m xg* m £jj its ''R g x^jgssx m ISTITUTION NOIinillSNl“"NV!NOSHllWSr S3 I H VH 8 ll“u B R AR I ES^SMITHSONIAN- INSTITUTION^ NO .v. £2 z £2 z w z «, ■ co «v. 2 ,< S ,.< . v, ^ " Z /wlX^ H .X /r.;/ A* z H CO |V o % 8fe^ I /J^g 3 ssPltt ^ co wW co - n|§lr ~ 2 t 3 I dvy a II2 LI B RAR I ESWSMITHS0NIAN2|NST!TUTl0NOTN0liniiiSNI^NVIN0SHilWSWS3 I dVd 8 11*" L! -5- w -?> _ co —■ to — O N^N. D-l^ ~ '<%S^ o — XJ/llS^Z Q ISTITUTION^NOlinillSNI^NVINOSHlISAIS S3 1 d Vd 8 11 B R AR 1 ES^SMITHSONIAN^INSTITUTION-" NO „ ^ I > g ^ I- Z . . «- 2 O m X £J XtSsTlriyJs. x'C'OQ \vX O _ /^oljN O m v >0*5?’ xjt-Ast'j^ m x^osv^x ^ m — co ~ co ± c/> ~ = aiavaan libraries Smithsonian institution NounuiSNi nvinoshiiws saiavaan lie 2 ^ - Z \ CO z CO . z < s < X I S /r^5x I < z ^ H wW/fo 2 H w¥4?// z co v *!&*&• co MavMA co |£fb sbl w y^fz: to X -A o X ^ Z.W-'* t VJ S "S^S'- > 2 '' >■ s >' 5 STITUTION N0IinillSNI_ N.VIN0SHlllAISi°S3 I d V H a n\ I B R A R I ES^SMiTHSONIAN^INSTITUTION NO, CO g-ZL -y > CO — C/5 — co CO o z siavaan libraries Smithsonian institution NounuiSNi nvinoshiiws saiavaan li co — to £ — to \ 5 co JSTITUTION NOlinillSNI NVIN0SH1UAIS SliaVHSll LIBRARIES SMITHSONIAN INSTITUTION NO v_. CO Z ^ CO Z ,...-. CO 2 CO S < Zcc^o^cX S .,, < S /CtSvJTx < mS ... X w > o VA ■/s/A > 2 a 2 3iavaail LIBRARIES^SMITHSONIAN INSTITUTION NOlinillSNI NVINOSHliWS SBiavaail^LI co X kSS co > z z CO ^STITUTION NOlinillSNI NVIN0SH1IINS S3ldVHail LIBRARIES SMITHSONIAN INSTITUTION NO z r- x„ z r~ z <“ 2 rn ^ (tj ^ CO '' ± c/5 — ± to 3 lava a ii libraries Smithsonian institution NounuiSNi nvinoshiiws saiavaan lie ? £2 ^ ^ z \ CO 2 _ 5 ^ 5 w ‘NVIN0SHlllAls'"S3iaVaan LIBRARIES^ SMITHSONIAN INSTITUTION „ NOIlfllllSNI NVINOSHIIIAIS . em ^ . (/) 'y ... W ^ 4 S^SMITHSONIAN^INSTITUTION^NOIiniliSNI^NVINOSHil^S^Sa I HVH 8 l“I_U B RAR I ES^SMITHSONIAN. • s I .J& i? 1 5 < m r- Kc^a^o/ m ^ Q ^V. D^/ _ O ll_lNVIN0SHimS^S3 I ava a I1JU B RAR I ES^ SMITHSONIAN^ INSTITUTION NOIinillSNI~NVINQSHlllAJS " ' lj%l ^ r= if £0 m x^vosv^x ^ m xiiyost^ S°°SMITHSON IAN ""*INSTITUTION^°NOiinillSN I ” NVINOSHIIIAIS S3 I a VH 9 II LI B R AR I ES SMITHSONIAN co z <, co z 05 z 'Vos*''JZ > Z c/> •* Z NVINOSHIIIAIS1" S3 I a va 0 II^LI B RAR I ES^SMITHSONIAN INSTITUTION NOIini!lSNI__NVINOSHillAIS w — tn ~ co z CO O - ” -^VA^ O S SM ITHSON IAN J INSTITUTION NOlinillSNI ^NVINOSHIIIAIS S3 IBVaS II LIBRARIES SMITHSONIAI' f~ z 2 z 2 co o yo > 33 — Yc? m xjw/ «; Njo>iiis2X m xj xco.in.sgx m __(/)” CO \ Z CO ni NvmosHims SBtavaan libraries Smithsonian institution NoiinmsNi nvinoshuws Z co Z co z > UJ ~Z " yj k. z co *S SMITHSONIAN INSTITUTION NOlinillSNI NVINOSHIIIAIS SiiaVHail LIBRARIES SMITHSONIAN co — to “ to INI NVINOSHIIIAIS S3iavaan LIBRARIES SMITHSONIAN INSTITUTION NOIlfUllSNI NVINOSHIIIAI! ~ z r- Z f“_z ' z 03 ~ JHfo. ra /-p- ^ t; m ^ xjvasv^x rn x^nos^^ ^ 8X1 ES SMITHSONIAN- INSTITUTION^NOIlDlllSNI *~NVINOSHIIl\S S3 I ava a 11 ^ B RAR I ES SMITHSONIA CO 2: V) z 2 00 THE PALAEONTOLOGICAL ASSOCIATION The Association was founded in 1957 to promote research in palaeontology and its allied sciences. COUNCIL 1984-1985 President : Professor C. Downie, Department of Geology, University of Sheffield, Sheffield SI 3JD Vice-Presidents'. Dr. J. C. W. Cope, Department of Geology, University College, Swansea SA2 8PP Dr. R. Riding, Department of Geology, University College, Cardiff CF1 1XL Treasurer : Dr. M. Romano, Department of Geology, University of Sheffield, Sheffield SI 3JD Membership Treasurer. Dr. A. T. Thomas, Department of Geological Sciences, University of Aston, Birmingham B4 7ET Institutional Membership Treasurer. Dr. A. R. Lord, Department of Geology, University College, London WC1E 6BT Secretary. Dr. P. W. Skelton, Department of Earth Sciences, Open University, Milton Keynes MK7 6AA Circular Reporter. Dr. D. J. Siveter, Department of Geology, University of Hull, Hull HU6 7RX Marketing Manager. Dr. R. J. Aldridge, Department of Geology, University of Nottingham, Nottingham NG7 2RD Editors Dr. D. E. G. Briggs, Department of Geology, Goldsmiths’ College, London SE8 3BU Dr. P. R. Crowther, Leicestershire Museums Service, Leicester LEI 6TD Professor L. B. Halstead, Department of Geology, University of Reading, Reading RG6 2AB Dr. R. Harland, British Geological Survey, Keyworth, Nottingham NG12 5GG Dr. T. J. Palmer, Department of Geology, University College of Wales, Aberystwyth SY23 2AX Other Members Dr. E. N. K. Clarkson, Edinburgh Dr. C. R. C. Paul, Liverpool Dr. D. Edwards, Cardiff Dr. A. B. Smith, London Dr. P. D. Lane, Keele Professor T. N. Taylor, Columbus Dr. A. W. Owen, Dundee Overseas Representatives Australia: Professor B. D. Webby, Department of Geology, The University, Sydney, N.S.W., 2006 Canada: Dr. B. S. Norford. Institute of Sedimentary and Petroleum Geology, 3303-33rd Street NW., Calgary, Alberta Japan : Dr. I. Hayami. University Museum, University of Tokyo, Hongo 7-3-1, Bunkyo-Ku, Tokyo New Zealand: Dr. G. R. Stevens, New Zealand Geological Survey, P.O. Box 30368, Lower Hutt U.S. A.: Dr. R. Cuffey, Department of Geology, Pennsylvania State University, Pennsylvania Professor A. J. Rowell, Department of Geology, University of Kansas, Lawrence, Kansas 66045 Professor N. M. Savage, Department of Geology, University of Oregon, Eugene, Oregon 97403 South America: Dr. O. A. Reig, Departamento de Ecologia, Universidad Simon Bolivar, Caracas 108. Venezuela MEMBERSHIP Membership is open to individuals and institutions on payment of the appropriate annual subscription. Rates for 1984 are: Institutional membership . . . £38-50 (U.S. $67.50) Ordinary membership . . . £18-00 (U.S. $31 ) Student membership . . . £1 1-50 (U.S. $20) Retired membership . . £9-00 (U.S. $16) There is no admission fee. Correspondence concerned with Institutional Membership should be addressed to Dr. A. R. Lord, Department of Geology, University College, Gower Street, London WOE 6BT, England. Student members are persons receiving full-time instruction at educational institutions recognized by the Council. On first applying for membership, an application form should be obtained from the Membership Treasurer, Dr. A. T. Thomas, Department of Geological Sciences, University of Aston, Gosta Green, Birmingham B4 7ET. Subscriptions cover one calendar year and are due each January; they should be sent to the Membership Treasurer. All members who join for 1984 will receive Palaeontology , Volume 27, Parts 1-4. All back numbers are still in print and may be ordered from Marston Book Services, P.O. Box 87, Oxford OX4 1 LB, England, at £19-50 (U.S. $34.50) per part (post free). Cover: The Coralline Crag (Pliocene) bryozoan Cribrilina sp. showing a group of ten feeding zooids each with an ovicell and paired adventitious avicularia on either side of the orifice. This specimen was figured as Lepralia punctata Hassall in G. Busk’s Palaeontographical Society monograph of Crag Polyzoa ( 1 859, pi. 4, fig. 1 ). It is reillustrated here by means of a new technique, scanning electron microscopy of the uncoated specimen using back-scattered electrons, x 75. CLASSIFICATION OF THE ECH INODERM ATA by ANDREW B. SMITH Abstract. A critical review of past attempts to classify echinoderms is presented and it is shown that, in retrospect, fossil groups have been incorporated into classifications in an arbitrary manner that has confused rather than clarified. The search for relationship through the recognition of pattern in character distribution has become progressively divorced from the production of classification schemes, and the most recent classifications are the most ambiguous about relationships. Furthermore, the increased knowledge we have gained about fossil echinoderms has added very little to our understanding of how extant groups are interrelated and, indeed, has sometimes been interpreted misleadingly. It is argued that fossils cannot generally provide insight into the relationships of living groups except where characters have been lost through developmental foreshortening. The most important taxonomic information that palaeontology can provide concerns the pattern of character acquisition within the stem group, although it can also be useful in providing the latest date by which a split occurred, and in checking statements of homology and identifying synapomorphic characters that have been lost in one or other sister group. It is concluded that the higher classification of the Echinodermata should be based first and foremost on the distribution of characters gleaned from the study of embryology and comparative anatomy in living echinoderms. Fossil groups can then be added to this classification in their appropriate place. An analysis of character distribution amongst the five extant classes of echinoderm shows that the Eleutherozoa form a monophyletic group whose primitive sister group is the Pelmatozoa. Within the Eleutherozoa, asteroids are the primitive sister group to the group (ophiuroids + echinoids + holothuroids) for which the name Cryptosyringida is proposed. The relationship of holothuroids within the cryptosyringids is more ambiguous but it is concluded that echinoids and holothuroids are sister groups and more closely related to one another than either is to the ophiuroids. A phylogenetic classification is proposed and this provides the primary framework into which fossil groups can be incorporated by using the concept of stem and crown groups. The position of principal fossil groups within this classification is briefly outlined and outstanding problems for future research are identified. Within the last few years, systematics, the study of biological classification in accordance with natural relationships, has undergone a rigorous scrutiny of its methodological basis. This debate has been fought largely, though not entirely, amongst zoologists and vertebrate palaeontologists and a vast literature now exists discussing the virtues and vices of phyletic, phenetic, and gradistic methods of classification. This debate has done nothing but good for the science of systematics and I feel that cladistic methodology has proved itself the most internally consistent and the most informative method of organizing data on character distribution. My interests he in unravelling the phylogeny of echinoderms and producing a classification that reflects this. The phylogeny can be inferred from analysis of character distribution which can be presented in the form of a branching diagram (cladogram) and the most informative classification is one that follows the hierarchical pattern revealed by the cladogram. However, I shall not discuss the merits of cladistics over other methods of classification since there is more than enough written on this subject already. Those unfamiliar with the ideas of cladistics and how they compare with more traditional methods may read one of the many books that have recently appeared on the subject (e.g. Eldridge and Cracraft 1980; Nelson and Platnick 1981; Wiley 1982). While arguments have raged in systematic zoology, the systematics of fossil invertebrates, as reflected in the pages of this journal, has continued much as before. Of course, some articles using cladistic methodology have appeared but there is still the prevailing feeling that the fossil record holds the key to understanding relationships. But this belief has been challenged. Fossils, it is said, are irrelevant in determining biological relationships (Kitt 1974; Lovtrup 1977) or play only a minor role [Palaeontology, Vol. 27, Part 3, 1984, pp. 431-459] 432 PALAEONTOLOGY, VOLUME 27 (Patterson 1982). If these challenges are correct then systematic palaeontologists must not only reassess their methodology but also their aims. This paper takes a critical look at the way in which echinoderms (particularly fossil echinoderms) have been classified, outlines what their fossil record can and cannot tell us, and suggests how they might be classified more informatively. Few, if any, of the ideas are new (although they have not been applied to fossil echinoderms before) but I feel it is important to make a clear statement of the methodology employed when recommending a fairly drastic change to the classification of echinoderms. THE EVOLUTION OF ECHINODERM CLASSIFICATION It is most instructive to follow the way in which the classification of echinoderms has altered as our knowledge of fossil echinoderms has improved. Echinoderms were not recognized as a natural group until 1791 when Bruguiere subdivided Linnaeus’s class Vermes in which they had previously been put. Bruguiere included asteroids, ophiuroids, and echinoids in his order Echinodermata but failed to recognize holothuroids as echinoderms. In 1801 Lamarck added holothuroids to the Echinodermata but grouped them with medusoid coelenterates in the class Radiata. Twenty years later Miller (1821) formally separated a group Crinoidea for stalked echinoderms that had previously been placed with the starfish in the group Stellerides. Thus by 1821 the five classes of living echinoderm had all been recognized. As an example of an early attempt to classify echinoderms I shall use the scheme proposed by Forbes (1841 ). He did not consider fossil forms and on the basis of comparative anatomy proposed the following grade classification: (i) Pinnigrada — Crinoideae (iv) Cirrhispinigrada— Echinidae (ii) Spinigrada — Ophiuroideae (v) Cirrhivermigrada — Holoturiadae (iii) Cirrhigrada — Asteriadae (vi) Vermigrada — Sipunculidae Forbes based his classification on what he identified as a ’progression of organization' from polyps to vermes starting with crinoids and ending with sipunculids and he referred to it as a procession through ‘forms gradually changing character’. This pre-Darwinian view of echinoderm relationships can be summarized as a gradistic tree (text-fig. 1a) and translated into a fully resolved gradogram (text-fig. 1b). Forbes then was very specific about how he thought echinoderm groups were related (though, of course, not necessarily correct) and used a classification scheme which reflected this. A Vermigrada Sipunculidae Cirrhivermigrada Holoturiadae Cirrhispinigrada Echinidae Cirrhigrada Asteriadae Spinigrada Ophiuroideae Pinnigrada Crinoideae text-fig. 1 . Interrelationships of echinoderm groups according to Forbes (1841). a, Forbes’s gradistic classification scheme, b, the gradogram derived from the classification. SMITH: ECHINODERM CLASSIFICATION 433 In the second half of the nineteenth century great strides were being made in both embryology and palaeontology and in 1900 Bather published a major account of Recent and fossil echinoderms in which he proposed the following classification: Grade A Pelmatozoa Class I Cystidea Class II Blastoidea Class IHCrinoidea Class IVEdrioasteroidea Grade B Eleutherozoa Class I Holothurioidea Class II Stelleroidea Subclass Asteroidea Subclass Ophiuroidea Class IIIEchinoidea Bather’s classification identifies three components within the Echinodermata (text-fig. 2a): a group Pelmatozoa, a group Eleutherozoa, and a group Stelleroidea. This is compatible with any of forty- five fully resolved statements of relationship. In addition, he presented a diagram which summarized his views on how these groups were related phylogenetically (text-fig. 2b) which can be transformed B Pelmatozoa Cystidea Edrioasteroidea Blastoidea Crinoidea — Holothurioidea > Eleutherozoa — Echinoidea — Stelleroidea text-fig. 2. Interrelationships of echinoderm groups according to Bather (1900). a, the information concerning relationship that is conveyed in Bather’s classification, b, his diagram showing how he thought the various groups were related to one another, c, a phylogram derived from b. 434 PALAEONTOLOGY, VOLUME 27 into a phylogram (text-fig. 2c). Comparing the classification and phylogeny shows that Bather used a classification that conveyed some but not all of the phylogenetic information. Bather’s scheme made three changes to the previous scheme of Forbes, two of which stem from the growth in knowledge about fossil echinoderms. Palaeontology showed that living crinoids were only a small remnant of a once much larger and more diverse group of stemmed echinoderms. Bather recognized three fossil groups in addition to crinoids, placing the whole lot in the subphylum Pelmatozoa. He also realized that the other living groups were more advanced in being unattached <3 B Protocoelomat a I first fixed ancestor Protopelmatozoa (?Thecoidea) / \ Cystoidea Crinc / Blastoidea text-fig. 3. Interrelationships of echinoderm groups according to MacBride (1906). a, the information concerning relationship that is conveyed in MacBride’s classification, b, his diagram of echinoderm phylogeny. c, a phylogram derived from b. SMITH: ECHINODERM CLASSIFICATION 435 and grouped them together in the subphylum Eleutherozoa. Secondly, Bather chose to group asteroids and ophiuroids together because fossils existed that were intermediate in form making any distinction based on character distribution in living groups unworkable. The illogicality of this view is discussed later. Finally, he reinterpreted holothuroids as the most primitive living eleutherozoans, not the most advanced, on the basis of embryological data. Bather recognized holothuroids to be ‘primitive with respect to Pelmatozoic structures, specialised as regards eleutherozoic’ but chose to emphasize the symplesiomorphic aspects of holothuroid development which, in retrospect, was a misjudgement. We can contrast the approach taken by Bather, who was a palaeontologist, with that of MacBride, an embryologist. MacBride (1906, 1914) considered the phylogenetic significance of echinoderm development without reference to the fossil record. He used a classification scheme that was identical to Bather’s except that asteroids and ophiuroids were separated at class level. His classification then identifies just two components (text-fig. 3a). MacBride ( 1906) illustrated how he believed the various echinoderm groups were related in a diagram (text-fig. 3b) which can be translated into a fully resolved phylogram (text-fig. 3c). Although MacBride chose a classification whose structure contained little of the information he had gleaned from embryology, he was able to make a positive contribution by reversing the position of asteroids and ophiuroids as set out by Forbes. He did this by recognizing that embryologically ophiuroids were more advanced than asteroids. Throughout this century palaeontologists have continued to discover and describe new fossil groups and in 1955 Hyman published her excellent review of echinoderms with the following classification: Subphylum Pelmatozoa Class Heterostelea Class Cystidea Class Blastoidea Class Edrioasteroidea Class Crinoidea Subphylum Eleutherozoa Class Holothuroidea Class Echinoidea Class Asteroidea Class Ophiuroidea Class Ophiocistioidea Two previously known fossil groups have been elevated to class level, the carpoids (Heterostelea) and the ophiocistioids making ten classes in all. In the text Hyman seems to accept MacBride’s views on echinoderm relationships yet the classification identifies just two categories higher than class level (text-fig. 4) and is consistent with over 1 1,000 possible fully resolved phylogenetic schemes. Thus it is relatively uninformative. Recently, there has been a dramatic increase in the number of minor fossil groups each containing a small number of distinctive species that have been elevated to high categorial rank. In the Treatise on text-fig. 4. The information concerning relationship that is conveyed in the classification of Hyman (1955). 436 PALAEONTOLOGY, VOLUME 27 Invertebrate Paleontology (Moore and Teichert 1978) a total of twenty-one classes, sixteen of which are extinct, are arranged into four subphyla as follows: Subphylum Homalozoa Class Ctenocystoidea Class Stylophora Class Homostelea Class Homoiostelea Subphylum Crinozoa Class Eocrinoidea Class Rhombifera Class Diploporita Class Blastoidea Class Parablastoidea Class Paracrinoidea Class Crinoidea Subphylum Asterozoa Class Stelleroidea Subclass Somasteroidea Subclass Asteroidea Subclass Ophiuroidea Subphylum Echinozoa Class Helicoplacoidea Class Camptostromatoidea Class Edrioasteroidea Class Edrioblastoidea Class Cyclocystoidea Class Ophiocistioidea Class Echinoidea Class Holothuroidea This classification relies heavily on the work of Fell (1945, 1962, 1963, 1965, 1967) who rejected embryology as a guide to relationships and in its place attempted to use fossils as the guiding criterion. The results of this departure can be seen in the marked increase in uncertainty about relationships. The information content contained in the classification has also decreased significantly (text-fig. 5). Just four components are recognized leaving five polychotomies and this scheme is consistent with over 2 x 101 1 possible statements of relationship! An alternative classification has been proposed by Sprinkle (1980) who recognized a fifth subphylum, as follows: Subphylum Crinozoa Class Crinoidea Class Paracrinoidea Subphylum Blastozoa Class Eocrinoidea Class Rhombifera Class Diploporita Class Parablastoidea Class Blastoidea Subphylum Asterozoa Class Asteroidea Class Ophiuroidea Subphylum Echinozoa Class Edrioasteroidea Class Edrioblastoidea Class Cyclocystoidea Class Helicoplacoidea Class Ophiocistioidea Class Echinoidea Class Holothuroidea Subphylum Homalozoa Class Stylopora Class Homoiostelea Class Homostelea Class Ctenocystoidea The information content of this classification is better, but only marginally so (text-fig. 6). Five components are identified leaving four unresolved polychotomies and the classification is consistent with over 1-7 x 109 different statements of relationship. So what conclusions are to be drawn from the way in which echinoderms have been classified in the past? Forbes provided a classification in which his ideas of relationship, as revealed by morphological organization, were clearly specified. Since then there has been a progressive decrease in the information about relationships that is incorporated into classification schemes, despite an increasing understanding of embryology and palaeontology. The growth of knowledge concerning embryology led to the construction of clearly defined phylogenetic hypotheses and corroborated all but one of Forbes’s findings. By showing that of the four extant classes of eleutherozoans, asteroids have the most generalized development and are therefore more primitive than ophiuroids, embryology made a positive contribution to our knowledge of relationships. The increased knowledge of the fossil record seems to have had no such beneficial effect. The result SMITH: ECHINODERM CLASSIFICATION 437 0 a ) ~o o >» o o c 0 O 03 o .c a o if) 03 <13 03 < r, > o E 0 1 03 03 03 o O E o X 03 0 "D O <0 >. O o o > O 03 0 "D O JZ o o X of discovering more and more fossil groups that have some characteristics that set them apart from living groups has been to add to the general confusion. By elevating these fossil groups to high taxonomic rank, the hierarchical arrangement of Linnaean classification has been largely destroyed and its most important attribute, its information content, greatly reduced. The most recent classifications are also the least specific about character distribution amongst the groups they recognize. The obvious question then arises— is our increasing uncertainty about relationships in echinoderms real or is it an artefact of the way in which data, particularly palaeontological data, have been handled? If the former is correct and the more fossils we continue to find the more confused our ideas of relationship become, then palaeontology can have nothing to contribute to this subject. However, the confusion that has arisen is attributable to two causes, misinterpretation of what the fossil record tells us and an inappropriate taxonomic methodology. The following two examples are given in illustration. 0 0 ■u O O O c 0 O 0 0 0 O E o X 0 0 "O O 0 CO 0 0 "D O O 0 0 CL 0 0 T3 O C O 438 PALAEONTOLOGY, VOLUME 27 1 . The position of ophiuroids in relation to other echinoderms. The clearest example of how the fossil record has been misinterpreted comes from the way in which ophiuroids have been linked with asteroids. Most zoologists who have considered the relationship of ophiuroids to other echinoderms have been so struck by the fact that ophiuroids and echinoids pass through very similar developmental stages that are advanced compared to those in asteroids, that they believe ophiuroids and echinoids to be more closely related (e.g. Hyman 1955). Yet amongst many palaeontologists from Bather onwards there has been a clear belief that the fossil record shows asteroids and ophiuroids to be more closely related and distinct from echinoids (text-fig. 7). This has led some palaeontologists to claim that embryology is misleading and best ignored (e.g. Fell 1967) whilst some zoologists flatly refuse to believe that the fossil record can be correct (e.g. Hyman 1955). What then does the fossil record show? Excellent work by Schondorf, Schuchert, and Spencer has shown that, in the lower Palaeozoic, asteroids and ophiuroids are much less distinct from one another (i.e. they have fewer autapomorphies) and that indeed there are some forms so generalized (primitive) in form that Forbes (1841) Palaeontology Embryology Bather (1900) Moore STeichert (1978) Sprinkle ( 1 980) MacBride (1906) text-fig. 7. The changing ideas of relationship amongst the five extant classes of echinoderm, in the form of phylograms. Forbes (1841) based his ideas primarily on comparative anatomy, MacBride (1906) on embryology. Input from palaeontology has actually made relation- ships less clear since, in recent classifications, there is an unresolved primary trichotomy. SMITH: ECHINODERM CLASSIFICATION 439 they can be interpreted as ancestors to both asteroids and ophiuroids. This led to the claim that because the fossil record proves that asteroids and ophiuroids stemmed from a common ancestor they must be more closely related compared to echinoids, whose origins are still largely unknown palaeontologically (e.g. Nichols 1968). It may come as some surprise then to realize that the two views are not contradictory but complementary. Both embryological and palaeontological observations are in complete agreement; it is only the interpretation placed on the palaeontological data that is at fault. The fossil record shows that asteroids and ophiuroids share a common ancestry— but this is also predicted from embryological evidence (see text-fig. 7). Embryological data make a further prediction: that echinoids and ophiuroids shared a common ancestor that was not also shared with asteroids, i.e. that some of the so-called primitive ophiuroids will turn out to be generalized enough to have been ancestral to both living ophiuroids and living echinoids. Surprisingly, the implications of the embryological data have not been realized before now and the absence of obvious fossil evidence has been taken as sufficient proof for rejecting the wealth of zoological data. The question which has never before been addressed and which is only now beginning to be investigated concerns which of the 'primitive ophiuroids’ are true ophiuroids and which are so generalized in morphology that they are best considered as ancestors to both living ophiuroids and living echinoids. 2. The classification of eocrinoids. As an example of what could be considered to be misdirected taxonomic endeavour I shall discuss the way in which eocrinoids, a primitive group of cystoids, have been classified. Eocrinoids were first recognized as a distinct group by Jaekel (1918), who believed them to be primitive crinoids. More recent work has clearly identified them as cystoids sensu lato ( = Blastozoans), and there seems to be complete agreement amongst all workers that eocrinoids are the most primitive group of cystoids from which all the other cystoid groups evolved: the ‘root stock’ of other cystoid groups to use gradistic terminology. More than any other pelmatozoan group, eocrinoids have been difficult to diagnose satisfactorily. For example, one of the most thorough and detailed reviews of the eocrinoids was carried out by Sprinkle (1973), yet his diagnosis for the Class Eocrinoidea is as follows: ‘Early blastozoan echinoderms having an irregularly adjacent or imbricately plated globular or flattened calyx, with or without epispires, an irregularly multiplated holdfast or a true stem as an attachment appendage [except for Lichenoididae], a primitive ambulacral system bearing normal or modified brachioles and usually little pentameral symmetry' (Sprinkle 1973, p. 58). The only unifying characteristic of this group of pelmatozoans seems to be ‘primitiveness’. Indeed, the Eocrinoidea includes a heterogeneous assemblage of species whose only similarity is that they lack the autapomorphic characteristics of the other, less ambiguously defined, cystoid groups. As such, they are simply what remains of the Cystoidea once species with diplopores (Diploporita), rhombs (Rhombifera), hydrospires (Blastoidea), and asymmetrical thecas with uniserial ambulacra and brachioles (Paracrinoidea) have been removed, and cannot possibly represent a natural (i.e. monophyletic) grouping. Largely because it is difficult to give any satisfactory diagnosis for the Eocrinoidea (because they are not a natural group) there has been a great deal of futile argument about precisely which species should be included in, and which rejected from, the ‘Class’ Eocrinoidea. Simply taking some of the changes that have been proposed since the Treatise (Ubaghs 1 967) will show how much disagreement exists. Paul (1968), for example, removed Macrocystella from the eocrinoids and grouped it with glyptocystitid rhombifera, but Sprinkle (1973) rejected it as a rhombiferan and returned it to the eocrinoids; arguments about this still continue. Springerocystis, Columbocystis, and Foerstecystis were removed from the eocrinoids by Sprinkle (1973) who placed them with paracrinoids. Parsley and Mintz(1975), however, objected to them being paracrinoids and returned them to the eocrinoids. Recently the coronates, which were originally grouped together with blastoids (Regnell 1945) and which were later transferred to inadunate crinoids by Fay (1978), have been added to the eocrinoids by Sprinkle (1979, 1980). Broadhead (1982) has added to the general confusion still further by rejecting all those species without epispires from the eocrinoids without making any positive contribution as to how the rejected taxa ought to be classified. 440 PALAEONTOLOGY, VOLUME 27 Broadhead (1982) Sprinkle (1973) Ubaghs (1967) (1980) w text-fig. 8. A phylogram showing the relationships of a number of cystoid groups taken from the phylogenetic tree given by Paul in Paul and Smith (1984). Three alternative views of what constitutes the ‘Class Eocrinoidea’ are shown. As eocrinoids are a paraphyletic grouping of primitive cystoids their boundaries are inevitably arbitrary. The arbitrary way in which eocrinoids have been grouped becomes obvious when the various alternative schemes are plotted on a phylogram of cystoid groups (text-fig. 8). Clearly there will always be arguments as to where boundaries are to be drawn for such a subjective and paraphyletic group as the ‘Eocrinoidea’. Such arguments about what constitutes a paraphyletic group are not only futile (since unnatural groups will always be arbitrary) but are a positive hindrance to discovering relationship amongst cystoids. An understanding of how the various cystoid groups are related will become much easier if the ‘Class’ Eocrinoidea is abandoned and its members allocated to appropriate monophyletic groups. Although this will necessitate the creation of new taxa or the redefinition of old taxa, it will lead to a much clearer and very much more precise view of cystoid evolution. Here, then, is an example where misdirected taxonomic endeavour has actually hindered growth of knowledge concerning the relationships of cystoid groups. CHARACTER DISTRIBUTION AND THE INFORMATION CONTENT OF CLASSIFICATIONS Before discussing the positive contribution that the fossil record can make to phylogenetic analysis, it is worth while outlining the concept of stem and crown groups which was first developed by Hennig (1966, 1981). Identifying pattern in the distribution of morphological characters is the SMITH: ECHINODERM CLASSIFICATION 441 essence of recognizing relationship. Derived characters shared amongst two or more species are synapomorphies that indicate phylogenetic kinship, those unique to one species or one group of species are referred to as autapomorphies. Obviously a character that is a synapo- morphy uniting a group of species can also be thought of as an autapomorphy of that group as a unit. The presence of derived characters shared amongst two or more species is usually taken as an indication of phylogenetic kinship. This is not to say that convergence does not occur. Convergence can only be recognized on the pattern of character distribution, since to suggest that a derived character found in two species is a convergent feature requires that at least two further derived characters are known that link one of those species to a group that does not include the other species. Evolutionary convergence is invoked where there is incongruence in character distribution and parsimony is used to determine which characters are true synapomorphs and which due to convergence. Any monophyletic group with both living and fossil species can be divided into two parts— a crown group and a stem group. The crown group contains the latest ancestor common to all living members of that group together with all of its descendants. They are recognizable as crown group members because they possess all of the synapomorphies that unite the living members and form a monophyletic group. The stem group contains only fossil species and is a paraphyletic assemblage. They are identified as stem group members since they will have at least one, but not all, of the autapomorphies of the crown group. In phylogenetic terms, the stem group consists of all those species to evolve after the group had separated from its living sister group but prior to the evolution of the latest common ancestor of the crown group. The importance of differentiating between crown and stem groups will become apparent later. A ‘natural’ classification scheme is best considered as a method of conveying information about character distribution. Both character distribution and the Linnaean system of classification have the form of a nested hierarchy. Maximum information about character distribution is conveyed when the hierarchical pattern of the classification exactly matches the pattern of character distribution. Unfortunately, past classification schemes have not been as informative as they might be and the recent predilection for erecting notional class status for small problematic groups of fossil echinoderms has had a most detrimental effect on the information content of classifications by destroying the hierarchical arrangement. In support of small fossil groups of high categorial rank Sprinkle (1975, 1980) and Paul (1979) have argued that it is a true reflection of an early diversity of form in echinoderm evolution. Even if this is so, it is no reason for elevating a large number of groups within one taxon to the same categorial rank since this is uninformative about character distribution within the higher taxon. Their preferred classifications are based not on the distribution of shared characteristics, but on the development of prominent autapomorphies (hence the necessity for a ‘class’ Eocrinoidea for all those cystoids left once other groups have been distinguished on autapomorphies). The presence of autapomorphies provides no information about the relationships with other groups. Unlike Breimer and Ubaghs (1974), it is not the taxonomic rank that 1 primarily object to but the purely subjective way in which a large number of groups are given the same rank within a large taxon. This procedure is not only arbitrary but makes no contribution to the search for pattern in character distribution and hence relationship. The illogicality of this approach can be illustrated by the recent creation of a sixth subphylum of echinoderms, Paracrinozoa, by Parsley and Mintz (1975). There are just eight genera of paracrinoid (seven when Parsley and Mintz erected the subphylum), all of which have a distinctively asymmetrical theca and uniserial free appendages. Prior to this paracrinoids had always been considered cystoids, but Parsley and Mintz thought that the group had characteristics which were in part cystoid (stem and theca) and in part crinoid (subvective system). Given that they are correct in their interpretation, then paracrinoids, crinoids, and cystoids must form a phylogenetically closely related group within the Echinodermata, a fact which Parsley and Mintz acknowledged. Yet by elevating the paracrinoids to subphylum rank they are in effect stating that it is as closely related morphologically to carpoids 442 PALAEONTOLOGY, VOLUME 27 (Homalozoa), sea stars (Asterozoa), and Echinozoa as it is to either cystoids (Blastozoa) or crinoids (text-fig. 9a). In my opinion Parsley and Mintz were mistaken in their identification of the free appendages as crinoid arms and pinnules. There is a great deal of confusion about the homology of pelmatozoan appendages which Paul and Smith (1984) have tried to clear up. In crinoids the entire subvective system is derived from ambulacra as a whole, whereas in cystoids many of the free appendages are brachioles derived from just cover-plate series. Paracrinoids have free or recumbent uniserial ambulacra (‘arms’) which give rise to free uniserial brachioles. Similar structures are known in other B Sprinkle (1973) 'Eocrinoids' Parsley & Mintz (1975) Paracrinoids text-fig. 9. The status of paracrinoids. A, the implied relationship of para- crinoids to other echinoderm groups in the classification proposed by Parsley and Mintz (1975). B, a cladogram for the better-known paracrinoids and some related ‘eocrinoids’ to show how analysis of character distribution leads to a clear statement about the status of paracrinoids within the cystoids. Characters 1-18 are stated in Table 1 . SMITH: ECHINODERM CLASSIFICATION 443 cystoids: uniserial appendages are found in diploporite cystoids while free or recumbent ambulacra with brachioles are found in many cystoids such as the eocrinoid Bockia, glyptocystitid rhombiferans, coronates, and blastoids. Sprinkle (1973) quite correctly pointed out that springero- cystid eocrinoids had an asymmetrical arrangement of ‘arm’ facets and a theca with stem and peristome offset as in paracrinoids. Cryptocrinites, another eocrinoid, has a similar asymmetric theca but has no discernible asymmetry of ‘arm’ facets. Thus, although paracrinoids are unusual in having brachioles arising from just one side of the ambulacrum their relationship as cystoids is to my mind unambiguous. A cladogram of character distribution (text-fig. 9b) can be constructed to suggest how paracrinoids relate to certain other cystoid groups. FOSSIL EVIDENCE IN DETERMINING RELATIONSHIPS AMONGST LIVING GROUPS The idea that relationship of living groups can be determined by looking at the fossil record is, at first glance, very appealing. After all, the fossil record is often thought to provide the only tangible evidence of evolution. And yet, if this is so, why has the advancement in palaeontological knowledge table 1. Character distribution for selected genera of paracrinoid and other cystoids as shown in text-fig. 9b Primitive Derived 1. Polyplated stalk 2. Basals undifferentiated 3. Ambulacra forming an integral part of the thecal wall 4. Periproct in C/D interray 5. Oral area flush with theca 6. Oral area composed of seven plates, six of which surround the peristome 7. Thecal plates numerous, new plates added by intercalation 8. Brachioles arise from both sides of the ambulacra 9. Peristome at apex of theca, opposite the stem 10. Ambulacra and brachioles biserial 11. Ambulacra more or less straight 12. Pentameral symmetry of rays 13. Globular or sac-like theca 14. Plates without internal pits 15. Plates smooth 16. Plates convex 17. Brachioles erect 18. Peristome exposed Holomeric stem composed of thin discoidal columnals Three basals (a) Ambulacra erect, exothecal, attached to facets close to the peristome ( b ) Ambulacra secondarily recumbent, overlying thecal plates Periproct lateral in B/C interray Oral area a spout-like projection (a) Oral area composed of six plates all surrounding the peristome (b) Oral area composed of four plates around the peristome (a) Thecal plates relatively few, not intercalated during growth ( b ) Thecal plates reduced to three cycles Brachioles arise from only one side of each ambulacrum Peristome offset; periproct at apex of theca, opposite the stem Ambulacra and brachioles uniserial Ambulacra curved in a solar direction Two primary rays: (a) unbranched; (b) both branched; (c) one only branched Biconvex theca Internal (respiratory) pits Plates strongly ornamented with radially arranged and internally excavated ribs Plates concave Brachioles recumbent Peristome covered by oral plates 444 PALAEONTOLOGY, VOLUME 27 of echinoderms not been reflected in an increased understanding about the interrelationships amongst living groups? Since 1900 many new fossil echinoderm groups have been described, yet taking just the five extant classes (text-fig. 7) we are no nearer understanding how they are interrelated than Bather (1900) was. Indeed, uncertainty has actually increased; whereas Bather accepted eleutherozoan echinoderms as a natural group, the failure of palaeontology to identify obvious intermediates between asterozoan and echinozoan eleutherozoans has resulted in less certainty about the relationship of these two groups (text-fig. 7). One can only conclude that historically, palaeontology has provided no input to the unravelling of relationship amongst living echinoderm classes. This, to some extent, may be because, until recently, there has not been the methodology to use the fossil record constructively, but it is also because the fossil record cannot by itself resolve problems of relationship. One of the difficulties of working with fossils is that only skeletal morphology is generally preserved. In comparison with the wealth of anatomical, genetic, biochemical, and embryological data available in living echinoderms, fossils can provide only a small part of that information. It is therefore not surprising that there is an increased uncertainty about affiliation amongst fossil groups. For example, in echinoderm classification the position of the radial water vessel, whether external or internal, is a character of some importance. Embryology shows quite unequivocally that the internal position of the radial water vessel is secondary and derived during development from an originally external position (MacBride 1914). Yet, as the radial water vessel is composed entirely of soft tissue, when we look at fossils it is open to argument where the radial water vessel was situated. Bather (1915), Ubaghs (1975), and myself (in Paul and Smith 1984) have all argued that in edrioasterids the radial water vessel lay external to the flooring plates. However, Bell (1975, 1977) has argued that edrioasterids had an internal radial water vessel. Although one or other side may present more convincing arguments, there is no way in which we can be absolutely certain unless a specimen with preserved soft tissue is found. Therefore, at least some characters that are crucial in identifying relationship amongst living echinoderms are absent or unprovable in fossil groups. Fossils preserve only a small proportion of all character attributes available in living groups. A second reason why palaeontology has had little or no impact on resolving relationships stems from the fact that fossils rarely contain a more informative pattern of character distribution than is present in extant groups. The following example will help to explain what is meant. Consider three extant groups A, B, and C each of which is quite distinct in having a number of autapomorphies. In addition, let us assume that only one synapomorphy ‘j’ can be discovered which identifies B and C as sister groups. Can we get more information from looking at the character distribution in fossils? As we go back in time the three groups will appear to become less distinct from one another as autapomorphic characters ‘disappear’. Eventually a point will come when groups B and C no longer exist as distinguishable taxa since their members are plesiomorphic with respect to all characters save for character ‘j’ which distinguishes them from group A members. So, although fossils may show that extant groups were more similar due to plesiomorphy in the past, the only characters which allow us to identify sister groups (synapomorphies) are very often already known from comparative anatomy of the living members. The fossil record simplifies by removing autapomorphies but cannot generally add to the number of synapomorphies. There are, of course, exceptions where the fossil record can show characters to be more general in distribution than might be suspected from living groups or might identify structures as homologous which are highly modified in living groups, and these are discussed below. In general, however, fossils contain a no more informative pattern of character distribution than is present in extant groups. Fossils provide information about their geological age from their stratigraphical occurrence, yet as Nelson and Platnick (1981) have argued this has no value on its own in determining relationship. Ideas on relationship are not based initially on stratigraphical occurrence but on comparative skeletal morphology. Where the stratigraphical sequence agrees with deductions based on comparative morphology then the fossil record is accepted as an adequate guide to relationships. Where comparative morphology and the stratigraphic record conflict then the fossil record is dismissed as incomplete. Clearly then, the fossil record on its own is no guide to relationship, since it SMITH: ECHINODERM CLASSIFICATION 445 is accepted as adequate when in agreement but rejected as inadequate when in conflict with comparative anatomy. All that can be claimed is that if the fossil record agrees with a hypothesis of relationship based on morphology (and one would hope that it might) then yet another piece of evidence has been added in support. If it conflicts then the hypothesis may still be correct, since the fossil record could be incomplete. Turning now to a practical example, clypeasteroids are believed to have evolved in the Tertiary and have an excellent fossil record. Here then is a group where one might reasonably expect the fossil record to provide additional evidence on how clypeasteroids are interrelated and from whence they originated. In order to simplify matters I shall just discuss three extant clypeasteroids, Clypeaster , Echinocyamus , and Echinarachnius , as representatives of the groups Clypeasterina, Fibulariina, and Scutellina respectively. Analysis of character distribution amongst these three clypeasteroids gives the cladogram in text-fig. 10. Outgroup comparison suggests that their closest living relatives are the cassiduloids (holectypoids are rejected since the character used by Durham et al. (1966) to unite holectypoids and clypeasteroids was the presence of a lantern, which is plesiomorphic). Morpho- logically, Echinocyamus is the least specialized of the three (i.e. it has the fewest autapomorphies) and both Clypeaster and Echinarachnius pass through a developmental stage in which they resemble fibu- lariids. It is therefore most parsimonious to assume that at some period in the past cassiduloids and clypeasteroids shared a common ancestor which they did not share with any other living group and that Echinocyamus , with its more generalized body plan, has diverged least from the latest common ancestor of living clypeasteroids. All so far has been deduced without reference to the fossil record. If fossil clypeasteroids are examined then we find species with either clypeasterinid, fibulariinid, or scutelinid autapomorphies, a few with characters common to both fibulariinid and scutelinids but without any autapomorphies of either group, and one genus, Togocyamus , which has a few basic clypeasteroid features but no autapomorphies of any one group or pair of groups. Togocyamus is, as was predicted from character distribution amongst extant groups, rather like Echinocyamus in shape and was originally classified as a fibulariid. However, from the description given by Kier (1982), Togocyamus clearly lacks all the advanced characteristics of perignathic girdle and pore arrangement that distinguish fibulariids from other groups. So far then the fossil record has simply confirmed what was already predicted from the living groups. What about the relationship of clypeasteroids to cassiduloids — can the fossil record provide evidence of transitionary forms linking these two groups? Here, however, we run into the basic problem of how to recognize a fossil as ancestral to the clypeasteroids when clypeasteroids are recognized by the presence of multiple ambulacral pores on adoral plates. All that we can be certain of is that the ancestor will have had the characteristics that are common to both cassiduloids and clypeasteroids, but none of the characteristics unique to clypeasteroids. Identifying Togocyamus as a primitive clypeasteroid has not made the relationship of clypeasteroids and cassiduloids any more obvious. The conclusions that are to be drawn from this example are threefold. First, palaeontology has corroborated the hypothesis of relationship based on living groups. Secondly, it has confirmed the statement on generality of characters since Togocyamus conforms to the concept of a primitive clypeasteroid based on character distribution amongst living groups. Thirdly, the recognition of fossils as primitive members of an extant group does not in this case lead to any clearer understanding about their relationship to other groups. The fossil record has only been able to corroborate what was already known about character distribution and has, as yet, provided no tangible link with cassiduloids. The evidence for clypeasteroid-cassiduloid relationship comes from character analysis of the living groups. So far I have tried to show that fossil echinoderms have done little more than corroborate hypotheses of relationship that can be deduced from the study of living groups. However, the fossil record does contain information on character distribution that is not available to neontologists and has a very positive role to play in the formulation of hypotheses of relationship as has clearly been shown by Patterson (1981 ). It is these positive aspects that are worth stressing since only through them will palaeontology be able to make a substantial contribution to our understanding of relationship. 