marginal denticles. About 14 teeth in maxilla; equivalent number in dentary. Crowns of cheek teeth squat, triangular, inflated bucally, flattened lingually; mesial and distal edges of crowns with small erect denticles of unvarying size. Replacement from lingual side, alternate; wear affects mesial and distal edges of cheek teeth separately.* About 15 dorsal vertebrae; sacrum of 5 vertebrae, with neural canal expanded ventrally within centra. Scapula tall, with projecting postero- dorsal corner and prominent ‘acromial’ process. Fore limb much smaller than hind limb. Humerus slightly shorter than scapula, with delto-pectoral crest confined to upper third. Radius and ulna slender, roughly equal in length, shorter than humerus. Manus diminutive, with probable formula of 2: 3:4: 3:0. Pelvis tetra-radiate ; ilium long, low, with pointed and deflexed anterior process, without ‘antitrochanter’. Postpubis long, narrow, rod-like, fairly straight; prepubis short, blade-like, twisted. Obturator process very high up on ischium. Acetabulum open, roofed above by lateral extension of ilium. Femur twice as long as humerus, with pendent fourth trochanter confined to proximal half; greater and lesser trochanters divided by deep cleft; distal condyles sub equal, not drawn out behind. Tibia stout, twisted, longer than femur; fibula slender, rod-like, slightly shorter than tibia but longer than femur. Metatarsus long and narrow; meta- tarsal I reduced, splintlike; metatarsal III equivalent to 55% of tibia length. Phalangeal formula of foot 2: 3:4: 5:0; digits clawed.
Size
It seems likely that Fabrosaurus australis is a dinosaur which did not attain any great size (rather than a foimi represented by immature specimens). All known specimens are of much the same size; the list includes the holotype (Ginsburg 1964), a nearly complete skull (Thulborn 1970n), the two individuals in assemblage B. 17 and parts of two undescribed individuals in the British Museum (Natural History).
In his discussion of Hypsilophodon, Swinton (1936) implied that the length of the humerus, relative to that of the scapula, might indicate the maturity of individual hypsilophodonts:
‘. . . scapula and humerus ... are almost equal in length in the new young specimen, but the former is definitely shorter than the latter in the adult, so that generally it may be said that this somewhat unusual condition in dinosaurian osteology is common to Thescelosaums neglectiis and Hypsilophodon.^
If scapula and humerus lengths really are equal in immature hypsilophodonts, whilst the humerus is longer in adults, comparisons of these bones should provide a rough working guide to the maturity of individual specimens. Table 1 shows that in Fabrosaurus the scapula is longer than the humerus, an arrangement which is totally irreconcilable with Swinton’s hypothesis. The scapula-humerus ratio in Fabrosaurus is comparable with that in the Upper Cretaceous hypsilophodont Parksosaurus (Sternberg 1940) and in the iguanodont Camptosaurus (Gilmore 1909). It is clear that this skeletal ratio is no sound criterion upon which to establish the relative maturity of hypsilophodont speci- mens. This ratio is rendered even more suspect when one considers that the dorsal margin of the scapula may, since it passes into the supra-scapula, be ossified to very different degrees in diflerent individuals.
* Details of tooth wear and replacement from Thulborn 1971.
52
PALAEONTOLOGY, VOLUME 15
Gallon (in press) maintains that in Hypsilophodon the humerus displays greater tor- sion with greater maturity of the individual. This seems to imply that the straight and untwisted humerus of Fabrosaurus signifies immaturity. Alternatively the lack of torsion affecting the Fabrosaurus humerus might well be interpreted as a primitive character, especially since this bone is only slightly twisted in pseudosuchians such as Euparkeria.
Since all the known specimens are of roughly similar size it seems reasonable to assume that Fabrosaurus was, in fact, a rather small ornithischian dinosaur. It is esti- mated that the smaller (more complete) individual in assemblage B. 17 had a maximum length (from snout to tail tip) of slightly less than one metre. The hind limb is com- parable in size with that of a living chicken {Gallus). The fore limb is very much shorter than the hind limb whilst the neck, which was relatively short, carried a quite large skull (perhaps 10 cm long). The tail is not well represented in the material but is shown in the reconstruction (text-fig. 13) as roughly equivalent in length to head, neck, and trunk combined (i.e. much as in other hypsilophodont ornithopods).
Locomotion
It is important to recognize that there are no specific features of skeletal construction which might serve as absolutely reliable criteria in distinguishing between bipedal and quadrupedal dinosaurs. A comparable situation exists in living lizards, where bipeds are indistinguishable from quadrupeds through examination of the skeleton alone. There are, however, numerous osteological characters in Fabrosaurus which are dis- tinctly suggestive of bipedalism and which, in conjunction, render the concept of Fabro- saurus as a biped quite acceptable. These adaptations for bipedalism are evident in almost every part of the skeleton. The entire skeleton is very lightly built. The slender, hollow and thin-walled limb bones resemble those of birds and of pterosaurs, where weight reduction would have been of critical importance. Lightening of the skeleton is most marked in advance of the hips — in the fenestrated skull (Thulborn 1970a), in the relatively short neck, in the small fore limbs and in the rather delicate construction of the presacral vertebrae. Such weight reduction in front of the hips is explicable when one considers that in a biped the whole body must be pivoted over the hips and that, in consequence, the tail alone must counter-balance the weight of the trunk, fore limbs, neck, and head. Further, the lack of dermal armour in Fabrosaurus would have con- tributed to the reduction of total body weight.
The almost horizontal zygapophysial faces of the vertebrae immediately preceding the sacrum would have prevented undue sagging of the vertebral column whilst the animal was in a bipedal pose. The ‘lumbar’ vertebrae of large and undoubtedly bipedal theropods {Tyrannosaurus, AUosaurus, and the like) frequently show traces of fusion, presumably with a similar functional basis. The lattice of ossified tendons, attached to the neural spines in front of and behind the sacrum, seems to have a similar purpose. Ostrom (1964) suggests that such a tendon system would have effected resistance to any sagging of the vertebral column.
The zygapophysial faces of the caudal vertebrae are practically vertical, indicating that flexures of the tail took place mainly within a vertical plane. Snyder (1949) has emphasized the importance of the tail in the bipedal locomotion of lizards such as Basiliscus and has shown that their bipedal faculties are seriously impaired when the tail is even partially amputated. This author points out that the tail is held clear of the
THULBORN: POST-CRANIAL SKELETON OF FABROSAURUS
53
ground (i.e. elevated in a vertical plane) during bipedal running. Snyder (1962) further makes it clear that such tail movements serve to counter-balance the weight of those parts of the body in advance of the hips. Vertical tail movements, in every way com- parable to those of bipedal lizards, would seem to have been of fundamental importance in Fabrosaurus. These tail flexures would not have been impaired by the ossified tendons since individual bundles of tendons would have ‘slipped’ relative to those crossing above and below.