446 PALAEONTOLOGY, VOLUME 27 OJ 0) Q. O Q. o O < O UU text-fig. 11. Cladogram for four of the five extant classes of echinoderm based on embryology and comparative anatomy. Characters 1 43 are given in Table 3. classes (text-fig. 1 1). It identifies crinoids as a primitive sister group to the other three and asteroids as the primitive sister group to ophiuroids plus echinoids. This is in full agreement with the conclusions of MacBride (1914) and Hyman (1955). The phylogenetic position of holothuroids is less obvious and needs careful analysis. Holothuroids share a number of derived characters with ophiuroids and echinoids and several more with only echinoids. However, there are a number of other derived characters that are common to asteroids, ophiuroids, and echinoids or only to ophiuroids and echinoids which are not found in holothuroids. To try to resolve the phylogenetic position of holothuroids the problem can be reduced to a number of three taxon problems and the alternatives compared. First, let us ignore ophiuroids and consider whether holothuroids or asteroids are phylogenetically the more closely related to echinoids. The alternative cladograms are given in text-fig. 12. It is quite evident that there are many more derived characters that suggest that holothuroids and echinoids are sister groups than suggest that asteroids and echinoids are sister groups. It is therefore worth examining the four characters that suggest asteroids and echinoids to be more closely related and which on the grounds of parsimony alone would be rejected. The presence of a genital rachis and multiple gonads, as found in asteroids and echinoids, is undoubtedly a derived character while the single gonad and gonopore of holothuroids is primitive. However, primitive stem group echinoids have but a single gonopore and, by inference, a single gonad (Smith 1 984); therefore the genital rachis text-fig. 12. A three taxon cladogram to resolve the relationship of holothuroids to asteroids and echinoids. Characters 1-43 are listed in Table 3. 450 PALAEONTOLOGY, VOLUME 27 table 3. Primitive and derived character states in extant echinoderm classes. The classes that possess derived character states are given in the third column Primitive Derived Occurrence 1. Skeleton absent 2. Larval development bilaterally symmetrical 3. Without radial symmetry 4. Larva without processes 5. Hydropore opening simple 6. Definitive anus opens lateroventrally 7. Adult attached 8. Ambulacral plates added at tip of radial water vessel 9. Tube feet arise directly from the radial water vessel 10. No articulating spines 1 1 . Larval vestibule formed 12. Aboral surface greatly enlarged 13. Entoneural nerve plexus present 14. Entoneural nerve plexus as primary motor coordination system 15. No hyponeural sinuses 16. Right hydrocoel present but vestigial in development 17. Tube feet without internal ampulla 18. Larva attaches by pre-oral lobe Calcite skeleton of stereom Development of right-hand side larval coeloms suppressed With radial (pentameral) symmetry Larva with incipient processes (auricularia) Hydropore opening a calcified body (madreporite) (a) No anus in adults ( b ) Definitive anus opens laterodorsally in B/C inter- radius (c) Definitive anus opens dorsally at site of larval anus Adult free-living (a) Radial water vessel tip associated with terminal plate; new ambulacral plates added adorally to terminal plate (b) Ambulacral plates wanting Tube feet arise from lateral branches of the radial water vessel Articulating spines No larval vestibule formed ( a ) Aboral and oral surfaces equally developed ( b ) Aboral surface greatly reduced Entoneural nerve plexus absent Ectoneural nerve plexus as primary motor coordination system Hyponeural sinuses present Right hydrocoel does not form during development Tube feet with internal ampulla Larva unattached Crinoids, Asteroids, Ophiuroids, Echinoids, Holothuroids Crinoids, Asteroids, Ophiuroids, Echinoids, Holothuroids Crinoids, Asteroids, Ophiuroids, Echinoids, Holothuroids Asteroids, Ophiuroids, Echinoids, Holothuroids Asteroids, Ophiuroids, Echinoids, Holothuroids Ophiuroids, some Asteroids Some Asteroids Echinoids, Holothuroids Asteroids, Ophiuroids, Echinoids, Holothuroids Asteroids, Ophiuroids, Echinoids Holothuroids Asteroids, Ophiuroids, Echinoids, Holothuroids Asteroids, Ophiuroids, Echinoids Asteroids Asteroids, Ophiuroids Echinoids, Holothuroids Ophiuroids, Echinoids, Holothuroids Asteroids, Ophiuroids, Echinoids, Holothuroids Asteroids, Ophiuroids, Echinoids, Holothuroids Crinoids, Holothuroids Asteroids, Echinoids, Holothuroids Ophiuroids, Echinoids, Holothuroids SMITH: ECHINODERM CLASSIFICATION 451 Primitive Derived Occurrence 19. Radial water vessel and nerve external Radial water vessel and nerve enclosed by epineural folds Ophiuroids, Echinoids, Holothuroids 20. Single internal gonad arising from genital stolon Multiple internal gonads arising from genital rachis surrounding axial complex Asteroids, Ophiuroids, Echinoids 21. No epineural sinuses Epineural sinuses present Ophiuroids, Echinoids, Holothuroids 22. Entomesoderm forms in gastrula from archenteron Entomesoderm starts to form in blastula from one side of the wall before embolic invagination Ophiuroids, Echinoids, Holothuroids 23. Radial water vessel grows radially Radial water vessel grows meridionally Echinoids, Holothuroids 24. Tube foot wall uncalcified Tube foot wall with spicules Echinoids, Holothuroids 25. Suckered tube feet without skeletal disc plates Suckered tube feet with skeletal disc plates Echinoids, Holothuroids 26. Larva lacks a mouth Larval mouth forms Asteroids, Ophiuroids, Echinoids, Holothuroids 27. Gonads internal Gonads external, on arms Crinoids 28. Ambulacra forming integral part of the theca Ambulacra extending free of theca as arms Crinoids 29. Larva with short processes Larva with elongate processes Ophiuroids, Echinoids 30. Larval processes not supported by calcite rods Larval processes supported by calcite rods Ophiuroids, Echinoids 31. Larval mouth retained as adult mouth but migrates to the left during development Larval mouth lost during development; adult mouth opens to the left of the larval mouth Ophiuroids, Echinoids 32. No peripharyngeal coelom Peripharyngeal coelom Echinoids, Holothuroids 33. No perianal coelom Perianal coelom Echinoids, Holothuroids 34. Haemal system rudimentary, an open lacuna network Haemal system extensive and well developed, with a rete mirabile Echinoids, Holothuroids 35. Axial complex fully developed Axial complex absent or vestigial Holothuroids 36. Adoralmost ambulacral ossicles forming a semi-flexible oral frame Adoralmost ambulacral ossicles modified into a muscular jaw apparatus Ophiuroids, Echinoids 37. Adoralmost ossicles remain an integral part of ambulacral plating Adoralmost ossicles internal and surround oesophagus Echinoids, Holothuroids 38. Radial ambulacral muscles mterossicular and segmented Radial ambulacral muscles internal and unsegmented Echinoids, Holothuroids 39. Ambulacral ossicles present Ambulacral ossicles lost Holothuroids 40. Blastopore remains as larval anus Blastopore closes after formation of archenteron Crinoids 41. No polian vesicles Polian vesicles Asteroids, Ophiuroids, Holothuroids 42. No Tiedemann’s bodies Tiedemann’s bodies Asteroids, Ophiuroids, Echinoids 43. Vestibule sealed off from exterior during development Vestibule remains open Ophiuroids 452 PALAEONTOLOGY, VOLUME 27 text-fig. 13. A three taxon cladogram to resolve the relationship of holothuroids to echinoids and ophiuroids. Characters 1 -43 are listed in Table 3. and multiple gonads must have evolved independently in echinoids and asteroids. This character can therefore be rejected as being a convergence on the basis of fossil evidence. Tiedemann’s bodies are present in ophiuroids as well as in asteroids and echinoids. Their absence in holothuroids may be a secondary loss, since these bodies perform the same function as the axial complex (Bachmann and Goldschmidt 1980) which is vestigial or absent in holothuroids. The two remaining characters common to asteroids and echinoids but not to holothuroids are the presence of terminal plates (oculars), which appear early in development, and the presence of articulating spines. Neither character carries much weight as the skeleton of holothuroids has become highly modified and is usually reduced to rudimentary spicules. The development of the few living holothuroids that retain a skeleton of thin imbricate plates has never been reported and it is therefore impossible to recognize these characters in holothuroids. In view of the outstanding evidence in favour of placing asteroids as the primitive sister group of holothuroids plus echinoids, it seems reasonable to assume that living holothuroids have lost both spines and apical plates as a consequence of the profound simplification of their body wall skeleton. The only question remaining then is, comparing ophiuroids, echinoids, and holothuroids, which pair is the more closely related? Derived characters exist that link holothuroids and echinoids and which link echinoids and ophiuroids but none exist linking ophiuroids and holothuroids. Therefore we need consider only two of the three possible cladograms (text-fig. 13). Both seem to be supported by a number of characters. However, of those identifying ophiuroids and echinoids as a group, four (the presence of a genital rachis and multiple gonads, Tiedemann’s bodies, terminal (ocular) plates, and articulating spines) have already been rejected on the strength of the preceding cladogram, and cannot be used. Of the remaining four characters, one, the presence of homologous ambulacral ossicles modified into a jaw apparatus, is questionable because holothuroids have such a modified and reduced larval skeleton that such a structure might easily have been lost. The internal calcareous ring may be homologous with some plates of the jaw apparatus but there is too little evidence to be certain. This character can be rejected on fossil evidence, however, since the stem group holothuroid Rotasaccas has a fully developed lantern which is in all details, save for tooth structure, identical with that of echinoids (Haude and Langenstrassen 1976). Three characters remain that are incongruent: the absence of a pluteus larva with elongate processes, the absence of skeletal rods supporting the larval processes, and the retention of the larval mouth throughout development. The first two characters are interconnected since the larval skeleton forms to support processes that develop in the pluteus larva to extend the ciliated bands. Neither the processes nor the skeleton are identical in echinoids and ophiuroids. The ophiopluteus has no pre-oral processes and the main locomotory processes that develop early on are the posterolateral ones, whereas in the echinopluteus, elongate pre-oral processes are present, and the main locomotory processes are the post-oral ones. The posterolateral processes either appear much later in development and remain small or are absent SMITH: ECHINODERM CLASSIFICATION 453 altogether. The ophiopluteus has just two centres of calcification from which calcite rods grow, one on either side, whereas the echinopluteus has five, two on the left, two on the right, and an anterior V-shaped rod for the pre-oral processes. There is therefore a distinct possibility that elongation of the small processes common to all eleutherozoan larvae occurred independently in ophiuroids and echinoids. The evidence concerning the phylogenetic position of holothuroids, although ambiguous, definitely tends to favour echinoids and holothuroids as being sister groups. If I have identified the ophiocistioid Rotasaccus correctly as a stem group holothuroid then the presence of a lantern so similar to that of echinoids in Rotasaccus convinces me that echinoids and holothuroids are sister groups and that holothuroids have undergone fairly major change through reduction of the body skeleton since the two groups became separated. However, it must be said that the available biochemical evidence concerning sterols (Bolker 1967; Goad et al. 1972), phosphorus carriers (Florkin 1952), and collagen (Matsumura et al. 1979) do not support this and indicate that echinoids and ophiuroids share a greater similarity. As I lack expertise in this field I cannot assess these data from a cladistic standpoint and therefore cannot tell what sort of similarity it is that echinoids and ophiuroids share. The result of analysing character distribution amongst living groups of echinoderms gives the nested hierarchy shown in text-fig. 14. If, following historical precedence, the five extant groups are given class status then the hierarchical pattern must dictate the higher classification of the Echinodermata. Names are available for all but one group. The Echinodermata can be divided into two subphyla, Pelmatozoa for the crinoids and Eleutherozoa for the asteroids, ophiuroids, echinoids, and holothuroids. (Haugh and Bell (1980) rejected the Eleutherozoa as a monophyletic taxon on the grounds that the ‘absence of stem’ was a non-character, an argument which comparative embryology refutes.) At superclass level we can use the name Asterozoa for the asteroids but no name has ever been proposed for the group comprising the Ophiuroidea, Echinoidea, and Holothuroidea. I therefore propose to name this group Cryptosyringida (derivation — Greek Kryptos, hidden; r- Pelmatozoa — , Eleutherozoa SUBPHYLUM ,-Asterozoa — , Cryptosyringida- SUPERCLASS Echinozoa , SUB-SUPERCLASS text-fig. 14. The most parsimonious cladogram for the five extant classes of echinoderm with a suggested hierarchical classification. Characters 1 43 are given in Table 3. 454 PALAEONTOLOGY, VOLUME 27 Syringos, a pipe or fistula, in allusion to the fact that the radial water vessel and radial nerve becomes covered during development). Finally, the echinoids and holothuroids are grouped together in the Echinozoa at sub-superclass level. Incorporating fossil groups into the primary classification The concept of crown and stem groups becomes indispensable when dealing with the classification of fossil groups. Monophyletic groups of living echinoderms belong to a whole series of increasingly more generalized crown groups of which the most narrowly defined contains only members of that group and no others. Fossil echinoderms also belong to a whole series of increasingly more generalized crown groups but, with one exception, all fossils also belong to a single stem group. In systematics the pattern of character distribution is used to determine at what level a species or group of species belong. Neontologists search for the most narrowly defined crown group whereas palaeontologists attempt to discover the unique stem group that each fossil belongs to. The level of generality for stem groups is as variable as it is for crown groups. For example, the Cretaceous cidarid Stereocidaris sceptifera belongs to the stem group of the genus Stereocidaris whereas the lower Cambrian helicoplacoid Helicop/acus gi/herti is so generalized that it is a member of the stem group of the Echinodermata. Each stem group may contain one or many members. In some cases the stem group might be quite small, as in echinoids where there are approximately 125 known stem group species but almost 7,000 crown group species. In other cases — for example, Pelmatozoa— the stem group is enormous compared with the crown group and includes all cystoids and all crinoids except the Articulata. The members of each stem group possess at least one but not all of the autapomorphies that define the crown group. It is therefore possible to arrange fossils in the stem group according to the distribution of crown group autapomorphies (see Patterson and Rosen 1977; Wiley 1979). The most primitive will have just one autapomorphy, the most advanced will have all but one. However, a few fossils will belong not to the stem group but to the crown group. These will have all the autapomorphies that define the crown group but none of the autapomorphies of any subdivision of the crown group and will include the first member of the crown group. Because groups can be distinguished as discrete entities only when they have evolved a new character, the maximum resolution that we can hope for is to distinguish one or a group of fossils at the appearance of each new crown group autapomorphy. Those fossils which all have the same crown group autapomorphies represent a monophyletic side branch from the stem line. The number of autapomorphies that can be identified limits the number of stem groups that can be identified, yet although this is presumably finite, there is no way of predicting how many can be recognized. Each side branch of the stem group (zwischenkategorien of Hennig 1969; plesion of Patterson and Rosen 1977) may contain only a single species or may contain a large number of species, in which case character distribution can be analysed to discover pattern and phylogenetic relationship within the side branch. Each side branch, being a monophyletic group, can be named and classified from the species level up. Their nominal categorial rank is unimportant and is best based on diversity or historical precedence. The groups which make up the stem group can then be listed in an order corresponding to the acquisition of crown group autapomorphies and incorporated into the primary classification as recommended by Wiley (1979). As palaeontologists are concerned with pattern recognition in stem groups, it is possible that having a name for each stem group might be quite useful for communicating precisely about which group of fossils are under investigation. To avoid further proliferation of names, it is probably best if they were referred to as stem group cidaroids, stem group isocrinids, etc., but if a widely used name is available I can see no objection to its being used. For example, when I analysed the stem group echinoids (Smith 1984) the traditional group Perischoechinoidea seemed to correspond more or less to the stem group and I suggested retaining Perischoechinoidea for the paraphyletic stem group of the Echinoidea. Some traditional fossil groups are truly monophyletic and can be incorporated into the stem group in their correct position. Others, however, turn out to be paraphyletic since they were defined on the SMITH: ECHINODERM CLASSIFICATION 455 absence of one or more crown group autapomorphies. These will eventually have to be abandoned in favour of groups which are more informative about character distribution. The status within this classification of the principal fossil groups of echinoderm recognized in the Treatise (Moore and Teichert 1978) will now be outlined: (i) Carpoids (‘Classes’ Ctenocystoidea and Stylophora and ‘Orders’ Soluta and Cincta). Carpoids are all basically asymmetric, without a trace of radial symmetry and either lack ambulacra or have a single exothecal appendage. They all have a single feature, their calcite endoskeleton, which they share with crown group echinoderms. However, Jefferies (1981) believes that Stylophora show evidence of gill slits and a post-anal tail and should therefore be classified as stem chordates. If this proves to be correct then the other carpoids may be stem chordates, stem echinoderms, or stem (chordates plus echinoderms). Further work is required to resolve the phylogenetic position of these groups and I shall not consider them further. (ii) Helicoplacoids (‘Class’ Helicoplacoidea). There are only two or possibly three genera of helicoplacoids and a handful of species. Their morphology and phylogenetic position has been discussed by Paul and Smith (1984). Helicoplacoids have a laterally positioned mouth, no oral/aboral differentiation of the skeleton and triradial ambulacra. They are stem group echinoderms and have been incorporated into the classification as a plesion with the nominal rank of family (Table 4). (iii) Camptostroma (‘Class’ Camptostromatoidea). The phylogenetic status of Camptostroma has also been discussed by Paul and Smith (1984). Camptostroma , represented by a single known species holds a rather important position in the cladogram (text-fig. 15) since it possesses all of the autapomorphies of crown group echinoderms but none of the autapomorphies of either Pelmatozoa or Eleutherozoa. It therefore belongs to the group in which the latest ancestor of crown group echinoderms would be placed. (iv) Cystoids (‘Superclass’ Cystoidea = ‘Subphylum’ Blastozoa). In recent years the cystoids have been split up into a number of high categorial taxa. Some of these are undoubtedly monophyletic (blastoids, paracrinoids) whereas others are apparently paraphyletic (rhombiferans, eocrinoids — see Paul and Smith 1984) and need to be reclassified in a more informative way. Previously the presence or absence of a single character (usually a respiratory structure) has been used to identify groups. Cystoids sensu lato are clearly a monophyletic group and their subvective system includes brachioles which are homologues of cover-plate series. The only crown pelmatozoan autapomorphies that they share with extant crinoids are the presence of an elongate dorsal stalk and, in some, the extension of ambulacra free from the thecal wall. They are the most primitive stem group pelmatozoans known. A phylogenetic classification of cystoids should be relatively straightforward and will require a careful analysis of character distribution. Cladistic analysis of this group has never been attempted and holds considerable promise for future research. Cystoids have been incorporated into the classification as a plesion with a nominal rank of Superclass. (v) Echmatocrinus (‘Subclass’ Echmatocrinea). The single species E. brachiatus , represented by some six specimens, is generally taken to be the most primitive member of the Class Crinoidea. It is more advanced than cystoids in that some at least of its ambulacra branch to produce multiple free arms, but it is primitive in comparison with other crinoids in lacking organized thecal plating or stem plating. The Class Crinoidea is monophyletic and corresponds to the crown group plus part of the stem group of the Pelmatozoa. Echmatocrinus is the most primitive-known crinoid and is incorporated into the classification as a plesion with generic rank. (vi) Palaeozoic crinoids (‘Subclasses’ Inadunata, Camerata, and Flexibilia). The structure of this part of the stem group is the least satisfactory. This is because, although the Camerata and Flexibilia are probably monophyletic groups, the Inadunata is unquestionably a paraphyletic grouping of ‘primitive’ crinoids that contains the ancestors of camerates, flexibles and articulates (crown group Pelmatozoa). A paraphyletic group such as the Inadunata can only be arbitrarily defined and is undesirable since paraphyletic groupings simply mask the pattern of character acquisition within the stem group. At present only the relative positions of the Camerata and Flexibilia can be shown in a cladogram (text-fig. 15). The inadunates include stem (Camerata + Flexibilia + Articulata), stem 456 PALAEONTOLOGY, VOLUME 27 (Flexibilia H- Articulata), and stem (Articulata). They represent one of the outstanding areas of ignorance in echinoderm phylogeny and future palaeontological research should be directed towards discovering the pattern of character distribution within inadunates and partitioning this grouping into monophyletic groups. With increasing understanding of the inadunates more plesion categories will be added to the classification in Table 4 between Echmatocrinus and the Articulata. Echinodermata Crown group echinoderms — Pelmatozoa , Crinoidea Cystoidea 4 „ — a — , Eleutherozoa . Cr y pt osy r ingi da text-fig. 15. Cladogram incorporating some of the more important fossil groups to show how they fit into the classification scheme. All fossil groups can be assigned to a stem group of one of the crown groups identified in text-fig. 14. Characters 1 25 as follows: 1, calcite endoskeleton of stereom; 2, biserial ambulacra forming integral part of body wall; 3, ambulacra arranged radially, around the mouth (triradial); 4, skeleton differentiated into dorsal and ventral surfaces; 5, pentaradial symmetry; 6, free appendages developed carrying extensions of the radial water vessels; 7, dorsal surface modified to form a stalk; 8, brachioles arise from ambulacra; 9, ambu- lacra extend free of the theca and carry extensions of major body coeloms; 10, ambulacra uniserial and branched; 11, cup plating clearly differentiated from stem plating; 12, arm plates incorporated into tegmen; stout, rigid tegmen; pinnate arms; 13, tegmen flexible with differentiated ambulacral and interambulacral zones; 14, some arm articulations muscular; arms pinnate; 15, mouth opens through tegmen; 16, anal plates lost from cup; 17, dorsal surface generally flat; adults primitively free-living; 18, epispires lost from ventral surface; 19, mouth frame flexible, composed of ambulacral ossicles only; 20, calcified madreporite; 21, cover-plates modified to adambulacral/lateral arm ossicles; 22, adoralmost ambulacral ossicles modified to form jaw apparatus; 23, radial water vessel enclosed; 24, meridional growth pattern; 25, wheel spicules in body wall. SMITH: ECHINODERM CLASSIFICATION 457 table 4. A phylogenetic classification of the Echinodermata (conventions as in Wiley 1979) Phylum Echinodermata plesion (Family) Helicoplacidae Subphylum Pelmatozoa plesion (Superclass) Cystoidea Class Crinoidea* plesion (Genus) Echmatocrinus plesion (Subclass) Camerata plesion (Subclass) Flexibilia Subclass Articulata Subphylum Eleutherozoa plesion (Genus) Stromatocystites plesion (Class) Edrioasteroidea Superclass Asterozoa Class Asteroidea Superclass Cryptosyringida Subsuperclass Ophiuroidea Subsuperclass Echinozoa Class Echinoidea Class Holothuroidea * Phylogenetic analysis of the Inadunata will add a number of plesions between Echmatocrinus and Articulata in the future. (vii) Edrioasteroids (‘Class’ Edrioasteroidea). Edrioasteroids are best considered as stem group Eleutherozoa. The most primitive members were unattached (e.g. Stromatocystites) and probably common ancestors to all Eleutherozoa. Most edrioasteroids, however, have a number of autapomorphies and represent a monophyletic side branch of the stem group. Most returned to a fixed mode of life attached via their aboral surface. Unlike pelmatozoans, those that elevated themselves above the sea floor did not develop an aboral stem but expanded their oral surface to become pedunculate. Edrioasteroids have been added to the classification as a plesion with nominal class status. (viii) Cyclocystoids (‘Class’ Cyclocystoidea). These form a small but diverse group characterized by a number of well-defined autapomorphies. They also share a number of autapomorphies with isorophid edrioasteroids, notably uniserial ambulacral flooring plates and a marginal ring with a single layer of peripheral platelets. I therefore now prefer to place them within the edrioasteroids as the sister group of the Isorophida and with a nominal rank of Order. (ix) Ophiocistioids (‘Class’ Ophiocistioidea). Ophiocistioids share a number of synapomorphies with the Echinozoa. The discovery of the Devonian ophiocistioid Rotasaccus by Haude and Langenstrassen (1976) was a most important find, since Rotasaccus has the body wall skeleton of a holothuroid but possesses an echinoid-type lantern. This provides evidence that stem group holothuroids possessed a lantern even though it has been lost in all living holothuroids. Ophiocistioids are undoubtedly a paraphyletic group and an analysis of character distribution within this group will lead to a better understanding of the early history of the Echinozoa. A simplified cladogram that incorporates the more important fossil groups is shown in text-fig. 1 5 and a scheme of classification derived from this cladogram is given in Table 4. I have followed the recommendation of Patterson and Rosen (1977) in giving plesion categories only nominal rank and the order in which plesions are listed is dictated by the pattern in the cladogram, as formally recommended by Wiley (1979). 458 PALAEONTOLOGY, VOLUME 27 Finally, I should like to point out some of the major gaps in our knowledge about the phylogeny of echinoderms. There remains a basic ignorance about the phylogenetic relationships of cystoid groups which a cladistic approach could help to dispel. More seriously, the classification of Palaeozoic crinoids is most unsatisfactory and a careful and searching look at the Inadunata is needed so that this grouping can be abandoned in favour of monophyletic (and therefore more informative) groups. Thirdly, the phylogeny of primitive ‘starfish’ has yet to be unravelled satisfactorily and promises to be a most rewarding task. The development of cladistic methodology, which is now such a powerful tool in determining relationships, has opened up new and exciting possibilities for making a real advance in our understanding of echinoderm phylogeny. Acknowledgements. I wish to thank Dr C. R. C. Paul, Liverpool University, and Dr R. P. S. Jefferies, British Museum (Natural History), for much helpful discussion and constructive criticism of this paper. REFERENCES bather, f. a. 1900. The Echinodermata. In lankester, e. r. (ed.). A treatise on zoology , part iii. A. and C. Black, London, 344 pp. 1915. Studies in Edrioasteroidea I-IX. Published by the author at ‘Fabo’, Marryat Road, Wimbledon, London, England. bell, B. M. 1976. A study of North American Edrioasteroidea. New York State Museum and Science Service, Memoir, 21,1 -447. 1977. Respiratory schemes in the class Edrioasteroidea. J. Paleont. 51, 619-632. bolker, H. I. 1967. Phylogenetic relationships of echinoderms: biochemical evidence. Nature, Lond. 213, 904-905. breimer, a. and ubaghs, G. 1974. A critical comment on the classification of the pelmatozoan echinoderms. Proc. Koninkl. Nederl. Akad. Wettenschap. Amsterdam (B), 78, 398-417. broadhead, t. w. 1982. Reappraisal of class Eocrinoidea (Echinodermata). In Lawrence, j. m. (ed.). Echinoderms: proceedings of the international conference, Tampa Bay , 125-131 . Durham, J. w. 1966. Classification. In moore, r. c. [ed.]. Treatise on invertebrate paleontology: part U, Echinodermata 3. The Geological Society of America and the University of Kansas Press, Lawrence, Kansas, U270-296. eldridge, n. and cracraft, j. 1980. Phylogenetic patterns and the evolutionary process. Columbia University Press, New York, 349 pp. fay, R. o. 1978. Order Coronata Jaekel, 1918. In moore, r. c. and teichert, c. (eds.). Treatise on invertebrate paleontology: part T, Echinodermata 2. The Geological Society of America and the University of Kansas Press, Lawrence, Kansas, T574-577. fell, h. b. 1948. Echinoderm embryology and the origin of chordates. Biol. Reviews, 23, 81-107. 1962. A classification of echinoderms. Tuatara , 10, 138-140. — 1963. Phylogeny of sea-stars. Phil. Trans. Roy. Soc. Lond. (B), 246, 381 -435. — 1965. The early evolution of the Echinozoa. Brevoria, 219, I 17. — 1967. Echinoderm ontogeny. In moore, r. c. (ed.). Treatise on invertebrate paleontology: part S, Echinodermata 1. The Geological Society of America and the University of Kansas Press, Lawrence, Kansas, S60-85. florkin, m. 1952. Caracteres biochimiques des categories supraspecifiquesde la systematique animale. Ann. Soc. Roy. Zool. Belgium, 83, 111-130. forbes, E. 1841. A history of British starfishes and other animals of the Class Echinodermata. London, 267 pp. goad, L. J., rubenstein, I. and smith, A. G. 1972. The sterols of echinoderms. Proc. R. Soc. Lond. (B), 180, 223-246. haude, R. and langenstrassen, F. 1976. Rotasaccus dentifer n.g., n. sp., ein devonischer Ophiocistioide (Echinodermata) mit ‘holothuroiden’ Wandskleriten und ‘echinoidem' Kauapparat. Palaont. Z. 50, 130-150. haugh, b. n. and bell, b. m. 1980. Classification schemes. In broadhead, t. w. and waters, j. a. (eds.). Echinoderms: notes for a short course. Studies in Geology, 3, 94-105. University of Tennessee, Department of Geological Sciences. hendler, g. 1978. Development of Amphioplus abditus (Verrill) (Echinodermata: Ophiuroidea). II, description and discussion of ophiuroid skeletal ontogeny and homologies. Biol. Bull. 154, 79-95. hennig, w. 1966. Phylogenetic systematics. University of Illinois Press, Urbana, 263 pp. SMITH: ECHINODERM CLASSIFICATION 459 hennig, w. 1981. Insect phylogeny. J. Wiley, New York, 528 pp. hyman, l. 1955. The invertebrates: Echinodermata. McGraw Hill Book Co., New York, 763 pp. jaekel, o. 1918. Phylogenie und System der Pelmatozoen. Palaont. Z. 3, 1-128. Jefferies, r. p. s. 1981. Fossil evidence on the origin of the chordates and echinoderms. Atti dei Convegni Lincei , 49, 487-561. kier, p. M. 1982. Rapid evolution in echinoids. Palaeontology , 25, 1-10. kitts, D. B. 1974. Palaeontology and evolutionary theory. Evolution , 28, 458-472. L0VTRUP, s. 1977. The phylogeny of Vertebrata. J. Wiley, London, 330 pp. macbride, e. w. 1906. Echinodermata. In harmer, s. f. and shipley, a. e. (eds.). The Cambridge natural history. MacMillan, London. — 1914. Text-book of embryology , volume 1, Invertebrata. MacMillan, London, 692 pp. matsumura, T., hasegawa, m. and shigei, m. 1979. Collagen biochemistry and phylogeny of echinoderms. Comp. Biochem. Physiol. 62B, 101 105. miller, J. s. 1821 .A natural history of the Crinoidea or lily-shaped animals , with observations on the genera Asteria , Euryale, Comatula and Marsupites. Bryan and Co., Bristol, 150 pp. moore, R. c. and teichert, c. (eds.). 1978. Treatise on invertebrate paleontology: part T. Echinodermata 2. The Geological Society of America and the University of Kansas Press, Lawrence, Kansas, 1027 pp. nelson, g. J. 1972. Comments on Hennig’s ‘Phylogenetic Systematics’ and its influence on ichthyology. Syst. Zool. 21, 364-374. — 1974. Classification as an expression of phylogenetic relationship. Ibid. 22, 344-359. — and platnick, n. i. 1981. Systematics and biogeography: cladistics and vicariance. Columbia University Press, New York, 556 pp. nichols, D. 1968. Echinoderms. Hutchinson University Library, London, 200 pp. parsley, R. l. and mintz, l. w. 1975. North American Paracrinoidea (Ordovician: Paracrinozoa, new: Echinodermata). Bull. Amer. Paleont. 68, 1-112. Patterson, c. 1981. Significance of fossils in determining evolutionary relationships. Ann. Rev. Ecol. Syst. 12, 195-223. — and rosen, D. e. 1977. Review of Ichthyodectiform and other Mesozoic teleost fishes and the theory and practice of classifying fossils. Bull. Amer. Mus. Nat. Hist. 158 (2), 85-172. Paul, c. r. c. 1968. Macrocystella Callaway, the earliest glyptocystitid cystoid. Palaeontology , 11, 580-600. — 1979. Early echinoderm radiation. In house, m. r. (ed.). Origin of major invertebrate groups. Academic Press, London, 415-434. — and smith, a. b. 1984. The early radiation and phylogeny of echinoderms. Biol. Reviews. 59 (4) (in press). regnell, g. 1945. Non-crinoid Pelmatozoa from the Palaeozoic of Sweden: a taxonomic study. Medd. Lunds Geol. Mineral. Inst. 108, 1-255. smith, a. b. 1984. Echinoid palaeobiology . George Allen and Unwin, London, 190 pp. sprinkle, j. 1973. Morphology and evolution of blastozoan echinoderms. Spec. Publ. Mus. Comp. Zool., Harvard Univ. 1 -248. — 1976. Classification and phylogeny of ‘pelmatozoan’ echinoderms. Syst. Zooi 25, 83-91. — 1 979. Convergence of Paleozoic stemmed echinoderms with crested calyces. Geol. Soc. Amer. Abstr. Prog. 1 (7), 522. — 1980. Origin of blastoids: new look at an old problem. Ibid. 12 (7), 528. ubaghs, G. 1967. Eocrinoidea. In moore, r. c. (ed.). Treatise on invertebrate Palaeontology: part S , Echinodermata I. The Geological Society of America and the University of Kansas Press, Lawrence, Kansas, S455-495. 1975. Early Paleozoic Echinoderms. Ann. Review Earth Plan. Sci. 3, 79-98. wiley, e. o. 1979. An annotated Lmnaean hierarchy, with comments on natural taxa and competing systems. Syst. Zool. 28, 308-337. — 1981. Phylogenetics: the theory and practice of phylogenetic systematics. John Wiley and Sons, New York, 439 pp. ANDREW B. SMITH Department of Palaeontology British Museum (Natural History) Manuscript received 5 September 1983 Cromwell Road Revised Manuscript received 18 January 1984 London SW7 5BD I 1 A MUSCLE ATTACHMENT PROPOSAL FOR SEPTAL FUNCTION IN MESOZOIC AMMONITES by R. A. HENDERSON Abstract. The unusual septal surface typical of lytoceratid ammonites is described from unique Western Australian specimens of the late Cretaceous Indopacific species Pseudophyllites indra (Forbes). Median dorsal septal recesses and inner septa combine to form a septate tunnel lying within the phragmocone; their free margins are complexly fluted like that of the septal periphery in contact with the outer shell wall. Functional analysis of the fluted septal recesses and inner septa suggest that they were not related to phragmocone strength but facilitated the attachment of adductor muscles. By analogy, a muscle attachment function is argued for the fluted septal periphery of P. indra and for the septal periphery of Mesozoic ammonites in general. The role of septal design in contributing necessary strength to phragmocone construction is re-evaluated and considered subordinate. Muscle attachment translocation during growth, a special problem for ectocochleate cephalopods, was accomplished by the release of muscle attachment and rapid forward movement of the ammonite animal in its shell. Muscles were re-attached along a narrow zone at the free margin of a newly formed septum, convolution of which enlarged the attachment surface. It is argued that connecting rings of the siphuncle were preformed in the body chamber prior to movement of the animal; location of the siphuncle, details of its construction, and the nature of associated structures are consistent with this proposal. The muscle-attachment hypothesis is further supported by shell microfabrics known for Mesozoic ammonites, including new data for Sciponoceras. Gross differences in shell form and ornamentation which separate Mesozoic ammonites and nautiloids are thought to be due to differences in growth style, necessitated by the manner in which muscle attachments were translocated in members of the two groups. Ammonites were one of the most common shelled-invertebrate groups in Mesozoic seas and their shells, characterized by completely fluted septa and a ventral siphuncle, have attracted functional comment throughout the history of invertebrate palaeontology. Raup (1966, 1967) has demonstrated that the outer shell form of planispiral ammonites can be generalized to a mere three parameters. No such simplicity is apparent for septa which show a bewildering array of morphologies among Mesozoic members of the subclass yet are taxonomically specific, showing no more variation at the species level than any other test attribute. Clearly their morphology was under close genetic control, seemingly at considerable expense relative to the genetic investment in other elements of shell morphology. By implication the septa, or the body surface they replicated, must have fulfilled an important functional role for the living animal. Further, the clear phyletic changes of septal design shown by discrete lineages of Mesozoic ammonites (Wiedmann and Kullman 1981) suggest the constant operation of selection pressures tuning septal morphology to its functional role. As reviewed by Westermann (1971) and Kennedy and Cobban (1976) a number of functions have been attributed to the fluting of ammonite septa since Owen ( 1 843) suggested that they were designed in such a way as to impart necessary additional strength to the shell. Owen's view has been frequently endorsed in the literature (Zittel 1884; Pfaff 1911; Ruzhentsev 1946; Arkell 1949; Kennedy and Cobban 1976) and according to Westermann (1975) is the consensus of present-day opinion. Maastrichtian specimens of P. indra (Forbes) from the Miria Marl, Carnarvon Basin, Western Australia show with exceptional clarity the extraordinary septal surface that typifies the Jurassic and Cretaceous suborder Lytoceratina. Its septal morphology is incompatible with the strength paradigm of function which is thereby brought into question. It is here re-evaluated for Mesozoic ammonites generally. An alternative proposal, that the complex morphology of the septal periphery was required for muscle attachment, is argued for P. indra and its general applicability to Mesozoic ammonites is examined. IPalaeontology, Vol. 27, Pari 3, 1984, pp. 461-486, pis. 48-49.] 