The five centra of the Fabrosaurus sacrum are distinguished through the great enlarge- ment of the neural canal within them. Ewer (1965) noted similar inflation of the neural canal in the whole ‘lumbar’ region of the pseudosuchian Euparkeria (from the 13th dorsal vertebra to the 1st caudal). It is evident that this peculiarity affects similar zones in Euparkeria and Fabrosaurus but that owing to the shortness of the sacrum in the former (only two vertebrae) it extends to include both posterior dorsal and anterior caudal vertebrae. The functional significance of the dilated neural canal is debatable. Ewer (op. cit.) maintains that these dilations in the centra might have been filled with non-nervous tissue intimately associated with the nerve cord (rather than with any expansion of the spinal cord itself). Such an arrangement might be paralleled in birds, where the sinus lumbosacralis is the associated tissue (Terni 1924). This hypothesis suffers, however, from some difficulties. Firstly, the avian sinus lumbosacralis lies dorsal to the nerve cord whilst the excavations in the Fabrosaurus vertebrae (and those of Euparkeria) are situated ventrally. Secondly, the function of the glycogen-filled sinus lumbosacralis is not readily apparent. It is unlikely that such glycogen-rich tissues could have provided fuel for muscular energy since (in birds at least) the main locomotor muscles derive fuel from deposits of fat, which provides twice the energy that glycogen does (George and Berger 1966). It is suggested here, in contrast, that the inflated neural canal in the sacral region did in fact accommodate nervous tissue— a genuine ganglionic expansion of the nerve cord. Dilations of the spinal cord have long been quoted in a variety of dinosaur sacra (e.g. see Marsh 1881; Seeley 1882) and Romer (1956) sug- gests that this local refinement of the central nervous system is to be related directly to the size of the hind limbs. In both Euparkeria and Eabrosaurus the hind limbs are considerably larger than the fore limbs. Yet sheer relative size of the hind limbs alone would not seem to account for any local expansion of the nerve cord in these animals, particularly when their over-all small size is recalled. It is important to recognize, how- ever, that Eabrosaurus is presumed to have been an habitual (if not obligatory) biped. When not running this animal must have walked slowly on the hind limbs alone. In such slow bipedal progression there regularly comes a point when contact with the ground is maintained by one foot alone (assuming that the tail would have been nearly clear of the ground). Thus for short periods, perhaps a second or so in duration, the animal must be poised on one foot and must, as a result, be prone to simply topple over. The only way in which this tendency can be counteracted is through delicate shifts of body weight so as to maintain equilibrium. This, in turn, demands perfect muscular control within and between the hind limbs and the major organ of balance, the tail. Such sophisticated muscular control might well have been governed by a local dilation of the nerve cord which, logically, would have been situated close to both hind limbs and the base of the tail (i.e. in the sacrum).
The proximal end of the humerus bears an extensive articular surface but lacks any
54
PALAEONTOLOGY, VOLUME 15
salient ‘head’. This indicates that the humerus had significant freedom of movement within the glenoid cavity. Many earlier dinosaur reconstructions have the proximal end of the humerus set firmly in the glenoid; in consequence the humerus is directed horizontally at right angles to the line of the vertebral column (see Casier 1960, for restorations of Hypsilophodon and Megalosaurus). Whilst this attitude was undoubtedly possible it seems more likely that in normal circumstances the humerus would have been closely applied to the side of the thorax and directed down and forwards. In this position the projecting proximo-medial corner of the humerus would have served as the functional ‘head’. Such humeral mobility suggests that the fore limb had no very great positive locomotor function; the humerus would surely have been too prone to dis- location for the fore limb to have sustained any sizeable proportion of the body-weight. This, once again, is indirectly suggestive of habitual bipedalism.
The Fabrosaurus hand, in relation to the foot in the same animal, is diminutive (com- pare text-figs. 7r and 12r). None of the bones of the hand shows the elongation which characterizes the foot bones. It is clear that this delicate hand could have had no sig- nificant locomotor function. In quadrupedal archosaurs (e.g. Alligator, Stegosaurus) the hand is generally only slightly smaller than the foot, despite a startling size-difference between the entire fore and hind limbs. It appears that the span of the hand relative to that of the foot (rather than the length of the fore limb against that of the hind limb) is of some use in deducing the probable mode of locomotion.
Strong pelvic and thigh muscles have been inferred for Fabrosaurus on the evidence of muscle scars on the femora and pelvic girdle bones. These muscles are matched in the related dinosaurs Thescelosaurus (Romer 1927) and Hypsilophodon (Galton 1969) and are comparable, in general terms, with those of birds and lizards. The principal locomotor agent in Fabrosaurus was doubtless a strong backwards thrust of the hind limb, generated by contraction of the powerful femoral adductor muscles. These adductor muscles (the coccygeo-femorales) extended from the fourth trochanter of the femur to areas of origin on the rear part of the ilium and the anterior tail vertebrae. During femoral adduction there would necessarily have been some tendency for the base of the tail to bend into a kink towards the approaching femur. This is admirably shown by Snyder (1962) in a figure of the lizard Crotaphytus. Such tail fiexure would have affected the efficiency of the adductor muscles to a considerable degree. It is likely that the vertical zygapophysial faces of the caudal vertebrae served to brace the tail in order to resist such lateral flexure.
The Fabrosaurus ilium is distinguished by its swollen and flared out acetabular mar- gin. This forms, in effect, an overhanging roof above the acetabulum. In other hypsilo- phodonts the supra-acetabular part of the ilium is flat or only very slightly inflated. Fabrosaurus seems, in fact, to be unique amongst ornithischians in this portion of its iliac morphology. Such roofing-over of the acetabulum may be matched only in the coelurosaur Coelophysis and in the problematical reptile Poposaurus (Colbert 1961). These three reptiles have, moreover, certain other features in common; they are all of late Triassic age and they all appear (with the possible exception of Poposaurus) to have been habitual bipeds. The supra-acetabular expansion is probably a specialization related to bipedalism. In a biped much of the mechanical thrust affecting the femur is directed upwards; hence a deepened and partially roofed-over acetabulum would have assisted in retaining the ‘head’ of the femur in place during locomotion. This arrange-
THULBORN: POST-CRANIAL SKELETON OF FABROSAURUS 55
ment may, in turn, be correlated with the rather weak development of the femoral ‘head’ in these forms.
The dorsal margin of the ilium is not at all everted or extended laterally and there is no trace of the ‘antitrochanter’ which occurs in larger ornithopods and in quadrupedal ornithischians. The functional significance of the ‘antitrochanter’ is obscure, though Romer (1927) suggests that its presence reflects some elaboration of the ilio-femoralis musculature. Lack of this structure from the Fabrosaunis ilium presumably points to an unspecialized arrangement of the ilio-femoralis (which holds the femoral ‘head’ in place and assists in elevating the thigh). The anterior process of the ilium is only slightly deflexed and has an acutely pointed tip (in contrast to the broadly spatulate tip observed in many iguanodonts). Most importantly the anterior process exhibits none of the arching seen in the ilia of hadrosaurs (e.g. Hypacrosaunis). Romer (op. cit.) relates such arching with reversion to quadrupedalism, suggesting that this flexure of the anterior process permitted passage for the ilio-femoralis internus muscle from the femur to the posterior thoracic region. Two ridges on the medial face of the anterior process (text-fig. 8c) seem to have strengthened this rather delicate structure. It may be noted, in this context, that the anterior process would have been subject to some lateral ‘pull’ through contraction of the attached ‘sartorius’ muscle.
The postpubis rivals the ischium in length and is also fairly straight; it lacks any of the downwards curvature which Romer correlates with quadrupedalism (1927). The large obturator foramen (text-fig. 9f-g) probably served as an exit for the obturator nerve and associated blood vessels. The foramen is very nearly encircled by bone and the small gap at the rear margin was doubtless filled with cartilage during life. This near- complete enclosure of the obturator foramen is characteristic of hypsilophodonts ; in many other ornithopods the foramen is widely open at the back and is little more than a notch.
The prepubis is so poorly known (text-fig. 9d) that it cannot be discussed in detail. Hence it is impossible to reconsider Romer’s hypothesis (1927) that the prepubis served principally as an abdominal support structure and that no musculature of any conse- quence was attached to it. Recently Gallon (1969) has suggested that the prepubis did not provide the main support for the abdomen and that some muscle (the pubo-tibialis or part of the pubo-ischio-femoralis internus) originated from its lateral surface.
The elongated and thin-walled bones of the Fabrosaurus hind limb are not unlike those of birds in general appearance and are distinctly suggestive of bipedal potential. The hind limb is somewhat unusual in that the tibia is considerably longer than the femur. Comparable predominance of tibia over femur is seen in related hypsilophodonts, in some pseudosuchians (e.g. Saltoposuchus) and in many coelurosaurs (e.g. CoeJophysis, Ornithomimus). The metatarsus of Fabrosaunis is similarly attenuated, the longest (third) metatarsal being equivalent to some 55 % of the tibia length. Such elongation of the hind limb is almost certainly indicative of habitual bipedalism, though this does not imply that forms with a relatively short tibia and metatarsus (e.g. Thescelosaurus, Euparkeria) were precluded from a similar mode of locomotion. Lengthening of the hind limb probably served to increase potential for rapid acceleration. Hildebrand (1959, 1961) has shown that acceleration is achieved, in cursorial mammals at least, by lengthening of the stride rather than by any increase in the number of limb strokes per minute. This suggests that the long hind limbs of Fabrosaurus increased this animal’s
56
PALAEONTOLOGY, VOLUME 15
ability to lengthen the stride and, in consequence, its capacity to achieve acceleration. It is probable that a dinosaur as small as Fabrosaurus could take off" into rapid bipedal flight from a stationary position, much as the lizard Basiliscus (Snyder 1962) or the domestic chicken.
There is no trace of the fifth digit in the Fabrosaurus foot. In all probability digit V was represented only by a splint-like vestige of the fifth metatarsal (i.e. as in Hypsilo- phodon and Thescelosaiirus). Digits II, III, and IV are the longest and stoutest and the Fabrosaurus foot may be considered as functionally tridactyl and rather bird-like (text-fig. 12r). The tridactyl foot would seem to have been a specialization related to bipedalism since it is encountered in other hypsilophodonts, in the pseudosuchian thecodonts (though digits I and V are not excessively reduced here) and in saurischian bipeds (e.g. Allosaurus, Ornithomimus). The second, third, and fourth digits of the Fabrosaurus foot were directed forwards and were somewhat splayed out whilst the shorter first digit was probably directed back and down as a heel-like ‘prop’. This interpretation is borne out by the orientation of the first metatarsal when the metatarsus is preserved in an undisturbed state (text-fig. 12b).
In the saurischian dinosaurs Colbert (1964) has indicated some relationships between certain skeletal proportions and the presumed mode of locomotion. Some of the con- clusions attained by this author may be extended to cover the early ornithischian now imder consideration. Colbert correlates a ‘dolichoiliac’ type of pelvis (i.e. one with a long, low ilium) with the development of ‘complete bipedalism’, citing Coelophysis, Compsognathus, and Ornithomimus as examples. This author suggests that ‘. . . the dolichoiliac pelvis . . . would furnish the muscular base for an efficient bipedalism’. A similar argument might be applied to Fabrosaurus: a powerful pelvic musculature, well-suited for bipedal locomotion, has been inferred from the evidence of muscle scars whilst the long and blade-like ilium could be accommodated without difficulty in Colbert’s ‘dolichoiliac’ category.
Colbert (op. cit.) also attempts to establish some line of distinction between saurischian bipeds and quadrupeds on the basis of disparity in size between the fore and hind limbs. Such an approach cannot, however, be utilized in the ornithischian dinosaurs; there are several cases where a limb ratio of bipedal aspect (with the fore limb very much shorter than the hind limb) is encountered within undoubted quadrupeds (e.g. Stegosaurus, Nodosaurus). Despite this difficulty it is still possible to distinguish purely bipedal ornithopods from those with a tendency to quadrupedalism — through comparisons of limb bone lengths within the hind limbs. In habitually bipedal types, such as the hypsilo- phodonts, the tibia is considerably longer than the femur and the metatarsus is equiva- lent to a significant proportion of the femoral length (66% in Fabrosaurus). In those ornithopods tending to quadrupedalism (notably iguanodonts and hadrosaurs) the tibia is not as long as the femur and the short, stout metatarsals constitute a foot of graviportal aspect.
It seems reasonable, in view of all this evidence, to envisage Fabrosaurus as a small and agile biped with distinct cursorial ability. In conclusion it may be pointed out that Fabrosaurus compares favourably with cursorial bipeds from elsewhere within the Reptilia — with Basiliscus and Chlamydosaurus (Snyder 1949, 1952, 1954, 1962), with Coelophysis (Colbert 1964), with Euparkeria (Ewer 1965) and with Velocipes (von Huene 1932).
THULBORN: POST-CRANIAL SKELETON OF FABROSAURUS 57
Ornithischian origins
Since all known Triassic ornithischians may be referred to the family Hypsilopho- dontidae it is clear that the whole question of ornithischian ancestry is bound up with the origin of this family in particular. This inferred monophyletic origin for the order Ornithischia is sustained by a remarkable homogeneity of structure throughout the group; important diagnostic features, such as the predentary bone and the biramous pubis, are encountered in even the most aberrant ornithischians. The presence of ornithischians in the Upper Trias certainly implies that the group came into existence at some considerably earlier date. This concept of an extremely early start to orni- thischian history, though unsupported by fossil evidence, is reinforced by the geographic and structural diversity of the known Triassic forms {Fabrosaurus, Lycorhinus, and Geranosaurus from southern Africa, Tatisaurus from China, Pisanosanrus from Argen- tina). Such diversity points, in turn, to some pre-Upper Triassic episode of adaptive radiation and dispersal.
Ornithischian origins are probably to be sought within the order Thecodontia. Romer (1966) distinguishes four suborders of thecodonts; Proterosuchia, Phytosauria, Aeto- sauria, and Pseudosuchia. Of these the proterosuchians and the phytosaurs may at once be discounted as possible near-ancestors of the Ornithischia on account of their re- markably specialized construction. The phytosaurs may also be rejected in view of their stratigraphic location (being mainly of late Triassic date these are contemporary with early ornithischians). The aetosaurs (including Stagonolepis and its allies) have recently been studied by Walker (1961). In some respects these reptiles are very similar to ornithischians; Walker (op. cit.) mentions especially the elongate external naris, loss of teeth from the front of the premaxilla, the generally reduced number of teeth, the forwardly inclined quadrate and the well developed dermal armour. Further, the existence in Stagonolepis of a horny sheath at the mandibular symphysis immediately calls to mind the ornithischian predentary bone. But the possibility that the aetosaurs might represent ornithischian ancestry must be discounted for two important reasons. Firstly, certain features of aetosaur construction are totally irreconcilable with orni- thischian conditions — principally the extremely specialized armour (whilst Triassic ornithischians are unarmoured), the very marked reduction of the lower temporal opening, the lateral situation of the upper temporal opening, and the typically thecodont pelvic girdle. Secondly, the aetosaurs are mainly of late Triassic age (i.e. contemporary with early ornithischians). It seems rather improbable, in view of these facts, that the aetosaurs could be involved in ornithischian history.
This leaves only the pseudosuchian thecodonts to be considered as possible orni- thischian ancestors. In discussing the Lower Triassic pseudosuchian Euparkeria Ewer (1965) supports the suggestion, advanced by Broom (1913), that the family Euparkeriidae probably represents the ancestry of all the major groups of later archosaurs, including the Ornithischia. It is, however, rather difficult to imagine the derivation of orni- thischians from hypothetical Euparkeria-like ancestors. This difficulty springs from fundamental differences in structure; Ewer (op. cit.) concludes that the Ornithischia arose ‘. . . from some form other than Euparkeria, differing from the latter in the struc- ture of both pelvis and ankle . . .’. Whilst it may be inferred that ornithischian ancestry extends back ultimately into the Euparkeriidae this still does not clarify the problem of ornithischian history between Lower and Upper Trias. Euparkeria and its allies display
58
PALAEONTOLOGY, VOLUME 15
no obvious tendency towards the ornithischian state of organization whilst the earliest known ornithischians exhibit relatively few primitive characters and might be regarded as ‘fully-fledged’ members of the Ornithischia. There are, in consequence, no apparent ‘intermediates’ between the Lower Triassic Euparkeria and the Upper Triassic Fabro- saurus.
|
TRIAS |
JURASSIC |
CRETACEOUS |
|
® hypsilophodont genera. |
STEGOSAUF |
IS |
|
cO^> 1 ^ ^O^’^ IGUANC |
HADROSAURS |
|
|
)DONTS |
||
|
^ I ^ / / / /'I thecodont / / | / / ^ |
||
|
ancestors \ |
||
|
CERATOPSIANS ANKYLOSAURS |
||
TEXT-no. 14. Outline of ornithischian phytogeny.
It has already been suggested that Fabrosaurus is a fairly direct antecedent of the hypsilophodonts of the Jurassic and Cretaceous (Thulborn 1970a). The structure of the post-cranial skeleton fully substantiates this assertion. So it is clear that Fabrosaurus represents the earliest known portion of a hypsilophodont stock which persisted through the greater part of the Mesozoic era. These hypsilophodonts appear to lie at the core of ornithischian history; they represent the ancestry, ultimately at least, of such groups as the iguanodonts, hadrosaurs, and ceratopsians (text-fig. 14). Hence there is some justification for regarding Fabrosaurus as a genuine ‘archetypal’ ornithischian. Amongst other Triassic ornithischians the Chinese Tatisaurus and the South American Pisano- saurus seem to be fairly close relatives of Fabrosaurus. The coeval Lycorhinus (Hetero- dontosaurus), from southern Africa, has a peculiar dentition which includes large ‘canine’ teeth (Crompton and Charig 1962; Thulborn 1970Z7, 1971). This genus appears to
THULBORN: POST-CRANIAL SKELETON OE FABROSAURUS 59
represent an extremely early and rather specialized hypsilophodont divergence which failed to survive the changes concomitant with the close of the Triassic period.