462 PALAEONTOLOGY, VOLUME 27 Location of specimens. The following abbreviations are used to denote the repositories of figured specimens: WAM Western Australian Museum, Perth; NMV— National Museum of Victoria, Melbourne; BM(NH)— British Museum (Natural History), London; OUM— Oxford University Museum, Oxford. SEPTAL SURFACE OF PSEU DO P HYLLITES INDRA Description The periphery of the septal surface in contact with the outer shell is complexly fluted to the fourth order. Convex flutes, corresponding to saddles of the suture, widen towards the periphery and arch forward in an adapertural sense. Concave flutes, corresponding to lobes of the suture, arch backwards towards the shell apex and narrow towards the periphery so that they form conical vaults below the outer shell. Minor flutes die out rapidly away from the periphery and at approximately the mid-line between the periphery and centre of the septum, first-order flutes disappear. The septal surface is then essentially planar across a narrow zone which is horseshoe-shaped in plan view. Inside this zone the septal surface is curved uniformly backwards to form a pronounced depression, here termed the septal recess, occupying the mid-dorsal septal field (PI. 48, figs. 2, 5; text-fig. 1). The connection zone septal recess inner septum text-fig. 1 . Septal architecture of Pseudophy Hites indra (Forbes), a , schematic representation of two successive septal surfaces (fluting omitted). Note the zone of connection, provided by the neck of the inner septum, between each septal recess and the main chamber which succeeds it. b , phragmocone cut in the median plane; note the two chamber systems and the interconnection between them provided by the necks of the inner septa which do not close on to the preceding septal recesses. explanation of plate 48 Figs. 1 -5. Pseudophyllites indra (Forbes) showing details of the septal surface. 1, 2, lateral and apertural view of phragmocone, WAM 60.44, xl. 3, latex cast from internal mould of septal recess, WAM 81.2500, xl. 4, latex cast from internal mould showing the fluted free margin of a septal recess and fluted inner septum. Note that the line of suture is continuous from the septal recess to the inner septum, both being part of a single septal surface, WAM 81.2433, x4. 5, internal mould of septal surface showing septal recess and inner septum, WAM 60.130, x 1. PLATE 48 HENDERSON, Septal function in ammonites 464 PALAEONTOLOGY, VOLUME 27 septal recess is in effect a short tube, terminating at the mouth of its equivalent on the preceding septum. Thus successive septal recesses link one septum to the next, forming a tunnel which extends throughout the phragmocone (PI. 49, fig. 1). The tunnel is, however, itself septate. A large limb subtended from the median ventral wall of each septal recess, here termed the inner septum, plugs the mouth of the septal recess formed by the previous septum (PI. 48, figs. 4, 5). In concert the septal recesses and inner septa combine to form an inner phragmocone analogous to the phragmocone proper which surrounds it. Chambers of the two phragmocone systems are linked, a zone of interconnection being provided in the necks of the inner septa which do not close against the mid-ventral sector of the subjacent septal recesses. Text-figure 1 provides a diagrammatic summary of the septal architecture. The free margin of the septal recess is complexly fluted (PI. 48, figs. 3, 4). Its third-order subdivisions splay out to form a fringe which reaches beyond the mouth of the previous septal recess, extending across the subjacent planar septal surface to the base of the first-order flutes of the outer perimeter (PI. 48, fig. 4). Fluting of the inner septum is also complex and represents a smaller scale equivalent to the structure of the outer septal periphery. Margins of the septal recess and inner septum are confluent and comprise the septal lobe of sutural nomenclature (Arkell 1957; Kullmann and Wiedmann 1970). The suture of the inner septum thus comprises a very large median saddle within the septal lobe proper which is formed by the free margin of the septal recess. The septal lobe is confluent with, and may be considered as an extension of, the internal lobe. A septal lobe is already present at a shell diameter of 8 mm, the smallest growth stage in the collection displaying a septal surface. Subsequent growth trends are isometric across the size range of septal surfaces available for measurement (text-fig. 2) The pattern of fluting associated with the septal recess and inner septum, like that of the septal periphery, was stabilized early in ontogeny. A detailed comparison of fluting represented at shell diameters of 44 mm and 1 62 mm shows it to be identical in all respects other than scale. Flutes associated with minor lobes within the septal lobe are more intricately divided and more elongate than those associated with the minor saddles so that the suture shows a marked polarity (PI. 48, fig. 4). Interpretation Structures associated with the septal lobe are striking features of Pseudophyllites phragmocones and represent a considerable genetic investment in shell architecture and in organization of the posterior EXPLANATION OF PLATE 49 Fig. 1 . Internal mould of Pseudophyllites indra (Forbes) showing two linked septal recesses, one showing the sutural trace of an internal septum. WAM 80.976, x 1 -2. Figs. 2, 3. Two halves of a juvenile shell of Nautilus pompilius (Linneus) in which the last septum is incomplete and represented by a rim of shell only. 2, uncoated and showing the mural zones of the two last septa. 3, coated with ammonium chloride and showing that the last septum is widely separated from the mural ridge which bounds the mural zone of the previous septum. OUM 14475, x 2. Figs. 4-10. Latex peels of internal moulds of ammonites showing structures of the inner shell surface. 4, Kitchinites sp. nov. showing midventral ridge on inner shell, NMV P31013, x 3. 5, Lytoceras cornucopia (Young and Bird) showing the posterior of the body chamber on the midventral line. Note the ridges extending a short distance into the body chamber and showing an indistinct lobate termination, BMNH 43902, x 3. 6, Phylloceras heterophyllum (J. Sowerby), showing midventral ridge with minor ridges converging on to it, OUM J20360, x 3. 7, Hamites maximus J. Sowerby showing faint ridges of shell forming a pair of rings straddling the mid-dorsal line of the body chamber immediately anterior to the last septum. Figured by Crick (1898, pi. 17, figs. 6, 7) who regarded them as muscle scars, BMNH C6802, x 3. 8, Gunnarites kalika (Stoliczka), showing midventral ridge, WAM 80.845, x 3. 9, Maorites densicostatus (Kilian and Reboul), showing midventral ridge, NMV P3 1023, x 3. 10, Gaudryceras kayei (Forbes), showing midventral ridge and subordinate ridges parallel to it, WAM 80.839, x 3. PLATE 49 % rf Sr- -~> jt/R*- HENDERSON, Septal function in ammonites 466 PALAEONTOLOGY, VOLUME 27 60 - <0 CO 50 - 0) o JD • • 4 20 40 60 whorl breadth • text-fig. 2. Plot of breadth of septal recesses against whorl breadth. Dots represent maximum values measured from the ex- tremities of the free margin of the septal recess where it is sutured to the preceding septal surface. Triangles represent minimum values measured where the septal recess ± is most constricted; they correspond to breadths of inner septa. body of the Pseudophyllites animal. Their retention as enduring features in the Tetragonites - Pseudophyllites stock, and in the Lytoceratina in general, indicates that they served some specific functional role. Given the comparability of fluting associated with the septal lobe with that of the external septal periphery, an integrated function or functions may be adduced for the septal surface in toto. Strength. Since phragmocone strength is the generally accepted functional role attributed to septal design, it is instructive to examine the septal lobe and its associated structures in this context. Fluting related to the septal lobe is entirely internal and therefore cannot have contributed to support of the outer shell. It might be argued that fluting associated with the inner septum contributed strength, and consequent economy in shell thickness, to the median dorsal part of the septal surface in toto. As discussed more fully below, each septal surface must, at some stage of phragmocone growth, have carried the hydrostatic load and it has been argued that septal architecture in ammonites reflects a response to this requirement. The internal septum might thus merely represent a somewhat bizarre elaboration of this theme. However, design of the flared and fluted base of the septal recess intersecting the preceding septal surface at a low angle along a deeply embayed line of suture which displays marked polarity, cannot be readily reconciled with shell strength as its functional role. Phyletic context. The septal lobe is a structure of considerable phyletic longevity. It is characteristic of the morphologically conservative suborder Lytoceratina whose range extends from early Jurassic to late Cretaceous (Kullmann and Wiedmann 1970). The earliest Pseudophyllites are of Santonian age and the genus is best known from the Campanian and Maastrichtian. Origins of the genus are clearly to be found in Tetragonites (see Kennedy and Klinger 1977), which also shows a well-developed septal lobe and which first appears in late Aptian times. The septal lobe was therefore a stable morphological feature of the Tetragonites Pseudophyllites stock for some 40 m.y. and was likely to have been stable over a much longer period, probably having appeared in an ancestral lytoceratid stock during the early Jurassic. The peculiarities of lytoceratid septal architecture cannot be considered as unique among ammonites but, as has long been recognized, merely represent an extreme development of the dorsal flute of the septum corresponding to the internal lobe of the suture. The septal lobe of sutural terminology represents an extension of the internal lobe onto the preceding septum (Wiedmann and Kullmann 1981). Thus the septal recess is homologous with the flute corresponding to the internal lobe which is ubiquitous among ammonites except for a few Devonian forms. The inner septum is HENDERSON: SEPTAL FUNCTION IN AMMONITES 467 homologous with a minor flute corresponding to a small median saddle contained within the internal lobe of most Mesozoic ammonites. Organization of associated soft-tissue. Disposition of soft-tissue associated with the formation of each septal recess and associated internal septum is replicated by the shell itself and is complex. The suture line at the free margin of the septal recess marks the intricate termination of a fringe of tissue which followed the shape shown by the distal portion of a septal recess, curving away from the main body of tissue (PI. 48, fig. 2; text-fig. 3). A space, presumably fluid-filled in life, must have lain between this fringe and the main body of the animal. The inner septum replicates the posterior termination of the animal’s soft parts which plugged the narrowest part of the preceding septal recess. Its complex suture represents branching arms of tissue that extended along the walls of the preceding septal recess (PI. 49, fig. 1). In moving forward during growth to the site of a new septum, all tissues were withdrawn from the septal recess. To accomplish this, considerable constriction of the fringe would have been necessary as the smallest cross-section of the septal recess is as little as 60 % of the area displayed by the associated fringe (PI. 48, fig. 3; text-figs. 2, 3). The space between the fringe and the main body of tissue would have assisted in accommodating the constriction. The fringe itself must have consisted of tissue with exceptional elastic properties for, having been compressed and distorted, the fringe recovered its original shape, fitting exactly to the mouth of the previously formed septal recess with the fine divisions of its delicately serrated margin reaching across the previously formed septal surface towards its periphery. Collagenous connective tissue organized as a pliant composite, or tissue capable of hydrostatic self-support, would possess the appropriate mechanical attributes (Wain- wright et al. 1975). Soft tissue lying adjacent to the septal periphery adopted a complex shape of identical type to that of the fringe and was laterally continuous with it. The same type of tissue must therefore have been located along the frilled periphery of the Pseudophy Hites septum. The episodic, growth-required stress imposed on the fringe of soft tissue lining the septal recess and producing marked changes in its shape suggests that forward movement of the animal within the shell itself was episodic. It seems likely that short periods of rapid forward movement and distortion of the fringe were followed by long resting periods when the fringe adopted its unstressed shape. Certain authors (Seilacher 1975; Bayer \911a) have contended that the posterior of the ammonite animal’s soft tissue functioned as a membrane surface prior to septal fabrication. In this model the membranal precursor of the septum is thought to have approximated initially to a planar surface inferred zone of muscle peripheral zone of last septum septal recess fringe inner septum text-fig. 3. Transverse longitudinal section drawn in the plane of maximum whorl breadth for the last two camerae and posterior body chamber of Pseudophy Hites indra (Forbes) showing the relationship of tissue to shell. 468 PALAEONTOLOGY, VOLUME 27 occupying the whorl cross-section and attached at a number of point locations on its periphery. Hydrostatic or muscular stress acting perpendicular to the initial planar membrane resulted in its deformation and the production of the complex, fluted surface later replicated in shell by secretion of the septum. Deformation was maximized at the periphery of the subsequent septum, with its free segments stretched between the point attachments. Orientation of the major folds in the suture line is thereby controlled by the direction of stress imposed on the membrane. Fluting of the fringe, of which the free margin of the septal recess is a replica, has a different orientation and cannot have resulted from the same mechanism. Here the sutural folds are orientated perpendicular to the inferred stress direction and are unrelated to it (see PI. 48, figs. 3, 4). Fluting shown by the septal recess is identical to, and laterally continuous with, that shown by the outer septal periphery and the inner septum. Furthermore, the entire septal surface was almost certainly the result of a single secretional episode. Clearly the type of tissue, its episodic movement and the secretory processes responsible for formation of the septal recess were general to the entire septal surface. STRENGTH PARADIGM OF AMMONITE SEPTAL FUNCTION RE-EVALUATED Accept ing that ammonite phragmocones functioned like that of living Nautilus and were filled with gas of about one atmospheric pressure which imparted bouyancy to the shell, they may be regarded as sealed vessels required to withstand hydrostatic pressure. Three aspects of phragmocone strength need to be considered: the strength of the outer shell which is supported by the septa, the strength of the septa themselves and the strength of the junctions between the outer shell and the septa on the line of suture. An initial concern, however, is the palaeobathymetry of ammonites and thereby the hydrostatic pressures to which their phragmocones were likely to have been subjected. Palaeobathymetry The depth habitat of ammonites is not easily evaluated because shells which are buoyant in life may be easily transported in death to a completely different environment (Reyment 1958). Post-mortem cessation of siphuncular osmotic pumping may have caused flooding of the phragmocone and sinking of the shell to a site of fossilization at much greater depth than that inhabited by the living animal. Alternatively post-mortem separation of body from shell may have caused the latter to float to the surface and drift inshore to a shallow water, even littoral, environment as is common for Nautilus (House 1973; Reyment 1973). The broad palaeoecological context of ammonites is best assessed from the types of sediment in which their remains occur and the nature of fossil assemblages of which they are part. More recent data support the conclusion of Miller and Furnish (1957) that ammonites are predominantly associated with sedimentary rocks of shallow-water origin. McKerrow (1978), in a comprehensive review of British fossil assemblages, many of which are also widely represented in Europe and elsewhere, recorded ammonites as essentially of neritic palaeoecology throughout their history. Casey and Rawson (1973) drew similar environmental conclusions for ammonite-bearing strata of the Jurassic and Cretaceous boreal realm, whilst Kauffman (1977) concluded that the vast epeiric sea which invaded the Western Interior of North America and sponsored an abundance of ammonites had a maximum depth of perhaps 300 m. Cretaceous sediments containing spectacularly diverse ammonite assemblages accumulated in passive continental margin, neritic settings on the border- lands of the Indian Ocean following the fragmentation of eastern Gondwana in Madagascar and southeastern Africa (Blant 1973; Kennedy and Klinger 1975; Basairie and Collignon 1956), Western Australia (Veevers and Johnstone 1974; Henderson and McNamara unpublished) and southern India (Kossmat 1898; Sastri et al. 1973). In comparison, records of ammonites from deep-water sedimentary environments are few. In addition to neritic assemblages, Scott (1940), Ziegler (1967), and Tanabe et al. (1978) have recorded distinctive assemblages typified by Phylloceratina and Lytoceratina from deeper water environments HENDERSON: SEPTAL FUNCTION IN AMMONITES 469 where they were considered to have had benthic or bathypelagic life habits. Phylloceratids and lytoceratids are however best known from shelf sediments and may have been shallow-water nectopelagic organisms whose range extended beyond the continental shelves (Birkelund 1965) and their deep-water associations could well be a post-mortem artefact (Kennedy and Cobban 1976; Tanabe 1979). Ammonites, like all macrofossils, are very rare in Mesozoic eugeosynclinal flysch (continental rise and trench) sediments such as those of the Franciscan assemblage in California (Irwin 1957) and the Torlesse Supergroup in New Zealand (Stevens and Speden 1978) whereas they are much more common in subjacent broadly coeval sedimentary rocks of the northern part of the Great Valley sequence (see also Ojakangas 1968; Matsumoto 1960) and Mirihiku Supergroups respectively where shallower water depositional environments prevailed. The record of ammonites in deep-sea, pelagic sediments is sparse but its interpretation is rendered equivocal by the dissolution of aragonite shells at depth as shown so clearly by pteropods in the deep- sea Cenozoic record (Kennett 1982). As reviewed by Bernoulli and Jenkyns (1974), Mesozoic ammonite and aptychus-bearing pelagic lithofacies are well known from onshore exposures in the Alpine-Mediterranean region where they are considered to be continental margin deposits which accumulated partly under neritic conditions within the photic zone and partially in deep-water environments. Many of the records of ammonite phragmocones and lamellaptychi from deep-sea drilling and dredging represent neritic deposits of Mesozoic continental margins (Renz et al. 1975; Wiedmann and Neugebauer 1978) or deep-water pelagic deposits which are not greatly distant from Mesozoic continental margins (Renz 1972, 1978, 1979a, 1979c, Wiedmann 1979). However, reports of lamellaptychi from the central Pacific (Renz 1973) and Atlantic sites distant from any continental mass (Renz 1977, 19796) show that some ammonites were fully oceanic. Lamellaptychi are characteristic of the Oppeliidae (Arkell et al. 1957), a cosmopolitan family which is well known from shallow-water sediments. All ammonite genera known to occur in deep-sea, pelagic sediments as phragmocones are also known from shallow water occurrences. Overall the lithofacies and associational data for ammonites support the view of Lehmann (1975) that they were predominantly of neritic palaeoecology and lived in water depths ranging to perhaps 300 m corresponding to a hydrostatic pressure of 30 bars. The contention of Mutvei (1975), based largely on considerations of functional morphology, that ammonites were predominantly denizens of the open oceans where they ranged into deep-water environments, finds no support in the fossil record. Strength of outer shell Ammonite phragmocone strength in this context can be assessed in an approximate way by reference to the engineering of thin-walled pressure vessels for which there is a sound theoretical design basis to contend with failure by buckling. As shown by Brownell and Young (1959) the strength of cylindrical vessels subjected to external pressure is governed by the diameter of the cylinder, the thickness of its wall, the spacing of internal strengthening rings or septa, and the elastic properties of the wall material as determined by Young’s Modulus. Segments of ammonite phragmocone may be considered as approximations to such vessels. Wall thickness and septal spacing can be measured directly. Young’s Modulus has been measured for Nautilus nacre (Wainwrighl et al. 1975) and the nacre which comprises the principal shell layer of ammonite phragmocones (Birkelund 1981) may be expected to have shown comparable elasticity. Very few ammonite phragmocones conform to a circular cross-section; a diameter which matches the widest arc of cross-sectional curvature and represents minimum strength seems an appropriate approximation to adopt in the analysis. Text-figure 4, which is adapted from Brownell and Young (1959, text-fig. 8.4), charts the relationship between cylindrical pressure vessel specifications and strain at failure. By employing Young’s Modulus for Nautilus nacre, the abscissa may be rescaled for stress. Also shown is the field to which ammonite phragmocones approximate and the stress imposed by a hydrostatic pressure of 50 bars. The strength indicated for ammonite phragmocones assuming planar septa departs from the stress field to which they were probably subjected (up to 30 bars), plus an appropriate safety margin, by an order of magnitude. Whilst it is admitted that the analogy between ammonite phragmocones 470 PALAEONTOLOGY, VOLUME 27 pressure in bars text-fig. 4. General chart showing relationship between diameter (d), thickness (t), distance between internal septa ( 1 ), and strain (e) at the point of failure by buckling for cylindrical thin-walled vessels subjected to external hydrostatic pressure (from Brownell and Young 1959). The field to which Mesozoic ammonites approximate, based on shell measurements for seventy-five taxa, is shaded. A value of Young’s Modulus (E), adopted as 44 G Nm 2 determined for Nautilus nacre by Wainwright et al. (1976), allows calculation of a scale in terms of pressure. Even with simple septa, Mesozoic ammonite phragmocones appear to have been immune to failure by buckling at the maximum pressure (50 bars) to which they were probably subjected in life. and cylindrical pressure vessels is a gross approximation, the disparity between indicated strength and expected stress is so marked that septal fluting, as a design to enhance strength by reducing the effective spacing between septa, seems to be completely unwarranted. Further, the plan of septal fluting shown by ordinary planispiral phragmocones seems inappropriate if support for the outer shell was the only design consideration. The presence of a convolute internal suture is commonplace in Jurassic and Cretaceous ammonites (see Wiedmann and Kullman 1981) and represents septal fluting which buttresses directly to the venter of the previous whorl, abutting a shell surface that is already fully supported. The internal lobe straddling the dorsal mid-line is especially noteworthy; it represents a major septal flute in almost all ammonites yet in the vast majority of phragmocones the dorsum is not part of the outer shell and can have carried no hydrostatic load. There are many ammonites for which fine divisions of the suture, reflecting fine fluting of the septal periphery, cannot have added any appreciable strength to the outer wall. This is true for most ceratitic and pseudoceratitic ammonites in which second-order sutural divisions at the base of the lobes are very fine. The same general argument applies to many Jurassic and Cretaceous ammonites with highly frilled sutures. Often in such forms the finest divisions, or more properly the flutes they represent, contribute little or nothing to minimizing the size of unsupported spans of outer shell. In HENDERSON: SEPTAL FUNCTION IN AMMONITES 471 addition, the distinctive shape of certain sutural elements is difficult to reconcile with the strength paradigm. A convenient example is provided by the distinctive, phyletically enduring phylloid terminations to saddles of the Phylloceratina. As shown by Bayer ( 1 977/?), the fluctuation in septal distance during growth displayed by several ammonites and the generally poor relationship between the septal spacing and shell form are anomalous in the context of strength. Many authors, most recently Wiedmann and Kuhnian ( 1981 ), have emphasized the conservatism of suture lines, and by implication, septal fluting. Patterns displayed by suture lines often allow stocks of common ancestry to be recognized whereas other aspects of shell morphology, including whorl profile, show wide variation within stocks and repeated convergence between stocks. Given that whorl profile is an important mechanical factor in strength of the outer shell, a close relationship between suture lines and whorl profiles would be predicted by the strength paradigm. Such is patently not the case. Strength of septum itself As noted by Raup and Stanley ( 1971 , p. 1 79) hydrostatic pressure would have affected both the outer shell and the last septum to form part of an evacuated chamber. Body fluids of the living animal would have been at the hydrostatic pressure and the living tissue would have had little strength or at least much greater elasticity than the enclosing shell. At some stage of ontogeny, therefore, each septum would have been obliged to accept the full hydrostatic load. Several authors, for example Westermann ( 1975) and Wainwright et al. (1976), have noted that the convex form of ammonite septa is generally favourable to withstanding hydrostatic stress. Septa are too complex to permit any rigorous strength analysis but some general observations are pertinent. The strengths of curved shells is inversely proportional to their radii of curvature and proportional to their thickness. Tensional and compressional strengths of the constructional material are additional factors; for Nautilus nacre compressional strength is approximately twice that of the tensional strength (Wainwright et al. 1976) and a similar contrast may be expected for the nacre of which ammonite septa are constructed. To conform to the strength paradigm, each subzone of a septum should be of equal strength. Since shell thickness does not vary between convex flutes and concave flutes, though these would have experienced compressional stress and tensional stress respectively, the two types of flute should show marked differences in radii of curvature. However, suture lines show that this is not the case; in almost all ammonites lobes and saddles are invariably of comparable widths. Proliferation of fluting towards the septal periphery resulted in shell economy, reduction in radii of curvature of the flutes being reciprocated by a reduction in septal thickness. Enhancement of the weight/strength relationship, however, was not required to maintain buoyancy. Calculations by Mutvei and Reyment (1973) and Heptonstall (1970) have shown that ammonite phragmocones possessed a buoyancy potential beyond the requirements of the living animal, so much so that several chambers must have remained flooded. Shell economy would have been desirable from a metabolic standpoint. If proliferation of fluting represented such a strategy, then it was not fully exploited. As well known to ammonite specialists, even slight abrasion of phragmocone moulds or steinkerns results in a considerable reduction in sutural complexity. In other words, fine-scale fluting is restricted to a very narrow zone at the periphery of septa. Radial lengthening of fine-scale flutes into a broader zone would seem desirable if shell economy was a matter of vital concern. Strength of suture Pfaff (1911) considered the suture between the septum and the outer shell to have been a zone of structural weakness and that linear elongation of the suture imparted necessary strength to the junction. He cited as evidence the allometric relationship between septal diameter and length of suture which he regarded as typical of ammonites. Since septal area, and therefore total hydrostatic load, increases in proportion to the square of its diameter whereas the circumference increases in 472 PALAEONTOLOGY, VOLUME 27 linear proportion to the diameter, such a relationship may be interpreted as necessary to maintain sutural strength. However, shear strength of the interface between septal nacre and the interior surface of the outer shell at the line of suture has never been mechanically evaluated and there is no a priori reason for assuming it to be a zone of special weakness. Further, as noted by Westermann (1975) if the junction was a zone of mechanical weakness the septum could be wedged out against the outer shell over a broad zone of contact. This would seem a more expedient solution to the problem rather than maximizing the length of contact by sutural frilling. Conclusion It would be in error wholly to deny the strength as a consideration in the design of ammonite septa; clearly there are several aspects of septal morphology that are consistent with such a function. Equally, it is evident that strength is not the only, nor probably the major, consideration. Septal elaboration in ammonites generally, as in Pseudophyllites and lytoceratids of similar shell architecture, must have served some other function. MUSCLE ATTACHMENT PARADIGM OF SEPTAL FUNCTION The idea that fluted margins of septa may have assisted in attaching the body of an ammonite to its shell was first noted last century and reintroduced by Spath (1919). It has subsequently been listed among alternatives for septal function in a number of reviews (for example, Raup and Stanley 1971; Westermann 1971; Kennedy and Cobban 1976) but the case has never been argued in detail. Accepting that the attachment of body to shell in the Mollusca is intimately related to muscle attachment, it is instructive to consider the interface between muscle and shell. The general mechanism which applies in molluscs has been described by Hubendick (1957) with important emendations from Nakahara and Bevelander (1970) and Tompa and Watabe (1976). The muscle base is attached to a layer of specialized mantle epithelium, the tendon layer, from which extracellular fibres extend into the adjacent shell surface. Organic extensions from the living tissue are thereby physically embedded in the shell. During growth-required translocation of muscle attachment it is thought that mantle cells adjacent to the leading edge of the muscle base become transformed into tendon cells and extend the tendon layer. New muscle tissue is added above the tendon layer addition and concomitantly part of the tendon layer and associated muscle is atrophied at the trailing edge of the muscle base. In this way muscle bases track across pre-existing mantle in steady, progressive movement as documented for the oyster Crassostrea by Galtsoff (1964). However, this mechanism cannot apply to ectocochleates in which the shell is much larger than the body. Here the entire mantle must move forward in the shell during growth because the shell is growing much faster than the body. The style of growth is more like an ecdysis with older portions of the shell progressively evacuated by the body rather than the steady, matched growth of body, mantle, and enclosing shell seen in most molluscs (text-fig. 5). It follows that muscle attachment cannot have been permanent with extracellular extensions embedded in the adjacent shell. If this were the case the mantle would have been rendered immobile, always pinned to the shell at the zone of muscle attachment. Rather, a temporary means of attachment must have applied, allowing the tendon layer to detach from the shell, move forward to a new site, and reattach. This type of muscle translocation is unique to ectocochleate cephalopods and is unstudied. Perhaps a cement, similar to that demonstrated by Bonar (1978) for muscle attachment of the nudibranch Phestilla to its larval shell and destroyed during metamophosis to a shell-less adult, provided the attachment mechanism. The presence of a thin ‘chitinous’ layer covering the shell on areas of muscle insertion (Griffin 1900) supports this view. However, the ease with which dead Nautilus are slipped from their shells suggests an organic means of attachment, maintained by living tissue, was also involved. HENDERSON: SEPTAL FUNCTION IN AMMONITES 473 text-fig. 5. Diagram contrasting mechanisms of muscle translocation shown by ectocochleate cephalopods, typified by Nautilus , and most other molluscs, typified by a bivalve. In Nautilus the shell is much larger than the body and enveloping mantle which must be moved forward in the shell, complementary with muscle translocation during growth. In the bivalve, mantle underlying the muscle bases is permanently bonded to the shell by extracellular organic fibres. Muscle translocation is achieved by growing new muscle tissue over pre-existing mantle at the advancing edge of the muscle base and complimentary wasting of muscle tissue at the trailing edge. Proposal for muscle attachment in Pseudophyllites indra The suture of Pseudophyllites is complex to the degree where its convolutions occupy all the available surface area of the outer shell save for a narrow zone adjacent to the mid-ventral line where the sides of the vental lobe do not quite meet (text-fig. 6a). Fine-scale fluting of the septa had reached its ultimate expression and was so from the early stages of shell growth as sutures are fully differentiated at shell diameters as small as 2 cm. Development of the septal recess and inner septum might be regarded as morphological strategies designed to extend the free septal margin and fine-scale fluting associated with it. 474 PALAEONTOLOGY, VOLUME 27 a b c text-fig. 6. Latex casts taken from natural internal moulds of ammonite phragmocones to show details of the inner shell surface, a , Pseudophyllites indra (Forbes) showing fine-scale ridges and grooves along the ventral mid- line and the intricate fluting of the septal periphery; WAM 60.130, x2. b, Pachydiscus sp. showing a broad median ventral ridge; WAM 81.2693, x 3. c, Phylloceras (Neophylloceras) meridianum Spath showing narrow median ventral ridge with indistinct ridges and grooves lateral to it; holotype BM(NH) C41320, x 2. As already discussed, inferences concerning the fringe of soft tissue associated with the septal recess in life suggest it was resilient in nature, capable of accepting stress by distortion followed by elastic recovery to its original shape. If the tissue possessed any appreciable strength, as might be expected from its elastic properties, it must have acted to hold the animal in its shell. A similar case can be argued for the tissue once associated with the fluted vaults lying beneath the outer shell at the septal periphery and at the margin of the inner septum. Fluted terminations of the vaults are widely splayed so that soft tissue contained in them must have been deformed when drawn forth during growth (PI. 48, fig. 3; text-fig. 3). Accepting that attachment of body to shell is typically associated with the musculature in molluscs, it may be that the fine-scale flutes represent sites of muscle attach- ment. Soft-part organization in ammonites is very poorly known and the nature of musculation can only be surmised from analogy with extant Nautilus. Mutvei (1957) has reviewed Nautilus musculature which is inserted on the annular elevation, a broad zone on the inner wall of the shell immediately in front of the last septum. Three distinct zones of muscle insertion are represented; the subepithelial muscle which originates from a narrow band at the posterior edge of the annular elevation and abutting the last septum; retractor muscles which originate from the lateral sectors of the annular HENDERSON: SEPTAL FUNCTION IN AMMONITES 475 elevation and contribute the bulk of the muscle attachment area; and the mantle musculature which originates from a narrow band at the anterior margin of the annular elevation. Although insertion of the retractor muscles is lateral, the muscles themselves lie in the dorsal part of the viscera each having a reniform cross section and terminating in the cephalic cartilage (see Mutvei 19646, text-fig. 2; Kennedy and Cobban 1976, text-fig. 5). If the free septal margins represented the location of muscle attachment in Pseudophyllites, two subzones are apparent. The free margin of the septal recess and periphery of the inner septum would have contributed a pair of muscles located dorsally in the visceral mass and each with a reniform cross-section. The septal periphery would have contributed a sheet of muscle to the outer body wall. Such a system of musculature is strikingly similar to that displayed by Nautilus , the paired muscles representing retractors and the peripheral muscular sheet representing the mantle and subepithelial muscles. The process by which the muscle attachment sites are translocated in Nautilus prior to the accretion of a new septum is poorly understood. Fine growth-lines on the retractor muscle scars show that these muscles were moved forward along the wall of the body chamber in very small increments, each one marked at the leading edge by a minute rim of prismatic myostracum. X-ray radiography by Ward et al. (1981) has shown that movement of the animal forward in its shell is episodic and rapid, taking place in less than 6 days in a cycle of chamber formation of over 70 days. In contrast, shell growth at the aperture proceeds at a slow and constant rate. Growth-required movement of the Pseudophyllites animal is also thought to have been rapid because it necessitated distortion of the soft-tissue at the posterior margin of the visceral mass. Disposition of tissue responsible for secreting the septal recess (text-fig. 3) may be cited in evidence; with initiation of forward movement of the animal in its shell the fringe must have suffered distortion in shape, increasingly so as it was drawn past the central constriction of the septal recess. It would not have recovered its original shape until the forward movement had stopped. Tissue filling the fluted vaults at the septal periphery would have experienced a similar pattern of distortion and recovery. It is suggested that muscle translocation was accomplished in a more dramatic way in Pseudophyllites where the muscles were inserted on the septal periphery, compared to Nautilus where the muscles are inserted on the wall of the body chamber. At the onset of movement, the muscles detached completely and were refastened to a newly accreted septum at the cessation of movement. This scenario suggests very rapid movement, perhaps only of a few hours, when body functions dependent on longitudinal musculature would have been suspended. Septal surface periphery as a template for muscle organization Accepting a need for temporary muscle attachment on the septum of a tubular, subepithelial muscle sheet as general for ammonites, the complex fluting at the septal periphery may be seen as a strategy for increasing the surface area of muscle attachment. This may have been required simply for bonding muscle to shell or, alternatively, as facilitating rapid secretion of the septal periphery prior to muscle insertion. The pattern of suture can therefore be regarded as expressing organization of the musculature; its complexity and diversity among ammonites reflects evolution of the longitudinal muscle system. Fluting associated with major sutural lobes in Mesozoic ammonites is invariably dendritic in form with larger flutes subdividing to smaller ones; in taxa with complex sutures, four orders of fluting are represented on the septa. Muscle fibres within the fine-scale flutes, and inserted near their junction with the outer shell, would thereby have been aggregated in dendritic fashion forming thick bands of muscle tissue extending forward from the major lobes with thinner bands of muscle tissue extending from the major saddles where the dendritic pattern is less well developed (text-fig. 7). Seen in this way, unusual shapes of saddle termination, characteristic of Phylloceratina and various stocks of pseudoceratitic ammonites find a ready explanation. They are merely by-products of the particular dendritic form adopted by the lobes. The progressive, overall elaboration of sutures, from goniatitic to ammonitic grade, may be considered as complementing a general advance in organization of the 476 PALAEONTOLOGY, VOLUME 27 text-fig. 7. a-e, typical Mesozoic ammonite suture lines illustrating their dendritic nature, all x 1 . Ceratitina: a, Hedenstroemia hedenstroemi (Keyserling) (redrawn from Spath 1934); b, Psilosturia mongolica (Diener) (redrawn from Diener 1895). Phylloceratina: c, Phylloceras ( Hypophylloceras ) onoense (Stanton) (redrawn from Matsumoto 1959). Ammonitina: d, Hypengonoceras wart/fi (Kossmat) (redrawn from Kossmat 1895); e, Puzosia planulata (Sowerby) (redrawn from Kossmat 1898)./. hypothetical reconstruction of body wall in relation to the last septum. Longitudinal muscle fibres originate at the margin of the septum resulting in a continuous muscular sheath which thickens adjacent to lobes of the suture where investment in muscle attachment is greatest. musculature. The appearance of fine subdivisions of the primary lobes in sutures of the intermediate ceratitic grade is readily explained as part of this trend. If the septal recess and inner septum anchored the retractor muscle in the Lytoceratina then its morphological equivalent in other suborders, the internal lobe, would also have served this function. The investment in retractor musculature in ammonites other than the Lytoceratina would seemingly have been very limited and the musculature would have comprised little more than a subepithelial sheet enveloping the body. Body-chamber shape suggests that many ammonites had elongate, worm- like bodies (Mutvei and Reyment 1973; Kennedy and Cobban 1976) curved