Acknowledgements. It is a pleasure to express my thanks to Dr. K. A. Kermack, of University College, London, who provided the material for this paper and offered many helpful suggestions. Mrs. F. Mussett has given me much useful advice on preparation techniques. This work has also received the benefit of discussions with Dr. A. J. Charig (British Museum, Natural History), Dr. P. M. Galton (Peabody Museum, Yale University) and Dr. P. L. Robinson (University College, London). Financial support came from the Natural Environment Research Council and, subsequently, from a research fellowship at the Department of Geology, University of Birmingham.
REFERENCES
BROOM, R. 1913. On the South African pseudosuchian Euparkeria and allied genera. Froc. Zool. Soc. Lond. 1913, 619-633.
CASAMIQUELA, R. M. 1967. Un nucvo dinosaurio ornitisquio triasico {Pisanosaums mertii; Ornithopoda) de la formacion Ischigualasto, Argentina. Ameghiniana, Rev. Asoc. Pal. Argent. 5, 47-64.
CASiER, E. 1960. Les Iguanodons de Bernissart. L’inst. roy. Sci. nat. Belg, Brussels, 134 pp.
COLBERT, E. H. 1961. The Triassic reptile, Poposaiims. Fieldiana (Geol.), 14, 59-78.
1964. Relationships of the saurischian dinosaurs. Amer. Mas. Novit. No. 2181, 1-24.
CROMPTON, A. w. and CHARIG, A. J. 1962. A new Ornithischian from the Upper Triassic of South Africa. Nature, 196, 1074-1077.
EWER, R. F. 1965. The anatomy of the Thecodont reptile Euparkeria capensis Broom. Phil. Trans. R. Soc. B248, 379-435.
GALTON, p. M. 1969. The pelvic musculature of the dinosaur Hypsilophodon (ReptUia: Ornithischia). Postilla, No. 131, 1-64.
(in press). On the anatomy of the ornithischian dinosaur Hypsilophodon foxii. Bull. Brit. Mus.
{Nat. Hist.) Geol.
GEORGE, J. c. and berger, a. j. 1966. Avian myology. Academic Press, New York and London, 500 pp. GILMORE, c. w. 1909. Osteology of the Jurassic reptile Camptosaurus, with a revision of the species of the genus, and descriptions of two new species. Proc. U.S. Nat. Mus. 36, 197-332.
GiNSBURG, L. 1964. Decouvcrtc d’un Scelidosaurien (Dinosaure ornithischien) dans le Trias superieur du Basutoland. C.r. hebd. Seanc. Acad. Sci. Paris, 258, 2366-2368.
HILDEBRAND, M. 1959. Motions of the running cheetah and horse. J. Mammal. 40, 481-495.
1961. Further studies on locomotion of the cheetah. Ibid. 42, 84-91.
HUENE, F. VON. 1932. Die fossile Reptil-Ordnung Saurischia, ihre Entwicklung und Geschichte. Monogr. Geol. Paldont. 4, 1-361.
MARSH, o. c. 1881. Principal characters of American Jurassic dinosaurs; Part iv: Spinal cord, pelvis and limbs of Stegosaurus. Amer. J. Sci. (3) 21, 167-170.
OSTROM, J. H. 1964. A reconsideration of the paleoecology of hadrosaurian dinosaurs. Ibid. 262, 975-997.
ROMER, A. s. 1927. The pelvic musculature of ornithischian dinosaurs. Acta Zool. Stockholm, 8, 225-275.
1956. Osteology of the Reptiles. Chicago, 111 pp.
1966. Vertebrate paleontology. 3rd edn. Chicago, 468 pp.
SEELEY, H. G. 1882. On Thecospondylus Horneri, a new Dinosaur from the Hastings Sand, indicated by the Sacrum and the Neural Canal of the Sacral Region. Q.J. Geol. Soc. 38, 457-460.
SIMMONS, D. J. 1965. The non-Therapsid reptiles of the Lufeng Basin, Yunnan, China. Fieldiana {Geol.), 15, 1-93.
SNYDER, R. c. 1949. Bipedal locomotion of the Lizard Basiliscus basiliscus. Copeia, 1949 (2), 129-137.
1952. Quadrupedal and bipedal locomotion of lizards. Ibid. 1952 (2), 64-70.
1954. The anatomy and function of the pelvic girdle and hindUmb in lizard locomotion. Amer. J.
Anat. 95, 1-46.
1962. Adaptations for bipedal locomotion of lizards. Amer. Zool. 2, 191-203.
60 PALAEONTOLOGY, VOLUME 15
STERNBERG, c. M. 1940. Thescelosaunis edmontonensis, n.sp., and classification of the Hypsilophodon- tidae. J. Paleont. 14, 481-494. (Abstract for this paper published separately in 1937; Geol. Soc. Amer. Proceedings for 1936, p. 375.)
swiNTON, w. E. 1936. Notes on the osteology of Hypsilophodon and on the family Hypsilophodontidae. Froc. Zool. Soc. Lond. 1936, 555-578.
TERNi, T. 1924. Ricerche sulla cosidetta sostanza gelatinosa (corpo glicogenico) del midollo lombo- sacrale degli Uccelli. Arch. ital. Anat. Embriol. 21, 55-86.
THULBORN, R. A. 1970fl. The skull of Fabrosaurus australis, a Triassic ornithischian dinosaur. Palaeon- tology, 13, 414-432.
19706. The systematic position of the Triassic ornithischian dinosaur Lycorhmus angustidens.
Zool. J. Linn. Soc. 49, 235-245.
1971. Tooth wear and jaw action in the Triassic ornithischian dinosaur Fabrosaurus. J. Zool.
164, 165-179.
WALKER, A. D. 1961. Triassic reptiles from the Elgin area: Stagonolepis, Dasygnathus and their allies. Phil. Trans. R. Soc. B244, 103-204.
R. A. THULBORN
Department of Geology University of Birmingham Birmingham, 15
Revised typescript received 18 June 1971
PERIODICITY STRUCTURES IN THE BIVALVE SHELL: ANALYSIS OF STUNTING IN CERASTODERMA EDULE FROM THE BURRY INLET (SOUTH WALES)
by G. E. FARROW
Abstract. Individuals of Cerastoderma edule from the Burry Inlet are smaller than those from any other British cockle population. This stunting results from the situation of the cockle flats near neap high water mark, and is due to arrested growth during periods in the tidal cycle when the shells were exposed to the air. This, together with the longer winter stoppages, leaves cockles from high water neap level some two-thirds of the size of their contemporaries from low water spring level. The external width of a winter ring and the sharpness of its incision into the shell profile become more acute higher up the shore. Greater prominence of the crossed-lamellar structure in the outer shell layer is associated with winter reduction in growth rate, the lamellae become markedly deflected if an actual stoppage, marked by a dark periostracal band, takes place. Application of these results to the geological record would lead to a precise differentiation of littoral environments and enable the recog- nition of stunting in fossils.
The Burry Inlet, like the Thames whose cockle populations have already been described (Farrow 1971), is an estuarine area and one of the three most important commercial sites in the country (Hancock and Urquhart 1966, fig. 1). Its environment is, however, different in several important respects :
{a) The unusually high situation of the cockle flats, caused by the marked silting up of the Inlet.
{b) The large tidal range (10 m at major spring tides).
(c) Greater instability of the substratum, caused by migration of the river channel and exposure to gales.
In this stringent habitat fully grown cockles rarely exceed 30 mm in length, compared with over 50 mm for examples from Barra in the Hebrides.
The present work owes much to the co-operation of the Burnham-on-Crouch Fisheries Laboratory whose staff have been engaged since 1958 in a study of the population dynamics of the Cerastoderma edule (L.) community. The present experiments on daily growth patterns thus effectively supplement a sound framework of macroshell growth analyses (Hancock and Simpson 1962; Hancock 1965, 1967). ‘Transect F’ of Hancock and Urquhart (1966, figs. 4-6) was levelled accurately and living cockles collected from stations 40 m apart. Specimens from this transect, preserved immediately after collec- tion and subsequently oven dried, were supplied by Dr. D. A. Hancock and these form the main basis of this paper. The location of the transect within the estuary is shown in text-fig. 1.
TRANSECT DATA
Abundance and distribution of age-groups. Text-fig. 2 shows the abundance of the various year-groups present along the sand-flat in February 1969. Cockles in their first, second, and sixth winters are most abundant; in the intervening years spatfalls
[Palaeontology, Vol. 15, Part 1, 1972, pp. 61-72, pis. 8-10.]
62
PALAEONTOLOGY, VOLUME 15
TEXT-FIG. 1. Map showing the distribution of Recent sediments and the location of Transect F within the Burry Inlet, based on Admiralty Chart
no. 1167. All cockles described in the text are from this transect.
FARROW: PERIODICITY STRUCTURES IN CERASTODERMA
63
were insignificant. Densities of up to 7000/m^ are recorded for the 1967 year-group and this spatfall was selected for study since it was the most widely distributed along the transect. The density distribution of the other year-groups is broadly similar. Size-frequency distribution of cockles with two winter rings. Text-fig. 3 shows the varia- tion in shell length of the 1967 year-group along the transect, sampled on 14 May 1969. Modal specimens from stations 250 m apart are illustrated for comparison on text-fig. 4. Densest settlement occurred just below neap high water mark where three stations yielded more than 5000/m^ Both higher than this, and below it where the steeper flank of the flat is reached, densities are much lower.
6
4
Winter
Ring
3
TEXT-FIG. 2. Curves of density distribution of the various age-groups of cockles collected from the transect in February 1969; maximum density shown = 7000/m^. Plots based on OT m^ quadrats at
40 m intervals.
ANALYSIS OF RESULTS
Two sets of samples were used in studying internal details of shell growth by means of acetate peels of radial valve sections: the first a random series of shells from the February survey; the second a complete series of shells, one being selected from a sample of six modal individuals sent from Burnham for each station (text-fig. 3), obtained on 14 May 1969. Since chance preservation of shells in the fossil record does not usually permit large numbers to be examined it is of some value to compare the
64
PALAEONTOLOGY, VOLUME 15
F21
15
20
25
30
I
►
F8
)
FO
15
Shell
Length
(mm)
TEXT-FIG. 3. Curve showing the size-frequency distribution of two-year-old cockles along the transect on 14 May 1969; greatest width corresponds to 1400 of one particular centimetre group/m^; highest density shown = 5500/m^
TEXT-FIG. 4. Diagram showing the shore profile of Transect F (from Hancock and Urquhart 1966, fig. 6). Cockles selected from the mode at four stations each separated by about 250 m demonstrate the striking reduction in shell length on the higher parts of the flats (which is also apparent in their relative size at the first winter ring). Black dots indicate the location of specimens illustrated on PI. 9; stipple density is proportional to density of cockle settlement.
FARROW: PERIODICITY STRUCTURES IN CERASTODERMA
65
results obtained. External measurements of shell length were carried out by the staff at Burnham on all shells collected during the May survey.
Relationship of shell length to position on tidal flat. The striking visual correlation may be seen on text-fig. 4; this is also apparent in the cockle’s relative size at the first winter ring. Text-fig. 5a shows that when plotted out arithmetically the rate of reduction in
TEXT-FIG. 5. Graphs illustrating the relationship between modal shell length and tidal height: a, plotted arithmetically; b, plotted logarithmically from the curve fitted to a. Chart Datum (C^) = —14 ft O.D. (Newlyn).
total growth increases sharply on the higher parts of the flats. Plotting points from the fitted curve of text-fig. 5a on a logarithmic scale resolves growth characteristics into two groups, both of which are related logarithmically to tidal height. The bulk of the population may be fitted to the curve :
y = 0T9+0-087X,
where y = log(20— tidal height in feet+Q),
X = shell length in mm; Q = Chart Datum.
F
C 8472
66
PALAEONTOLOGY, VOLUME 15
This relationship seems to hold on the flats down to a point (text-fig. 5b) corresponding to a tidal height of about 1 1 feet, after which a second function is operative. However, text-fig. 3 indicates that values of modal shell length on the lower part of the transect may be in error owing to sparsity of individuals, so that it would be unwise to attach undue significance to these two categories. Nevertheless it seems unlikely that tidal cover per se can explain the observed variation completely. The fact that the rate of reduction in growth increases on the most level part of the flats may imply that friction causes a decrease in tidal current velocity here, and hence also in suspended matter on which the cockles feed. Greater turbidity in this region may also be important.
The theoretical limit of cockle growth may be determined from the above relation- ship at the point where x = 0. The resulting tidal height of 18-5 feet is only inches above the highest level at which cockles are recorded; which proves that the higher Burry cockles must be ‘stunted’ to a degree which approaches the maximum theo- retically possible.
Winter rings. The internal form of a cockle’s first winter ring is governed strongly by its position on the shore; this is seen in both the random (PI. 8) and modal (PI. 9) samples. Because of poor growth during the initial months the high shore shell F. 0 is still very thin during its first winter (PI. 8a) and may already truly be referred to as ‘stunted’ by reference to the longer, much thicker shells from lower on the shore ; the inner complex crossed-lamellar layer is particularly thin compared with F. 14 for example (PI. 8b). The most striking difference, however, is seen in the external width of the winter ring and in the sharpness of its incision into the generally smooth outer shell profile. Shells from the vicinity of low water springs show a relatively shallow trough beneath which growth appears to have been continuous (Pis. 8c, 9e). Moving up the shore this trough becomes more acute, its base being marked by a dark periostracal band extending back into the shell structure. This marks a stoppage of growth when the mantle was strongly retracted. Because of this stoppage, growth before and after is relatively rapid and produces a sharp nick in the highest shore shells (Pis. 8a, 9a), but the lower shore examples show a gradual diminution followed by a gradual increase which produces only a shallow indentation (PI. 9d, e). These wider areas are often more noticeable externally (e.g. F. 21, text-fig. 4) and create an illusion of greater winter susceptibility, in complete contradiction to the internal evidence.
One particular aspect of the shell structure is associated with any interruption of the normal pattern of calcification, being especially apparent during the winter. The crossed
EXPLANATION OF PLATE 8
Acetate peels showing daily increments during the first autumn and winter in three specimens of Cerastodenna edide selected at random from different stations along Transect F, Burry Inlet :
A. High-shore specimen (F. 0) showing disturbance ring and sharp winter ring: shell mueh thinner than lower shore individuals, especially the inner complex crossed-lamellar layer.
B. Mid-shore specimen (F. 14) showing markedly cyclical growth pattern, broader winter ring and thicker shell.
c. Low-shore specimen (F. 21) showing continuous, uniform growth and very broad winter ring; because of the mueh greater thickness of this shell only part of the outer shell layer can be illustrated.
Dark lines in a and b extending baek from the winter rings into the shell structure mark a cessation of growth.
Palaeontology, Vol. 15
PLATE 8
FARROW, Periodicity structures in Cerastodenna
FARROW: PERIODICITY STRUCTURES IN CERASTODERMA
67
lamellae become extremely prominent, appearing to transgress the growth lines; immediately before the stoppage they extend strongly outwards (PI. 9a) ; after it they appear inconspicuous. Marked deflections in the orientation of the lamellae take place across the dark periostracal band.