through more than one volution of the shell in some forms. In the heteromorphs the body was often elongate and was required to adopt different shapes as determined by changes in shell coiling that occurred during ontogeny. Body shape in all such forms is compatible with a tubular muscle sheath, less so with a pair of powerful retractors. For retractors to be efficient, their curvature needs to be limited, as in Nautilus HENDERSON: SEPTAL FUNCTION IN AMMONITES All (see Mutvei 19646; Kennedy and Cobban 1976). It might therefore be expected that the Lytoceratina should display short bodies accommodated in inflated, rapidly expanding shells of evolute form, as indeed is typical of the group (see Arkell el al. 1957). It is of interest to note that shell fluting in response to organization of the musculature was independently developed, in an incipient way, by members of the ‘nautiloid’ orders Oncoceratoida and Discoceratoida during Ordovician time. In these groups the annular zone of muscle insertion at the base of the body chamber is ornamented with short ridges and intervening hollows. A larger paired muscle platform, thought to have been the site of retractor insertion, lies on the midline and is generally regarded as ventral (Sweet 1959; Teichert 1964). On some shells the ridges buttress the septum (text-fig. 8) and although they originate from the shell wall, they may be likened to the small septal flutes which buttress the shell wall in ammonites of the Triassic family Sageceratidae or the Cretaceous family Sphenodiscidae. text-fig. 8. Internal mould of the Upper Ordovician Oncoceratid nautiloid Diestoceras sp. drawn from Sweet 1959, p. 42, figs. 6, 7. a, lateral view, b, ventral view; x 1. The muscle attachment zone (annulus) lies on the posterior surface of the body chamber immediately adapertural of the last septum and carries nodes which increased the surface area for muscle attachment. Shell structure in Sciponoceras If the septal periphery represents a zone of muscle insertion, it should carry prismatic shell which is the characteristic myostracum of molluscs. The septal periphery, because of its convoluted nature, is often difficult to examine for micro-fabrics. Juvenile S. glaessneri Wright from Bathurst Island, northern Australia are ideal for this purpose, displaying both simple sutures and empty phragmo- cones in which the original shell fabrics are perfectly preserved. As shown in text-figure 9 the junction between septum and shell wall is complex. A band of prismatic shell, here termed the preseptal prismatic zone, precedes the septum and is clearly distinct from the inner prismatic layer of the shell wall. It is narrow and sharply defined in the saddles, becoming broader and more diffuse in the lobes (text-fig. 9a). A second layer, of nacre, comprises the septum proper and wedges out against the inner prismatic layer of the shell wall (text-fig. 9c). A third band of shell with prismatic microstructure and 478 PALAEONTOLOGY, VOLUME 27 here termed the post-septal prismatic zone lies on the apron of the septal nacreous layer lapping on to the adjacent inner prismatic layer of the shell wall (text-fig 9 6, c). In the saddles, where the angle between septum and outer shell is obtuse, its development is unconstrained and it forms a ridge. In the lobes, however, it lies in an acute angle between septum and outer wall and no such ridge is developed. The postseptal prismatic zone of S. glaessneri is regarded as the location of subepithelial muscle insertion. Related structures have been identified in a number of studies of ammonite shell microstructure. Howarth (1975) observed that septal nacre in the Dactylioceratidae is replaced by prismatic shell, thickened to form a ridge, at the junction of septum and outer shell. According to Kuliki (1979, pi. 44, fig. 2), a similar transition marks the margin of septa in Quenstedtoceras. Birkelund and Hansen (1974, 1975) described a prismatic layer on the adapertural face of septa in Hypophylloceras. It appears to thicken at the septal periphery (Birkelund and Hansen 1974, pi. 9, fig. 2) gradually replacing nacre at the confluence of septum with shell wall (Birkelund and Hansen 1975, pi. 1). Well-preserved internal moulds of Cretaceous ammonites often show a faint gutter immediately adapertural of each septum in the crests of saddles (text-fig. 9 d\ see also Kennedy and Cobban 1976, pi. 2, fig. 3 a). The gutter reflects a ridge of shell similar to that of S. glaessneri and probably represents the postseptal prismatic zone. The preseptal prismatic zone is represented in Nautilus as a ridge of shell lying in the angle each septum makes with the shell wall. Mutvei (1964«) recorded it as the supraseptal ridge but, as his interpretation of ectocochleate orientation has not been widely accepted, this term is not used here. In addition to Nautilus , Blind (1975, 1976, 1980) recognized it in some Palaeozoic nautiloids and in a few taxa of ammonites. According to Blind, the preseptal prismatic layer corresponds to the mural ridge, an annular band of prismatic shell deposited as myostracum to a band of subepithelial muscle attachment which abuts the adapertural edge of the septum where it finally wedges out against the shell wall. Blind supposed that the mural ridge served a double function: first, as a zone of muscle attachment and secondly as a precursor elevation of shell to which the posterior of the animal’s body is fitted prior to the secretion of a new septum. This suggestion was also adopted by Ward et al. (1981) who documented chamber formation in living Nautilus by X-ray radiography. Contrary to these views, however, the preseptal prismatic zone and the mural ridge of Nautilus are quite separate, unrelated structures (PI. 49, figs. 2, 3). The preseptal prismatic zone is here considered to represent an immediate precursor to secretion of the septum proper. It is thought that when the animal had moved forward to the position of a new septum, the band of myoadhesive epithelium at the posterior margin of its body established adhesion by secreting the preseptal prismatic zone. With the body thus stabilized in position, a new septum was secreted beginning at the periphery as reported by Willey (1902, p. 749). The preseptal prismatic zone of Sciponoceras is thought to have served a similar role. A portion of the myoadhesive epithelium at the perimeter of the posterior, septa-forming part of the mantle adhered to the shell wall and secreted prismatic shell. The nacreous layer of the septum proper was then rapidly added, followed by a second zone of prismatic shell, the preseptal zone, which marked the stable location of muscle insertion until such time as a new septum needed to be formed. Shell wall markings interpreted as muscle scars If the septal periphery is to be regarded as the principal site of muscle attachment, consideration must be given to various markings on internal moulds of ammonite body chambers that have been interpreted as muscle scar impressions by a number of workers (see Kennedy and Cobban 1976, p. 10). The best-known structures consist of two narrow gutters enclosing a pair of matched shapes on the mould, symmetrically disposed on its dorsal surface. In planispiral ammonites each gutter encloses a semi-elliptical area abutting the last septum whereas in the heteromorphs each gutter typically forms a ring, completely enclosing an oval or bean-shaped area which lies some distance in front of the last septum (PI. 49, fig. 7). The areas so defined have generally been interpreted as the sites of retractor muscle insertion, comparable to those of Nautilus (Kennedy and Cobban 1976). In Nautilus the retractor scars are bounded adaperturely by an annular ridge of shell to which subepithelial muscles HENDERSON: SEPTAL FUNCTION IN AMMONITES 479 text-fig. 9. a-c, microstructure at the septal periphery of juvenile Sciponoceras glaessneri Wright, a , plan view of septal junction with outer shell adjacent to a saddle of the suture; x 30. b, detail of (a) as indicated showing the preseptal prismatic zone (A), the nacreous septal periphery (B) and the post-septal prismatic zone (C); x 180. c, broken shell surface in the saddle of a suture showing the shell wall (thin inner prismatic layer and thick nacreous layer), the preseptal prismatic layer (A), the nacreous septal periphery (B) and the postseptal prismatic zone (C); note the superpositional relationships of the preseptal zone, septum proper and postseptal zone; x 225. t/, dorsal lobe of Baculites ovatus Say on an internal mould showing a well-developed gutter immediately adapertural of the suture proper, interpreted as a mould of the postseptal prismatic zone. 480 PALAEONTOLOGY, VOLUME 27 were attached (Mutvei 1957). The gutter on ammonite body chamber moulds, or rather the ridge of shell it replicated, is not homologous with the annular ridge of Nautilus. As clearly shown by the heteromorphs, in which it forms a complete ring, it cannot represent the site of subepithelial muscle insertion. The retractor scars in Nautilus carry a layer of prismatic myostracum and are clearly marked by growth lines, neither of which have been identified on their supposed counterparts in ammonites. Perhaps the markings represent the imprint of internal organs which pressed against the mantle and had their outlines recorded on the shell. A variety of other structures have been described from internal moulds as representing zones of muscle attachment on the outer shell, most recently by Jordan (1968). They include lobate markings lying immediately adapertural to the ventral saddle of the last suture (sometimes showing as a continuous track along the siphuncular line), large tongue-shaped markings on the flanks of body chambers, and annular markings encircling the posterior of the body chamber. They are much less common than the paired dorsal structures and are commonly defined by staining or by slight differences in texture of mould surfaces. Their interpretation is considered here to be conjectural. Some of the ventral structures may record growth of the siphuncle according to the proposals outlined below. It should be noted that acceptance of such structures as muscle scars does not wholly invalidate the model of septal muscle attachment argued here. They could be viewed as additional sites of muscle attachment, separate from those of the subepithelial sheath. Growth model for siphuncle The model of septal muscle insertion, in which the animal episodically detached its muscles and moved rapidly forward in its shell to a new site of septal fabrication and muscle insertion, has implications for the siphuncle. New siphuncle could not have grown at such a rapid rate and a length of siphuncle, sufficient for the increment of forward movement, would need to have been performed. This problem does not apply to Nautilus where forward movement of the body, although episodic, is thought to take several days (Ward et al. 1981). Only the hard sheath of ammonite siphuncles is preserved, consisting of connecting rings (horny organic tissue perhaps mineralized in some taxa) and calcareous funnels at the ventral margins of septa formed by an extension of the septal nacre (septal necks) or by independent structural elements (false septal necks and auxiliary deposits). The disposition of these various elements shows considerable variety among different ammonite stocks (Birkelund 1981). Any proposal for the growth of the siphuncle and fabrication of its protective coating must account for the following features shown by the sheath and the midventral line of the inner shell wall: 1 . The range of geometries and relationships of its structural components. 2. Location of the connecting rings on the ventral midline in contact with the shell wall. In some taxa the septal necks are also in contact with the shell wall but in many others they lie free. 3. The prochoanitic projection of septal necks. 4. Projection of the connecting rings into the living chamber in some phylloceratid taxa as reported by Drushchits and Doguzhayeva (1974), Kuliki (1979), and Westermann (1982). text-fig. 10. Morphology of the siphuncular sheath and tissue inferred to have been responsible for its formation, a-c, inferred disposition of tissue at the posteroventral margin of the ammonite animal, a , soft parts adjacent to the tip of the ventral saddle of the last septum showing the siphuncle and the parting in the mantle above the siphuncle. b and r, cross-section and longitudinal section of (a) showing invagination of the mantle to form a sleeve, d , e, diagrams showing components of the siphuncular sheath comprising septal necks (dashed) connecting rings (black) and false septal necks and auxiliary deposits (blank); adapted from Birkelund (1981) and orientated with the adapertural portions uppermost, d, Saghalinites. e, Tetragonites. /, Phylloceras ( Hypo - phylloceras). g , /;, decoupling of mantle sleeve and secretion of hard parts as the ammonite animal moves forward in its shell during growth, shown in longitudinal section. Mantle of the sleeve secretes the septal neck while mantle of the siphuncle itself secretes the connecting ring. HENDERSON: SEPTAL FUNCTION IN AMMONITES 481 m irn mitt 482 PALAEONTOLOGY, VOLUME 27 5. The existence of horny membranes which connect the siphuncular sheath to the shell wall and to the septa in a variety of taxa (Grandjean 1910; Westermann 1971; Erben and Reid 1971; Kuliki 1979). 6. The presence of fine longitudinal grooves and ridges which show on the inner shell surface along the midventral line on many phragmocones and is doubtless related to the formation of the siphuncle (PI. 49, figs. 3-5, 7-9, text-fig. 7). On other phragmocones this zone of the inner shell is smooth. Given such specifications, the mode of growth of the siphuncle is severely constrained. A siphuncular growth model involving the preformation necessary for rapid forward movement of the animal proposed here is consistent with the observed morphology of the sheath. It is thought that the siphuncular mantle extended along the ventral midline at the posterior of the body and invaginated itself to form an outer sleeve which opened along the midline as a slit (text-fig. 10). At its posterior margin, the sleeve was continuous with the septal-secreting part of the mantle. Growth of the siphuncle and its sleeve is envisaged to have been slow and constant with their zone of formation migrating forward with respect to the remainder of the body. Prior to forward advance of the body at the initiation of an episode of chamber formation, the new length of preformed siphuncle was detached from the outer sleeve via the slit and adpressed to the shell wall by contraction of the musculature of the body wall. In some taxa, horny membranes that accreted on the mantle of the sleeve braced the siphuncle to the shell wall and to adjacent parts of the septum (Kuliki 1979; Birkelund 1981). They have been most commonly described from close proximity to septa in those taxa where the septal neck is not in contact with the shell wall and a small part of the adjacent connecting ring is suspended within the chamber. A more general means of attachment of the siphuncle was provided when its mantle began secreting the connecting ring, thereby cementing it to the shell wall. The midventral markings on the inner shell wall represent shell secretion at the lip of the slit in the sleeve. A median ridge is its common expression (PI. 49, figs. 4-6, 8-10; text-fig. 66, c) and the adjacent set of fine markings may represent shell secretion by folds in the mantle which formed as tissue near the slit was compressed by enlargement of the siphuncle beneath. The faint lobate markings lying on the midventral line and some distance in front of a septum (PI. 49, fig. 3) which shows on rare specimens may mark the zone where the siphuncular mantle invaginated to form the sleeve rather than a site of muscle insertion as some authors would have it. Secretion of a new septum followed movement of the body, slipping past the newly emplaced segment of siphuncle (text-fig. 1 0g, h). The posterior part of the siphuncular sleeve secreted the septal neck which was structurally continuous with the septum. False septal necks and auxiliary deposits (text-fig. 10c,/) are considered to be mineralized parts of the connecting ring proximal to the new septum and to have been secreted by the mantle of the siphuncle rather than that of the sleeve. The connecting ring inclusive of its mineralized parts is invariably constricted in the septal neck (text-fig. 10(7-/) indicating that the final part of each segment of connecting ring formed later than the septal neck to which it relates. The segmented nature of the connecting ring, with individual segments joined together within the septal neck or at its termination (text-fig. 10 d-f), show that its secretion was episodic, with a hiatus following the completion of each new septum. Mesozoic ammonites show marked changes in siphuncular organization during growth. Very early growth stages commonly display a morphology like that of Nautilus , with retrochoanitic septal necks and connecting rings suspended free within the chambers (Spath 1950; Kuliki 1979). The siphuncular organization of adults proposed here could easily have been developed from such a precursor by gradually moving the position of the siphuncle and by progressively developing an evagination of its mantle. Shell shape and ornament The type of growth experienced by the ectocochleate animal influences the gross nature of its shell In nautiloids the body movement is gradual and the muscle insertions track across the inner shell wall. Here the body can expand its girth by gradual growth to fit a shell form in which there is a rapid increase in expansion rate. Ornate shells with corrugated inner shell surfaces are not conducive to the translocation of muscles by this mechanism. HENDERSON: SEPTAL FUNCTION IN AMMONITES 483 The converse is held to be true for ammonites. Here the animal moves rapidly forward in the shell and is therefore required to expand its girth rapidly to fit the new whorl dimensions. It might be expected that ammonite shells should generally show more gradual whorl expansion rates than nautiloids. A comparison of shell shape in Mesozoic ammonites and nautiloids provided by Ward (1980) shows that this is indeed the case. Transposition of the ammonite body would have been unconstrained by irregularities in the shell wall because muscle tracking was not involved. The striking contrast between Mesozoic ammonites and nautiloids in the general development of ornament thereby finds a ready explanation. Acknowledgements. Critical reviews of the manuscript by Drs. G. R. Chancellor, N. J. Morris, W. J. Kennedy, P. W. Skelton and Mr. C. W. Wright resulted in its improvement and are gratefully acknowledged. M uch of the work was completed at the University Museum and Department of Geology and Mineralogy, Oxford where Dr. W. J. Kennedy and Professor E. A. Vincent arranged the provision of facilities. Drs. T. A. Darragh and P. A. Jell (National Museum of Victoria), Dr. K. J. McNamara and Mr. G. W. Kendrick (Western Australian Museum) and Dr. M. K. Howarth and Mr. D. Phillips (British Museum Natural History) arranged access to specimens in their charge. Financial support for the project in the years 1982- 1983 was provided by the Australian Research Grants Scheme. REFERENCES arkell, w. j. 1949. Jurassic ammonites in 1949. Sci. Prog. 147, 401-417. — 1957. Aptychi. In moore, r. c. (ed.). Treatise on Invertebrate Paleontology. New York and Lawrence, Geol. Soc. Am., pt. L, Mollusca 4, Cephalopoda, Ammonoidea. L437-440. kummel, b. and wright, c. w. 1957. Systematic descriptions. Ibid., LI 29-437. bayer, u. 1977a. Cephalopoden-Septen. Teil I : Konstructionsmorphologie des Ammoniten-Septums. NeuesJb. Geol. Palaont. Abh. 154, 290-366. — 1977ft. Cephalopoden-Septen. Teil 2: Regelmechanismen im Gehauseund Septenbauder Ammoniten. Ibid. 155, 162-215. Bernoulli, d. and jenkyns, h. c. 1974. Alpine, Mediterranean and central Atlantic Mesozoic facies in relation to the early evolution of the Tethys. In dott, r. h. and shaver, r. h. (eds.). Modern and Ancient Geosynclinal Sedimentation. Soc. econ. Palaeont. Mineral. Spec. Publ. 19, 129-160. besairie, h. and collignon, m. 1956. Le Systeme Cretace a Madagascar. Trav. Bur. geol. Madagascar , 77, 66 pp. birkelund, t. 1965. Ammonites from the Upper Cretaceous of West Greenland. Medd. Grnland. 179, 1-192. — 1981. Ammonite shell structure. In house, m. r. and senior, j. r. (eds.). The Ammonoidea. The Evolution , Classification , Mode of Life and Geological Usefulness of a Major Fossil Group. Systematics Association Special Volume No. 18; Academic Press, London. 177-214. hansen, h. j. 1974. Shell ultrastructures of some Maastrichtian Ammonoidea and Coleoidea and their taxonomic implications. Biol. skr. Dan. Vid. Selsk. 20, 1-34. — 1975. Further remarks on the post-embryonic Hypophylloceras shell. Bull. geol. Soc. Denmark , 24, 87-92. blant, g. 1973. Structure et paleogeographie du littoral meridonale et oriental de l’Afrique. In blant, g. (ed.). Sedimentary Basins of the African Coasts , 2. South and East Coasts , 193-231. Assoc. Afr. geol. Surv., Paris. blind, w. 1975. Uber die Entstehung und Funktion der Lobenlinie bei Ammonoideen. Palaeont. Z. 49, 254- 267. — 1976. Die ontogenetische Entwicklung von Nautilus pompilius (Linne). Palaeontographica (A), 153, 117-160. — 1980. Uber Anlage und Ausformung von Cephalopoden-Septen. Neues Jb. Geol. Palaont. Abh. 160, 217-240. bonar, d. b. 1978. Fine structure of muscle insertions on the larval shell and operculum of the nudibranch Phestilla sibogae (Mollusca: Gastropoda) before and during metamorphosis. Tissue and Cell , 10, 143-152. brownell, l. e. and young, e. h. 1959. Process Equipment Design. Vessel Design. John Wiley and Sons, New York. Pp. 408. casey, r. and rawson, p. f. 1973. A review of the boreal Lower Cretaceous. In casey, r. and rawson, p. f. (eds. ). The Boreal Lower Cretaceous. Seel House Press, Liverpool. Pp. 415-430. 484 PALAEONTOLOGY, VOLUME 27 crick, G. c. 1898. On the muscular attachment of the animal to its shell in some fossil Cephalopoda (Ammonoidea). Trans. Linn. Soc. (2) 7, 71-1 13. diener, c. 1895. The cephalopods of the Muschelkalk. Mem. geol. Surv. India Palaeont. indica , ser. 15, 2 (2), 1-118. druschchits, v. v. and doguzhayeva, l. a. 1974. About certain features of morphogenesis of Phylloceratidae and Lytoceratidae (Ammonoidea). Paleont. Zh. 1974, 42-53. erben, h. k. and reid, r. e. h. 1971. Ultrastructure of shell, origin of conellae and siphuncular membranes in an ammonite. Biomineralisation , 3, 22-31. galtsoff, p. s. 1964. The American oyster Crassostrea virginica Gmelin. Fishery Bull. Fish. Wildl. Serv. U.S. 64, 1-480. grandjean, F. 1910. Le siphon des ammonites et des belemnites. Bull. Soc. geol. France, 10, 496-519. griffin, l. e. 1900. Anatomy of Nautilus pompilius. Mem. Nat. Acad. Sci. Wash. 8, 103-197. heptonstall, w. b. 1970. Buoyancy control in ammonoids. Lethaia, 3, 317-328. house, M. r. 1973. An analysis of Devonian goniatite distributions. Spec. Pap. Palaeont. 12, 305-317. howarth, m. k. 1975. The shell structure of the Liassic ammonite family Dactylioceratidae. Bull. Br. Mus. nat. Hist. Geol. 26, 45-67. hubendick, b. 1957. On the molluscan adhesive epithelium. Ark. Zoo!. 11,31 -36. irwin, w. p. 1957. Franciscan Group in Coast Ranges and its equivalents in Sacramento Valley, California. Bull. Am. Ass. Petrol. Geol. 41, 2284-2297. Jordan, r. 1968. Zur Anatomie mesozoischer Ammoniten nach der Strukture-lementen der Gehause- Innenwald. Beih. geol. Jh. 77, 1-64. kauffmann, e. G. 1977. Geological and biological overview: Western Interior Cretaceous basin. Mountain Geol. 14, 75-99. Kennedy, w. j. and cobban, w. a. 1976. Aspects of ammonite biology, biogeography and biostratigraphy. Spec. Pap. Palaeont. 17, 94 pp. — and klinger, h. c. 1975. Cretaceous faunas from Zululand and Natal, South Africa. Introduction, Stratigraphy. Bull. Br. Mus. nat. Hist. Geol. 25, 263-315. — 1977. Cretaceous faunas from Zululand and Natal, South Africa. The ammonite family Tetra- gonitidae Hyatt, 1900. Ann. S. Afr. Mus. 73, 149-197. kennett, j. p. 1982. Marine Geology. Prentice Hall, Englewood Cliffs, New Jersey. Pp. 813. kossmat, f. 1895. Untersuchungen uber die sudindische Kreideformation Erster Theil. Beitr. Palaeont. Geol. Ost.-Ung. 9, 97-203. — 1898. Untersuchungen uber die sudindische Kreideformation Dritter Theil. Beitr. Palaeont. Geol. Ost.-Ung. 11,89-152. kullmann, J. and wiedmann, J. 1970. Significance of sutures in phylogeny of Ammonoidea. Paleont. Contr. Unix. Kansas Pap. 47, 1 -32. kuliki, c. 1979. The ammonite shell, its structure, development and biological significance. Palaeont. Pol. 39, 97-142. lehmann, u. L. 1975. Uber Nahrung und Ernahrungsweise von Ammoniten. Palaont. Z. 43, 187-195. matsumoto, t. 1959. Cretaceous ammonites from the Upper Chitina Valley, Alaska. Mem. Fac. Sci. Kyushu Unix. ser. D, 8, 49-90. — 1960. Upper Cretaceous ammonites of California. Part III. Mem. Fac. Sci. Kyushu Unix. ser. D, Spec. Vol. II, 1-204. mckerrow, w. s. (ed.). 1978. The Ecology of Fossils. The MIT Press, Cambridge, Massachusetts. Pp. 384. miller, a. K. and furnish, w. m. 1957. Introduction to Ammonoidea. In moore, r. c. (ed.). Treatise on Inxertebrate Paleontology. New York and Lawrence, Geol. Soc. Am., pt. L, Mollusca 4, Cephalopoda, Ammonoidea. LI -2. mutvei, h. 1957. On the relations of the principal muscles in the shell in Nautilus and some fossil nautiloids Ark. Miner. Geol. 2, 219-254. 1 964«. On the shells of Nautilus and Spirula with notes on the shell secretion in non-cephalopod molluscs. Ark. Zool. 16, 221-278. 1 9646. Remarks on the anatomy of Recent and fossil cephalophods. Stockholm Contr. Geol. 11, 79 102. 1975. The mode of life in ammonoids. Palaont. Z. 49, 196-202. and reyment, R. A. 1973. Buoyancy control and siphuncle function in ammonoids. Palaeontology , 16, 623-636. nakahara, h. and bevelander, g. 1970. An electron microscope study of the muscle attachment in the mullusc Pinctada radiata. Texas Rep. Biol. Med. 28, 279-286. HENDERSON: SEPTAL FUNCTION IN AMMONITES 485 ojakangas, R. w. 1968. Cretaceous sedimentation, Sacramento Valley, California. Geol. Soc. Am. Bull. 79, 973-1008. OWEN, R. 1843. Lectures on the Comparative Anatomy and Physiology of the Invertebrate Animals, Delivered at the Royal College of Surgeons in 1843. Longman, Brown, Green, and Longmans, London. Pp. 392. pfaff, e. 1911. Ober Form und Bau der Ammonitensepten und ihre Beziehungen zur Suturlinie. Jber. niedersachs. geol. Ver. Hannover , 4, 207-223. raup, d. m. 1966. Geometric analysis of shell coiling: general problems. ,/. Paleont. 40, 1 178-1 190. — 1967. Geometric analysis of shell coiling: coiling in ammonoids. Ibid., 41, 43-65. and Stanley, s. M. 1971 . Principles of Paleontology. W. H. Freeman, San Francisco. Pp. 388. renz, o. 1972. Aptychi (Ammonoidea) from the Upper Jurassic to Lower Cretaceous of the Western North Atlantic. (Site 105, Leg. 1 1, DSDP.) Initial Reps. Deep Sea Drilling Project , 1 1, 607-629. — 1973. Two lamellaptychi (Ammonoidea) from the Magellan Rise in the Central Pacific. Ibid. 17, 895-901. — 1977. Aptychi (Ammonoidea) from the Late Jurassic and Early Cretaceous of the Eastern Atlantic, DSDP Site 367. Ibid. 41,499-513. — 1978. Aptychi (Ammonoidea) from the Early Cretaceous of the Blake-Bahama Basin, Leg 44, Hole 391C DSCP. Ibid. 44, 899-906. — 1979a. Lower Cretaceous Ammonoidea from the northern Atlantic, Leg 47B, Hole 398D, DSDP. Ibid. 43,(2), 361-369. — 1979ft. Aptychi (Ammonoidea) and ammonites from the Lower Cretaceous of the western Bermuda Rise, Leg 43, Site 387, DSDP. Ibid., 591 597. — 1979c. Ammonoidea from the Lower Cretaceous of Hole 402A in the Bay of Biscay, DSDP Leg 48. Ibid., 48, 641-645. imlay, r., Lancelot, y. and ryan, w. b. f. 1975. Ammonite-rich Oxfordian limestones from the base of the continental slope off northwest Africa. Eclogae geol. Helv. 68, 431-448. reyment, r. a. 1958. Some factors in the distribution of fossil cephalopods. Stockholm Contrib. Geol. 1, 97-184. — 1973. Factors in the distribution of fossil cephalopods. 3. Experiments with exact models of certain shell types. Geol. Inst. Univ. Uppsala Bull. N.s. 4, 7-41 . ruzhentsev, v. e. 1946. The evolution and functional significance of the septa of ammonites. Izvestiya AN SSSR Ser. Biol. 6, 675- 706. [In Russian], sastri, v. v., sinha, r. n., singh, G. and murti, k. v. s. 1973. Stratigraphy and tectonics of sedimentary basins on east coast of peninsular India. Bull. Am. Ass. Petrol Geol. 57, 655-678. scott, g. 1 940. Paleoecological factors controlling the distribution and mode of life of Cretaceous ammonoids in the Texas area. J. Paleont. 14, 299-323. seilacher, a. 1975. Mechanische Simulation und funktionelle Evolution des Ammoniten-Septums. Paldont. Z. 49, 268-286. spath, l. f. 1919. Notes on ammonites. Geol. Mag. 56, 27-35,65-74, 1 15-122, 170-177, 220-225. — 1934. Catalogue of the Fossil Cephalopoda in the British Museum (Natural History), Part 4, The Ammonoidea of the Trias. London. 1 -521. 1950. The study of ammonites in thin median sections. Geol. Mag. 87, 77-84. stevens, G. r. and speden, i. g. 1978. New Zealand. In moullade, m. and nairn, a. e. m. (eds.). The Phanerozoic Geology of the World. II. The Mesozoic , A, 251-328. Elsevier Scientific Publishing Company, Amsterdam, Oxford, New York. sweet, w. c. 1959. Muscle-attachment impressions in some Paleozoic nautiloid cephalopods. ./. Paleont. 33, 293-304. tanabe, k. 1979. Palaeoecological analysis of ammonoid assemblages in the Turonian Scaphites facies of Hokkaido, Japan. Palaeontology, 22, 609-630. — obata, i. and futakami, m. 1978. Analysis of ammonoid assemblages in the Upper Turonian of the Manji area, central Hokkaido. Bull. Nat. Sci. Mus. Tokyo Ser. C, 4, 37-62. teichert, c. 1964. Endoceratoidea. Actinoceratoidea, Nautiloidea. Morphology of hard parts. In moore, r. c. (ed.). Treatise on Invertebrate Paleontology. New York and Lawrence, Geol. Soc. Am. pt. K, Mollusca 3. K 13-53. tompa, a. s. and watabe, n. 1976. Ultrastructural investigation of the mechanism of muscle attachment in the gastropod shell. J. Morph. 149, 339-352. veevers, j. j. and Johnstone, m. h. 1974. Comparative stratigraphy and structure of the Western Australian margin and the adjacent deep ocean floor. Initial Reps. Deep Sea Drilling Project, 27, 571 -585. 486 PALAEONTOLOGY, VOLUME 27 wain wright, s. a., biggs, w. d., currey, j. d. and gosline, j. m. 1976. Mechanical Design in Organisms. Edward Arnold, London. Pp. 423. ward, p. 1980. Comparative shell shape distributions in Jurassic-Cretaceous ammonites and Jurassic-Tertiary nautiloids. Paleobiology , 6, 32 43. — greenwald, L., and magnier, Y. 1981 . The chamber formation cycle in Nautilus macromphalus. Ibid. 7, 481-483. westermann, G. E. G. 1971. Form structure and function of shell and siphuncle in coiled Mesozoic ammonoids. Life Sci. Contr. roy. Out. Mas. 78, 1-39. 1975. Model for origin, function and fabrication of fluted cephalopod septa. Palaont. Z. 49, 235-253. — 1982. The connecting rings of Nautilus and Mesozoic ammonids: implications for ammonite bathymetry. Lethaia, 15, 373-384. wiedmann, j. 1979. Early Cretaceous molluscs from DSDP Hole 397A off northwest Africa. Initial Reps. Deep Sea Drilling Project 47 (1), 283-287. and kullmann, j. 1981. Ammonite sutures in ontogeny and phylogeny. In house, m. r. and senior, j. r. (eds.). The Ammonoidea. The Evolution , Classification , Mode of Life and Geological Usefulness of a Major Fossil Group , 215-255. The Systematics Association Special Volume No. 18. Academic Press, London. — and neugebauer, j. 1978. Lower Cretaceous ammonites from the South Atlantic Leg 40 (DSDP), their stratigraphic value and sedimentologic properties. Initial Reps. Deep Sea Drilling Project, 40, 709-734. willey, a. 1902. Contribution to the natural history of the pearly Nautilus. In: Zoological Results Based on Material from New Britain, New Guinea, Loyally Islands and Elsewhere, pt. 6, 69 1-830. Cambridge University Press, Cambridge. wright, c. w. 1957. In moore, r. c. (ed.). Treatise on Invertebrate Paleontology, L185-465. New York and Lawrence, Geol. Soc. Am., pt. L, Mollusca 4, Cephalopoda, Ammonoidea. ziegler, b. 1967. Ammoniten-Okologie am Beispiel des Oberjura. Geol. Rundsch. 56, 439-464. zittel, K. A. 1884. Handbuch der Palaeontologie. Abt. 1, Band 2. Berlin. Pp. 893. R. A. HENDERSON Geology Department James Cook University of North Queensland Townsville Q481 1 Australia Typescript received 5 July 1983 Revised typescript received 17 November 1983 PRINCIPAL FLORAS OF PALAEOZOIC MARINE CALCAREOUS ALGAE by Boris chuvashov and Robert riding Abstract. The stratigraphic distribution of eighteen groups of fossils commonly assigned to the calcareous algae reveals three major floras in shallow marine carbonate deposits of Palaeozoic age: (1) Cambrian flora, (2) Ordovician flora, (3) Carboniferous flora. The Cambrian flora appears abruptly near the Precambrian- Cambrian boundary and is dominated by cyanophytes. The Ordovician flora appears quickly during the lower and middle Ordovician and is dominated by chlorophytes, ?rhodophytes, and problematic groups. The Carboniferous flora appears gradually, mainly during the Carboniferous, and is dominated by rhodophytes, chlorophytes, and problematic groups. Important extinctions occurred near the ends of the Devonian, Carboniferous, and Permian. The succession of floras is reflected in the changing sedimentological roles of Palaeozoic calcareous algae. Cambrian reefs are dominated by Epiphyton- Renalcis assemblages which reappear briefly in the Devonian. During most of the middle Palaeozoic algae are subordinate to metazoan reef-builders, but Solenoporaceae, Rothpletzella, and Wetheredella are nevertheless important locally. Following a hiatus during the lower Carboniferous, Donezella , Ungdarella , phylloid algae, and Tubiphytes were important reef-builders. Skeletal oncoids built by Girvanella , Hedstroemia , Ortone/la , and Rothpletzella , together with Solenopora rhodoliths, are common at many levels in the Palaeozoic, but skeletal stromatolites are generally rare. Nodules formed by Archaeolithophyllum and Cuneiphycus occur in the upper Palaeozoic. Sand- and gravel-size fragments, mainly of chlorophytes and rhodophytes, increase in abundance from the Ordovician onwards. no ocobeHHocTHM CTpaTHrpac|)HHecKoro pacnpocTpaHemm BoceMHa^maTir rpynn OKaMeHejiocTeir, o6mhho OTHOCHMbIX K H3BCCTKOBbIM BOflOpOCJTHM, pa3AH4aK>TCA TpH OCHOBHbIX KOMnjieKCa B MejTKOBOAHblX MOpCKHX Kap6oHaTHbix OTJio>KeHiiax naneo3oa: (1) KeMSpuncKuu KOMnneKc; (2) opaobhkckhh KOMnneicc; (3) KaMeH- HoyrojibHbiu KoivmjieKc. KeM6pnHCKHH KOMnneKC BOAopocnefi noHBJiaeTCH bhahmo b6jih3h hhachch rpammbi KeMbpua; b ero cocTaBe flOMHHupyroT uuaHO(})HTbi. Opaobhkckhh KOMnneicc noABHACA 6biCTpo b cpeAHeM opAOBHKe h npegcTaBjreH npeHMymecTBeHHo xnopocjmTaMH, poAocjjHTaMir (?) h npobjieMaTHHHbiMH rpynnaMH. KaMeH- HoyrojrbHan (jmopa (j)opMHpyeTca nocTeneHHO, rnaBHbiM o6pa30M, b TeneHUH KapboHa. B ee cocTaBe AOMHHupoBaAH poAO(j)HTbi, XAopo(j)HTbi h npobAeMaTHHHbie rpynnbi. Ba*Hbie H3MeHeHna nponcxoAHAii b KOHue AeBOHa, Kapbona h nepMu. YcTaHOBAeHHaa nocAeAOBaTenbHOCTb b pa3BHTnn BOAopocneu oTpaacaeTca b roMeHemm ceAHMeHTOAorn- uecKoro 3HaHeHHH naAeo3oircKHx H3BecTKOBbix BOAOpocAeii. CpeAH KeMbpuHCKHx pu(j)OB AOMHHupoBaAO coobmecTBO poaob Epiphyton- Renalcis, KOTopoe 3aTeM Ha KopoTKoe BpeMa noaBAaeTca BHOBb b acbohc. B TeueHue cpeAHero naAeo3oa boaopocah, KaK pH(j)o- o6pa30BaTeAH, 6biAH noAHHHeHbi MeTa3oa, ho Solenoporaceae, Rothpletzella w Wetheredella nrpaAH MecTaMH BaacHym poAb b co3AaHHH ocaAKOB. nocae HH*Hero KapboHa, b TeueHMe KOToporo nopoAoobpa3yiomee 3HaueHHe H3BecTKOBbix BOAopocAeir 3aMeTHO naAaeT, Donezella, Ungdarella , <(>HAAOMAHbie boaopocah, a raK/KC Tubiphytes bbiAH BaacHbiMir pi«j)oobpa30BaTeAaMH. CxeAeTHbie ohkoham, nocTpoeHHbie Girvanella, Hedstroemia , Ortone/la h Rothpletzella coBMecTHO c Solenopora-pojxomwdiwm aBAarorca obbiuHbiMH Ha mhothx ypoBHax naAeo3oa, ho CKeAerHbie CTpoMaTOAHTbi obbiHHO peAKH. >KeABaKH, obpa30BaHHbie c yaaciHCM Archaeolithophyllum n Cuneiphycus BCTpenaroTca b BepxHeM naAeo3oe. 3epna necnaHOH h rpaBHHHoft pa3MepHOCTH obpa30BaHbi, rAaBHbiM obpa30M, 3a cueT 3eAeHbix h KpacHbix BOAopocAefl, yBeAHAHBaroTca KOAHuecTBeHHo b ocaAKax c no3AHero opAOBHKa. We present a general overview of the stratigraphic distribution of calcareous algae during the Palaeozoic. Our aims are to discern broad patterns of calcareous algal evolution and to evaluate briefly how these are reflected in the sedimentological importance of these fossils. We have IPalaeontology, Vol. 27, Part 3, 1984, pp. 487-500.| 488 PALAEONTOLOGY, VOLUME 27 incorporated available data from North America and one or two other areas, but most of our information is derived from work in Europe and the USSR. In order to present this we have divided the many genera involved into a number of groups which have a broad base within current systematic schemes. The problems of affinity in Palaeozoic calcareous algae are well known (Riding 1977a), but remain largely unresolved. They are mainly responsible for uncertainty concerning the systematics of these fossils. We have selected groups which have some degree of morphological similarity. In some cases their affinities are clear, in others doubtful. Some groups include members which are possibly not algae. In this paper we attempt to encompass all groups which are commonly regarded as algae, even if we personally have doubts concerning such an attribution. However, we have neglected some small groups represented by only a few genera. It would be a major undertaking to plot accurately and comprehensively the distribution of the large number of genera involved, and such a compilation would necessitate substantial taxonomic revision. Our aim here is to review the changing composition of these floras during the Palaeozoic in a very broad way in order to assess general patterns. Thus, these results are preliminary and doubtless imperfect with respect both to the groups selected and their ranges. In particular we have recognized the fewest possible number of major groups, and this has involved a degree of ‘lumping’ which will be open to criticism. Nevertheless, we believe that this procedure enhances, rather than detracts from, the validity of the patterns elucidated here. € i OSD i iii C P i i Solenoporaceae Hedstroemia-Ortonella Girvanella RenaScis-Shuguria Epiphytales-Cambrinales Moniliporellaceae Rothpletzella Receptaculitales Dasycladales Codiaceae/ Udoteaceae Wetheredella Kamaena - Donezella Beresella Ungdarella -Stacheia phylloid algae Archaeolithophyllum - Cuneiphycus Tuhiphytes Gymnocodiaceae -► text-fig. 1. Stratigraphic ranges of eighteen major calcareous algal groups during the Palaeozoic. Generic names indicate groups, not individual genera (see Table 1), except in the case of Tubiphytes. Ranges are drawn from the base of the sub-period (early, middle, late) in which the first member of a group appears, to the top of the sub-period in which the last member occurs. Arrow indicates that group continues into the Mesozoic. Length of periods is based upon Harland et al. (1982). CHUVASHOV AND RIDING: PALAEOZOIC MARINE CALCAREOUS ALGAE 489 table 1. Eighteen major groups of Palaeozoic calcareous algae and possible calcareous algae, showing their main characters and the affinities confidently or currently attributed to them. The references give sources of further information but in many cases do not cite the authors of the groups or of the named genera themselves. Group Contents Characters Affinities References 1. Solenoporaceae Solenopora Dybowski, Parachaetetes Deninger, etc. Massive, tabular, hemispherical or nodular skeleton composed of closely packed cellular filaments, sometimes possibly containing sporangia Probable rhodophytes; possibly related to the Corallinaceae Johnson (I960); Maslov (1962) 2. Hedstroemia- Ortonella group Bolomaella Korde, Bevocastria, Garwood, Garwoodia Wood, Hestroemia Rothpletz, Ortonella Garwood, etc. Fan-like bundles of branched tubes of varying cross-sectional shape. In some genera the tubes are closely packed and share walls, in others the tubes are separate. Branching ranges from dichothomous, to irregular and multiple Several of the genera resemble extant calcareous cyanophytes Wray (1977, pp. 38-39) 3. Girvanella group Batenevia Korde, Botominella Reitlinger, Cladogirvanella Ott, Girvanella Nicholson and Etheridge, Obruchevella Reitlinger, etc. Narrow simple tubes of constant diameter and without cross- partitions. Tubes may be straight, sinuous, irregularly tangled, or spiral, and may be arranged in tightly woven, cable-like bundles or loose masses Probably filamentous cyanophytes Wray (1977, pp. 36-37) 4. Rena leis - Shugaria group Gemma Luchina, [zhella Antropov, Renalcis Vologdin, Shuguria Antropov, Tarthinia Drosdova, etc. Cloud-like forms consisting of clusters of a few or many hollow, bubble-like compart- ments, sometimes arranged in short branched series Possibly cyanophytes Drosdova (1980, pp. 14-19); Wray (1977, p. 40) 5. Epiphytales- Cambrinales Cambrina Korde, Epiphyton Bornemann, Gordonophyton Korde, Potentillina Korde, Tubomorphophyton Korde, etc. Dendritic solid micritic, tubiform or chambered thalli. Branches narrow; branching dichothomous or irregular Possibly cyanophytes or rhodophytes Korde (1973, pp. 125-212); Luchinina (1975) 6. Codiaceae/ Udoteaceae Dimorphosiphon Hoeg Lancicula Maslov , Litanaia Maslov, Palaeoporella Stolley, etc. Entire or segmented thallus, may be branched, internally divided into cortex and medulla consisting of numerous branched tubes Chlorophytes Gnilovskaya (1972, pp. 79-100); Shuysky (1973, pp. 61 -80) 7. Rothpletzella group Flabellia Shuysky, Halysis Hoeg, Rothpletzella Wood Flat, curved, or encrusting sheets of juxtaposed tubes which branch dichothomously in one plane Microproblematica, often regarded as cyanophytes or chlorophytes Flugel and Wolf (1969) 490 PALAEONTOLOGY, VOLUME 27 Group Contents Characters Affinities References 8. Wetheredella group Aphralysia Garwood, Asphaltina Mamet, Sphaeroporella Antropov, Wetheredella Wood, etc. Short tubes, possibly branched, with blister-like or rounded cross-sections; fibrous wall structure sometimes with additional micritic layer Microproblematica sometimes regarded as chlorophytes, foraminifers, or worms Ischenkoand Radionova (1981); Mamet and Roux (1975, pp. 156-166) 9. Receptaculitales Calathium Billings, Ischadites Murchison, Receptaculites Deshayes, etc. Large, hollow, pear- or sack- like bodies, usually open at one end, with double-walls and faceted outer surfaces Problematica, often referred to the Chlorophyta Nitecki (1972); Rietschel (1969); Zhuravleva and Myagkova (1981) 10. Dasycladales I. II. Dasyporella Stolley, Rhahdoporella Stolley, Vermiporella Stolley, etc. in middle Palaeozoic; Diplopora Schafhaiitl, Epimastopora Pia, Globiferoporella Tchuvashov, Macroporella Pia, Mizzia Schubert, etc. in upper Palaeozoic Hollow sack- or stick-like algae, usually large and erect, sometimes segmented; relatively thick walls pierced by simple or branched pores Palaeozoic forms show clear separation into middle Palaeozoic and upper Palaeozoic assemblages (I and II) Chlorophyta Pia (1920); Elliott (1972); Shuysky (1973, pp. 80-87), Chuvashov (1974) 1 1 . Kamaena- Donezella group Donezella Maslov, Jansaella Mamet and Roux, Kamaena Antropov, Palaeoberesella Mamet and Roux, etc. Branched, mainly dichotho- mously, septate tubes with finely porous or fibrous wall- structure. Septa may be entire or incomplete Microproblematica, often regarded as chlorophytes or rhodophytes, sometimes as foraminifers Maslov (1956); Antropov (1967); Mamet and Roux (1974) 12. Beresella group Beresella Machaev, Dvinella Khvorova, Uraloporella Korde, etc. Moderately large, straight to sinuous, branched tubes; some- times septate; wall relatively thick with pores which may be simple or branched Microproblematica, but commonly regarded as dasycladaleans Korde et al. (1963, p. 211, p. 217) 13. Phylloid algae Anchicodium Johnson, Eugonophyllum Konishi and Wray, Ivanovia Khvorova, Neoanchicodium Endo, etc. Large, thin, wavy, leaf-like thallus; cortex thick, usually porous, sometimes containing ‘sporangia’; medulla filamentous, usually poorly preserved Commonly regarded as codiaceans/udoteaceans; some genera are comparable with Squamariaceae (Peys- soneliaceae, Rhodophyta) Wray (1968; 1977, pp. 52- 54) 14. Archaeolitho- phyllum- Cuneiphycus group Archaeolithophyllum Johnson, Cuneiphycus Johnson, Eflugelia. Encrusting or hemispherical masses; coarse cellular construction, occasionally bearing conceptacles Rhodophyta (Archaeolithophyllum) and Microproblematica Wray (1977, pp. 71-72) CHUVASHOV AND RIDING: PALAEOZOIC MARINE CALCAREOUS ALGAE 491 Group Contents Characters Affinities References 1 5. Ungdarella Stacheia group Aoujgalia Termier and Termier, Epistacheoides Petryk and Mamet, Fourstonella Cummings, Komia Korde, Stacheia Brady, Stacheoides Cummings, Ungdarella Maslov, etc. Encrusting or erect, rod-like, branched fossils with cellular construction; cells sometimes aligned in coarse rope-like strands Microproblematica, often regarded as rhodophytes Maslov (1962); Petryk and Mamet (1972); Massa and Vachard (1979) 16. Gymnocodiaceae Gymnocodium Pia, Permocalculus Elliott, etc. Cylindrical or sack-like fossils, sometimes segmented; cortex thin, with external pores; medulla filamentous. Sporangia may be present Chlorophytes or rhodophytes Elliott (1955); Korde (1965) 17. Tubiphytes group Tubiphytes Maslov Large erect, irregular, or encrusting skeletons with a dense, dark, pseudo-cellular construction showing concentric bands; irregular internal tubes often present Microproblematicum, sometimes regarded as a cyanophyte; possibly a sponge (E. Ott, pers. comm. 1982) Maslov (1956); Fliigel (1977, pp. 324 325, p. 339) 18. Moniliporellaceae Ansoporella Gnilovskaya, Contexta Gnilovskaya, MonUiporeUa Gnilovskaya, etc. Cylindrical, nodular, or irregular fossils with hollow interior; sometimes segmented; thick wall consisting of cellular filaments Rhodophytes or chlorophytes Gnilovskaya (1972, pp. 100 126) GROUPS OF CALCAREOUS ALGAE In Table 1 we list the certain and the equivocal algal groups whose stratigraphic distribution we have plotted in text-fig. 1. Each group is distinguished by either a supra-generic name or by one or two genera representing typical or well-known elements of the group. We list a few additional generic examples which normally represent only a small fraction of the total for each group. In addition we indicate the likely affinities (or doubts concerning affinities) for each group. Pertinent references are also given. STRATIGRAPHIC DISTRIBUTION The distribution chart of major calcareous algal groups (text-fig. 1) has been compiled from the literature cited in the table and from our personal experience of working with these fossils. It shows the first and last occurrences of each group as a whole, and not of the named genera alone. For example, Hedstroemia and Ortonella are principally Silurian and Carboniferous fossils respectively, but other members of the Hedstroemia-Ortonella group, as it is defined here, such as Botomaella , occur in the Cambrian and related types like Cayeuxici Frollo occur in the Mesozoic. 