Disturbance rings. The problem of separating disturbance rings from those of annual origin can be tackled statistically by external growth-ring analysis (Craig and Hallam 1963), though this is a somewhat lengthy procedure. Preparation of an acetate peel of a shell takes only five minutes, and from it the two can be distinguished at a glance even at low magnification. PI. 8a shows an autumnal disturbance ring in the first year of growth; PI. 10a an autumnal disturbance during the second year on the same specimen. Although the interruptions may produce dark bands extending back into the shell structure in the same fashion as truly annual rings, albeit fainter, the disturbance rings are readily diagnosed by their suddenness; revealed by deep, narrow slots in the outer shell profile (rather than the gentle trough-like depressions associated with annual rings) ; and revealed by a stable background of high daily increments on either side of the disturbance (well seen in PI. 10a). Thus, in contrast to the pattern of daily growth preceding winter stoppages, there is no hint in the micro structure of the disturbance to come. Subsequent growth is at a rate identical to that prior to the stoppage, and this is a reflection of some kind of physical rather than biological disturbance, for with a spawning ring, for example, resumption of growth is gradual and background values may not be attained for some weeks (Pannella and MacClintock 1968, PI. 6, fig. 2).
Comparison of high and low shore cockles shows that physical disturbances are commonest on the higher parts of the flats; the external expression of an autumn disturbance ring in the first year can be seen on F. 8, text-fig. 4. PI. 10 shows growth during the second autumn for shells from three stations along the transect: F. 0 has a pronounced disturbance ring, but the indentation of the outer shell profile of F. 14 at the same period is scarcely perceptible; F. 21 from low water springs exhibits very uniform growth throughout the period. Physically induced disturbances are severe only during the autumn (Flouse and Farrow 1968; also for cockles inhabiting sand in the Thames, Farrow 1971). The reason for this is to be sought in the equinoctial tides, as is explained in the following section.
Cyclicity in the pattern of daily growth. Study of microstructural periodicities demon- strates the close control of tidal cyclicity on the growth of cockles from different transect stations. Again this is shown as well by random samples (text-fig. 6) as by individuals carefully selected from the mode at each station (text-fig. 7). On text-fig. 6 are plotted second year summer and autumn daily increments for the random series of cockles. A 29-day tidally controlled cyclicity can be identified, even in the low shore F. 17, but this becomes attenuated during the autumn with growth stoppages occurring at the beginning and end of each cycle in the high shore F. 0. The modal specimens (text-fig. 7) show that the severity of such stoppages decreases as low water mark is approached; further, it shows that this effect is pronounced only around the time of the equinoctial tides. The probable reason for this is illustrated on text-fig. 8 and outlined briefly below.
The higher shore cockles are situated very close to mean high water of neap tides (text-fig. 4), which means that at certain seasons when the tidal range is small the flowing
68
PALAEONTOLOGY, VOLUME 15
tide never reaches them and they may be left high and dry for days on end. The degree to which observed disturbances correlate with predicted occurrences of very low high water is shown on text-fig. 8. Even at other periods some individuals may be covered by only a few inches of water, and during the summer this evidently leads to a dis- turbance of growth (text-figs. 6, 7). Throughout the year it is the resumption of vigorous growth following neap-tide deceleration which produces the marked cyclicity. The reduced effect nearer low water springs enables more continuous growth to take place,
•’T V- "n- •’U-
TEXT-FIG. 6. Daily increment plots for three shells selected at random from different transect stations, showing summer and autumn growth in the second year. The amplitude of the cyclical growth pattern is considerably greater in the higher shore specunen F. 0 than in F. 17; in the autumn the pattern becomes attenuated as a result of periodic growth disturbances. Shells collected in February 1969; maximum increment = 1 50 /xm.
though in the second year individual daily increments are not necessarily greater here than higher on the shore, for shells like F. 21 have already reached a comparatively large size as a consequence of much more prolonged growth during their first autumn than higher shore shells (text-fig. 7) which are still quite small. It is thus possible to relate the amplitude of the cyclicity to position on the shore, values for each lunar cycle being tabulated in Table 1. Variation in the amplitude and attenuation of the cyclicity is thus the key to understanding why the higher shore specimens are stunted.
PALAEOECOLOGICAL IMPLICATIONS
Two aspects of the Burry Inlet experiment are of especial value to the palaeontologist. First, all specimens studied were collected in life position and the results obtained can thus be applied readily to in situ infaunal fossils. Second, two series of shells were analysed side by side; a random collection of single cockles from unselected transect stations and shells from a quantitative survey which were statistically selected from the modal shell length value for regularly spaced stations. The fact that the results could be
EXPLANATION OF PLATE 9
Acetate peels showing cross sections of the first winter ring in modal specimens of Cerastoderma edule from successively lower along Transect F: (a) F. 0 (b) F. 8 (c) F. 12 (d) F. 17 (e) F. 21. Trans- gressive crossed-lamellar structure is particularly evident in the higher shore individuals immediately prior to the winter stoppage. Progressive increase in width of the winter ring with increasing depth is well shown: in the lowest example growth appears to have been continuous, but dark periostracal bands indicating stoppage seem to be present in the remainder. Position of the stations is indicated on text-fig. 4 : Scale bar = 200 ixm.
Palaeontology, Vol. 15
PLATE 9
F12
FI 7
FARROW, Periodicity structures in Cerastoderma
H H
70
PALAEONTOLOGY, VOLUME 15
compared closely is encouraging for the extrapolation into the fossil record where numbers sufficient for statistical work are but rarely found.
The palaeoecological implications of the work are best assessed on two levels. First, that of the individual mollusc. Five minutes spent preparing an acetate peel for the internal analysis of any one shell enables many incidents in the life of that shell to be recognized. Winter rings can be distinguished readily from disturbance rings by their external profile and transgressive crossed-lamellar structure, thus enabling the true life span to be calculated : the season of greatest disturbance may be compared with the season of death : the month with lowest winter water temperature can be obtained from the correlation of pronounced autumnal growth cyclicity with the equinoctial tides.
Tidal Range
Daily Growth Increments number per cycle days stoppage
days HW below critical level (18')
TEXT-FIG. 8. Diagram illustrating the probable correlation between the occurrence of very low high tides and the incidence of growth disturbances in cockles from the higher parts of the flats during late summer and autumn of the second year. Daily increments = F. 0 (modal) : tidal values for each spring and neap tide from Whitaker's Almanac (1968) corrected according to Hancock and Urquhart (1966,
p. 15).
Secondly, implications may be considered at the community level. Here it should prove possible to assess the degree of mixing which has taken place between different faunal elements in dead shell accumulations by utilizing discrepancies in, for example, winter ring characteristics or in the amplitude of growth cycles. In considerations of possible
EXPLANATION OF PLATE 10
Acetate peels showing daily growth increments during the second autumn for the three shells illustrated on Plate 8 ;
A. High-shore specimen (F. 0) showing cyclical growth pattern and a pronounced disturbance ring, before and after which growth increments are large (cf. winter rings).
B. Mid-shore specimen (F. 14) showing cyclical growth but with only a slight indentation in the smooth outer shell profile.
c. Low-shore specimen (F. 21) showing the uniform thickness of the growth increments: a cyclicity in the type rather than the thickness of the diurnal bands is apparent if the plate is viewed from the side at a low angle.
Palaeontology, Vol. 15
PLATE 10
FARROW, Periodicity structures in Cerastodenna
F21
FARROW: PERIODICITY STRUCTURES IN CERASTODERMA
71
Stunting in any fauna internal analysis of periodicity structures may reveal its cause, whether it be a prolonged winter cessation of growth or repeated physical disturbance caused by periodic subaerial exposure.
The most novel application of this experiment to the fossil record would undoubtedly lie in palaeotidal analysis. It would be most exciting to extend the semi-quantitative assessment of intertidal environments using amplitude measurements, of the type pre- sented in Table 1, in conjunction with the equation formulated from text-fig. 5. Even extending the use of periodicity structures into sub-tidal regimes is likely to prove
TABLE 1 . Variation in amplitude of tidally controlled growth cycles in Cerastodenna edule from stations at different heights along Transect F, Burry Inlet, S. Wales
|
Transect |
Tidal height |
||||||
|
number 20 June |
19 July |
18 Aug. |
16 Sept. 15 Oct. |
Mean |
of Stations |
||
|
amplitude |
(inftA- |
||||||
|
Upper shore |
{May-Sept.) |
MLWS) |
|||||
|
F. 0 10 |
19 |
11 |
13 |
8 |
8 |
13-3 |
18-8 |
|
F. 8* 13 |
15 |
8 |
14 |
10 |
7 |
12-5 |
18-7 |
|
F. 11/12 5 |
7 |
8 |
10 |
4 |
5 |
7-5 |
17-5 |
|
F. 17 5 |
5 |
6 |
7 |
4 |
5 |
5-7 |
160 |
Lower shore
Amplitude = maximum daily increment— minimum daily increment per lunar monthly cycle (mm x 160); all values represent the average of measurements from text-figs. 6 and 7 except F. 8 * which was obtained from text-fig. 7 only. MEWS (Mean low water spring tide) = —15 ft O.D. (Newlyn).
rewarding to judge from the work of Rhoads and Pannella (1970, p. 158, fig. 9) who have illustrated the extremely uniform growth of deep-water molluscs, where spawning dis- turbances replace winter temperature and physical disturbance rings as the most conspicuous interruptions of the normal calcification process. Periodicity in the type rather than the thickness of diurnal increments seems to characterize the growth of those molluscs living in sublittoral habitats which are nevertheless influenced by tidal flux: compare Cerastodenna edule from extreme low water springs (PI. 10c, viewing at a low angle) with Tridacna squamosa (Pannella and MacClintock 1968, pi. 7) and Nucula proxima from —6m (Rhoads and Pannella 1970, fig. 9b). With further examples to act as standards molluscan growth characteristics could become most useful indicators of contemporary water depth in ancient seas. Before having confidence in such esti- mates, however, additional experiments on living molluses must be undertaken. First, there is the question of how widespread is the tidally controlled cyclicity characteristic of the Burry Inlet cockles; it is certainly dilficult to perceive in the Thames fauna where it is swamped by other ecological variables (Farrow 1971). Future studies should be undertaken not only on a wider range of cockle habitats but on a wider spectrum of organisms, for there is no reason to suppose that Cerastodenna should be more prone to physical growth controls than any other carbonate-secreting animal.
Acknowledgements. This work forms part of a project financed by the Natural Envffonment Research Council which was carried out under the general direction of Professor M. R. House. Without the generous co-operation of Dr. D. A. Hancock and Mr. A. C. Simpson of the Fisheries Research Laboratory, Burnham-on-Crouch, this paper could not have been written. Mr. K. G. Walker prepared and photographed the acetate peels. An early version of the manuscript was read by Dr. L. F. Penny, whose suggestion of additional mathematical treatment has resulted in major improvement.
72
PALAEONTOLOGY, VOLUME 15
REFERENCES
CRAIG, G. Y. and HALLAM, A. 1963. Size-frcqucncy and growth-ring analyses of Mytilus edulis and Cardiurn edide and their palaeoecological significance. Palaeontology, 6, 731-750.
FARROW, G. E. 1971. Periodicity structures in the bivalve shell: experiments to estabhsh growth controls in Cerastoderma edide from the Thames Estuary. Ibid. 14, 571-588.
HANCOCK, D. A. 1965. Graphical estimation of growth parameters. J. Cons. perm. int. Explor. Mer. 22, 77-90.
1967. Growth and mesh selection in the edible cockle {Cardiurn edule L.) J. Appl. Ecol. 4, 137-
157.
and SIMPSON, a. c. 1962. Parameters of marine invertebrate populations. In The Exploitation of
Natural Animal Populations. Ed. E. D. le gren and m. w. holdgate, Oxford, 29-50.
and URQUHART, A. E. 1966. The fishery for cockles {Cardiurn edule L.) in the Burry Inlet, South
Wales. Fishery Investigations, ser. ii, 25, 1-32.
HOUSE, M. R. and farrow, g. e. 1968. Daily growth banding in the shell of the cockle, Cardiurn edule. Nature, 219, 1384-1386.
PANNELLA, G. and MACLiNTOCK, c. 1968. Biological and environmental rhythms reflected in moUuscan shell growth. In Palaeontological Society Memoir 2, 64-79.
RHOADS, D. c. and PANNELLA, G. 1970. The use of molluscan shell growth patterns in ecology and palaeoecology. Lethaia, 3, 143-161.
G. E. FARROW
Department of Geology The University
Typescript received 3 March 1971 Hull
THE MECHANICAL PROPERTIES OF BIVALVE (MOLLUSCA) SHELL STRUCTURES
by JOHN D. TAYLOR and MARTIN LAYMAN
Abstract. Bivalve shells are composed of a two phased composite material consisting of calcium carbonate and a largely protein matrix. The two phases are arranged into a number of distinct shell structures; these occur in discrete layers and their occurrence appears to be correlated with mode of life. Some mechanical properties of individual shell structures were tested; these included compression, bending, impact, and microhardness tests. Density and matrix content were also determined. Some structures were nearly twice as strong as bone. The relative strength is apparently related to the size of the microstructural units rather than to the matrix content, which is low. The possible functional significance of the various shell structures is discussed but it is difficult to see why any structure apart from nacre, which is both the strongest and the phylogenetically oldest, has been evolved.
Recently much attention has been given to the mechanical properties of bone (Evans 1957; Currey 1964 and, with a good review, 1970; Bell 1969; and many others) but little attention has as yet been paid to other calcified tissues such as mollusc shells. In spite of the recent aetivity in the study of shell structures the only worker to have con- sidered the microstructure of mollusc shells from a mechanical functional point of view is Wainwright (1969). Having studied the microstructure of molluscan shell materials for some years (Taylor et al. 1969; Kennedy et al. 1969) we were impressed by an apparent correlation between the type of shell structure and the mode of life of the animal concerned. We thus decided to investigate the mechanical properties of these materials in relation to their possible functional significance. The limitations imposed by the mechanical properties of the shell materials may have influenced the course of molluscan evolution; for instance this study might help to explain why certain possible shell coiling forms have never been utilized in nature.
The bivalve shell has obviously important functions in the protection of the animal, the maintenance of the mantle cavity and the support of the organs within it. In addition the shell plays an important part in the burrowing and boring processes (Trueman 1968). The shell is as functional as the more widely studied structures such as gills, siphons, stomachs, etc. Wainwright (1969) has stated that the meehanical function of the shell depends upon its ability to resist deformation and failure under environmental stresses; and that two main factors in shell architecture, shape, and construction materials, are involved in determining shell strength.
SHELL STRUCTURES
The bivalve shell, like bone (Currey 1964), may be considered as a material consisting of two phases retaining their separate identities (Wainwright 1969). The phases are crystalline calcium carbonate in the form of ealcite or aragonite, and an organic matrix consisting largely of fibrous protein. The phases are arranged into various distinct fabrics which are recurrent throughout the Bivalvia and other molluscan classes. The mineralogy and micromorphology of these shell structures have been described in some
[Palaeontology, Vol. 15, Part 1, 1972, pp. 73-87.]
74
PALAEONTOLOGY, VOLUME 15
detail by Schmidt (1924), B0ggild (1930), Wada (1961), Wilbur (1964), Wilbur and Simkiss (1967), and Taylor et al. (1969 and in press).
The shell structures found in the Bivalvia belong to six main arrangements briefly described below and illustrated diagrammatically in fig. \a-h. Further details can be found in the references cited; the nomenclature is largely retained from Boggild (1930).
Simple prismatic structure consists of columnar crystals, polygonal in section, up to 200 ^tm in length and 9-80 /xm in width, but the size is very variable. Each prism is sur- rounded by a sheath of matrix. The prisms are aligned normal to the shell exterior and are usually found as an outer shell layer. Composite prismatic structure consists of very small needle-like crystals 2 /xm in width and up to 10 /xm in length radiating from a central axis which is aligned parallel to the shell exterior. This structure is found only as an outer shell layer. Nacreous structure consists of tablet-like crystallites 2-10 /xm in length and 0-4-3 /xm in thickness, which are arranged in sheets and in section have the appearance of a brick wall. Another variety of nacre has the crystallites arranged into columns (lenticular nacre). Nacreous structures are usually found in the middle and inner layers of shells. In foliated structure the crystalline units are lath-like crystallites 2-4 /xm in width, 0-2-0- 5 /xm in thickness and up to at least 20 /xm in length, and are arranged in side to side contact into irregular sheets which have the same general orientation towards the shell margin and lie subparallel to the inner shell surface. Crossed-lamellar structure consists of lath-like crystals 5 /xm in width and up to 20 /xm in length arranged into lamellae. The lamellae are of variable size but some can be seen with the naked eye ; in adjacent lamellae the crystallites are aligned in opposing direc- tions. Complex crossed-lamellar structure is rather similar to crossed-lamellar, but con- sists of an intergrowth of blocks of crystallites arranged with four principal orientations. Homogeneous structure consists of small granular crystallites up to 5 /xm in diameter with no obvious crystal form. The shell material deposited beneath the muscle attachment areas, the myostracum, has an irregularly prismatic structure.