492 PALAEONTOLOGY, VOLUME 27 The pattern of stratigraphic distribution of the major groups (text-fig. 1 ) allows three distinct floras to be recognized (table 1; text-fig. 2): (1) Cambrian flora, (2) Ordovician flora, (3) Carboniferous flora. The Cambrian flora is dominated by cyanophytes (Hedstroemia-OrtoneRa group, Girvanella group) and possible cyanophytes (Epiphytales-Cambrinales, Renalcis-Shuguria group); the possible rhodophyte group, Solenoporaceae, is present but rare. This flora ranges intact into the upper Devonian but then loses the Epiphytales-Cambrinales and Renalcis-Shuguria groups. The remaining elements continue beyond the Permian-Triassic boundary. The Ordovician flora is dominated by Codiaceae/Udoteaceae, Dasycladales, the possible rhodophyte Monilioporellaceae, the possible chlorophyte Receptaculitales, the problematic Rothpletzella and € OSD CP 5 text-fig. 2. Three principal floras of Palaeozoic marine cal- careous algae (from text-fig. 1) showing ranges, general com- position, and rates of appear- ance. The Cambrian, Ordovi- cian, and Carboniferous floras appeared over periods of 5, 50, and 100 Ma respectively. Length of periods is based upon Harland et al. (1982). CHUVASHOV AND RIDING: PALAEOZOIC MARINE CALCAREOUS ALGAE 493 Wetheredella groups, together with the major elements of the Cambrian flora. Its new, characteristic, elements originated mainly in the lower and middle Ordovician and persist to the upper Devonian or beyond. The Carboniferous flora is dominated by the problematic Kamaena-Donezella group, the possible chlorophyte Beresella group, a new assemblage of dasycladaleans, the possible rhodophyte Ungdarella-Stacheia group, the chlorophyte or rhodophyte phylloid algae and Gymnocodiaceae, the partly rhodophyte Archaeolithophyllum-Cuneiphycus group, and the doubtfully algal Tubiphytes group, together with elements of the Cambrian and Ordovician Devonian floras which survived an important phase of extinction near the Devonian-Carboniferous boundary. The Carboniferous flora was introduced episodically, mainly during the lower and middle Carboniferous, with the Kamaena- Donezella group appearing earlier (in the middle Devonian) and the Tubiphytes group and Gymnocodiaceae later (in the upper Carboniferous and early Permian respectively). Most of these new upper Palaeozoic groups did not survive into the Mesozoic. SEDIMENTOLOGICAL ROLES The importance of calcareous algae as producers of loose and in situ sediment is clear in Recent carbonate environments. It is equally recognizable in its effects upon limestone deposition from the first appearance of calcareous algae near the base of the Cambrian. The sedimentological roles of particular algal groups depend essentially upon the morphology and mode of growth of the algae, and the succession of algal floras has in turn imprinted a stratigraphic pattern upon their sedimentary products. The reef-building algae of the Cambrian differ in size, shape, and effects from those of the Carboniferous, the abundance of algal skeletal fragments changes through time, and the types of nodule-forming algae also change. The broad patterns of these variations are controlled by evolution and extinctions more than by sedimentary processes and palaeo- geography. In order to show these changes we have plotted qualitative assessments of the relative importance of Palaeozoic algal groups in the following roles: reef-building; stromatolite, oncoid, and rhodolith formation; and the production of recognizable, usually sand- to gravel-size, fragmentary (broken or disaggregated) material (text-figs. 3-7). Reef-building The three floras recognized here are clearly reflected in Palaeozoic reef-building (text-fig. 3). The Cambrian algal reef-builders belong mainly to two groups: the Epiphytales-Cambrinales and Renalcis-Shuguria. Angulo cellular ia Vologdin, omitted from these distribution charts because it constitutes a taxonomically small group, is also locally an important reef-builder (see Riding and Voronova 1982). These are all small but abundant fossils and commonly exceed archaeocyathans in volumetric importance in the early lower Cambrian (James and Debrenne 1980). Ordovician-Devonian algal reef-builders are more diverse and, in general, none has the individual importance of those in the Cambrian. Receptaculitaleans are rather rare fossils, and Hedstroemia is at present only known to be important in reefs in the Silurian (Riding and Watts 1981). Rothpletzella and Wetheredella form thick crusts (Copper 1976), but usually on large metazoan reef builders such as stromatoporoids, corals, and bryozoans. Solenoporaceans form the largest individual skeletons but are, nevertheless, usually subordinate to metazoans (Harland 1981). On the whole, a variety of algal groups is locally conspicuous in middle Palaeozoic reefs, but they are usually only accessory to larger and more abundant metazoans. A curious feature of middle Palaeozoic algal history is the reappearance of ‘Cambrian’ reef-building genera in the Devonian, especially the upper Devonian. These are members of the Epiphytales and Renalcis-Shuguria groups and their return to prominance at this level, after insignificance from the middle Ordovician to early Devonian, has so far defied satisfactory explanation. 494 PALAEONTOLOGY, VOLUME 27 Epiphytales-Cambrinales Renalcis-Shuguria Rothpletzella Wetheredella Solenoporaceae Receptaculitales Hedstroemia - Ortonella phylloid algae Kamaena - Donezella Ungdarella-Stacheia Tubiphyles text-fig. 3. Reef-forming cal- careous algae during the Palaeo- zoic: estimated relative import- ance. Groups with minor roles are omitted. There is a hiatus in the early Carboniferous with few algal reef-builders following the demise of some of the Cambrian-Devonian groups. Only members of the Renalcis-Shuguria group have been reported as common reef constituents at this level. However, the phase of algal evolution which took place in the lower to middle Carboniferous yielded several important groups which filled this gap. In particular Donezella (Rich 1967, Riding 1979) and Ungdarella-Komia (Freeman 1964) are important mound-builders or, at least, mound-associates in the middle Carboniferous and Pennsylvanian. Phylloid algae also created bioherms from the middle Carboniferous to early Permian (Wilson 1975) and Tubiphytes is important from the upper Carboniferous until the Triassic (Fliigel 1977). Stromatolites Skeletal stromatolites, i.e. stromatolites formed by calcareous algae rather than by algae which merely trap and bind sediment (Riding 19776) are, so far, only known to be common in parts of the Ordovician, Devonian, and Carboniferous (text-fig. 4). Rothpletzella and Wetheredella form stromatolitic crusts on metazoan skeletons in upper Ordovician reefs (Copper 1976). Rothpletzella forms stromatolitic caps on stabilized oncoids on the fore-reef slope of the upper Devonian Canning Basin reefs in Western Australia (Playford, Cockburn, Druce and Wray 1976). Girvanella builds stromatolites in the Devonian of the Ural Mountains, USSR, and Ortonella and Bevocastria build stromatolites in the lower Carboniferous of the Scottish border country. Great Britain (Garwood 1931). € O S D C P Rothpletzella Girvanella Hedstroemia - Ortonella text-fig. 4. Stromatolite-forming cal- careous algae during the Palaeozoic: estimated relative importance. Groups with minor roles are omitted. CHUVASHOV AND RIDING: PALAEOZOIC MARINE CALCAREOUS ALGAE 495 text-fig. 5. Oncoid-forming calcareous algae during the Palaeozoic: estimated relative importance. Groups with minor roles are omitted. Girvanella Rothpletzella Wetheredella Hedstroemia - Ortonella Kamaena - Donezella P Oncoids Skeletal oncoids are much more widespread than skeletal stromatolites in the Palaeozoic (text-fig. 5). Girvanella forms oncoids from the Cambrian to Carboniferous, and in association with Nubecularia it was responsible for Osagia nodules in the Pennsylvanian (Johnson 1946). Rothpletzella and Wetheredella are often mutually associated in oncoids from the Ordovician to Devonian. Hedstroemia forms oncoids in the Silurian of Gotland and Bevocastria , Ortonella and Garwoodia are involved in oncoid formation, as well as stromatolite formation in the Lower Carboniferous of Britain. In addition, Donezella forms encrusted nodules in the middle Carboniferous. Rhodoliths Solenoporacean nodules are common during the Ordovician and Silurian (Johnson 1960) and are also locally abundant in the Upper Palaeozoic (Belka 1979). Archaeolithophyllum and Cuneiphycus form nodules in the upper Carboniferous (text-fig. 6). Fragments The presence of algal skeletal debris in shallow marine Palaeozoic limestones reflects the history of fragile or jointed specimens which were readily broken or disaggregated into sand- or gravel-size pieces (text-fig. 7). Algae, like the modern codiacean/udoteacean Penicillus , which may have disaggregated after death into mud- and silt-size particles leave no readily recognizable trace because the resulting particles are too small for their origin to be recognized. At times during the Palaeozoic, as in subsequent geological eras, chlorophytes produced large quantities of calcareous debris. This sedimentological role commenced in the Ordovician when codiaceans/udoteaceans and dasycladalean fragments are also associated with those of moniliporel- laceans. Codiacean/udoteacean debris is also locally common in the Devonian, but is more rare in the Carboniferous when this role was mainly occupied by dasycladaleans. In the Cambrian, algal fragments are rare. This is a result both of the absence of calcareous chlorophytes and the fact that the common cyanophytes were generally firmly attached reef-builders. If the latter were broken from their substrates they produced small, micritic fragments difficult to distinguish from peloids. Nevertheless, in the Devonian, members of the Renalcis-Shugaria group are locally common as transported grains, as are Girvanella and Rothpletzella. The latter are also minor components of near-reef sediments in the Ordovician and Silurian. Kamaena-Donezella group fragments are common at various levels from middle Devonian to upper Carboniferous, but the text-fig. 6. Rhodolith-forming calcareous algae during the Palaeozoic: estimated relative importance. Groups with minor roles are omitted. Solenoporaceae Archaeolithophyllum - Cuneiphycus 496 PALAEONTOLOGY, VOLUME 27 Moniliporellaceae Dasycladales Codiaceae/Udoteaceae Renalcis-Shuguria Girvanella Rothpletzella Kamaena - Donezella phylloid algae Beresella Ungdarella-Stacheia Tubiphytes Gymnocodiaceae € OSD C P text-fig. 7. Debris-producing calcareous algae during the Palaeozoic: estimated relative importance. Groups with minor roles are omitted. principal increase in algal debris in the upper Palaeozoic took place in the middle Carboniferous when fragments of phylloid algae, plus the Beresella , Ungdarella-Stacheia , and Tubiphytes groups, combined with those of the new assemblage of dasycladaleans to produce loose material which often dominated shallow marine carbonate microfacies. The Gymnocodiaceae added to this, especially in the upper Permian. DISCUSSION Floras The Cambrian calcareous algal flora was dominated by cyanophytes and possible cyanophytes: Solenoporaceae were relatively rare. The Ordovician and Carboniferous floras are both more mixed. If groups whose affinities are unclear are not considered, then the resulting picture of algal evolution is that calcareous cyanophytes appeared in the Cambrian, chlorophytes (codiaceans/udoteaceans and dasycladaleans) in the Ordovician, and rhodophytes ( Archaeolithophyllum ) in the Carboniferous. If we take possible affinity into consideration, this time-distribution does not change for cyanophytes and chlorophytes but calcareous rhodophytes may be present from the early Cambrian (Soleno- poraceae, Epiphytales). It is clearly a matter of current importance for research to attempt to clarify the affinities of these and other possible algal groups in the Palaeozoic. Diversity of major algal groups increases from the Cambrian (five groups) to the Ordovician (eleven groups). Subsequent increase is slight, rising to twelve groups in the Devonian, fourteen in the Carboniferous, and falling slightly to thirteen in the Permian (text-fig. 8). The resulting sigmoidal curve resembles the pattern of exponential diversification followed by equilibrium derived for marine metazoan orders during the Phanerozoic (Sepkoski 1978, fig. 9). Evolutionary events , extinctions , ranges The three principal algal floras recognized here were introduced in the earliest Cambrian, Ordovician, and Devonian-Carboniferous respectively. Of these, the first event near the Precambrian-Cambrian CHUVASHOV AND RIDING: PALAEOZOIC MARINE CALCAREOUS ALGAE 497 boundary was abrupt, the Ordovician event more gradual, and that in the Devonian-Carboniferous slow (text-fig. 2). In the Nemakit Daldyn Formation of late Precambrian or early Cambrian age in northern Siberia members of the Hedstroemia-Ortonella , Girvanella , Epiphytales-Cambrinales, and Renalcis-Shuguria groups appear together virtually synchronously (Riding and Voronova, in prep.) and are joined, probably within 5 Ma, by solenoporaceans. The Ordovician event spanned approximately 50 Ma from early to late Ordovician, and the third phase of evolution was a slow episodic appearance of groups over a period approaching 100 Ma between the middle Devonian and early Permian (text-fig. 2). Nevertheless, both the last two events show some concentration, first in the lower-middle Ordovician and secondly in the lower-middle Carboniferous. text-fig. 8. Number of major algal groups present in each Palaeozoic period. Data from text-fig. 1. Shading indicates first appearances. Extinctions, at group-level, were concentrated near the Devonian-Carboniferous boundary, in the early Permian, and at the Permian-Triassic boundary (text-fig. 9). It is noteworthy that each flora has nearly similar numbers of groups persisting into the Mesozoic: three in the case of the Cambrian flora, two each for the other two floras (text-fig. 1). In fact, most of the groups appearing during the Cambrian and Ordovician are very long-ranging, five out of the eleven continuing not only into the Mesozoic but also into the Cenozoic. The groups appearing during the Devonian to Permian are relatively short-ranging, only Tubiphytes (which is also of doubtfully algal affinity) surviving into the Triassic. Sediment ology Cyanophytes locally dominated Cambrian reefs. In comparison Ordovician-Devonian algae were nearly always subordinate to metazoan reef-builders, although Solenopora , Rothpletzella , and Wetheredella can be important in the Ordovician and Silurian, and the ‘Cambrian’ Epiphyton- Renalcis association reappears in upper Devonian bioherms. Elowever, nodules, including oncoids and Solenopora rhodoliths, are more common in the middle Palaeozoic than in either the Cambrian or Permian. The poor algal contribution to reef-building in the lower Carboniferous could be due to the scarcity of bioherms generally at this level, other than waulsortian mounds. It was middle Carboniferous expansion of the groups containing Donezella , UngdareUa-Komia , phylloid algae, and Tubiphytes which provided the new algal reef-builders for the upper Palaeozoic. Stromatolites built by calcareous algae are generally uncommon in the Palaeozoic. 498 PALAEONTOLOGY, VOLUME 27 S D C P ! I i I 2 V/////////A 3 evolutionary extinction events phases 1 W77\ 2 0 30 m ® to •*-> e © E i3 ® m (0 E Solenoporaceae Rothpletzella Wetheredella Renalcis- Epiphyton phyiloid, Tubiphytes Epiphyton-RenalcisC^y^S O Ungdarella - Donezella mainly chlorpphyte & rhodophyte debris; © Girvanella Solenopora Rothpletzella -wetheredella <5> Archaeolithophyllum- Cuneiphycus Solenopora Ortonella text-fig. 9. Summary of calcareous algal floras during the Palaeozoic, showing phases of development and extinction, and sedimentological roles. Although the patterns of reef-, nodule-, and stromatolite-formation by calcareous algae show sharp variations during the Palaeozoic, that of debris production is a relatively simple trend of increase throughout the era (text-fig. 9). Essentially this reflects the history of various fragile chlorophyte and rhodophyte groups. Acknowledgements. This work is an outcome of discussion facilitated by the USSR Academy of Sciences at the Palaeontological Institute, Moscow, in September 1982, during a visit by Riding sponsored by the Academy and the Royal Society. We are grateful to Eleonora Radionova for advice on several algal groups. Larisa Voronova provided working space. Graham Elliott and Donald Toomey read the manuscript and made valuable suggestions for its improvement. REFERENCES antropov, i. a. 1967. Devonian and Lower Carboniferous (Tournaisian) deposits of the central part of the Russian Platform. Acad. Sci. USSR , Siberian Branch , Inst. Geol. Geophys. 118-125. [In Russian.] belka, z. 1979. Shallow-water Solenoporaceae and their environmental adaptation. Upper Permian of the Holy Cross Mts. Bull. Centre Rech. Explor.-Prod. Elf- Aquitaine, 3, 443 452. CHUVASHOV AND RIDING: PALAEOZOIC MARINE CALCAREOUS ALGAE 499 chuvashov, b. i. 1974. Permian calcareous algae of the Urals. In Algae, brachiopods and miospores from Permian deposits of the west Urals. Acad. Sci. USSR , Ural Sci. Centre , Trans. Inst. Geol. Geochem. Sverdlovsk , 3-76. [In Russian.] copper, p. 1976. The cyanophyte Wetheredella in Ordovician reefs and off-reef sediments. Lethaia , 9, 273-281 . drosdova, n. a. 1980. Algae in Lower Cambrian organic mounds of west Mongolia. Trans. Joint Soviet- Mongolian Palaeontological Expedition , 10, 140 pp. [In Russian.] elliott, G. f. 1955. The Permian calcareous alga Gymnocodium. Micropalaeontology , 1, 83-90. — 1972. Lower Palaeozoic green algae from southern Scotland, and their evolutionary significance. Bull. Brit. Mus. Nat. Hist. Geology , 22, 355-376. flugel, e. 1977. Environmental models for Upper Palaeozoic benthic calcareous algal communities. In flugel, e. (ed.). Fossil algae , recent results and developments , 314-343. Springer, Berlin. — and wolf, k. h. 1969. 'Sphaerocodien’ (Algen) aus dem Devon von Deutschland, Marokko und Australien. N. Jb. Geol. Palaont. Mb. 2, 88 103. freeman, t. 1974. Algal limestones of the Marble Falls Formation (Lower Pennsylvanian), central Texas. Geol. Soc. Amer. Bull. 75, 669-676. garwood, e. j. 1931. The Tuedian beds of northern Cumberland and Roxburghshire east of Liddel Water. Quart. J. Geol. Soc. London , 87, 97-159. GNILOVSKAYA, m. b. 1972. Calcareous algae of the middle and late Ordovician of eastern Kazakhstan. Acad. Sci. USSR, Inst. Precambrian Geol. and Geochronology. Nauka, Leningrad 196 pp. [In Russian.] harland, t. L. 1981. Middle Ordovician reefs of Norway. Lethaia , 14, 169-188. HARLAND, W. B., COX, A. V., LLEWELLYN, P. G., PICKTON, C. A. G., SMITH, A. G. and WALTERS, R. 1982. A geologic time scale. 131 pp. Cambridge University Press. ischenko, a. a. and radionova, E. p. 1981. On the morphology and systematic position of the genus Wetheredella Wood, 1948. In Questions of Micro-palaeontology. Trans. Acad. Sci. USSR , Geol. Inst. 24, 140 151. [In Russian.] james, N. p. and debrenne, f. 1980. Lower Cambrian bioherms: pioneer reefs of the Phanerozoic. Acta Palaeont. Polonica , 25, 655-668. Johnson, j. H. 1946. Lime-secreting algae from the Pennsyvanian and Permian of Kansas. Geol. Soc. Amer. Bull. 57, 1087- 1 120. — 1960. Palaeozoic Solenoporaceae and related red algae. Quart. Colorado School Mines , 55, 1 77. korde, k. b. 1965. Algae. In ruzencev, v. e. and sarycheva, t. g. (eds.). The development and change of marine organisms at the boundary between the Palaeozoic and Mesozoic. Trans. Palaeont. Inst. USSR , 108, 268-284, 414-429. [In Russian.] korde, k. b. 1973. Cambrian algae. Trans. Palaeont. Inst. USSR. 139, 1-349. [In Russian.] — KULIK, E. L., MASLOV, V. P., MOSKALENKO, T. A. and NAUMOVA, S. N. 1963. Chloiophyta. In ORLOV, YU. A. (ed.). Fundamentals of Palaeontology , 15, 198-223. [In Russian.] luchinina, v. a. 1975. Palaeoalgological characteristics of the early Cambrian of the Siberian Platform. Trans. Acad. Sci. USSR , Siberian Branch , Inst. Geol. Geophys. 216, 1-99. [In Russian.] mamet, b. and roux, a. 1974. Sur quelques algues tubulaires scalariformes de la Tethys Paleozoique. Rev. Micropaleonto/ogie, 17, 134-156. — and 1975. Algues Devoniennes et Carboniferes de la Tethys occidentale. Troisieme partie. Rev. de Micropaleonto/ogie 18, 134-187. maslov, v. p. 1956. Calcareous algae of the U.S.S.R. Trans. Acad. Sci. USSR , Geol. Inst. 160, I 301. [In Russian.] 1962. Fossil red algae of the USSR. Trans. Acad. Sci. USSR , Geol. Inst. 53, 1-221. [In Russian.] massa, d. and vachard, d. 1979. Le Carbonifere de Libye occidentale: biostratigraphie et micropaleontologie. Rev. Inst. Francois du Petrole, 34, 1-65. nitecki, m. h. 1972. North American Silurian receptaculitid algae. Fieldiana Geology , 28, 1-108. petryk, a. a. and mamet, b. l. 1972. Lower Carboniferous algal microflora, southwestern Alberta, Can. J. Earth Sci. 9, 767 -802. pia, j. 1920. Die Siphoneae verticillatae vom Karbon bis zur Kreide. Abh. Zool.-Bot. Ges. Wien, 11, 1-263. playford, p. e., cockburn, a. e., druce, E. c. and wray, j. l. 1976. Devonian stromatolites from the Canning Basin. In Walter, m. r. (ed.). Stromatolites , 543-564. Elsevier, Amsterdam. rich, m. 1967. Donezella and Dvinella, widespread algae in Lower and Middle Pennsylvanian rocks in east- central Nevada and west-central Utah. J. Paleont. 41, 973-980. riding, r. 1977a. Problems of affinity in Palaeozoic calcareous algae. In flugel, e. (ed.). Fossil algae , recent results and developments, 202-2 1 1 . Springer, Berlin. 500 PALAEONTOLOGY, VOLUME 27 riding, R. 19776. Skeletal stromatolites. In flugel, e. (ed.). Fossil algae , recent results and developments , 57-60. Springer, Berlin. — 1979. Donezella bioherms in the Carboniferous of the southern Cantabrian Mountains, Spain. Bull. Centre Reck. Explor .-Prod. Elf- Aquitaine, 3, 787-794. and voronova, l. g. 1 982. Recent freshwater oscillatoriacean analogue of the Lower Palaeozoic calcareous alga Angulocellularia. Lethaia , 15, 105 114. (in prep.). Calcareous algae from the late Precambrian-early Cambrian Nemakit-Daldyn Formation, northern Siberia. and watts, n. 1981. Silurian algal reef crest in Gotland. Naturwissenschaften, 68, 91. rietschel, s. 1969. Die Receptaculiten. Senckenbergiana Lethaea, 50, 465-517. sepkoski, j. j. 1978. A kinetic model of Phanerozoic taxonomic diversity. I. Analysis of marine orders. Paleobiology , 4, 223-251. shuysky, v. p. 1973. Calcareous reef-building algae of the Lower Devonian of the Urals. Acad. Sci. USSR, Ural. Sci. Centre, Inst. Geol. Geochem. Nauka, Moscow. 156 pp. [In Russian.] wilson, j. L. 1975. Carbonate facies in geologic history. Springer, Berlin. 471 pp. wray, j. l. 1968. Late Palaeozoic pylloid algal limestones in the United States. Proc. 23 Intern. Geol. Congr., Prague , 8, 113-119. 1977. Calcareous algae. Elsevier, Amsterdam. 185 pp. Zhuravleva, i. t. and myagkova, e. i. 1981. Some data for studying Archeata. In sokolov, b. s. (ed.). Problematica of the Phanerozoic. Acad. Sci. USSR , Siberian Branch , Inst. Geol. GeophysA \ -74. [In Russian.] BORIS CHUVASHOV Institute of Geology and Geochemistry Ural Scientific Centre USSR Academy of Sciences 620219 Sverdlovsk USSR ROBERT RIDING Department of Geology University College Manuscript received 7 May 1983 Cardiff CF1 1XL Revised manuscript received 21 November 1983 UK PALAEOECOLOGY OF MARGINAL MARINE SEDIMENTARY CYCLES IN THE ALBIAN BEAR RIVER FORMATION OF SOUTH-WESTERN WYOMING by FRANZ T. FURSICH and ERLE G. KAUFFMAN Abstract. The Middle to Late Albian Bear River Formation of south-western Wyoming includes a cyclic sequence of fine-grained sediments with numerous shell beds comprised of abundant, low diversity fresh- and brackish-water faunas. These record the initial marine transgression of the Cretaceous in this region (Skull Creek-Kiowa tectono-eustatic cycle) and are interpreted as part of an extensive embayment with limited marine influence. Biostratinomic data suggest low rates of sedimentation, frequently shifting environments, and, in the coquinas, reworking by storms to account for mixing of meso- to oligohaline and freshwater faunas. Five discrete, repetitive benthic associations are documented for the freshwater and two associations with four subsets for the brackish-water facies. They were controlled in their distribution largely by substrate, temperature, and oxygen levels in freshwater and by substrate and salinity in brackish water. Size/frequency curves of brackish species document seasonal fluctuations in salinity (tertiary cycles). Repetitive successions of facies and faunas record regressive sequences (secondary cycles), whilst fluctuations in the relative dominance of fresh- or brackish-water conditions within bundles of regressive sequences reveal a still higher order of cyclicity (primary cycles) within the Bear River Formation. These cycles are partly of climatic origin, partly autocyclic. During the early Cretaceous (middle Late Albian), a narrow seaway which extended from the Proto-Gulf of Mexico to the Circum-Boreal Sea was first established in the Western Interior Basin of North America (Eicher 1960), and has been named the Skull Creek Seaway by McGookey et al. (1972, p. 200). Near its centre, grey to black, finely laminated, commonly organic-rich shale facies (e.g. Thermopolis Shale, Mowry Shale) characterize the seaway. Toward the west, these offshore marine facies grade into a zone of marginal facies representing deposition in estuaries, lagoons, and bays. These in turn grade further west into deposits of low-lying flood plains (text-fig. 1). Isotopic studies suggest that even in the centre of the seaway subnormal salinity existed during Skull Creek time. The Middle to Late Albian Bear River Formation represents the western marginal marine facies deposited early in transgression of the northern arm of the Skull Creek Sea (e.g. Young 1969) before a connection was established across Colorado with the warm-water incursion of the Interior Seaway from the Gulf of Mexico. The Bear River Formation comprises fluvial channel and overbank sandstones and, in its centre, lagoon or bay fill sequences of shales, siltstones, and silty carbonates. They contain, in places, low-diversity but highly abundant molluscan-dominated faunas indicative of fresh- and brackish-water conditions. The depositional environments of these fine-grained facies, and especially of the numerous small-scale sedimentary cycles within them, are poorly understood. The Bear River Formation overlies the non-marine Smoot Formation of the Gannett Group and is in turn overlain by marine shales of the Aspen Formation. Towards the south-east, along the depositional strike, rocks of the Bear River Formation grade into those of the Dakota Formation (e.g. Horstman 1966). Due to widespread thrusting, which commonly cuts the Bear River, the thickness of the formation is difficult to evaluate. It is usually given as 1 75 to 1750 m, depending on the region (e.g. Stanton 1892;Veatch 1907; Wilmarth 1938). The type section (Stanton 1892, fig. 1 ) is near the site of the former Bear River City, about 12 miles south-east of Evanston in south-western Wyoming. This section, first described by Meek (1873) and interpreted in more detail by Stanton [Palaeontology, Vol. 27, Part 3, 1984, pp. 501-536.] Palinspastic reconstruction after Royse et al. (1975). FURSICH AND KAUFFMAN: ALBIAN PALAEOECOLOG Y 503 (1892), is now very poorly and incompletely exposed. A largely complete, freshly exposed section exists along a gravel road leading south past Myers Reservoir (text-fig. 1)1-5 miles to the west of the type section and about 0-5 miles east of Bear River. This section, subsequently referred to as the Bear River Reference Section, displays in its middle to upper part thin coals, carbonaceous shales, siltstones, thin sandstones, and thin, resistant beds of silty limestone which contain well-preserved fossils representing the characteristic ‘Bear River fauna’ (Stanton 1892). Cyclic sedimentation and palaeoenvironments are reflected at three different levels of magnitude by both facies and faunas in this part of the section. Second-level cycles are most obvious. Within each secondary cycle the facies change up-section from shelly silts and silty limestones, to silty and carbonaceous shales, to coals, and finally to silty or marly freshwater limestones. The benthic fauna changes accordingly in composition from brackish-water-dominated associations to freshwater-dominated associations. It is the purpose of this paper to analyse the fauna ecologically, to describe in detail the cyclic changes of the facies and faunas, and to speculate on the origin of these cycles at all levels of development. HISTORY OF RESEARCH Although frequently discussed in the second half of the last century, little modern work has been done on the Bear River Formation. The rocks were discovered by Engelmann in 1859 near the mouth of the Sulphur Creek (type area), and first mentioned by Meek and Engelmann ( 1860). In 1869 Hayden gave them the name Bear River Group, after the nearby Bear River. In the following twenty-five years, there was a lively debate about the age of the rocks. Originally thought to be Eocene in age (Meek and Engelmann 1 860; Meek 1 860), their Cretaceous age was only gradually recognized (e.g. Engelmann 1876; White 1891). Stanton (1892) established beyond doubt the Cretaceous age of the rocks and their position below the Colorado Formation. White (1895) reviewed earlier literature, in particular the age controversy, described in detail the faunas, discussed aspects of their ecology, and concluded that the Bear River Formation was ‘deposited in a brackish water lake or sea which was more or less completely cut off from open marine waters’ (pp. 68-69). Stanton (1892) had already recognized the occurrence of the Bear River fauna at other localities in south- western Wyoming (e.g. Twin Creek, Ham’s Fork, Cokeville) (see also Veatch 1907). Later on rocks with similar faunas were discovered as far as northern Wyoming (e.g. LaRocque and Edwards 1954; Wanless et al. 1 955) and, in part, described as westward equivalents of the Bear River Formation (Rubey 1973). Due to the lack of index fossils the age question was not yet settled and, in the middle of this century, the discussion was continued by Yen (1952, 1954) who, noting the similarity of the Bear River Pyrgulifera to those from Cenomanian localities in Europe, assumed a Cenomanian age for the rock unit, while Cobban and Reeside (1952) assigned the Bear River Formation to the Middle Albian Inoceramus comancheanus zone. Today, a Middle to Late Albian age is generally accepted on grounds of facies relationships, and the Bear River Formation is thought to correspond in parts to the Thermopolis Shale (e.g. Eicher 1960; Haun and Barlow 1962; Young 1969; Kauffman et al. 1976). THE BEAR RIVER REFERENCE SECTION Along the dirt road leading from U.S. 1 50 toward Myers Reservoir, a large part of the Bear River Formation is exposed. The rocks are overturned and faulted several times so that it was not possible to measure one continuous section. The lower part of the exposure consists of greenish to reddish clays and silts with lenticular intercalations of silt- and sandstones and rare layers of silty limestone. Fossils are very rare and consist of poorly preserved freshwater gastropods. The environment is most likely a low-lying coastal or flood plain dotted with lakes and small streams. This sequence is overlain by a series of silts, shell beds, highly carbonaceous shales with thin coal-seams, and impure limestones (text-fig. 1). Fossils are abundant at numerous horizons, often forming text-fig. 2. Key to symbols used in text-figs. 1, 12-17, 19. KEY an COAL = FINE LAMINATION BENTONITE S' TROUGH CROSSBEDDING ED CLAY RIPPLE CROSSBEDDING ED SILT 'ZT' SHELLS n SAND T PLANT FRAGMENTS F -1 CARBONACEOUS UA ROOTS ra CALCAREOUS 1 1 BIOTURBATION 504 PALAEONTOLOGY, VOLUME 27 shell beds or beds of shell hash. Only the lower portion (42 m) and part of the upper half (20 m) of the fossil-rich strata were measured in detail, the remaining part being identical. The fossiliferous beds are finally overlain by unfossiliferous silty shales. Facies types. Fine-grained sandstones occur near the base of the measured section only. They have an erosive base and exhibit, in places, large-scale trough cross-bedding and small-scale ripple cross-bedding. Occasionally, levels with rip-up clasts are present near the base. The tops of the sandstones are rooted and occasionally bioturbated (small vertical tubes and Chondrites-like traces). Siltstones do not exhibit any sedimentary structures, but are sometimes bioturbated. They are usually rich in plant debris and occasionally contain scattered shells or thin shell bands. Silty clays and clayey silts are widespread and can be subdivided into the following types: (a) laminated shales indicative of little or no disturbance of the sediment/water interface. Rarely, thin shell bands are intercalated, (b) Carbonaceous silty clays and clayey silts with a high percentage of plant debris, (c) Blocky silty clay and clayey silt, unfossiliferous or with scattered shells only. All three types may contain thin bands of jarosite. Shell beds are very common and represent the following three types: (a) beds of shells and shell hash; the matrix is silty clay or clayey silt. Carbonized plant fragments are common in some beds. Shells are usually of brackish-water origin and largely disarticulated. ( b ) Beds of shelly silty limestone or calcareous siltstone. Most shells are of brackish-water origin. Disarticulated shells dominate, although sometimes individuals are found in life position. Beds of type a and b may have an erosive base and vary in thickness laterally, (c) Beds of shelly silty limestone or marly siltstone. Most shells are small gastropods of freshwater origin. Frequently the beds are highly carbonaceous and may contain coal fragments; occasionally they are rooted. Coal-seams several centimetres in thickness occur throughout the section. They usually alternate with thin layers of highly carbonaceous silty clay. Bentonite layers are common. In the measured section, twenty-three layers have been encountered, the thickest measuring 25 cm. Occasionally they have an irregular base. Facies sequence. The vertical succession of the various facies types is not random, but cyclic. As a rule, the base of each cycle is characterized by thin, bioclast-supported shelly limestone beds, less frequently by shelly silty clay. These may alternate with beds of silty clay and clayey sill in which shells are scattered or absent. At this level poorly fossiliferous siltstones may also be found. Then follows a zone of carbonaceous silty clays, sometimes laminated and often associated with thin coal-seams. Intercalated between these and often terminating the cycles are beds of mud-supported silty or marly limestones that frequently contain plant or lignite fragments and sometimes roots. The cycles range from 50 (a very incomplete cycle) to 500 cm in thickness, most of them being 150 to 300 cm thick. Fauna. The fauna of the Bear River Formation is generally of low diversity but high abundance and this is also true of the Bear River section. The faunas from the type locality and other localities were first described by Meek (1860, 1870a, b) and more fully by White (1895), but not all Meek’s species have been found (for list of species see Table 1). In addition, some are interpreted here as representing only variants of one and the same species. For example, Corbula engelmanni Meek (White 1895, p. 40, pi. 4, figs. 10-1 1) is clearly the juvenile form of Ursirivus pyriformis (Meek). Similarly the two species of Pyrgulifera ( P humerosa and P. stantoni ) described by White (1895, p. 55, pi. 8, figs. 1 11; pi. 9, figs. 1-8) are here considered only variants of the same species (as also recognized by White) with P. humerosa Meek having priority. A similar case can be made for Mesoneritina naticiformis (White) and M. stantoni (White), the latter being regarded as a junior synonym of the former. Later descriptions of the macrofauna, in particular the gastropods— although from other localities— are by Yen (1951, 1954). The microfauna and microflora were described by Jones (1893), White (1895), and more recently by Peck (1951) and Peck and Craig (1962). The latter authors list thirteen species of ostracods and charophytes from the Bear River Formation. Palynological investigations were carried out by Tingey (1978). In the Bear River section gastropods dominate in number of species (twenty) and in abundance. Both fresh- and brackish-water species are present, as is the case among the less diverse bivalves (seven species). In most cases, fresh- and brackish-water species are found in the same bed, but beds with only freshwater or only brackish-water faunal elements are also present. In many beds several species of ostracods and charophytes occur. It was soon recognized that the Bear River Formation represents a marginal marine environment and White ( 1 895) distinguished clearly between a freshwater and a brackish-water fauna. The assignment of ancient species to a certain salinity regime is particularly difficult in the case of the Bear River fauna where mixing plays a significant role. Criteria such as the preferred occurrence with undoubtedly freshwater forms (e.g. unionids) or brackish-marine forms (e.g. Crassostrea , Ursirivus , Brachidontes) were used to establish the broad salinity range FURSICH AND KAUFFMAN: ALBIAN PALAEOECOLOG Y 505 table 1. List of species found in the Bear River Formation at the Bear River Locality. Familial classification of gastropods based mainly on Yen (1951, 1954) Freshwater Bivalves: Unionidae Gastropods: Valvatidae Viviparidae Neritidae Amnicolidae Pleuroceratidae Cyclophoridae Lymnaeidae Brackish water Bivalves: Corbiculidae Corbulidae Mytilidae Ostreidae Gastropods: Pleuroceratidae Ellobiidae Loxopleurus belliplicatus (Meek) Protelliptio (Plesielliptio) vetustus (Meek) Valvata praecursoris (White) Lioplacodes stachei (White) Viviparus couesi White Campeloma macrospira Meek Mesoneritina naticiformis (White) Parateinostoma occultum (White) Parateinostoma latense (White) Parateinostoma cf. P. altispirale Yen Tornatellina?’ isoclina White Pachychiloides cleburni (White) Pachychiloides turriculus (White) Pachychiloides chrysalis (Meek) Pachychiloides chrysalloideus (White) Pachychiloides macilentus (White) Goniobasis sp. Pseudarinia sp. Lymnaea nitidula (Meek) Veloritina durkeei (Meek) Ursirivus pyriformis (Meek) corbulid sp. A Brachidontes multilinigera (Meek) Crassostrea soleniscus (Meek) Pyrgulifera burner osa (Meek) Rhytophorus meeki White Zaptychius haldemani (White) of doubtful species. Thus Pyrgulifera humerosa is regarded as a brackish species (in contrast to Yen 1952; LaRocque and Edwards 1954) as is Rhytophorus meeki (see Table 1). The vertical distribution of the fauna, like the sediment, reveals a cyclic pattern; within a cycle the relative abundance of freshwater forms in individual beds invariably increases towards the top. At the base, shell beds are dominated by brackish-water bivalves whilst at the top brackish-faunal elements are usually missing or do not constitute more than 5% of the fauna, the rest being small freshwater gastropods and unionid bivalves. SIGNIFICANCE OF ENVIRONMENTAL CONDENSATION FOR THE PALAEOECOLOGICAL INTERPRETATION OF MARGINAL MARINE FAUNAS Palaeosynecological interpretation of faunas requires that they have undergone only minimal disturbance. This is particularly true of marginal marine faunas where environments and faunas may change drastically across short lateral distances and transport lead to mixing of communities from different biotopes. On the other hand, relatively uniform marginal marine environments such as large protected lagoons and bays may exhibit only insignificant lateral faunal mixing (e.g. Peterson 1976). Of far greater importance in such environments is a process called environmental condensation (Fiirsich 1975), whereby faunas representing different environments in time are telescoped into one stratigraphic horizon. Prerequisite for such a process is a low rate of sedimentation and rapid change in environmental parameters such as seasonal or larger scale fluctuations in 506 PALAEONTOLOGY, VOLUME 27 salinity in connection with monsoonal type climates or variations in freshwater discharge of rivers. Under such circumstances fresh- and brackish-water to marine faunas, both autochthonous and partly even in life position, can become mingled in one single bed. The recognition of environmental condensation is relatively easy where ecologically incompatible faunas such as marine and freshwater elements are mixed. If this mixing involves faunas representing different brackish- water regimes, it is more difficult to recognize. Interpretation of such condensed faunas as relics of a single former community will lead to erroneous conclusions with regard to faunal composition, diversity, and evenness and consequently to incorrect ecological inferences (see also Peterson 1977). Recognition of environmental condensation is therefore crucial for any ecological analysis of Recent and ancient faunas. The following features may facilitate the recognition of environmental condensation: (1) differences in sediment fill of shells; (2) preservational differences (e.g. degree of abrasion, breakage, encrustation) where they are not due to differences in life habits (such as infaunal versus epifaunal); (3) a less pronounced repetition of assemblages; (4) ecological incompatibility; and (5) morphological differences among individuals of the same species, such as co-occurrence of dwarfed and normal-sized, or thin- and thick-shelled individuals. Environmental condensation played a significant role in shaping the faunal assemblages of the Bear River Formation. BIOSTRATINOMIC AND SYNECOLOGICAL ANALYSIS OF THE BENTHIC FAUNA Methods of study Most of the fossiliferous part of the exposure was measured in detail and eighty-six bulk samples representing over 1 0,000 specimens of the benthic fauna were collected. The samples were broken up in the laboratory and the fossils counted as described in Fursich (1977). In addition, the percentage of fragmentation among shells was noted and the right/left valve ratio calculated. Where possible, the size of the dominant faunal elements ( Ursirivus , Pyrgulifera, Veloritina) was measured. Two aspects of faunal diversity were calculated: species thxt-kig. 3. Typical example of in situ reworked shell bed. Bear River Reference Locality. Polished section, x 1. FURSICH AND KAUFFMAN: ALBIAN PALAEOECOLOG Y 507 richness, expressed by the number of species present, and evenness which was calculated as D = 1 \Zpi 2 whereby pt is the relative frequency of the /th species (MacArthur 1972, p. 197). As diagenetic distortion of 'the fauna can be excluded with confidence (aragonitic faunal elements are invariably preserved, although aragonite has been replaced by calcite or silica) and selective transport was not a major factor, the fossil assemblages can be regarded as relics of former communities. Samples from individual beds frequently contain both fresh- and brackish- water elements, which were analysed separately. Biostratinomy The most conspicuous features of the Bear River Formation are the shell beds. Varying in thickness between 3 and 40 cm, they are usually packed with shells of bivalves and gastropods (text-fig. 3). Most shells are disarticulated; only 4 to 12% of brackish-water bivalves in the various associations (see below) are still articulated. This figure is higher for freshwater unionids: over 90% may still be articulated in individual samples. Only very rarely are individuals encountered in life position: the unionids Loxopleurus and Protelliptio occur predominantly in growth positions within two beds, the corbulid Ursirivus in another. In most shell beds a large proportion of the fauna is fragmented, in particular small specimens. In shell beds dominated by brackish-water bivalves estimated frag- mentation percentage is generally in the range of 75 to 95; only rarely the percentage is 10 to 50. In contrast, shell beds dominated by freshwater elements have a lower fragmentation percentage, usually between 50 and 75, occasionally even zero. The dense packing in most shell beds is expressed by a variety of biofabrics. Particularly common are nesting (text-fig. 3) or small-scale oblique imbrication. In contrast to the usual random orientation of shells within the shell bed, shells exhibit a preferred convex-up orientation near the top. The size range of shells and shell fragments is very large in most shell beds and individual species show often bi- or even polymodal size/frequency distribution patterns. The right/left valve ratios of dominant faunal elements ( Ursirivus , V eloritina) are surprisingly close to 1 in most samples (text-fig. 4). The matrix of many shell beds is not homogeneous. For example, in a silt or clayey-silt matrix, silty carbonaceous clay may be found under shells, in pockets, or as thin discontinuous or rarely continuous layers. a O OO O OOOOGDO Og OD eg O OO O > < > LU To ' 12 ' 14 b •o o cm o o o • o 06 T 08 T 10 ' li" d 04 ' 16 ' 18 o 8 0 o O o o* * omtod o RIGHT VALVE o'.8 10 ' 12 ' 14 ' 0.8 1.0 1.2 1.4 1.6 1.8 • VELORITINA o URSIRIVUS text-fig. 4. Right/left valve ratios of Ursirivus pyriformis and Veloritina durkeei in samples representing four subsets of the brackish-water U. pyriformis association. Minimum count per sample, 25. a, U. pyriformis / Pyrgulifera humerosa subset; b , U. pyriformis/ V. durkeei subset; c, P. burner osa subset; <7, U. pyriformis/ Crassostrea soleniscus subset. 