The structures described above are found in discrete shell layers ; certain combinations of structures are recurrent and show a distribution related to the probable phylogenetic history of the class.
The morphology of the organic matrix has been extensively studied by Gregoire (1967 with references), and in bivalves mostly consists of lace-like sheets which surround and in some cases are contained within the crystallites. The protein of the matrix resembles the keratin-myosin-epidermin-fibrin group of fibrous proteins (Degens et al. 1967; Wilbur and Simkiss 1968). The variation in amino-acid composition and amino sugars may be related to both phylogenetic and environmental effects. Proteins of the shell matrix group are characterized by a high degree of cross-linkage, a feature which will have an effect on mechanical properties and resistance to disaggregation. Variation in the amount and type of cross-linkage in the various shell structures has not yet been studied, and any possible effects upon shell strength are unknown. Sur- prisingly little is known of the total matrix content of the structural types. Hare and Abelson (1964) gave some general results which indicated a total protein content of 0-l%-5%, varying between various shell structures. The work of Hudson (1967), although based upon more exact layer separation and documentation, examined too few structural types to be of use in the present context.
TAYLOR AND LAYMAN: MECHANICAL PROPERTIES OF BIVALVE SHELL 75
h
TEXT-FIG. 1. Diagrammatic rep- resentation of the textures of bivalve microstructures as seen in sections normal to the shell surface.
(a) Simple prismatic structure
(b) Composite prismatic struc- ture
(c) Sheet nacreous structure id) Lenticular nacreous struc- ture
(e) Foliated structure (/) Crossed-lamellar structure (g) Complex crossed-lameUar structure
(b) Homogeneous structure
76
PALAEONTOLOGY, VOLUME 15
METHODS AND MATERIALS
The fresh specimens used in the tests were supplied from Plymouth and Millport marine laboratories, with the exception of Tridacna maxima which was collected at Malindi, Kenya. All dry and preserved specimens were from the collections of the British Museum (Natural History).
Microhardness. Tests were made on shell layers from seventy species from a wide variety of habitats and geographical localities and exhibiting all the shell structural types. A list of species and localities is available on request.
Specimens of separate shell layers were mounted in quick-setting resin, ground flat and polished to 3 ju.m. When the layer to be tested was very thin it was mounted on the surface of the resin and tested without further treatment. Indentations were made with an Akashi microhardometer, a standard Vickers diamond pyramid indenter, and a load of 500 g. Several tests were made for each specimen and the average taken. The material tested was at least five times thicker than the depth of indentation; tests were made at more than five times the indentation diagonal from another indentation or the edge of the specimen. Initially tests were made upon Mytilus edulis to determine the hardness variation within a layer and the effects of age and orientation upon hardness. Fresh, dry, and formalin-preserved specimens were tested in a preliminary survey. Little difference was found between wet and dry, so dry specimens were mostly used in the survey. Considerable variation was found in formalin-preserved specimens.
Compression tests. Test specimens of dimensions 8 X T5 x 1 -5 mm were cut using a Capco Q. 35 cutting machine, which produces parallel cuts and ensures accuracy to 0 025 mm. The specimens were glued to a Sindyano base which was attached to a base allowing 90° rotation. The cutting machine was lubricated and cooled by mineral oil which might conceivably penetrate the specimens, but this was unavoidable. The specimens were carefully washed after cutting and fresh specimens kept under water until tested.
The length to diameter ratio of the specimens was high, and this may have produced slight bowing which could reduce the values of compressive strengths obtained and also have some effect upon the modulus of elasticity. However, in producing longer specimens the stresses during cutting were reduced. There was little plastic deformation produced by the tests and as the specimen ends were cut parallel the tendency to bow was reduced. The convenience of the larger specimens outweighed the effect of buckling, which was considered to be small. The results are valid for comparative purposes even if the absolute values may have a small error.
Testing was carried out using an Instron, an accurate machine with a high elastic stiffness, upon which a load versus compression graph is automatically plotted. A crosshead speed of 0 05 cm/minute was used throughout testing. Both fresh and dry specimens were tested to fracture and the results plotted on a stress/strain curve. Similar specimens were tested to a load below fracture and then the cross head velocity reversed and the load removed. Griffith’s cracks on the specimen surface may influence the fracture strength; although the specimens appeared satisfactory visually the cutting pro- cess may have caused some surface deformation. A sample of nacre without visible banding was there- fore polished on diamond paste to 1 jum, and tested for comparison with the unpolished specimens.
Bend tests. Bivalve shells are brittle, and with the equipment available it was not possible to cut specimens in a suitable shape for direct tensile tests. Thus, as in ceramics, the modulus of rupture as determined through bend tests was used to give an indication of tensile properties.
For bivalves a three point test was used for convenience, although the superiority of the four point test is recognized. The dimensions of the test specimens were 20 mm in length, 5 nun in width, and T5 mm in depth. The length between the lower knife edges was 16 mm. The cutting of the specimens was carried out on a Capco cutting machine similar to that used for compression tests. Dry and fresh wet specimens were tested. The bending was carried out on a three point test rig with an Instron testing machine. Displacement of the specimen at the load point was automatically plotted against load and the specimens tested until fracture.
Impact tests. There was no standard impact testing machine suitable for the testing of bivalve shells.
TAYLOR AND LAYMAN; MECHANICAL PROPERTIES OF BIVALVE SHELL 77
Either the machine was too large and lacking in sensitivity or the specimen size and shape were un- suitable. Consequently a simple test machine was constructed which, although unable to give absolute values of the energy absorbed, could give a comparison of the impact resistance of the various shell structures. The apparatus was modified from a crystal cleaver mounted in a wooden frame, with a hammer head replacing the cleaver blade and the specimens held against two blunt edges of metal. Portions of fresh shells containing one or more shell layers and periostracum were tested but as the shells were of different thickness, curvature, and ornamentation little direct information was obtained on shell structures. To obtain results for individual shell layers specimens of single structural types were cut to 5 X 1-5 X 1-5 mm on a Capco cutting machine and tested in both the fresh and dry states.
Density. The densities of individual shell structures were measured in two ways; by a standard weighing method and by a titration method using heavy liquids (Embrey 1969). The results obtained were closely comparable.
Total organic nitrogen content. A Kjeldahl digestion method was used, followed by steam distillation of the alkali treated digest. Initially this technique was applied on a semi-micro scale using up to 100 mg of shell. In view of the variability of the small quantities of nitrogen detected in some samples the amount of shell subsequently used was increased tenfold.
Pieces of individual shell structures were separated out, care being taken to remove all the peri- ostracum. The shell was then digested over low heat with 6 ml of 50% sulphuric acid containing 1 % selenium dioxide plus a small crystal of cupric sulphate. Prior to steam distillation into OOIN sulphuric acid the digest was made alkaline by the addition of 14 ml of ION sodium hydroxide. The quantities involved were within the scope of Quickfit semi-micro apparatus, and although the variation between samples of the same piece of shell structure was still high the limits appeared to narrow with the increased quantity of shell used.
Microstrnctiires. Microstructures were studied by acetate peels of polished and etched sections of shells and by reflected light microscopy of polished surfaces. Surfaces and sections were also examined by scanning electron microscopy.
RESULTS
Compression tests. A graph of load versus compressive strain was automatically plotted during compression testing and then replotted as a stress/strain diagram (text-figs. 2-6). After minor adjustments (bedding down) most specimens exhibited a virtually linear relationship (text-fig. 2). This signifies elastic behaviour with the material obeying Hooke’s law. The modulus of elasticity was obtained from the slope of the stress/strain curve. Deformation was elastic almost up to fracture, with possibly a small amount of ‘plastic’ deformation just before the point of fracture. Slight local deviations were observed in some curves ; these were small and made no difference to the over-all form of the plot but indicated that the mechan- isms of deformation, although apparently corresponding