508 PALAEONTOLOGY, VOLUME 27 Discussion At a first glance the shell beds appear to have undergone extensive transport resulting in significant distortion of original benthic communities. The high percentage of fragmentation, the largely disarticulated valves, and mixing of fresh- and brackish-water faunas as well as biofabrics clearly indicate reworking of the shells. However, the biostratinomic data do not support extensive lateral transport: lack of size sorting, bimodal size/frequency distributions, and the right/left valve ratios all favour within habitat reworking. Faunal mixing was most likely caused by rapidly changing environmental conditions rather than by lateral mixing, and accentuated by local reworking. The case for environmental condensation is strengthened by the occurrence, in some of the shell beds, of 3- to 5-mm thick layers of silty clay that contain only freshwater species whilst the remainder of the bed consists of largely or only brackish-water species. During preparation of the shell beds, these thin layers could not be separated effectively from the rest, leading to samples that exhibit a mixed fauna. Another example of environmental condensation is a bed of shelly dark-grey silty clay in which large Pyrgulifera and articulated Ursirivus, some in life position, are found between a host of small disarticulated Ursirivus and freshwater gastropods. In this case, at least three different environmental situations are recorded: a freshwater environment represented by the Lioplacodes stachei association (see below); a marginal brackish environment represented by juvenile members of the Ursirivus pyriformis association that were killed off before reaching maturity; and a more favourable brackish environment in which members of the U. pyriformis association reached normal adult size. Differences in the matrix and faunal composition within single shell beds show that they represent a period of time during which sedimentation regime and environmental conditions changed at least once, if not several times. The extensive reworking was most likely caused by waves in connection with storms. Accordingly, the shell beds are interpreted to represent an environment below fair weather, but above storm-wave base, and thus not exposed to constant reworking as evidenced by the lack of widespread abrasion and the, albeit rare, individuals preserved in life position. As the biostratinomic analysis shows, the fauna has not undergone significant lateral transport and can be taken to represent relics of former benthic communities. Two problems, however, remain. First, how far did the extensive breakage act selectively, thereby distorting the original relative abundance of taxa with hard parts? Apart from Brachidontes multilinigera and Crassostrea soleniscus all brackish-water species have relatively thick shells and even small specimens appear fairly sturdy. The likelihood of fragmentation not only depends on shell size and thickness, however, but also on shell structure and crystal size, and it is very difficult to evaluate the combination of these features with regard to breakage. Observations on shell beds indicate, as one would expect, that small individuals have indeed undergone more breakage than large ones (see Hallam 1967) and thus distorted size frequency curves to some degree. The fact that in twenty out of twenty-four samples of the U. pyriformis association right valves are more numerous than left valves, although usually only barely so, probably does not reflect selective transport as the species is nearly equivalved; it is more likely that left valves were slightly more prone to fragmentation than right valves. No differences with regard to fragmentation were noted between Veloritina , Ursirivus , and Pyrgulifera. In the case of the extremely thin Brachidontes and the thin to moderately thick Crassostrea , differential breakage was taken into account when the relative abundance of species was established. Nearly all freshwater gastropods were small and thin shelled, in contrast to the large and thick-shelled unionids. However, in both groups the percentage of fragmented shells was relatively low and apparently none of the two groups experienced preferential breakage, except that in most gastropods parts of the aperture and last whorl were damaged. The second problem is the vertical mixing of different communities which may drastically alter faunal composition and diversity (e.g. Peterson 1977; Fiirsich 1978). The fresh- and brackish-water faunas are separated relatively easily, but what about environmental condensation within the fresh- or brackish-water regime? Of the eight brackish-water species only three ( Ursirivus , Veloritina , and Pyrgulifera) are abundant, two more occur in moderate to low numbers ( Brachidontes , Crassostrea), whilst the remaining three are rare. The three abundant species and Crassostrea recur in sets with different relative abundances and these sets have been grouped into associations. The low and fairly FURSICH AND KAUFFMAN: ALBIAN P AL AEOECOLOG Y 509 constant diversity, the lack of preservational differences, and ecological links between species (e.g. the large size of Veloritina in the U. pyriformisjC. soleniscus subset as opposed to its predominantly small size in the P. humerosa subset) do not favour extensive environmental condensation within the brackish regime. Furthermore, the salinity range represented by the brackish fauna is thought to be relatively small. In the freshwater fauna five recurring sets of species were recognized, four of which are closely related and differ largely in the relative abundance of dominant species. There is no evidence to suggest or discount condensation of different faunas. THE BRACKISH-WATER FAUNA As has been demonstrated above, the benthic fauna of the Bear River section represents relics of former communities. In the following, five repetitive sets belonging to two associations (Table 2) are table 2. Composition of the brackish-water associations and subsets. EC— epifaunal cemented; SI— semi-infaunal; SHI — shallow infaunal; EM — epifaunal mobile; S— suspension-feeder; H- herbivore; HD — herbivorous detritus-feeder. Rel. abundance Presence percentage Rank position Life habit Feeding mode A. Crassostrea soleniscus association (1 sample; 73 specimens) C. soleniscus 100 100 10 EC S Ursirivus pyriformis association: B. U. pyriformis! C. soleniscus subset ( 1 1 samples; 1054 specimens) U. pyriformis 660 100 M SHI S Veloritina durkeei 15-4 100 2-4 SHI S Pyrgulifera humerosa 13-5 100 2-7 EM HD C. soleniscus 3-4 100 3-8 EC S Brachidontes multilinigera 0-8 18-2 SI S C. U. pyriformisj V. durkeei subset (8 samples; 907 specimens) U. pyriformis 67-8 100 M SHI s V. durkeei 21-3 100 1-9 SHI s P. humerosa 10-2 87-5 30 EM HD B. multilinigera 0-4 25 SI S Rhytophorus meeki 01 12-5 EM H corbulid sp. A 01 12-5 SHI S D. U. pyriformisj P. humerosa subset (24 samples; 2,1 54 specimens) U. pyriformis 69-6 100 10 SHI S P. humerosa 22-1 100 20 EM HD V. durkeei 7-4 70-8 31 SHI S B. multilinigera 1-3 20-8 SI S corbulid sp. A 0-2 4-1 SHI s Zaptychius haldemani 004 4-1 EM H R. meeki 004 41 EM H E. P. humerosa subset (13 samples; 761 specimens) P. humerosa 75-9 100 10 EM HD U. pyriformis 201 92-3 2-2 SHI S V. durkeei 4-2 76-9 3-2 SHI s B. multilinigera 2-9 38-4 SI s R. meeki 0-6 15-4 EM H corbulid sp. A 0-3 7-7 SHI S 510 PALAEONTOLOGY, VOLUME 27 briefly described and interpreted ecologically with particular regard to their salinity ranges. The sets were defined by using presence/absence as well as relative abundance data. The dominant faunal elements are shown in text-fig. 5. The Ursirivus pyriformis association The U. pyril'ormis/C. soleniscus subset. Represented by eleven samples and 1,054 specimens, this subset is dominated numerically by the corbulid U . pyriformis, followed by the corbiculid bivalve V. durkeei and the gastropod P. burner osa. C. soleniscus represents only 3 -4 % of the fauna, but is present in each sample. The mean number of species is 4-3. Only 4-4% of the bivalves are articulated and fragmentation varies between 75 and 95%. The subset occurs in shell beds; the matrix ranges from clayey silt to silty micrite. In one sample, coal fragments were common. The mean right/left valve ratio was 118 for both Ursirivus and Veloritina (text-fig. 4). The size distribution of Ursirivus was measured in four samples and was invariably bimodai with one peak at 4-6 mm height and the other, broader, at 18-24 mm. Both peaks were pronounced in three cases, while in the fourth the larger size range dominated by far. For Veloritina and Pyrgulifera , the size distribution could be established in one case each; the two species also exhibit a bimodai distribution pattern with peaks at 3-6 and 15-21 mm in diameter (Pyrgulifera), and 5-10 and 25-40 mm height ( Veloritina). In both cases the larger size group dominated. Veloritina reaches a relatively large size in most samples. Brachidontes , occurring only in two samples, reached its largest size within the Bear River section within one sample, but is small in the second. In the eleven samples of the subset, freshwater faunal elements ranged from 0 to 25% of the total fauna with a mean of 6-6%. The U. pyriformis/V. durkeei subset. A second subset occurring in eight samples with 907 specimens was also dominated by U. pyriformis, followed by V. durkeei. P. humerosa occurs in seven of the eight samples and constitutes 10-2% in terms of relative abundance. The remaining three species (B. multilinigera, Rhytophorus meeki, and the corbulid sp. A) are rare and occur only in few of the samples. The mean number of species in the association is 3-4. Of the bivalves, 1 1% are articulated; the fragmentation percentage ranges from 50 to 95 and is 90 for most samples. The sediment is either clayey silt, silty clay, or silty limestone (in one case with plant fragments) and the shell density is high in all cases. The mean right/left valve ratio is higher than in the preceding subset, both for Ursirivus ( 1 26) and Veloritina (T51), in the latter case possibly indicative of either selective transport or breakage. text-fig. 5. Dominant brackish-faunal elements of the Bear River Formation, a, Ursirivus pyriformis (Meek); b, Veloritina durkeei (Meek); c, Brachidontes multilinigera (Meek); d, Pyrgulifera humerosa (Meek); all x 1. Bear River Reference Locality. FURSICH AND KAUFFMAN: ALBIAN PALAEOECOLOG Y 511 The size/frequency distribution of Ursirivus is bimodal in four cases with the smaller size group usually more pronounced; in one case, only one peak, at 3-6 mm is present. A similar pattern (one bimodal, one unimodal) is exhibited by Veloritina. Brachidontes is small to tiny in all samples. Veloritina exhibits a wide size range except in one sample where all individuals are tiny. In this particular sample, all other faunal elements are equally of small size. The percentage of freshwater elements in the total fauna of each sample ranges from 0 to 60 with a mean of 131. The U. pyriformis/P. humerosa subset. The U. pyriformisjP . humerosa subset is very widespread in the Bear River section, represented by twenty-four samples and over 2,000 individuals. U. pyriformis is by far the dominant species followed by P. humerosa and V. durkeei. The remaining four species are rare (corbulid sp. A and the gastropods Zaptychius haldemani and R. meeki ) or encountered occasionally ( B . multilinigera ). The mean number of species is 31. Of the bivalves, 1 1 -5% are articulated; percentage of fragmentation varies from 30 to 95 and in most samples is close to the latter. In one sample, U . pyriformis was found in life position. The mean right/left valve ratio of specimens of Ursirivus is 1 ■ 18 with individual values ranging from TO to T41 (text-fig. 4). The sediment range in which the U. pyriformis subset occurs is like that of the two preceding subsets except that about 16% of the samples occur in silt. About 20% of the sediments are carbonaceous containing either plant debris or wood fragments. The size/frequency distribution of Ursirivus could be established in eighteen samples. Except in one case, where only small individuals are present, the curves are bimodal. Both size clusters are either roughly equal or show a dominance of smaller forms. The size/frequency distribution of Pyrgulifera was unimodal in one case, most specimens belonging to the 12 to 21 mm size range, and polymodal in a second case with peaks at 3-6 mm (pronounced), 12-15 mm, and 24-27 mm in diameter. Ursirivus and Pyrgulifera are generally large, whilst Veloritina is small in some samples and occurs in a wide size range in others. In some samples, all specimens of Ursirivus are small as are all other faunal elements. Brachidontes is tiny in one sample, but relatively large in others. The relative abundance of freshwater elements in samples ranges from 0 to 89-5% with a mean of 19-9%. The P. humerosa subset. Seven hundred and sixty-one specimens in thirteen samples constitute the subset which is strongly dominated by the pleuroceratid gastropod P. humerosa. U. pyriformis is also common, occurring in all but one sample. The remaining four species ( V. durkeei , B. multilinigera , R. meeki , and corbulid sp. A) are all uncommon or rare. In individual samples, two to five species occur; the mean number of species is 3.3. Of the bivalves, only 9-5% are articulated. The right/left valve ratio of Ursirivus , measured in five samples, is extremely close to one (mean IT) with individual values ranging from TO to T33. The percentage of fragmented shells varies from 10 to 95 but is usually between 50 and 90. Most samples occur in argillaceous silt or calcareous siltstone; some are found in silt, silty clay, or siliceous fine sandy siltstone. Two samples were rich in plant debris; bioturbation (Chondrites- like burrows) was encountered in one case. In three samples, the size/frequency curves of Ursirivus were bimodal, with small specimens strongly dominating. In contrast, Pyrgulifera was represented predominantly by large specimens, three out of the four samples having a bimodal distribution pattern, the fourth a unimodal one. In general, specimens of Ursirivus often did not reach maximum size. Specimens of Brachidontes and Veloritina are small or tiny in all except one sample. Freshwater elements constitute from T5 to 87% of the total fauna of individual samples with a mean of 51 -6. The Crassostrea soleniscus association Only represented by seventy-three specimens in one sample, the C. soleniscus association nevertheless appears to represent a true recurring association, as it has been mentioned from several horizons at different localities in the Bear River Formation, usually in thin, mono-, or near monotypic shell layers. At the Bear River locality, C. soleniscus alone occurs near the top of the measured section in a silty clayey limestone. The oysters occur in thin patches not more than 10 cm high and are preserved in situ. Over two-thirds of the shells are articulated and encrust each other. Life orientation of the specimens was horizontal to oblique. The shells of Crassostrea are thin to moderately thick and relatively small (most specimens being less than 7 cm in height). In contrast to most Recent or fossil oyster patch reefs, the shells are not encrusted or bored. There is no freshwater fauna associated with the oysters. Discussion Apart from the monotypic C. soleniscus association, the remaining four sets of species are similar in composition and are best grouped within a single association, the Ursirivus pyriformis association. 512 PALAEONTOLOGY, VOLUME 27 Within this major association, however, the relative abundance of individual species varies considerably. Samples in which Ursirivus strongly dominates are found as well as samples in which Pyrgulifera accounts for more than 80% of the fauna. Crassostrea occurs only in some samples. Veloritina accounts for a quarter of the individuals in some samples, but is rare in others. Among marine ecologists there are two schools: one which regards marine benthic communities as discrete entities with sharp boundaries, and one which recognizes only species gradients and regards communities as artificial subdivisions of such gradients (e.g. Mills 1 969). Community boundaries are usually shaped by the nature of environmental gradients. Sharp gradients such as the sudden change from a soft to a hard substrate will result in discrete community boundaries, weak gradients in turn will result in gradual replacement of species and therefore at most in blurred boundaries. Within the brackish-water faunas from the Bear River relative abundance and, to some extent, species composition changed along a gradient of an overriding ecological parameter. In an attempt to learn more about this parameter, the U. pyriformis association was subdivided into four subsets which are artificial in that they grade into each other, but are interpreted as occupying different positions along a continuous environmental gradient. Thus, we do not suggest that these subsets represent relics of discrete communities. They are simply the means by which we can illustrate changes in community composition and structure along an environmental gradient. Indeed, that these subsets appear to have occupied different positions along an environmental gradient is supported by the fact that size/frequency distribution, size range, diversity, and percentage of freshwater elements in samples differ systematically between the various subsets. Autecology. All species could apparently tolerate low salinities as indicated by their close association with non-marine sediments and faunas. Ursirivus pyriformis (text-fig. 5a) is a nearly equivalve corbulid with a tapering posterior. This, and the presence of a shallow pallial sinus (Vokes 1945) suggests that the bivalve lived as a shallow burrower in the sediment with the anteroposterior axis in a more or less vertical position. Like other corbulids it was a suspension-feeder. Corbulids have several adaptations (such as being able to close their valves very tightly) to withstand environmental fluctuations, especially with regard to salinity, oxygen level, and temperature (e.g. Lewy and Samtleben 1979) and are consequently eurytopic. This is true not only of Recent, but also of fossil species which are particularly common in marginal marine environments (e.g. Fiirsich 1981) where they typically occur in large numbers in low diversity assemblages. Being slow burrowers they are preferentially found in low-energy environments with fine-grained substrates, and this was clearly also the preferred habitat of U. pyriformis. Corbulid sp. A is a small, 3-4 mm long species that exhibits the sharp posterior ridge seen in many corbulid species. As no hinge line was seen, a generic designation was impossible. Most likely this fairly rare species lived, like other corbulids, as a shallow infaunal suspension-feeder. The corbiculid bivalve V. durkeei (text-fig. 5b) belongs to a family whose Recent members either five in fresh or brackish waters. The pallial fine of this trigonal species is posteriorly truncated and the species most likely possessed a pair of short siphons and lived as a suspension-feeder close to the depositional interface. Species of Veloritina and Corbicula are characteristic of marginal marine environments elsewhere in the Cretaceous of the Western Interior Basin (e.g. in the Fox Hills Formation of northern Colorado) and may there form monotypic shell beds. In the Bear River Formation, V. durkeei is generally less abundant than Ursirivus or Pyrgulifera. The oyster Crassostrea soleniscus lived in small clusters as an epifaunal-cemented suspension- feeder. Elsewhere (e.g. in the Cenomanian Woodbine Formation of Texas; Stephenson 1952) it reaches a considerably larger size and may form extensive patch reefs. The relatively small and thin valves of the Bear River occurrence as well as the small size of the patches most likely indicate that there the species lived close to the limit of its environmental range. B. multilinigera (text-fig. 5c) is extremely thin-shelled and consequently frequently fragmented. Judging from its cross-sectional shape, it appears to have lived semi-infaunally as an endobyssate suspension-feeder. The small to tiny size of most specimens compared with occurrences elsewhere in the Cretaceous suggest a largely unfavourable environment for this species. Of the three gastropods regarded as having lived in brackish rather than freshwater, P. humerosa FURSICH AND KAUFFMAN: ALBIAN PAL AEOECOLOGY 513 (text-fig. 5 d) is by far the most abundant, whilst R. meeki and Z. haldemani are rare. P. humerosa most likely fed on plant detritus. Little is known about the ecology of Mesozoic gastropods and more detailed interpretation would be speculative. Nature of environmental gradients. It is clear that the fauna lived on or in a moderately soft substrate. No specific substrate preferences have been noted, but all associations occur in a range of fine-grained substrates, and individual species appear to have been fairly eurytopic. The scarcity of sedimentary structures, the presence of laminated shales, and the fine-grained nature of the sediments indicate a low-energy environment except when, during storms, the wave base was lowered and shells accumulated in beds. This does not necessarily imply a great water depth. On the contrary, the presence of root horizons and coal layers suggests deposition in water depth of less than 5 m for most of the time. Other environmental parameters exerting a major influence on the distribution of benthic faunas are the oxygen level, variations in food supply, temperature, and salinity. There are no indications of anoxic conditions at or near the sediment/water interface with the possible exception of the laminated, often carbonaceous silty clays and clayey silts that are usually devoid of fauna and may well represent periods of poor oxygenation. Variations in temperature are difficult to assess, but may have been pronounced in the shallow extensive water body represented by parts of the Bear River Formation. Such variations would favour eurytopic, opportunistic species, but exclude most stenolopic forms. Variations in food supply, characteristic of estuaries, are apparently less pronounced in lagoons, which are frequently rich in nutrients (e.g. Mee 1978). Marginal marine environments are strongly influenced by extreme salinity values or salinity fluctuations. The close association of freshwater and marine species in the Bear River Formation shows that salinity was the major factor controlling distribution and growth of the benthic fauna possibly amplified by fluctuations in temperature, resulting in a high stress environment. Coal beds, freshwater gastropods, unionid bivalves, and the absence of features indicative of hypersaline conditions suggest that salinity values were lower than normal marine and that one end of the salinity spectrum was represented by the freshwater environment. As stenohaline faunal elements are totally lacking (this is true also of the microfauna where foraminifera are absent and only fresh- and brackish-water ostracods occur) the highest salinity values probably were considerably below normal marine values. This is not surprising as even the offshore Skull Creek Seaway supported only a restricted fauna possibly influenced by low salinities (e.g. Eicher 1962). Considering the very low species diversity of all samples, it is safe to assume that the faunas did not live in waters of a salinity much higher than the mid-mesohaline regime (about 12%). Distribution of the fauna along the salinity gradient . Within meso- and oligohaline regimes of Recent estuaries, the fauna can be classified as euryhaline opportunists or estuarine endemics with freshwater species encroaching within the 1 to 2%0 range (Boesch 1977). Considering that U. pyriformis, V. durkeei , and P. humerosa are confined to the marginal marine Bear River environments and are not known from anywhere else in North America, it is most likely that they were endemic within this salinity range and specifically adapted to life in lowered and fluctuating salinity regimes. This is corroborated by the fact that all these species reach a large size and are fairly thick shelled. In contrast, B. multilinigera and C. soleniscus are known to occur in more saline waters elsewhere and, when occurring in the Bear River Formation, are relatively small and thin-shelled. They may be more correctly classified as euryhaline opportunists close to their environmental limit. Using information on absolute size of individual taxa in connection with data on species richness and evenness, it is possible to arrange the five sets of species along a salinity gradient (text-fig. 6). Accordingly, the U. pyriformis) C. soleniscus subset occupies the upper end of the salinity scale, followed by the U. pyriformis/ V. durkeei and U. pyriformis/ P. humerosa subsets. These three subsets and the monospecific C. soleniscus association are thought to have occupied the lower part of the mesohaline regime, whilst the P. humerosa subset occupied the oligohaline regime. Within this proposed sequence the evenness declines as does species richness (except in the case of the P. humerosa subset). Ursirivus and Veloritina decrease in maximum size towards lower salinity whereas the size of Pyrgulifera remains unchanged. The proposed arrangement of associations is 514 PALAEONTOLOGY, VOLUME 27 corroborated by the increase of the mean percentage of freshwater elements along the gradient, thus substantiating the decrease of marine influence. The exact position, along this gradient, of the C. soleniscus association is difficult to establish. According to its species richness and evenness values, it would have to occupy the zone closest to the freshwater edge. However, this seems unlikely, as Crassostrea was not found outside the U. pyriformisjC. soleniscus subset. The monospecific oyster patches more likely occupied a position near the upper part of the salinity range. The lack of other species may be caused by other factors than salinity. For example, the biogenic hard substrate would have prevented burrowers such as Veloritina and Ursirivus. Lowered salinities versus salinity fluctuations: the significance of size/ frequency curves. Recent environments rarely exhibit long-term stable salinity reductions. Fluctuations may be seasonal (variations in river discharge, monsoon-like rainy seasons), diurnal (caused by tides), or random (e.g. caused by tropical rain storms, removal of barriers). Interpreting size/frequency distribution patterns of dominant taxa, the wave length and amplitude of such fluctuations can be estimated. Text-fig. 7 shows variations in the size/frequency pattern of Ursirivus. Veloritina and Pyrgulifera show similar curves. The pattern is bimodal or unimodal and small or large individuals may dominate. The first mode is in the size classes 0-9 mm ( Veloritina ) or 3-9 mm (Ursirivus and Pyrgulifera ), whilst the second mode lies between 24 and 40 mm ( Veloritina ), 15-27 mm ( Ursirivus ), and 12-21 mm (Pyrgulifera). The second size group is nearly always broader than the first one. The first mode most likely represents juveniles, whilst the second mode reflects adult populations. The size/frequency distributions do not include post-larval mortality, as those shells were too small to be recovered during the mechanical breaking up of the samples. Some individuals may be stunted, exhibiting extreme crowding of growth lines, but the majority are not. Compared with data from Recent bivalves (e.g. Hallam 1967) the smaller size group most likely represents one season’s growth. Samples with only the smaller size group present thus consist of individuals that were killed after roughly one year. The reason for the death was probably a drastic reduction in salinity and this would imply strong seasonal salinity fluctuations. Size/frequency distribution patterns with a strong dominance of juveniles in the population can also be caused by biological factors, in particular predation. An example to indicate that predation may play a significant role in marginal marine environments is the high abundance of blue crabs feeding on oyster banks in Chesapeake Bay (Levinton 1982, p. 345). Within the Bear River faunas signs of predation are extremely rare (some cases of repaired shells have been found). Predators such as starfish are unrepresented by isolated skeletal elements. The second important group preying on mollusks are crustaceans, in particular crabs. They usually leave marks on the shell when prying them open which should be preserved in the fossil record. Apart from the very rare cases of repaired shells— obviously representing unsuccessful attacks by predators such as crabs— no shell damage indicative of predation was encountered. Another indication that at least some individuals were not killed by predation is that they are still articulated. Finally, the occasional occurrence of severe growth restrictions in larger individuals at the size represented by the juvenile peak (text-fig. 8) strongly suggests that environmental rather than biological factors caused the death of many individuals after one season’s growth. Another explanation of the size/frequency pattern would be migration of individuals from a nursery area to their adult habitat. This is apparently common among shallow water and intertidal text-fig. 6. Inferred distribution of brackish-water associations along a salinity gradient. Note that species richness, evenness, and size of dominant species decline, whilst the relative abundance of freshwater elements in individual beds increases. For species richness and evenness both ranges and means are given. CR.: Crassostrea soleniscus association; Ursirivus pyriformis association — urs./cr.: U. pyriformisjC. soleniscus subset; urs./vel.: U. pyriformis/ Veloritina durkeei subset; urs./pyr.: U. pyriformis/ Pyrgulifera humerosa subset; pyrgulifera: P. humerosa subset. FURSICH AND KAUFFMAN: ALBIAN P A L A EOECO LOG Y 515 nsp Veloritina ID N Ursirivus c/) Pyrgulifera smal I medium associations CR. URS./CR. URS./VEL. URS./PYR. PYRGULIFERA 20 15 I m e s o h a -*ifniMniiiy 10 i n e SALINITY 5 0 %o I oligohaline I 516 16 8 8 8 16 8 24 16- 8- 40 32 24 16 8 24 16 8 TEXl fresl at ri PALAEONTOLOGY, VOLUME 27 12 38 —i 1 1 1 1 1 1 15 18 21 24 27 30mm 3 < — 1 — > — i — 1 — 100 50 FRESHWATER FAUNA 07- fig. 7. Examples of size/frequency histograms of Ursirivus pyriformis related to the percentage of water faunal elements in individual samples. Each histogram is representative of several samples (numbers ht side of text-fig). Note that the relative increase of large individuals is matched by a decrease in the relative abundance of freshwater faunal elements. FORSICH AND KAUFFMAN: ALBIAN PA L AEOECOLOG Y 517 text-fig. 8. Growth restriction in Ursirivus pyriformis. a, juvenile specimen, x 3-5; b, c, adult specimens, x 3-2. Bear River Reference Locality. species and by no means confined to mobile organisms (for recent summary see Cadee 1982). However, the clear relationship between size/frequency distribution and influence of freshwater (expressed by the percentage of freshwater faunal elements; see text-figs. 7 and 9) makes salinity the overriding factor explaining mortality patterns of the brackish-water mollusks. Some samples exhibit a unimodal distribution at the adult size range or a bimodal distribution with the adult population strongly dominating. In this case, the salinity fluctuation was apparently not severe enough to lead to the death of all animals. However, these samples have probably undergone some differential breakage resulting in an under-representation of small individuals. Numerous intermediate cases illustrate variations in the amplitude of the salinity fluctuations. No samples represent single generations but hundreds or even thousands of generations, and what we see is therefore the cumulative effect of salinity fluctuations of varying amplitudes. In order to kill off a well-adapted brackish-water fauna, freshwater influx must have been very high. This can be demonstrated by the presence of an essentially autochthonous freshwater fauna. As one would expect, there is a relation between the percentage of freshwater elements and the size distribution pattern (text-fig. 7). Whilst the curves in text-fig. 7 show signs of bias— the lack or scarcity of juveniles in text-fig. la-c is difficult to explain in samples that represent numerous generations (e.g. Hallam 1972) — the cumulative histograms for Ursirivus , Veloritina, and Pyrgulij era from beds with different proportions of fresh- and brackish-water elements (text-fig. 9) are more realistic. Along a gradient representing a decrease in the relative abundance of freshwater faunal elements, the relative proportion, in the histograms, of larger size groups representing adults increases. Assuming the same degree of distortion, in all samples, by under-representation of small individuals, this would indicate that conditions for survival into adulthood were clearly more favourable in beds that exhibit no or only little freshwater influence. The main factor governing size frequency distribution of the brackish-water species would therefore be fluctuations in salinity. Assuming salinity fluctuations to be largely seasonal, the following model is proposed (text-fig. 10): If the amplitude of salinity fluctuations were fairly high, ranging from freshwater to the mesohaline regime, the brackish fauna would not be able to survive and be represented only by juveniles. Were the fluctuations less pronounced, the brackish-water fauna might be able to survive seasonally reduced salinity values and be represented by juveniles and adults. Less pronounced fluctuations near the brackish/freshwater interface may also favour the establishment of a freshwater fauna over a longer period of time (text-fig. 10b). Text-fig. 10c presents a model of the water body most easily 518 PALAEONTOLOGY, VOLUME 27 URSIRIVUS 50 um text-fig. 9. Cumulative size/frequency histograms of the dominant brackish-water species arranged according to percentage of freshwater faunal elements in the same bed. FURSICH AND KAUFFMAN: ALBIAN PALAEOECOLOG Y 519 text-fig. 10. Model of the relationship between seasonal salinity fluctuations and size/frequency distribution of brackish-water species in an extensive embayment. a, model of salinity fluctuations and resulting histogram indicative of complete juvenile mortality; b, model of less regular salinity fluctuations resulting in histogram indicative of ‘normal’ juvenile mortality; c, palaeogeographic model of salinity zonation within an extensive embayment lacking barriers. 520 PALAEONTOLOGY, VOLUME 27 producing such distribution patterns: an extensive embayment is connected with the open sea at one end, while rivers enter it at the other. It is plausible to assume a predominantly freshwater mass near the river mouths and in neighbouring shallow areas— provided the climate is humid to subhumid. In contrast, a predominantly brackish-water mass would characterize the region close to the open sea and in the deeper parts of the embayment (Barnes 1 980). In between these two areas there would be a zone in which salinity fluctuates greatly from brackish to fresh, depending on the amount of the seasonal freshwater input (be it by rivers or rain). Considering that, during Middle Albian time, the Skull Creek Seaway did not connect southwards to the Proto-Gulf of Mexico, but the area in south- western Wyoming represented a cul-de-sac (text-fig. 1) with generally lowered salinity values even offshore, it is not necessary to invoke the existence of extensive barrier island systems to close off the highly brackish to fresh embayment from the open sea. It is envisaged that a salinity gradient without barriers could have been effective provided the freshwater influx was high and mixing of water masses was insignificant. THE FRESHWATER FAUNA The freshwater fauna is represented by small, thin-shelled gastropods and large, thick-shelled unionid bivalves. Five associations and one assemblage have been recognized on grounds of presence/absence and relative abundance of important taxa (Table 3). A modified x2 test performed on the relative abundances indicates that the associations are significantly different from each other at the 99% level. Dominant faunal elements are shown in text-fig. 1 1. text-fig. 1 1 . Dominant faunal elements of the freshwater, a, Loxopleurus belliplicatus ( Meek), x 1 ; b, Protelliptio vetustus (Meek ), x l;c, Lioplacodes stachei (White), x 3; d, Mesoneritina naticiformis (White), x 3. Bear River Reference Locality. FORSICH AND KAUFFMAN: ALBIAN P A L A EOECOLOG Y 521 table 3. Composition of the five freshwater associations. EM— epifaunal mobile; SI — semi- infaunal; H — herbivore; S — suspension-feeder. Relative abundance Presence percentage Rank position Life habit Feeding mode A. Lioplacodes stachei association (6 samples; 815 specimens) L. stachei 84-8 100 10 EM H Mesoneritina naticiformis 4-4 100 2-8 EM H Protelliptio vetustus 4-8 50 SI S Campeloma macrospira 3-4 33-3 EM H Loxopleurus belliplicatus 11 16-7 SI S Parateinostoma occultum 0-5 33-3 EM H ‘ Tornatellina isoclina 0-5 33-3 EM H Lymnaea nitidula 0-4 33-3 EM H Parateinostoma latense 01 16-7 EM H B. Lioplacodes stachei! P. occultum association (3 samples, 477 specimens) L. stachei 68-8 100 10 EM H Pachychiloides macilentus 21 -9 100 2-3 EM H Parateinostoma occultum 5-2 100 2-7 EM H P. latense 2-9 100 4-0 EM H Valvata praecursoris 0-2 33 EM H M. naticiformis 0-2 33 EM H Pachychiloides cleburni 0-2 33 EM H Protelliptio vetustus 0-2 33 EM H C. L. stachei! Pachychiloides macilentus association (14 samples, 2,607 specimens) L. stachei 711 100 1-2 EM H Pachychiloides macilentus 9-9 100 2-8 EM H M. naticiformis 10-3 85-7 3-4 EM H Protelliptio vetustus 2-5 64-3 SI S V. praecursoris 2-1 57-1 EM H Goniobasis sp. 0-8 14-3 EM H C. macrospira 0-7 42-8 EM H Loxopleurus belliplicatus 0-6 64-3 SI S Viviparus couesi 0-5 21-4 EM H Lymnaea nitidula 0-4 28-5 EM H Parateinostoma occultum 0-3 14-3 EM H P. latense 0-3 21-4 EM H Pachychiloides cleburni 01 7-1 EM H D. M. naticiformis! L. stachei association (8 samples, 1,016 specimens) M. naticiformis 701 100 1-0 EM H Lioplacodes stachei 191 100 2-1 EM H P. macilentus 4-9 100 3-1 EM H C. macrospira 1-4 25 EM H Protelliptio vetustus 1-2 37-5 SI S Pachychiloides cleburni 11 25 EM H Loxopleurus belliplicatus 0-3 25 SI S V. couesi 0-3 25 EM H P. chrysalis 0-3 25 EM H Lymnaea nitidula 0-2 12-5 EM H Parateinostoma occultum 0-2 12-5 EM H E. Pachychiloides chrysalis association (4 sam pies; 692 specimens) P. chrysalis 53-5 100 1-5 EM H M . naticiformis 19 9 100 2-2 EM H P. cleburni 19-4 100 2-5 EM H P. macilentus 7-2 100 3-7 EM H P. chrysalloideus 10 25 EM H Lioplacodes stachei 0-4 50 EM H 522 PALAEONTOLOGY, VOLUME 27 The Lioplacodes stachei association In six samples with 8 1 5 specimens, the viviparid gastropod L. stachei represents over 80% of the fauna. Six other gastropods occur in low to moderate numbers ( Mesoneritina naticiformis, Campeloma macrospira, Para- teinostoma la lensc\ P. occultum, Lymnaea nitidula , ‘ Tornatellina ’ isoclina), only one of them (A/, naticiformis) in all collections. Apart from the viviparid C. macrospira all gastropods are small to tiny (most of them not exceeding 1 cm in height). In contrast, the two species of unionids, Protelliptio vetustus and Loxopleurus belliplicatus , are large, measuring between 4 and 6 cm in height. They are more abundant than in other freshwater associations accounting for 4-8% (P. vetustus) and 11% ( L . belliplicatus) in terms of relative abundance. The number of species varies from three to six; the mean number of species is 4-2. Of the unionids, 50% are articulated. Percentage of fragmented shells ranges from 10 to 95, and in most samples is between 20 and 50. The association occurs in a range of sediments; in silty clays, silt, and, above all, in silty marls and limestones. Two samples are highly carbonaceous containing either coal fragments or plant debris. The percentage of brackish-water faunas in the total fauna ranges from 0 to 82, with a mean value of 1 5-3. The Lioplacodes stachei/Parateinostoma occultum association Three collections with All specimens were grouped in this association. L. stachei is the dominant species, followed by the pleuroceratid gastropod Pachychiloid.es macilentus. Two amnicolid gastropods, Parateinostoma occultum and P. Intense, are also present in each sample, albeit in lower numbers. Three more gastropods ( Valvata praecursoris, M. naticiformis, and Pachychiloides cleburni) are rare, as is the unionid Protelliptio vetustus. The number of species varies from five to six (mean 5-3). Fragmentation of shells ranges from 10 to 90%. Two collections are from argillaceous silt, the third from carbonaceous marly siltstone. The percentage of brackish-water elements in the total fauna ranges from 1-5 to 10T (mean 9-6%). The L. stachei/Pachychiloides macilentus association This is the commonest association in the freshwater regime, represented by 2607 specimens in fourteen col- lections. L. stachei is the dominant element followed by M. naticiformis and P. macilentus. Only L. stachei and P. macilentus occur in all samples. Eight gastropods occur rarely or sporadically only (Table 3), among them the relatively large Viviparus couesi and C. macrospira. Both unionid species are present with Protelliptio vetustus (2-5%) being far more common than Loxopleurus belliplicatus (0 6%). The number of species varies from four to nine with a mean of 6.1. 60-7% of the unionids are still articulated. Between 10 and 90% of the shells are fragmented (in most samples, however, not more than 20 to 50%). The fauna occurs in marly silts or calcareous siltstones. Over half the beds are carbonaceous with either abundant plant debris or coal fragments. In one case rootlets occur. The percentage of brackish elements of the total fauna of individual samples ranges from 0 to 85, with a mean of 15-6. The M. naticiformis/Lioplacodes stachei association In eight collections with 1016 specimens the neritid gastropod M. naticiformis dominates in terms of relative abundance followed by Lioplacodes stachei and Pachychiloides macilentus. The three species occur in all col- lections while six other gastropods (P. chrysalis, P. cleburni, Parateinostoma occultum, Lymnaea nitidula , C. macrospira, and V. couesi) and the two unionids are rare and occur only in one to three samples. The number of species varies from four to eight (mean 51). Only 25% of the unionids are articulated. The percentage of fragmentation ranges from 20 to 90 and most commonly is around 75%. In most samples the sediment is argillaceous silt, less commonly marly or calcareous siltstone. Nearly half the samples are carbonaceous containing plant debris or coal fragments. The percentage of brackish-water elements in individual samples is relatively high, ranging from 1-3 to 74%, with a mean of 38-9%. The Pachychiloides chrysalis association The P. chrysalis association differs drastically in species composition from the other associations. Found in four samples and represented by 692 specimens; it is dominated by the pleuroceratid gastropods P. chrysalis, P. cleburni, P. macilentus, and the neritid M. naticiformis. These four species occur in all samples, while the rare Lioplacodes stachei and P. chrysalloideus are found in only one or two collections. The number of species ranges from four to six (mean 4-8). Unionids are absent from this association. Fragmentation varies from 30 to 90%. The sediments are carbonaceous, non-calcareous, and range from argillaceous silt to silt and fine sandy siliceous siltstone. Two of the samples are bioturbated. The burrows consist of branching tubes, 1 mm in diameter, which are filled with dark, clayey silt. They resemble Chondrites, but their branching pattern is more irregular. The percentage of brackish forms within individual samples ranges from 6-9 to 24-7 (mean 16-9). FORSICH AND KAUFFMAN: ALBIAN PALAEOECOLOGY 523 The Valvata praecursoris and corbulid sp. A assemblages Found only in one sample, it is not known whether these assemblages represent recurrent associations. Occurring in low density in dark-grey shaley silty clay together with thin streaks of comminuted shell debris, the freshwater assemblage appears to be autochthonous. It consists of six species of small gastropods, with the tiny V. praecursoris most abundant (58-5%), followed by the amnicolid Parateinostoma cf. altipsirale (15-4%), the pleuroceratids Pachychiloides chrysalis (10-8%), and P. chrysalloideus (9-2%). M. naticiformis and Parateino- stoma occultum account for the remaining 61%. Fish scales are found occasionally in the sediment that also contains a brackish-water assemblage (33% of the total fauna) that is dominated by the small corbulid bivalve sp. A (31 -2%), followed by Veloritina durkeei (28-1%), Brachidontes multilinigera (25%), and Ursirivus pyriformis (15-6%). The two corbulid species are small, Veloritina and Brachidontes tiny. Discussion Autochthonous nature of the freshwater associations. Few biostratinomic data are available to indicate transport or the lack of it in the freshwater fauna. In most samples, unionid bivalves are predomi- nantly articulated and in two cases they have been found in life position. Although transport of recently dead articulated unionids downstream into a nearshore area might appear plausible, it is unlikely considering the large and thick shells of both species (see also White 1 895). The thin-shelled gastropods similarly underwent only little lateral transport, if any. The fragile shells would break very easily and the percentage of shell fragments would be much higher than it is. In most gastropods the aperture is damaged, but that can be explained by in situ reworking. Furthermore, the autecology of several species (see below) indicates that the fauna lived in the embayment and was not transported downstream from rivers. Finally, the pronounced relationship of some freshwater associations to silty limestones cannot be explained by transport. The freshwater fauna is therefore regarded as autochthonous or, at the most, parautochthonous. Autecology. The unionids Loxopleurus belliplicatus and Protelliptio (Plesielliptio) vetustus exhibit features such as curved ventral margins and a well-expanded posterior which are more typical of species living in large lakes than in swiftly flowing rivers (e.g. van der Schalie 1938; Eager 1948; Tevesz and Carter 1980). Little is known about the life habits of Recent unionids and even less about fossil forms. It is, however, likely that both species lived partially buried in the sediment. Similarly, ecological studies of fossil freshwater gastropods are scant and little information is available. Most likely, all freshwater gastropods of the Bear River section were herbivorous, feeding either on live algae or on plant detritus. Recent members of the families represented by Bear River species vary considerably in their salinity requirements. For example, most members of the family Ncritidae inhabit marine to brackish habitats at present day, but the genus Neritina tends to invade freshwater. A very transitional position is also assumed for the Cretaceous M. naticiformis as it occurs in large numbers where the percentage of brackish-water faunal elements is relatively high. This species is considered to have been able to live not only in freshwater, but also in oligohaline waters. Recent amnicolid gastropods also invade brackish water. How far this was true of the Cretaceous Parateino- stoma is unknown. Recent members of the family Valvatidae and Viviparidae, although able to live in waters with a salinity of 2-3 permille (e.g. Ankel 1936) are typical representatives of freshwater environments and it appears safe to assume a similar habitat for Cretaceous species. In viviparids, the mantle cavity shows many features which may be associated with a muddy habitat (Fretter and Graham 1962, p. 594). This also fits the habitat preference of the Cretaceous Viviparus couesi. Lymnaea nitidula belongs to the freshwater pulmonate family Lymnaeidae and is the only representative of this group at the Bear River section. In analysing similar faunas from the Cretaceous of Lincoln County, Wyoming, Yen (1951) concluded that they lived in relatively shallow, low-energy environments with an abundant aquatic vegetation. A freshwater assemblage from the Bear River Formation at Fossil Cut, south-east of Evanston (Wyoming), was interpreted by him (Yen 1954) as having lived in the lower part of the littoral zone (7-10 m deep), in a more or less closed and quiet bay with rich vegetation. Analysis of the associations. Four of the five associations are closely related, with Lioplacodes stachei being present in moderate to high numbers in all samples. However, as in the case of 524 PALAEONTOLOGY, VOLUME 27 I I I I I I I I I I I L. STACHEI ASS. L. STACHEI / P L. STACHEI / P MA- M. NATICIFORMIS PCHRYSALIS OCCULTUS ASS. CILENTUS ASS. /L. STACHEI ASS. ASS. text-fig. 12. Evenness, species richness, and substrate relationships of the five freshwater associations. For key of substrate see text-fig. 2. £//w/Wvws-dominated brackish associations, the relative abundance of species is thought to reflect subtle differences in the environment. Text-fig. 12 shows some ecological features of the five asso- ciations such as species richness, evenness values, and substrate relationships. All associations exhibit low species richness and evenness values suggesting that the faunas did not live under optimal conditions. However, they cannot be arranged along an environmental gradient as in the case of the brackish-water faunas. Faunal composition and diversity were probably influenced by several factors, not necessarily related. Substrate conditions appear to have influenced faunal distribution to some degree. Thus, the Pachychiloides chrysalis association occurs in coarser sediments than the others and the substrate was never calcareous. This biotope may have been close to river mouths and a slightly higher energy level than in the other associations probably prevailed. In contrast, the Lioplacodes stachei/P. macilentus association occurs nearly exclusively in calcareous siltstones or silty marlstones. Another factor governing faunal distribution was salinity. According to Remane and Schlieper (1971) some freshwater species are able to tolerate waters of the oligohaline or even lower mesohaline regime. Segerstrale (1957, p. 779) records a number of freshwater gastropods from the Baltic Sea which live in oligo- or even low mesohaline coastal waters. Many of them are greatly reduced in size. Even some unionid bivalves tolerate salinities up to 3%0 in parts of the Baltic Sea (Segerstrale 1957; see also Remane 1958 for summary of the salinity tolerances of freshwater organisms). It is therefore FURSICH AND KAUFFMAN: ALBIAN PALAEOECOLOGY 525 likely that several freshwater gastropods from the Bear River Formation could tolerate at least short- term small-scale salinity fluctuations. This may have been particularly true of M . naticiformis (see above). Consequently, the M. naticiformis / L. stachei association is thought to have lived closest to the brackish-water edge and was able to survive short-term incursions of oligohaline waters. This implies that there might have been a spatial and temporal overlap of members of this association with members of the oligohaline Pyrgulifera humerosa subset. The position of the M. naticiformis! L. stachei association close to the freshwater/brackish-water interface is supported by the relatively high percentage of brackish-water faunal elements in the total fauna as compared to that of other associations. The L. stachei association appears to have lived furthest away from the freshwater/brackish-water interface: in three out of six samples no brackish faunas occur, in two more the brackish fauna accounts for less than 10%, and only in one sample is the brackish fauna very abundant (81 -5%). Other factors not readily recognized from the substrate must have exerted a strong influence on the freshwater fauna. Most likely they were water chemistry (e.g. concentration of Ca-ions), temperature and oxygen level. High temperature fluctuations combined with periods of poor oxygenation are very common in large, very shallow-water bodies, restricting the fauna to eurytopic species. That it was a high stress environment can also be demonstrated by the size of several faunal elements. At Shell Hollow, a locality of the Bear River Formation several miles north-west of Evanston (Wyoming), the gastropods L. stachei , Pachychiloides chrysalis , P. chrysalhideus , and P. macilentus were about 20 to 25% larger than at the Bear River locality (Shell Hollow data from unsorted collections in the Henderson Museum of the University of Colorado at Boulder). This implies that at least these species (data on other freshwater gastropods were not available) did not grow to their maximum size at the latter locality. Low oxygen conditions were most likely responsible for the small to minute size of all specimens in the Va/vata praecursoris and corbulid sp. A. assemblages, which occur in a dark-grey laminated silty clay. Summary of depositional environment. The freshwater faunas lived in a large shallow body of freshwater that was in direct contact with brackish water. Most associations lived in low-energy environments on soft substrate that was richly vegetated. The close association of coal-seams and freshwater faunas indicate that water depth was extremely shallow. Only one association appears to have preferred a somewhat higher energy level and coarser substrate, probably close to the rivers emptying into the embayment. The embayment was subject to fluctuations in temperature and oxygen levels that restricted the fauna to eurytopic species. ANALYSIS OF CYCLES In the Bear River section both sediment and benthic faunas occur in cycles. The base of each cycle is defined by shell beds dominated by brackish-water fauna, the top by coals, highly carbonaceous sediments and/or calcareous beds dominated by freshwater faunas. The boundaries between cycles are marked by the drastic change in the proportion of fresh- to brackish-water faunal elements. Text-figs. 13-15 show the lithological change, the variation in the proportion of fresh- to brackish-water species, the variation in mean size of the dominant brackish-water species ( Ursirivus ) and the distribution of fresh and brackish-water associations throughout the Bear River section. Although quantitative data are not available for each bed, the pattern is quite clear and can be interpreted as a series of regressive sequences each followed by relatively rapid transgressive pulses. In most cases, the transgressive phase did not leave any sedimentary or faunal record, but occasionally the change from freshwater-dominated faunas at the top of a cycle to brackish-water-dominated faunas at the base of the next cycle is represented by intermediate stages: part of the transgressive phase. The amplitude of the cycles may vary. Beds with exclusively brackish-water faunas may grade into beds with only freshwater elements at the top of each cycle. Alternatively, beds at the base may already exhibit a strong freshwater influence and beds at the top may still have a high percentage of brackish-water elements. Even where no faunas are preserved, the regressive nature of the cycles can be demonstrated easily: thus, at the base of text-fig. 13, carbonaceous sediments intercalated with thin seams of coal are overlain by a fine-grained channel sandstone which is bioturbated and rooted at the top and overlain by silty clays that grade into highly carbonaceous shales. This sequence can be interpreted as fluvial channel sands giving way to overbank deposits and finally swamps. Strong [Text continues on page 530.] 526 PALAEONTOLOGY, VOLUME 27 text-fig. 13. Sedimentological and palaeoecological data (percentage of freshwater faunal elements, mean size of Ursirivus pyriformis, and distribution of benthic associations) of the Reference Section of the Bear River Formation. For key to symbols see text-fig. 2. C— clay; Si— silt; S- fine-sand; C^Crassostrea soleniscus association; UC — Ursirivus pyriformis/ C. soleniscus subset; UV — U. pyriformis I Veloritina durkeei subset; UP — U. pyriformis I Pyrgulif era humerosa subset; P —P. humerosa subset; Pa — Pachychiloides chrysalis association; ML — Mesoneritina naticiformis\Lioplacodes stachei association; LP —L. stachei/ Parateinostoma occullum association; L — L. stachei association; LPm L. stachei/ Pachychiloides macilentus association; Va — Valvata praecursoris assemblage. Empty rectangles are semi-quantitative field estimations of the percentage of freshwater elements. FURSICH AND KAUFFMAN: ALBIAN PALAEOECOLOGY 527 associations text-fig. 14. Continuation of section of text-fig. 13. 528 PALAEONTOLOGY, VOLUME 27 text-fig. 1 5. Upper part of marginal marine section at the Bear River Reference Locality. For key to symbols see text-fig. 2. Note disruption of section near top due to faulting. FURSICH AND KAUFFMAN: ALBIAN PALAEOECOLOGY 529 SIZE RANGE text-fig. 16. Detail of one regressive cycle. Both top of preceding and base of succeeding cycle have been included for comparison. Note increase in percentage of freshwater elements, changes in faunal composition, changes in the mean size of dominant brackish-water species, and size/frequency histograms towards the top of the cycle. For key to symbols see text-fig. 2. c— clay; s — silt; V — Veloritina durkeei ; U — Ursirivus pyriformis ; P — Pyrgulifera humerosa; B — Brachidontes multilinigera ; C — Crassostrea soleniscus; L — Lioplacodes stachei; M — Mesoneritina naticiformis; PM — Pachychiloides macilentus; PO — Parateinostoma occultum; PL — P. latense; PC — P. cleburni. 530 PALAEONTOLOGY, VOLUME 27 fluvial influence can only be recognized near the base of the measured section, while most other cycles record regression within a shallow embayment. The distribution of the brackish-water associations follows the cyclic pattern, although with several exemptions. The oligohaline Pyrgulifera humerosa subset is more commonly found near the top of cycles than at their base (text-figs. 13-15). Similarly, the mid-mesohaline U. pyriformis/C. soleniscus and U. pyriformis / Veloritina durkeei subsets are more often found near the base. The freshwater associations show a fairly random distribution pattern except for the L. stachei association that usually occurs toward the top of the cycles. In thirteen cases the mean size of U. pyriformis is larger near the base of the cycles than towards the top; in four cases it is not. Thus there is a general size decrease of Ursirivus within a cycle, notwithstanding a significant variability. This is true also of Veloritina and, although less pronouncedly, of Pyrgulifera (text-fig. 16). The picture may be distorted by preferential destruction of smaller valves. faunal composition size range text-fig. 17. Detail of one regressive cycle including base of the next cycle. Note increase in percentage of freshwater faunal elements, changes in the mean size of Pyrgulifera and Ursirivus, and dominance of juvenile individuals of Ursirivus near top of the cycle. For key to symbols see text-fig. 2. FURSICH AND KAUFFMAN: ALBIAN PALAEOECOLOGY 531 The lack of an anticipated neat correlation between changes in size and benthic associations and the regressive cycles can be explained by the overprint of seasonal fluctuations on the cyclic pattern. The resulting pattern is therefore one of interference of two cycles of different magnitudes. In text-fig. 1 7 the size/frequency distribution of Ursirivus within one cycle is plotted as are mean size of Pyrgulifera and Ursirivus. As expected a strong peak of juveniles indicative of mass mortality during freshwater interludes shows up when the percentage of fresh-water faunas among the total fauna increases. However, there are also several examples of high juvenile mortality among the brackish-water species near the base of cycles where the percentage of freshwater faunas is low. In these cases the freshwater phase killed the brackish fauna, but did not last long enough for the establishment of a freshwater fauna. Similarly, presence of adult Ursirivus in time-averaged brackish-water populations of beds dominated by freshwater elements can be explained by occasional less severe salinity reductions which enabled the brackish fauna to survive for more than one season. text-fig. 18. Model illustrating the overprint of seasonal cycles on the salinity reduction within one regressive cycle. Note that in reality thousands of seasonal cycles may be involved instead of the twelve shown here. Text-fig. 18 presents a model of the salinity changes within cycles combining seasonal fluctuations with a general reduction in salinity. Both parameters can be reconstructed by using size/frequency data, proportion of fresh- and brackish-water faunas in one bed, and the nature of the brackish associations. The cycles in text-figs. 13-15 vary in thickness, but also according to the dominant salinity regime. Groups of cycles in which brackish conditions dominate for most of each cycle followed by a short period of dominantly freshwater conditions alternate with groups where the brackish-water dominated phase is short and fresh-water conditions prevail for most of the time. The groups are composed of two to four cycles. Unfortunately for several cycles the data are insufficient to allow a more precise description of this feature. It seems, however, that yet another order of cyclicity is recorded by the fauna, but not by any features of the sediment. INTERPRETATION OF CYCLES Cyclicity recorded in the sedimentary or fossil record is either caused by climatic factors, by movements of the earth’s crust in the broadest sense (be it by spreading of mid-oceanic ridges or small-scale tectonic movements), or by factors connected with mechanisms of sediment transport and deposition within basins (e.g. articles in Merriam 1964, Duff el at. 1967, Einsele and Seilacher 1982). 532 PALAEONTOLOGY, VOLUME 27 There can be little doubt that the seasonal cyclicity found in the Bear River section, expressed by the size/frequency curves of brackish-water organisms, is of climatic origin and records the alternation of dry and wet seasons similar to monsoon seasons in today’s tropical and subtropical belts. This is in agreement with Tingey (1978) who, based on palynological data, postulated a subtropical to warm temperate climate for south-western Wyoming during the middle and late Albian. The origin of the secondary cycles can be less easily assessed. What is recorded is the gradual infilling and freshening of a large embayment. The tectonic hypothesis would require rhythmic pulses of subsidence that would lead to a rapid transgression across the embayment and subsequent gradual infilling with sediment. Another variation of the tectonic model assumes increased uplift of the hinterland (ancestral Rockies) which would result in increased supply of sediment to deltas. Reworking of sediments by longshore currents would then lead to increased formation of barrier bar systems which could restrict marine influence in the embayment and finally seal it oft' completely. However, there are no signs of extensive barrier systems along the shorelines of the Skull Creek Seaway which apparently was bordered by very low-energy shore-lines. Moreover, the regressive cycles in the embayment are not coarsening-upward cycles, though an increase in the rate of sedimentation toward the top of the cycles is indicated in many cases by the decrease in the thickness and abundance of shell beds and by the lower density of shells within them. Assuming a climatic origin of the cycles, with an increase of rainfall, there should be increased run off and consequently increased erosion and sediment supply to coastal waters. The same process of increased formation of barrier bars could then operate as in the tectonic model. As mentioned before, however, the existence of extensive barrier sands is unlikely. Increased rainfall could have resulted in a more extensive sheet of freshwater across the embayment and caused a long-term shift of the freshwater/brackish-water interface towards the open sea. The third possibility is that the regressive pattern is autocyclic, caused by the switching of major distributaries within a deltaic system. This would cause sediment influx into the embayment to cease and, assuming subsidence to continue, would result in rapid transgression. From the available evidence it is impossible to decide which model is correct. An autocyclic explanation for the regressive sequences is, however, the simplest model and is therefore favoured. The cause of the primary cyclicity, which is characterized by the relative duration of fresh- to brackish-water conditions within secondary cycles, is even more open to speculation. It may have been the result of climatic or tectonic factors, possibly expressed by slight eustatic fluctuations in sea- level. In this case, an allocyclic mechanism is more likely. Text-fig. 19 presents a summary of the three orders of cycles. Whilst the tertiary cycles represent fluctuations on the scale of 10° years, the secondary cycles probably are in the range of 103 to 104 years, and the primary cycles possibly present periods of 104 to 105 years. Unfortunately, biostratigraphic and chronostratigraphic control on the Bear River section is not available. The time ranges for the three orders of cycles must therefore be regarded as very tentative. To what extent the primary cyclicity, should it prove to be of climatic origin, can be related to cycles of the earth’s orbit (Milankovich 1930) remains unknown. CONCLUSIONS 1. Part of the Bear River Formation represents an extensive embayment in which fine-grained sediments ranging from silty clays to silts and silty limestones accumulated in a predominantly low- energy environment. Within this sequence, numerous shell beds with a highly abundant but low diversity fresh- and brackish-water fauna of bivalves and gastropods occur. 2. Biostratinomic data favour local reworking by storms rather than lateral transport by currents as the origin of the shell beds. Mixing of fresh- and brackish-water faunas in the same bed was caused by rapidly shifting environments and not by mixing of faunas from neighbouring habitats. 3. Sedimentological and palaeosynecological analyses demonstrate the presence of cycles that start with shell beds of predominantly brackish origin and end with thin seams of coal or beds of silty limestone dominated by freshwater faunal elements. FURSICH AND KAUFFMAN: ALBIAN PALAEOECOLOGY 533 4. Two benthic associations, one of them with four subsets, are recognized in the brackish-water fauna. They can be arranged along a salinity gradient ranging from mid-mesohaline (about 1 2%0) to the freshwater edge. Along this gradient, species evenness and richness drops and most faunal elements decrease in size. Five associations are recognized in the freshwater fauna. They are dominated by small gastropods, exhibit some substrate control, and, at least in one case, were probably able to invade oligohaline waters. 5. Size/frequency distribution patterns of the brackish bivalves Ursirivus, Veloritina, and the gastropod Pyrgulifera point to strong seasonal salinity fluctuations which, in many cases, caused a high juvenile mortality. 6. Altogether, three orders of cycles are recognized in the Bear River section of probably the following magnitudes: (a) 10° years (seasonal), recorded by size/frequency curves within shell beds; (b) 1 03— 1 04 years, expressed by regressive sedimentary sequences and a consistent change from brackish to freshwater-dominated biota; and (c) 104-105 years, expressed by the dominance of fresh or brackish conditions within bundles of regressive sequences. Whilst the tertiary (seasonal) cyclicity is of climatic origin, the nature of the secondary cycles is probably autocyclic. The origin of the primary cyclicity remains unclear. Primary cycle text-fig. 19. The three orders of cycles in the Bear River Formation at the Bear River Reference Locality. The terms primary, secondary, and tertiary cycles are used in order to avoid confusion with first, second, and third order cycles of Vail et al. (1977). b— brackish;/ — freshwater. 534 PALAEONTOLOGY, VOLUME 27 Acknowledgements. The work was carried out during the tenure of a Heisenberg Fellowship (F.T.F). F.T.F also acknowledges the hospitality of the Department of Geological Sciences, University of Colorado, Boulder, during 1982-1983. We thank T. A. Ryer and J. Horne, Research Planning Institute, Boulder, R. Gustavson and R. Diner, Geology Department, U.C., Boulder, and J. Hanley, U.S.G.S., Denver, for useful discussions. We are also very grateful to K. Flessa, University of Arizona, Tucson, for discussions of ecological and statistical problems and for performing the statistical analysis. R. Diner critically read the manuscript and made valuable suggestions. T. Ryer and D. Lawrence (Yale University) assisted in locating the new reference section. F. Hock, Munich, did the photographic work. The study was supported financially by a grant from the Deutsche Forschungsgemeinschaft (Fu 131/4-1). REFERENCES ankel, w. e. 1936. Prosobranchia. In grimpe, g. and wagler, e. (eds.). Die Tierwelt dev Nord- undOstsee , 9b, 1-240. Akademische Verlagsgesellschaft, Leipzig. barnes, R. s. K. 1980. Coasted lagoons. Cambridge University Press, Cambridge, 106 pp. boesch, d. f. 1977. A new look at the zonation of benthos along the estuarine gradient. In coull, b. c. (ed.). Ecology of marine benthos , 245-266. University of South Carolina Press, Columbia, S.C. cadee, G. c. 1982. Low juvenile mortality in brachiopods, some comments. Publ. Versl. Netli. Inst. Sea Res. 1982-3,29 pp. cobban, w. a. and reeside, j. b., jr. 1952. Correlation of the Cretaceous Formations of the Western Interior of the United States. Bull. Geol. Soc. Am. 63, 101 1 1044. duff, p. mcl. d., hallam, a. and walton, e. k. (eds.) 1967. Cyclic sedimentation. Dev. Sediment. 10, 1-280. eager, R. M. c. 1948. Variation in shape of shell with respect to ecological station: A review dealing with Recent Unionidae and certain species of the Anthracosiidae in Upper Carboniferous times. Proc. Roy. Soc. Edinburgh , B63, 130-148. eicher, d. l. 1960. Stratigraphy and micropaleontology of the Thermopolis Shale. Bull. Peabody Mus. Nat. Hist. 15, 126 pp. 1962. Biostratigraphy of the Thermopolis, Muddy, and Shell Creek Formations. Wyo. Geol. Ass. 17 th Ann. Fid Conf. Guide book , 72-93. einsele, G. and seilacher, A. (eds.) 1982. Cyclic and event stratification. Springer, Berlin, 536 pp. engelmann, h. 1876. Report on the geology of the country between Fort Leavenworth, K.T., and the Sierra Nevada near Carson Valley. In simpson, j. h. Report of exploration across the Great Basin of the Territory of Utah in 1859 , 247-336. Washington, D.C. fretter, v. and graham, a. 1962. British Prosobranch Molluscs. Ray Society, London, 755 pp. fursich, f. t. 1975. Trace fossils as environmental indicators in the Coralhan of England and Normandy. Lethaia. 8, 151 172. 1977. Corallian (Upper Jurassic) marine benthic associations from England and Normandy. Palae- ontology, 20, 337-385. — 1978. The influence of faunal condensation and mixing on the preservation of fossil benthic communities. Lethaia, 11, 243-250. — 1981. Salinity controlled benthic associations from the Upper Jurassic of Portugal. Ibid. 14, 203-223. hallam, a. 1967. The interpretation of size-frequency distributions in molluscan death assemblages. Palaeontology , 10,25-42. — 1972. Models involving population dynamics. In schopf, t. j. m. (ed.). Models in paleobiology , 62-80. Freeman, Cooper and Co., San Francisco. haun, j. d. and barlow, j. a., jr. 1962. Lower Cretaceous stratigraphy of Wyoming. Wyo. Geol. Ass. 17th Ann. Fid Conf. Guidebook , 15-22. hayden, F. v. 1 869. Preliminary Field Report of the United States Geological Survey of Colorado and New Mexico. Washington, D.C., 155 pp. horstman, a. w. 1966. Correlation of the Lower and Lower Upper Cretaceous rocks of a part of southwestern Wyoming, Utah and Colorado. Unpublished Ph.D. thesis. University of Colorado, Boulder. jones, t. r. 1893. On some fossil Ostracoda from southwest Wyoming, and from Utah, U.S.A. Geol. Mag. 3, 145-148. kauffman, E. G., cobban, w. A. and eicher, D. L. 1 976. Albian through Lower Coniacean strata, biostratigraphy and principal events. Western Interior United States. Ann. Mus. Hist. Nat. Nice , 4, xxiii, 1-52. larocque, a. and edwards, c. d. 1954. Bear River and Aspen Formations in Willow Creek, Teton County, Wyoming. Bull. Geol. Soc. Am. 65, 315-326. FURSICH AND KAUFFMAN: ALBIAN PALAEOECOLOGY 535 levinton, J. s. 1982. Marine ecology. Prentice-Hall Inc., Englewood ClifTs, N.J., 526 pp. lewy, z. and samtleben, c. 1979. Functional morphology and palaeontological significance of the conchiolin layers in corbulid pelecypods. Lethaia, 12, 341-351. macarthur, r. h. 1 972. Geographical Ecology. Harper and Row, New York, 269 pp. MCGOOKEY, D. P., HAMM, J. D., HALE, L. A., GOODELL, H. G., MCCUBBIN, D. G., WEIMER, R. J. and WULF. G. R. 1972. Cretaceous System. Geol. atlas of the Rocky Mountain region , 190-228. Rocky Mount. Ass. Geol., Denver. mee, l. d. 1978. Coastal lagoons. In Chemical oceanography (2nd edn.), 7, 441-490. Academic Press, London. meek, F. b. 1 860. Descriptions of new fossil remains collected in Nebraska and Utah by the exploring expeditions under the command of Capt. J. H. Simpson of U.S. Topographical Engineers. Proc. Philad. Acad. Sci. for I860, 308-315. — 1870a. A preliminary list of fossils collected by Dr. Hayden in Colorado, New Mexico, and California, with brief descriptions of a few of the new species. Proc. Am. Phil. Soc. 1 1, 430-431. — 18706. Collections from the Bear River estuary beds. Rept. U.S. Geol. Expl. 40th Par. 3, 462 466. — 1873. Preliminary paleontological report, consisting of lists and descriptions of fossils, with remarks on the ages of the rocks in which they were found. Sixth Ann. Rept. U.S. Geol. Surv. Terr, for 1872 , 431 -518. — and engelmann, h. 1860. Notice of geological discoveries made by Capt. J. H. Simpson, Topographical Engineers, U.S. Army, in his recent explorations across the continent. Proc. Philad. Acad. Sci. for I860, 126-131. merriam, p. f. (ed.) 1964. Symposium on cyclic sedimentation. Bull. Kansas Geol. Surv. 169, vols. 1 and 2, 636 pp. milankovich, m. 1930. Mathematische Klimalehre und astronomische Theorie cler Klimaschwankungen. Handbuch der Klimatologie, 1A. Gebr. Borntraeger, Berlin, 176 pp. mills, e. l. 1969. The community concept in marine zoology, with comments on continua and instability in some marine communities: a review. J. Fish. Res. Bd Canada , 26, 1415-1428. peck, R. E. 1951. A new ostracode genus from the Cretaceous Bear River Formation. J. Paleont. 25, 575-577. — and craig, w. w. 1962. Lower Cretaceous non-marine ostracods and charophytes of Wyoming and adjacent areas. Wyo. Geol. Ass. 17th Ann. Fid Conf. Guide book , 33-43. peterson, c. h. 1976. Relative abundance of living and dead molluscs in two Californian lagoons. Lethaia, 9, 137-148. — 1977. The paleoecological significance of undetected short-term temporal variability. J. Paleont. 51, 976-981. remane, a. 1958. Okologie des Brackwassers. In remane, a. and schlieper, c. Die Biologie des Brackwassers. Die Binnengewasser , 22, 1-216. Schweizerbart, Stuttgart. — and schlieper, c. 1971. The biology of brackish water. Wiley-Interscience, New York, 372 pp. royse, f., Warner, m. a. and reese, d. l. 1975. Thrust belt structural geometry and related stratigraphic problems. Wyoming-Idaho-Northern Utah. In bolyard, d. w. (ed.). Deep drilling frontiers of the Central Rocky Mountains, 41-54. Rocky Mount. Ass. Geol., Denver, Co. rubey, w. w. 1973. New Cretaceous formations in the western Wyoming thrust belt. Bull. U.S. Geol. Surv. 1372-1, 35 pp. SCHAlie, h. van der 1938. The Najad fauna of the Huron River in southeastern Michigan. Misc. Pub!. Univ. Mich. Mus. Zool. 40, 1 -83. segerstrale, s. g. 1957. Baltic Sea. In hedgpeth, j. w. (ed.). Treatise on marine ecology and paleoecology, vol. 1 , Ecology. Geol. Soc. Amer. Mem. 67, 751 -800. stanton, t. w. 1892. The stratigraphic position of the Bear River Formation. Am. J. Sci. 43, 98-1 15. Stephenson, l. w. 1952. Larger invertebrate fossils of the Woodbine Formation (Cenomanian) of Texas. Prof. Pap. U.S. Geol. Surv., 242, 226 pp. tevesz, m. j. s. and carter, j. g. 1980. Environmental relationships of shell form and structure of unionacean bivalves. In rhoads, d. c. and lutz, r. a. (eds.). Skeletal growth of aquatic organisms, 295-322. Plenum Press, New York. thomas, h. d. 1962. Some problems of the earlier Cretaceous rocks of Wyoming. Wyo. Geol. Ass. 1 7th Ann. Fid Conf. Guide book, 28-32. tingey, j. c. 1978. Palynology of the Lower Cretaceous Bear River Formation in the Overthrust Belt of south- western Wyoming. Unpublished Ph.D. thesis, Michigan State University. VAIL, P. R., MITCHUM, R. M., THOMPSON, S., TODD, R. G., SANGREE, J. B., WIDMIER, J. M., BUBB, J. N. and HATLELID, w. G. 1977. Seismic stratigraphy and global changes in sea level. Am. Ass. Petrol. Geol. Mem. 16,49-212. veatch, a. c. 1907. Geography and geology of a portion of southwestern Wyoming, with special reference to coal and oil. Prof. Pap. U.S. Geol. Surv. 56, 178 pp. 536 PALAEONTOLOGY, VOLUME 27 voices, h. E. 1945. Supraspecific groups of the pelecypod family Corbulidae. Ball. Am. Mus. Nat. Hist. 86, art. 1, 32 pp. wanless, h. r., belknap, r. l. and foster, h. 1955. Palaeozoic and Mesozoic Rocks of Gros Ventre, Teton, Hoback and Snake River Ranges, Wyoming. Mem. Geol. Soc. Am. 63, 90 pp. white, c. a. 1891. Correlation papers. Cretaceous. Bull. U.S. Geol. Surv. 82, 273 pp. 1895. The Bear River Formation and its characteristic fauna. Ibid. 128, 108 pp. wilmarth, M. G. 1938. Lexicon of geological names of the United States (including Alaska). Ibid. 896, 2396 pp., 2 vols. yen, t.-c. 1951. Freshwater mollusks of Cretaceous age from Montana and Wyoming. Prof. Pap. U.S. Geol. Surv. 233-A, 20 pp. — 1952. Age of the Bear River Formation, Wyoming. Bull. Geol. Soc. Am. 63, 757-763. — 1954. Nonmarine mollusks of Late Cretaceous age from Wyoming, Utah and Colorado. Prof. Pap. U.S. Geol. Surv. 254-B, 45-66. young, r. G. 1969. Lower Cretaceous of Wyoming and the southern Rockies. Bull. Wyo. Geol. y4s.s. Earth Sci. 2, 21-35. F. T. FURSICH Institut fur Palaontologie und historische Geologie der Universitat Richard-Wagner-Str. 10/11 D-8000 Miinchen 2, West Germany Typescript received 24 May 1983 Revised typescript received 22 October 1983 E. G. KAUFFMAN Dept, of Geological Sciences University of Colorado Boulder, Co. 80309, U.S.A. SCLEROCHRONOLOGY AND CARBONATE PRODUCTION IN SOME UPPER JURASSIC REEF CORALS by OMER E. ALI Abstract. Annual banding evident from epithecal increments and associated internal structural changes in phaceloid and massive Oxfordian corals show a range in growth rate from 5 to 10 mm ym1 in branching colonies ( Thecosmilia ) and 1 .5 to 3 mm yr1 for massive colonies (Thamnasteria, Fungiastraea , Isastraea). High- and low- density growth bands are identified in massive colonies. The denser part of each couplet is consistently the broader, in contrast with that of most shallow-water modern corals. This is interpreted as due to high local turbidity. The formation time for two sections is estimated with gross carbonate production of 2000 to 3300 g CaC03 m2 ym1. Growth bands in organisms are of great interest to biologists and palaeontologists because, where their periodicity can be determined, a means is provided to estimate growth rate and age. Although growth bands have been studied in several groups of marine invertebrates most interest has been shown in scleractinian reef-building corals because of the extensive distribution of modern corals and ancient coralliferous sediments. Hudson et al. (1976) have coined the term sclerochronology to describe coral growth band studies, in comparison with the well-known term dendrochronology (e.g. Jefferson 1982). Research has been mainly concerned with the nature of the growth banding, its ecological significance and applications in the biological (Buddemeier and Kinzie 1976) and geophysical sciences (Rosenberg and Runcorn 1975; Scrutton 1978). No growth studies have hitherto been made of Mesozoic corals but several British upper Jurassic corals display banding on well-preserved epithecae or in longitudinal section. The purpose of the present study is to examine the nature and periodicity of growth bands in the common branching and massive corals from the Oxfordian of England, and to assess their environmental implications, the time represented by particular coralliferous units, and the rate of carbonate production. Less- common genera such as Rhabdophyllia and Montlivaltia, which are seldom well preserved, and several genera recorded only from Steeple Ashton (Negus and Beauvais 1979) are not considered in any detail. Other invertebrates associated with coralliferous units, such as oysters, species of Chlamys, Lithophaga , and the alga Solenopora also show growth bands suggestive of annual periodicity but details are not included here. Although there are many papers concerned with special aspects of the Oxfordian rocks of England, Arkell (1933, 1947) provides accounts of the distribution and stratigraphy of the outcrops and Cope et al. ( 1 980) provides correlations. Material on which this paper is based was collected in the course of an investigation of coralliferous units in the Corallian (Oxfordian) of England (Ali, unpublished Ph.D. thesis. University of Reading, 1978) from the following: Shellingford Cross-Roads Quarry (SU 327941) (Arkell 1947, p. 87), Kingsdown Farm (temporary section) (SU 175885), Cumnor Hill by-pass (temporary section) (SP 465040) (see Arkell 1947, p. 89), Headington Cross-Roads Quarry (SP 550064) (Arkell 1947, p. 94), Steeple Ashton (ST 9057) (Negus and Beauvais, 1979); Yorkshire (Wright 1972), Ayton Quarry (TA 002856), Crossgate Quarry, Seamer (TA 028843), Pockley Quarry (SE 635846), Nunnington railway cutting (SE 649787), Stonegrave Quarry (SE 648787). [Palaeontology, Vol. 27, Part 3, 1984, pp. 537-548.] 538 PALAEONTOLOGY, VOLUME 27 text-fig. 1. Fungiastrea arachnoides. a, photomicrograph of unstained thin-section with neomorphic preservation, showing banding and a few more complete upward-tapering septa, x 25. D, positive print from stained longitudinal thin-section showing dark zones with thick septa and light zones with poor preservation of skeletal elements, x 6. Shellingford Cross-Roads Quarry (Reading University 14876a, b). b, c, e, f. Thecosmilia annularis, b, latex peel from external mould of partly decorticated specimen, x 2. c, longitudinal section with tabular dissepiments marked (for comparison with text-fig. 3b), xl.5. Shellingford Cross-Roads Quarry (RU 14879). E, periodic development of tabular (marked) and vesicular dissepiments, x 2. Cumnor By-pass. F, external mould of partly decorticated specimen showing major concentric markings associated with abrupt thickening of septa, x 1 -2. Shellingford Cross-Roads Quarry (RU 14877). Bar scales: A, 0-5 mm; D, I 0 mm. ALI: UPPER JURASSIC CORALS 539 GROWTH BANDS MASSIVE CORALS Thamnasteria concinna (Goldfuss), Fungiastraea arachnoides (Parkinson), and Isastraea explanata (Goldfuss). The three species of massive corals show distinct banding on longitudinal sections and on differentially weathered material, mostly as an alternation of a broader dark zone with thicker septa and a narrower zone with less distinct and thinner skeletal elements (text-fig. I a, d). The two zones are similar to bands recognized in modern scleractinian reef corals by several authors (e.g. Knutson et al. 1972) and referred to as high- and low-density bands. Using x-radiographic techniques Knutson et al. (1972), Buddemeier (1974), Buddemeier et al. (1974), MacIntyre and Smith (1974), Dodge et al. (1974), Dodge and Thomson (1974), and Weber et al. (1975) showed that a high- and low-density couplet is deposited yearly by many tropical corals. In the Oxfordian material the couplets have been diagenetically enhanced so that they are visually distinct. Both zones have undergone neomorphic replacement by calcite or ferroan calcite, but the skeletal relics are best preserved in the lower darker zone and relatively uncommon in the lighter zone (text-fig. 1a). Ferroan calcite is mainly in the upper zone. In thin section the change from one zone to another is generally sharp and distinct due to the contrasting degree of recrystallization. The differential recrystallization may be related to primary skeletal thickness and composition and to differences in the original skeletal porosity. An attempt was made to determine whether any variation in non-carbonate (clay) inclusions now occurs between the couplet zones. Selected specimens showing clear banding were analysed for Si, Ca, and Fe using electron probe. The results (in Ali unpublished Ph.D. thesis. University of Reading, 1978) show that there is no clear correlation between the slight variations in these constituents and position of individual bands. Couplet widths were measured on sectioned specimens and peels where preservation of the skeletal elements was deemed satisfactory. Poor relic preservation often overemphasizes the light zone when it may be questioned which zone is equivalent to the high- and low-density band of modern corals. The light zone was also prone to recrystallization with void stage. Dr. D. Kinsey (Australian Institute of Marine Science, Townsville) suggested to Dr. R. Goldring that the low-density band with its inferred greater amount of organic matter (Highsmith 1979) might fossilize relatively better than the high-density band, but this does not seem to have been the case. No significant variation in width of the high- and low-density bands has been noted between the bottom and top of colonies investigated, although there is an irregular variation through and laterally across a colony. The zones are occasionally of equal width but the ratio of the width of the light zone to that of the dark zone is never greater than 10 (text-fig. 2). This contrasts with modern shallow-water reef genera where it is the dense band that is text-fig. 2. Graphical representation to illustrate vari- ation in ratio (L/H) of low- to high-density bands in the modern corals Platygyra sp. and Montastrea annularis from Indo-Pacific and Caribbean localities • (data from Weber et al. 1975) and the ratio of light to dark zones in Thamnasteria , Fungiastraea , and Isastraea in the Oxfordian of England □. Colonies measured 540 PALAEONTOLOGY, VOLUME 27 thinner (e.g. Knutson et al. 1972; MacIntyre and Smith 1974; Baker and Weber 1975; Weber et al. 1975). Weber et al. (1975) show that the ratio low density/high density is always greater than 1 0 and up to 2-8. But Highsmith (1979) and Hudson (1981) find ratios of less than 10 associated with growth in deeper water. BRANCHING CORAL Thecosmilia annularis (Fleming). T. annularis is a common phaceloid coral in most sections. Incremental growth is evident on the epitheca and there are related structural changes to septa and dissepiments. (a) Epithecal banding. Where the epitheca is well preserved, or on good external moulds a regular banding of major concentric markings separated by areas with finer incremental ridges may be seen. Whilst the major growth bands (4 to 6 in 50 mm) are well defined and easy to follow, the fine bands (more than 200 in 10 mm) are generally indistinct and very difficult to trace on the material available. It has not been possible to establish the number of fine bands between major growth bands. ( b ) Structural changes. The major banding (above) is expressed internally by a sharp thickening of the septa (text-fig. 1b, f) which gradually thin upwards to the next thickening. This is best seen on partly decorticated specimens preserved as external moulds. The arrangement of tabular vesicular dissepiments shows a close relationship to the major growth banding (text-figs, lc, E, 3). Following formation of a tabular dissepiment, the marginal vesicular dissepiments show a regular change mostly realized by transverse contraction towards the text-fig. 3. Thecosmilia annularis, a, longitudinal section to show regular thickening and thinning of septa, b , longitudinal sections to show structural changes in dissepiments. ® Position of major epithecal banding, x tabular dissepiments (Reading University, 14878). ALI: UPPER JURASSIC CORALS 541 margin, abaxially. At the same time successively shorter dissepiments are introduced axially which maintain the general form of the calyx. Nevertheless, the introduction of a tabular dissepiment, which reached to about four- fifths the transverse length of a septum, led to a deepening and increased acuity of the thecal cone. The vesicular dissepiments are frequently thickened and crowded below the periodically introduced tabular dissepiment. This cyclicity or succession of dissepiment arrangement is rarely perfect, but the spacing of the tabular dissepiments corresponds with that of the major growth bands of the epitheca. No such structural changes have yet been described in modern phaceloid corals but the general changes in thickness of septa and dissepiments are similar to those observed in modern massive corals and the Oxfordian genera. Ma (1934) described similar changes in the vesicular elements of modern and subfossil plocoid Favia speciosa though MacIntyre and Smith (1974) commented that dissepiment spacing does not differ between highl- and low-density bands of Pavona gigantea. It seems reasonable to suppose that the periodic changes in Thecosmilia do represent annual changes and that the fine incremental ridges on the epitheca represent daily growth increments. DISCUSSION There can now be no doubt that each couplet of growth bands in modern corals represents an annual skeletal increment (Knutson et al. 1972; MacIntyre and Smith 1974; Moore and Krishnaswami 1974; Dodge and Thomson 1974; Weber et al. 1975). But what is not yet clear, as Scrutton (1978) and Scrutton and Powell (1980) discuss, are the environmental factors that contribute to the formation and seasonal timing of bands of dilferent densities. (There is also as yet no detailed geochemical analysis of the differences between the different density bands and it would be premature to analyse, e.g. 02 isotope, the banding of fossil corals.) Attempts to determine growth factors have, so far, led to somewhat contradictory conclusions. Some investigators (Dodge and Thomson 1974; Knutson and Buddemeier 1973; Buddemeier et al. 1974) suggest that high density is associated with seasonal low water temperature, whereas others (MacIntyre and Smith 1974; Weber and White 1974; Weber et al. 1975; Hudson et al. 1976; Isdale 1977) report that high skeletal density is correlated with periods of high water temperature. Buddemeier (1974) has correlated the dense bands with times of high seasonal rainfall and hence lower-light intensity. Stearn et al. ( 1 977), using seasonal variations in the Barbados environment, suggest that the high-density bands are formed in the autumn in response to relatively abrupt decreases in the available light and the low-density bands are formed in the spring and summer. The paired zones in Oxfordian corals must represent periods of significantly different conditions of skeletal gowth but it is difficult to draw from the fossil record what environmental factors led to the formation of different densities. England during the upper Jurassic was at about 40 N. (Smith et al. 1981) and a marked seasonality of climate is likely. If the negative correlation between skeletal density in corals and available light is accepted, then several characteristics of the corals may be explained. The contrast in thickness ratios of low- to high-density growth between the Oxfordian corals and Platygyra and Montastrea (Weber et al. 1975) might be explained by presuming generally reduced light conditions in the local ancient seas because of either greater water depth or turbidity (see Highsmith 1979 for discussion). This would have allowed growth of a wider band of high-skeletal density and a much thinner band of low-skeletal density. In Montastrea (Baker and Weber 1975) the relative thickness of the band changes with depth. Whilst at depths of less than 18 0 m the less-dense band is consistently more than twice the width of the denser band, at greater depths this value changes to less than 0-5 coinciding with a sharp change in linear growth rate, skeletal density, and growth form. However, the actual thickness of the denser band shows relatively little change with depth. Hudson (1981) reported decrease in growth-rate of Montastrea in deeper fore-reef locations. The local reduced light intensity may better be attributed to high turbidity in the Corallian reefs (Ali, in preparation). Evidence in support of this is the absence of calcareous algae and foraminiferal encrusters, except for rare occurrences on massive corals. Although detritus trapped within the coral skeleton is widespread (e.g. Ali 1983), it is not possible to prove that in any instance it was introduced during growth as described by Bernard et al. (1974). 542 PALAEONTOLOGY, VOLUME 27 GROWTH RATES In reviewing growth rates of modern scleractinian corals Buddemeier and Kinzie (1976) conclude that linear growth-rate ranges from 4 to 20 mm yr~* and that normal average growth rate is about 10-12 mm ym1. They mention examples of faster growth and the extreme growth rate exhibited by Acropora. Dodge and Vaisnys (1977) give a vertical growth rate of about 3-5 mm yi^1 for Bermuda corals. Such gowth rates for modern corals may be used as a background for the study of growth rate of related forms from the fossil record. The growth rate of the branching corals Thecosmilia in southern England (text-fig. 4) and Rhabdophyllia have been obtained by examination of the epithecal banding. (77 annularis is not common in most of the coralliferous localities in Yorkshire and the few specimens collected show only an indistinct banding.) Few measurements were available from Steeple Ashton (though many specimens were examined), but an appreciably lower growth rate is indicated. R. phillipsi is not common in the Oxfordian of southern England, and where it occurs, rarely shows well-defined periodic bands. Specimens from a temporary section at Cumnor Hill (SP 465041) show a banding which suggests rates of about 5-0-6-5 mm yr-1. The same range of values is suggested by specimens from Yorkshire localities. Shellingford Cumnor Kingsdown Steeple Ashton text-fig. 4. Mean and standard deviation of growth rates of thecosmilian colonies from four localities in the Oxfordian with number of measurements made on each colony. Growth rates of the massive corals were determined from measurements of the high- and low- density couplets in thin section and cut surfaces. Similar values have been obtained by measurement of epithecal banding. In general there is considerable variation in the thickness of the bands though this is matched by a similar variation in the thickness of individual bands over a colony. Thamnasteria concinna (text-fig. 5) has the lowest growth rate of the three species but the values show the least variation between colonies and between localities. The growth rate of Fungiastraea arachnoides is higher but the variability in the rate is greater. The relatively poor preservation of Isastraea explanata has allowed fewer measurements to be made. Values of growth rate vary from 2-5 mm ym1 to about 5-0 mm yr -1. Individual variability in linear growth rate, of much higher magnitude, is reported in modern scleractinians by Lewis et al. (1968) for Acropora and other corals, and by Weber and White ( 1 974) for Platygyra. Weber and White suggest that individual variations in growth rate among the different members of a population are probably attributable to a combination of environmental and genetic factors. Dodge et al. (1974) who studied the effect of sediment suspension on growth of Montastrea annularis from Jamaica, noted a decrease in the variability of growth with increasing resuspension. They speculated that the ability of the coral to respond to other favourable, or at least less-limiting, environmental variables is reduced by high resuspension. This factor may have applied to the Oxfordian corals under discussion. A LI: UPPER JURASSIC CORALS 543 text-fig. 5. Mean growth rates and standard deviation of colonies of Thamnasteria concinna, Fungiastraea arachnoides , and Isastraea ex- planata from Oxfordian localities with number of measurements made on each colony. Cumnor Kingsdown Stonegrave Pockley Crossgates ® T. concinna O F. arachnoides □ I. explonolo When compared with modern scleractinians the average growth rates of the massive fossil species are low. However, corals such as Agaracia (3-5 mm ym\ Stearn el al. 1977), Solenastrea (1-5 mm yr~\ Moore and Krishnaswami 1974), and Siderastrea (4.1 mm yr~‘ Stearn et al. 1977) are amongst the slow-growing genera. There are no data available on modern phaceloid genera such as Caulastrea and Astraemorpha , which are morphologically more similar to Thecosmilia and Rhabdophyllia than ramose thamnasteroid genera. Vaughan (1915) noted that, in general, the more massive and denser the skeleton the slower the growth, whilst the more ramose and porous the skeleton the more rapid the growth. The growth rates of some Dinantian rugose corals (Johnson and Nudds 1975) are, surprisingly, relatively high (40-60 mm ym1 ): though Scrutton and Powell ( 1981 ) quote 5- 1 8 mm yr_1 for Silurian favositids. Extension of growth-rate analysis to estimates of geological time represented by actual sections is fraught with difficulties but two sections were selected in the Oxfordian to attempt to determine the time taken for their formation. Sections at Shellingford Cross-Road Quarry and Ayton Quarry (text- fig. 6) are sufficiently extensive to locate profiles where growth was uninterrupted or where successive generations of coral colonies could be traced. This method was adopted by Hoffmeister and Multer (1964) to estimate growth rate of the Pleistocene Key Largo coral reef of the Florida Keys. An account of Shellingford Cross-Roads Quarry is being prepared. Arkell (1947, p. 87) gave a general description and Ali ( 1 977) described the effects of penecontemporaneous erosion and mantles of shelly biosparite, and smectite seams which interrupted coral growth. There are three intervals of smectites and three intervals of biosparite but the interruptions ("hazards’) to coral growth and 544 PALAEONTOLOGY, VOLUME 27 NW SE text-fig. 6. Successive sections used for estimating growth rate of the coralliferous unit at Shellingford Cross-Roads Quarry (above). Sections chosen at about 50-m intervals along quarry face. Lithology and coral distribution at Ayton Quarry (below). sedimentation were not continuous along the section. When specimens of massive coral are sectioned further local growth discontinuities are evident that are not visible in the field. But such interruptions probably represent only a few years at most. The estimated age for the coralliferous unit is 359 years giving an average growth rate of 3.7 mm yrW (This estimate is based on seven sections: data in Ah, unpublished Ph.D. thesis, University of Reading, 1978.) The section at Ayton Quarry was described and figured by Blake and Hudleston (1877) and referred to by Wright (1972). The coralliferous unit is composed of biolithite with lens-shaped colonies of 1 0 m height forming discrete but closely packed hemisphaeroids averaging 30 cm across, the roundheads of Blake and Hudleston (1877), with little matrix (biomicrite). There are rare colonies of Rhabdophyllia. The growth rate of Fungiastraea arachnoides forming the column in the centre is about 2-4 mm yr 1 and the thickness of this column is about 1-3 m. The minimum period of time required for the formation of the coralliferous unit at Ayton is about 550-600 years. CARBONATE PRODUCTION Calcium carbonate production today varies greatly from temperate to tropical environments (Chave et al. 1972; Smith 1970, 1971; Stearn et at. 1977; Bosence 1980). The data for these studies are based on some assumptions and estimations because of the complexity of the processes within reefs and carbonate banks. As well as the skeletons of the corals and other frame-builders, their epizoans and associated biota, and the sedimentary particles produced by physical and biological erosion and dispersed within and without the reef by currents and waves must also be evaluated. ALI: UPPER JURASSIC CORALS 545 Chave et al. (1972) defined potential production as the amount of calcium carbonate produced by a single organism, or colony of organisms per unit area of the surface covered by that organism. This definition is useful in providing a unit of measurement by which carbonate-producing abilities of different organisms can be compared. With fossil corals the problem is aggravated because not only is it seldom possible to trace time surfaces with any degree of accuracy through a unit, but primary production has been modified subsequently by diagenesis, solution, and precipitation of cements. text-fig. 7. Part of a thecosmilian cross- section to illustrate method used for esti- mating proportion of skeleton per unit area of corallite. The method to determine skeletal density of the coral species is shown in text-fig. 7. Successive septa were individually traced, using a low-power projection microscope, on to paper, the paper being moved to give a cumulative value of the septal cross-section areas. Epitheca and dissepiments were then added and the total cross-sectional area determined. It is assumed that Thecosmilia had a circular cross-section. The average diameter at Shellingford is about 2-7 cm and aragonite S.G. is 2-94 (Goreau 1963). The skeleton in Thecosmilia occupies about 31-5 % of the cross-sectional area. Skeletal proportions of massive corals were determined by the same method and the mass of CaC03 per unit volume of the organism (g/cm3) obtained for each species. Potential production, expressed as CaC03 mass per unit volume x growth rate (cm ym1) is given in Table 1. It is clear from the above data that the differences in linear growth rates for the Corallian species are represented by much smaller differences in actual calcification rate because the mass per unit volume of the slower-growing colonies is much higher than in those with faster growth rates. The estimated values of potential production of massive coral species from Shellingford and from Ayton (Table 1 ) are quite similar to productivity values of some modern scleractinians off Barbados determined by a different method (Stearn et at. 1977), though the calcification rate of Thecosmilia is considerably greater. The gross production is the amount of CaCOs produced per unit area of the reef (Chave et al. 1972). It is obtained by summing the product of potential production of each organism in a given reef, times the proportion of the reef area covered by the organism. The approximate coverage by 546 PALAEONTOLOGY, VOLUME 27 table 1. Potential production of coral species and gross production at Sliellingford Cross-Roads Quarry and Ayton Quarry (a) ( b ) (a x 2-94) (a x b x 2-94) (d) skeleton growth rate mass per unit potential prodn. mean coverage* gross prodn. per sq. cm cm ym1 vol. g/crn3 g CaC03 m2 ym1 per unit area a x b x 2-9 x d Shellingford, Oxfordshire Thecosmilia annularis Thamnasteria 0-315 0-98 0-926 9074-8 0-301 2731-5 concinna 0-588 0-104 1-729 1797-8 0-091 163-6 Fungiastraea arachnoides Isastraea 0-424 0-235 1-246 2929-4 1 0-128 367-7 explanata 0-312 0-28 0-917 2567-6 ( 3262-8 g CaC03 m2 ym1 Ayton, Yorkshire Fungiastraea arachnoides Thamnasteria 0-424 0-24 1-246 2990-4 0-4 1196-2 concinna 0-588 1-04 1-729 2421 0-28 677-8 Rhabdophyllia phillipsi 0-58 0-5 0-90 4500 0-02y 90-0 1964 0 g Cal 03 m2 ym1 x— based on seven sections (data in Ali, unpublished thesis for Ph.D., University of Reading 1978) Isastraea (2%) included with T. arachnoides. y — estimated. corals at Shellingford is shown in Table 1, together with an estimate of the gross calcium carbonate production of the coralliferous unit. (Where coral cover is extensive gross production on modern reefs is a reasonable approximation of net production: the carbonate permanently retained by the reef after allowance for carbonate dissolution and mechanical gains and losses. In the ancient examples discussed it is appropriate to take net production as equivalent to gross production.) The coralliferous unit at Ayton is composed mainly of Thamnasteria and Fungiastrea , but branching Rhabdophyllia also occurs forming about 2% of the rock. Estimates of calcium carbonate production for this unit are made by the same method and values are shown in Table 1 . Differences in productivity between the two sections are due to small differences in growth rates but mainly to differences in productivity of the species present. DISCUSSION AND CONCLUSIONS In upper Jurassic times scleractinians were still at an early stage of their evolution though some, such as the microsolenids (now extinct) had achieved a high degree of integration. Further, Jurassic corals in northwest Europe mostly formed localized banks or patch reefs on an extensive shelf area rather than fringing reefs. Such banks were subdued structures with amplitudes that could mostly be measured in centimetres. During its formation the coralliferous unit at Shellingford Cross-Roads Quarry probably consisted of scattered low domes of massive corals amongst short heads of candle- like thecosmilians and Rhaxella sponges emerging from a muddy sediment, with an associated fauna of vagile regular echinoids, brittle stars and small gastropods, and attached oysters and other bivalves as the preservable elements. A LI: UPPER JURASSIC CORALS 547 Thus it is surprising that this study shows a style of growth banding and values of growth rate and carbonate productivity that compare well with what is known from areas of modern tropical corals, and productivity an order of magnitude higher than temperate coralline algae (Bosence 1980). Indeed, the values are peculiarly close to the 3-5 kg/CaC03/m2 ym1 characteristic of tropical Pacific reef flats (Kinsey 1979, unpublished Ph.D. thesis. University of Hawaii, quoted by Grigg 1982). The possibility that reefs in the Jurassic extended well beyond the 40° latitude to which they are restricted today (Beauvais 1973) cannot be excluded. But the present study is only preliminary and until similar studies have been carried out, particularly on the diverse coral faunas of the northern margin of Tethys, it would be unwise to consider these results as typical for the Mesozoic. The need for more detailed work on modern corals that can be appled to fossil corals is also evident. Acknowledgements. Drs. D. W. J. Bosence (London), C. T. Scrutton (Newcastle upon Tyne) and J. H. Hudson (Miami) reviewed drafts and I am most grateful for their constructive criticism and advice. The work on which this study is based was carried out at the University of Reading with the aid of a grant from the British Council which is gratefully acknowledged. Dr. R. Goldring (University of Reading) prepared the final version. REFERENCES all o. e. 1977. Jurassic hazards to coral growth. Geol. Mag. 1 14, 63-64. — 1983. Microsolenid corals as rock-formers in the Corallian (Upper Jurassic) of England. Ibid. 120, 375-380. arkell, w. J. 1933. The Jurassic System in Great Britain. Clarendon Press, Oxford. Pp. 681. — 1947. The Geology of Oxford. Clarendon Press, Oxford. Pp. 267. baker, p. A. and weber, J. N. 1975. Coral growth rate: variation with depth. Earth planet. Sci. Lett. 27, 57-61. barnard, l. a., macintyre, i. g. and pierce, j. w. 1974. Possible environmental index in tropical reef corals. Nature, 252, 219-220. beauvais, l. 1973. Upper Jurassic hermatypic corals. In hallam, a. (ed.) Atlas of Palaeobiogeography, 317-328. Elsevier, Amsterdam. blake, j. f. and hudleston, w. h. 1877. The Corallian rocks of England. Q. Jl geol. Soc. Lond. 33, 260-405. bosence, d. w. j. 1980. Sedimentary facies, production rates and facies models for recent coralline algal gravels, Co. Galway, Ireland. Geol. J. 15, 91-1 1 1 . buddemeier, r. w. 1974. Environmental controls over annual and lunar monthly cycles in hermatypic coral calcification. Proc. 2nd int. Symp. on Cora! Reefs, 2, 259-267. Great Barrier Reef Committee, Brisbane. — maragos, J. E. and knutson, d. k. 1974. Radiographic studies of reef coral exoskeletons: rates and patterns of coral growth. J. exp. mar. Biol. Ecol 14, 179-200. — and kinzie, R. A., hi. 1976. Coral growth. Oceanogr. Mar. Biol. Ann. Rev. 14, 183-225. CHAVE, K. E., smith, s. V. and roy, K. s. 1972. Carbonate production by coral reefs. Mar. Geol. 12, 123-140. cope, j. w. c., duff, k. l., parsons, c. f., torrens, h. s., Wimbledon, w. a. and wright, j. k. 1980. A correlation of Jurassic rocks in the British Isles, Part 2, middle and upper Jurassic. Geol. Soc. Lond. Spec. Paper , 15, 109 pp. dodge, r. e., aller, r. c. and Thomson, j. 1974. Coral growth related to resuspension of bottom sediments. Nature, 247, 574-577. — and Thomson, j. 1974. The natural radiochemical and growth records in contemporary hermatypic corals from the Atlantic and Caribbean. Earth Planet. Sci. Lett. 23, 313-322. — and vaisnys, j. r. 1977. Coral populations and growth patterns: responses to sedimentation and turbidity associated with dredging. J. mar. Res. 35(4), 715-730. goreau, t. f. 1963. Calcium carbonate deposition by coralline algae and corals in relation to their roles as reef builders. Ann. N.Y. Acad. Sci. 109, 127-163. grigg, r. w. 1982. Darwin Point: a threshold for atoll formation. Coral Reefs, 1, 29-34. highsmith, r. c. 1979. Coral growth rates and environmental control of density banding. J. exp. mar. Biol. Ecol 37, 105-125. hoffmeister, j. e. and multer, h. g. 1964. Growth rate estimates of a Pleistocene coral reef of Florida. Bull. geol. Soc. Amer. 75, 353-358. Hudson, J. h. 1981. Growth rates in Montastraea annularis', a record of environmental change in Key Largo coral reef marine sanctuary, Florida. Bull. Mar. Sci. 31, 444-459. — shinn, e. a., halley, r. b. and lidz, b. 1976. Sclerochronology — a tool for interpreting past environments. Geology , 4, 361-364. 548 PALAEONTOLOGY, VOLUME 27 isdale, p. 1977. Variation in growth rate of hermatypic corals in a uniform environment. Proc. 3rd. int. Coral Reef Symp. 403-408. Rosenthiel School of Marine and Atmospheric Science, Miami. jefferson, t. h. 1982. Fossil forests from the lower Cretaceous of Alexander Island, Antarctica. Palaeontology, 25, 681-708. Johnson, g. a. l. and nudds, j. r. 1975. Carboniferous coral geochronometers. In rosenberg, g. d. and runcorn, s. K. (eds.) Growth rhythms and the history of the earth's rotation, 27-41 . Wiley, London. knutson, d. k. and buddemeier, r. w. 1973. Distribution of radionuclides in reef corals: opportunity for data retrieval and study of effects. In Radioactive Contamination of the Marine Environment, 735-746. Vienna International Atomic Energy Agency. — smith, s. v. 1972. Coral chronometers: seasonal growth bands in reef corals. Science, 177, 270-272. lewis, j. b., axelsen, F., goodbody, i., page, c. and chislett, g. 1968. Comparative growth rates of some reef corals in the Caribbean. Marine Science Manuscript report no. 10. McGill University. Pp. 260. ma, T. Y. H. 1934. On the seasonal change of growth in a reef coral, Favia speciosa (Dana), and the water- temperature of the Japanese Seas during the latest geological times. Proc. Imp. Acad. Japan (Tokyo), 10, 353-356. macintyre, i. G. and smith, s. v. 1974. X-Radiographic studies of skeletal development in coral colonies. Proc. 2nd Int. Symp. Coral Reefs, 2, 277-287 . Great Barrier Reef Committee, Brisbane. moore, w. s. and krishnaswami, s. 1974. Correlation of x- Radiography revealed banding in corals with radiometric growth rates. Ibid. 269-276. negus, p. e. and beauvais, l. 1979. The corals of Steeple Ashton (English Upper Oxfordian), Wiltshire. Proc. geol. Assoc. 90, 213-227. rosenberg, G. D. and runcorn, s. K. (eds.) 1975. Growth rhythms and the history of the Earth's Rotation. Wiley. Pp. xvi + 559. scrutton, c. T. 1978. Periodic growth features in fossil organisms and the length of the day and month. In brosche, p. and sundermann, j. (eds.) Tidal Friction and the Earth's Rotation, 154-196. Springer. — powell, j. h. 1980. Periodic development of dimetrism in some favositid corals. Acta Palaeont. Polonica. 25, 477-491. smith, a. g., hurley, a. m. and briden, j. c. 1981. Phanerozoic palaeocontinental world maps. Cambridge University Press. Pp. 1620. smith, s. v. 1970. Calcium carbonate budget of southern Californian Borderland, Hawaii Inst. Geoph. Rpt., HIG 70 1 1, 174 pp. — 1971. Budget of calcium carbonate, Southern California continental Borderland. J. sedim. Petrol 41, 798-808. stearn, c. w., scoffin, t. p. and martindale, w. 1977. Calcium carbonate budget of a fringing reef on the west coast of Barbados. 1. Zonation and productivity. Bull. mar. Sci. 27, 479-510. vaughan, t. w. 1915. The geologic significance of the growth rate of the Floridian and Bahaman shoal-water corals. J. Wash. Acad. Sci. 5, 59 1 600. weber, j. n. and white, e. w. 1974. Activation energy for skeletal aragonite deposited by the hermatypic scleractinian coral Platygyra spp. Mar. Biol. 26, 253-259. — weber, p. h. 1975. Correlation of density banding in reef coral skeletons with environmental parameters: the basis for interpretation of chronological records preserved in the coralla of corals. Paleobiology , 1, 137-149. wright, J. K. 1972. The stratigraphy of the Yorkshire Corallian. Proc. Yorks Geol. Soc. 39, 225-266. omer e. ali Department of Geology Typescript received 15 February 1983 University of Khartoum Revised typescript received 6 October 1983 Khartoum CONSTRUCTION AND PRESERVATION OF TWO MODERN CORALLINE ALGAL REEFS, ST. CROIX, CARIBBEAN by DANIEL W. J. BOSENCE abstract. The internal structures of two coralline algal reefs from St. Croix are described. The primary framebuilders are Lithophyllum congestion , which dominates in exposed mid-intertidal situations, and Porolithon pachydermum from the high intertidal. Secondary frameworks are constructed by one of the following corallines: Tenarea, Lithothamnium ruptile , Mesophyllum syntrophicum, Litliopliyllwn congestion , and Neogoniolithon sp., together with Honiotrema and vermetid gastropods. The environmental preferences of these corallines and their recognition in slabbed reef sections permits a reconstruction of past reef morphologies and environments. Predictable ecological successions are found within preserved coralline sequences which correspond with previous settlement plate experiments. The main agents of reef destruction are sponge and echinoid bioerosion. Inter-reef sediments are winnowed by wave currents and reflect the composition of surrounding coral reefs in addition to debris from the coralline algal reefs. A relatively low proportion of coralline algal debris in sediments around the reefs is thought to result from deposition of silt-sized sponge chips of corallines in quieter water elsewhere. Internal reef sediments reflect the composition of the reef constructors. This work has arisen from research on Tertiary coralline limestones of the Tethyan region (Bosence and Pedley 1982; Bosence 1983a, 6), during which it became evident that the interpretation of coralline limestones is limited by the paucity of studies on actuopalaeontology of Recent tropical coralline buildups. The longevity of coralline algal species together with their narrow environmental tolerances makes them valuable palaeoenvironmental indicators in the Tertiary (Adey et al. 1982; Bosence and Pedley 1982). Similarly, late Palaeozoic ancestral coralline reefs require descriptions of Recent counterparts for detailed interpretations. It is here shown that there are two main components involved in reef construction. The major reef framework is constructed by primary framebuilders which support later encrustations by secondary framebuilders. The characteristics of present-day species of coralline can be recognized in thin sections from the slabbed reef material, and this permits reconstruction of the form and the ecological succession in ancient coralline reefs. Study of the modern reef sediments shows that those around the reef do not accurately reflect reef composition, but that those from the internal reef cavities are similar in composition to the reefs. The St. Croix algal reefs (algal ridges) described here (text-fig. 1) were selected for investigation because of the extensive previous work on the present-day corallines by Walter H. Adey (Smithsonian Institution) and his co-workers. Adey (1975) describes the setting, morphology, large- scale internal features (text-figs. 2, 3), and Holocene development of the reefs. Adey and Vassar (1975) describe the colonization, succession, and growth rates of the corallines from a series of settling plate experiments. The environmental control on morphology of the main ridge-building alga Lithophyllum congestion is studied by Steneck and Adey (1976). The systematics and morphology of the corallines are to be described by Adey (pers. comm.). Detailed discussion of the relevance of these earlier studies to this project is included in the appropriate parts of this paper. The ecology of the coralline algae (Adey 1 975; Adey and Vassar 1975) provides the information for interpretation of the internal growth fabrics described below in this paper. In exposed situations in eastern St. Croix, a pavement of Acropora palmetto exists in depths of 1 -2 m. These coral pavements are coated by crusts of the coralline algae Neogoniolithon spp. and Porolithon pachydermum. If the IPalaeontology, Vol. 27, Part 3, 1984, pp. 549-574, pis. 50-52.| 550 PALAEONTOLOGY, VOLUME 27 pavements extend to low-water level then L. congestion dominates up to 20-30 cm above m.I.w.s. tides. Above this height P. pachydermum is the major reef constructor. In the many reef cavities and overhangs the following corallines are important in reef building: Lithothamnium ruptile, Mesophyllum syntrophicum, Neogoniolithon sp., and Tenarea. In addition, the foraminifer Homotrema and vermetid gastropods are common. Two coralline ridges or algal reefs were selected for this study on the basis of those drilled by Adey (1975). Isaacs Reef is located offthe exposed south-eastern shore of St. Croix (text-figs. 1, 2, 4b) and is composed of a number of fused reefs or boilers. The reefs are very cavernous with a number of walls and pillars supporting a roof and crest. Present-day coralline growth is luxuriant (text-fig. 4d, e) and construction by Lithophyllum congestion and P. pachydermum has built the reef to a height of up to 50 cm above m.I.w.s. tides (Adey 1975). Adey and Vassar (1975) give a maximum accretion rate of 5 mm/year for reef crest environments of exposed ridges. Adey’s (1975) drilling indicates coralline build-up over a previous A. palmata reef dated at the base as 4,040 yrs. b.p. The Acropora continues today in 3-4 m of water in front of the algal reefs. Coralline frameworks extending to a depth of 2-5 m were recovered. For the present study a large slot 1 m deep and extending back 1 -5 m from the reef crest was excavated to provide a continuous cross-section (text-figs. 2, 5). BOSENCE: CARIBBEAN CORALLINE ALGAL REEFS 551 text-fig. 2. Plan and cross-section of Isaacs Reef showing location of algal reefs and excavated reef section (after Adey 1975). • sediment samples. 552 PALAEONTOLOGY, VOLUME 27 Shark Reef, in contrast, occurs in the more sheltered Boiler Bay of north-eastern St. Croix (text- figs. 1, 3, 4a). It is further protected by a newly (c. 500 yrs. b.p.) emergent barrier reef extending across the bay (Adey 1975). Shark Reef was one of the most thoroughly studied of Adey’s reefs (Adey 1975, figs. 39, 40). A large A. palmata stand (dated at 2,900 yrs. b.p.) is overgrown by about 1-5 m of coralline reef which extends today to a height of about 15 cm above m.l.w.s. tides. Because of the text-fig. 3. Plan and cross-section of Shark Reef showing location of algal reef and excavated reef section; symbols as for text-fig. 2 (after Adey 1975). BOSENCE: CARIBBEAN CORALLINE ALGAL REEFS 553 reduced hydraulic energy conditions there is little Lithophyllum and Porolithon growth on Shark Reef today (text-fig. 4c) although these algae are important in reef sections (text-fig. 6). Present-day surfaces are covered with non-calcified algae (Abbott et al. 1974) and penetrated by large borings of the echinoid Echinometra. Adey and Vassar ( 1975) have shown present-day accretion rates on these reef crests to be only 1-2 mm/year which apparently cannot keep pace with the heavy Echinometra boring (Adey 1975). As with Isaacs Reef a large slot normal to the reef crest was excavated (1-4 m deep and 1 -2 m back from the crest), adjacent to and extending a previous slot of Adey (1975, fig. 40). The depth/age relationships for these two reefs (Adey 1975) indicate an age of 2,200 yrs. b.p. for the base of my Isaacs Reef section and 3,300 yrs. for the base of my Shark Reef section. The slabbed reef sections (text-figs. 5, 6; Table 1) indicate five major components within the reefs. Primary reef frameworks are constructed by L. congestum and Porolithon. Secondary reef frameworks comprise text-fig. 4. a, Shark Reef (arrowed); Boiler Bay, looking north-east. Note waves breaking on barrier reef across mouth of bay and small waves on beach, b, Isaacs Reef, looking south-west. Coralline algal ridges are clearly visible between waves, c, underwater photograph of Shark Reef crest, showing sparse crustose algal growth, white areas scraped by parrot-fish, and Echinometra in borings; x 0T2. d, surface view of Isaacs Reef crest, showing branching heads of Lithophyllum congestum and columnar growths of Porolithon pachydermum (upper centre, white). Darker tufts are fleshy algae (right of hammer). Echinometra borings (2-5 cm diam.) occur throughout the area; x0-12. e, underwater photograph of front wall of Isaacs Reef (c. 1 m depth), illustrating competitive intergrowth of coralline crusts and vermetid (V) in secondary framework; note Lithotrya (L) borings; x 1. REEF CREST 554 PALAEONTOLOGY, VOLUME 27 text-fig. 5. Section through Isaacs Reef crest, simplified from tracings of impregnated, cut, and polished reef blocks. Reef block labels refer to crest (IRC), wall (IRW), pillar (IRP), and floor (IRF) locations. BOSENCE: CARIBBEAN CORALLINE ALGAL REEFS 555 table 1. Point counts from grid on slabbed reef surfaces (text-figs. 5, 6) indicating percentage abundance of frameworks, borers, sediment, and cavities (‘Total’ column includes counts from slabs and cavities; ‘Preserved’ column is recalculated for the preserved slabs only). Isaacs Reef (1378 counts) Total Preserved Shark Reef (2272 counts) Total Preserved Primary Frameworks 15 41 10 29 Lithophyllum congestum 5 14 2 7 Porolithon 10 27 5 14 Millepora — — 3 8 Secondary Frameworks 1 2 8 20 Tenarea-Homotrema — — 4 10 Sponge chambers 3 9 5 14 Cemented sediment 20 55 14 37 Cavities 61 62 later crusts growing on and within the reef. Borings of sponges and other organisms account for 3% and 5% of Isaacs and Shark Reefs, respectively. Fourthly, cemented internal sediment forms a large part of the sections. Finally, reef cavities account for the ma jority of the cross-sectional areas (61 % and 62% of Isaacs and Shark Reefs, respectively). METHODS The material for this study was collected during January 1980. Large blocks of reef were removed by hammer, chisel, and bar to excavate cuttings normal to the reef crest. Work on Shark Reef was straightforward because of low energy conditions but Isaacs Reef was continuously pounded by 1 -0 to 1 - 5 m waves and ropes were used to maintain a position on the reefs. The reassembled blocks were slabbed along one plane, impregnated with Araldite resin, ground, and polished. Text-figs. 5 and 6 were traced from these prepared surfaces. Thin sections were prepared from further impregnated chips of both reefs for identification of the corallines, successional data, and study of the internal fabrics and sediments. Specimens of reef material were fractured, or polished, and etched, coated with gold for investigation under a JEOL JM 35 S.E.M. operating at 25 kv. Inter-reef sediments were collected with a scoop and sieved at half phi intervals. For all samples, subsamples from each sieve were amalgamated and impregnated. Composition was determined by point counts made on stained acetate peels. THE CORALLINE ALGAE An important part of this investigation involved the identification of the recently preserved coralline algae. The identifications were based on Adey’s representative collection of coralline thin sections together with the original descriptions from decalcified material. Measurements from micrographs provided data on crust thickness, hypothallus thickness, hypothallus and perithallus cell sizes and shapes, heterocyst and conceptacle size and shape for the eight species of coralline from the Cruzan reefs (Table 2). Six of these species were found to be common or abundant on the reefs: Lithothamnium ruptile Foslie, Mesophyllum syntrophicum (Foslie) Adey comb, nov., Lithophyllum congestum (Foslie) Foslie, Porolithon pachydermum Foslie, Neogoniolithon megacarpum n.s. Adey, 556 PALAEONTOLOGY, VOLUME 27 and Tenarea sp. The remainder were rare: Porolithon antillarium Foslie and Neogoniolithon imbrication n.s. Adey. One other common coralline in the reef sections is represented by thin monostromatic crusts with roughly cubic cells (c. 10 x 12 ^m). These occur as early colonizers of hard substrates in the reefs (see discussion below on ecological succession). They possibly correspond to Adey’s genus Leptoporolithon which is a common early colonizer on these reefs. Leptoporolithon occurs as thin REEF CREST secondary framework text-fig. 6. Section through Shark Reef crest, simplified from tracings of impregnated, cut, and polished reef blocks. Reef block labels refer to crest (SRC), wall (SRW), pillar (SRP), and floor (SRF) locations. table 2. Morphology of coralline algae. Measurement of each feature in microns expressed as mean (standard deviation), and range. BOSENCE: CARIBBEAN CORALLINE ALGAL REEFS 557 Xi rd dv OV o VO co Oi w m w | s cd O PP c 6 o T3 £ O C W cd £ £ O g w o C ^ 03 <§■* .S O a JS ~ 3 O ' — CiO t— XJ .5 o M-/ £ £ JS 03 oC ^ tin VO X filaments of 2-3 13(3-3)10-15 558 PALAEONTOLOGY, VOLUME 27 monostromatic crusts or developing a single-layered perithallus with heterocysts similar to Porolithon (Adey and Vassar 1975). No perithallus or conceptacles are present in the preserved monostromatic crusts. Because of the uncertainty over identification of these thalli they are here labelled 'monostromatic crusts’. Full morphological details of each coralline are to be treated by Adey (pers. comm.). The diagnostic features of each species as seen in thin sections are tabulated in Table 2 and illustrated where relevant in Plates 50-52. REEF CONSTRUCTORS Primary frameworks Lithophyllum congestion framework. Branching L. congestion may be abundant about 20 cm below to about 20 cm above m.l.w.s. tides in exposed ridges. The most exposed and therefore the highest ridges have the greatest cover of L. congestion (Adey 1975; Steneck and Adey 1976). Of the two ridges investigated in the present study only Isaacs Reef supports live branching L. congestion (text-fig. 4d). This occurs as a 60 cm thick band on the outer crest of the reef. The lower part of this band has up to 2 cm thick crusts bearing incipient branches of L. congestion overgrowing old bored, infilled, and cemented reef (text-fig. 5, slab IRW1; PI. 51, fig. 6). This represents new growth over a small slot cut by Adey in 1973 and gives an actual vertical crust accretion rate for crusts of 1 mm/yr. The branching Lithophyllum framework occurs as 5-3% of the area of the section (but 14% of the area of preserved reef — see table 1). The framework is preserved within the crest and the lower parts of the reef crest blocks (text-fig. 5, slabs IRC3-IRC7). Scattered areas are also preserved within the reef pillars and reef cavity floor (slabs IRP1 -IRP2, IRF1). In contrast. Shark Reef, which has a similar porosity of just over 60%, has large areas of well-preserved branching framework in the front wall (text-fig. 6, SRW1; PI. 50, fig. 3) but only small patches in other interior regions. The percentage area occupied by the framework is 2-5% or 14% of the preserved reef (area of slabbed surface). L. congestion is preserved as an in situ irregular branching framework with or without a crustose base. Many areas are clearly remnants of previously more extensive frameworks because the margins are bioeroded (PI. 50, fig. I) and replaced by a secondary framework (PI. 50, fig. 5) or cemented sediment. In addition to the areas of framework traced from the slabs, most thin sections contain small erosional relics of branching or crustose L. congestum. EXPLANATION OF PLATE 50 Fig. 1. Polished slab (upper part of IRC6, text-fig. 5) illustrating preserved columnar growth of Porolithon pachydermum succeeding Lithophyllum congestum (lower right). Note well-cemented internal sediment (grey) within centimetres of upper growth surface of reef, x 0-5. Fig. 2. Bases of articulate corallines ( Ampliiroa ) in laminar crust of P. pachydermum. Note horizontal rows of heterocysts (arrow) within perithallial tissue. Thin section micrograph from block SRC3 (text-fig. 6), x 26. Fig. 3. Section of lower lip of Shark Reef crest (SRC1, text-fig. 6). Branching L. congestum framework (upper right) is firstly overgrown by thick secondary framework of Mesophyllum -Tenarea- If omotrema-vennetid association then Tenure a Homotrema - ve r m e t i d association. Echinometra boring (upper right) has geopetal infill of uncemented sediment, x 0-3. Fig. 4. Thin-section micrograph showing detail of Tenarea- Homotrema framework with Homotrema (bottom centre) overgrown by successive crusts of Tenarea. Block SRW2 (text-fig. 6), x 10. Top to right of specimen. Fig. 5. Vertical slabbed section from front wall of Shark Reef (SRW1, text-fig. 6) illustrating well-preserved L. congestum branching framework, small area of Porolithon framework (lower right) and Echinometra borings partly infilled with secondary framework (top centre). Secondary framework overgrows primary framework in lower left. Grey areas are hard cemented internal sediment, x 0-25. Fig. 6. Thin-section micrograph of L. congestum framework illustrating branch fusion. Where filaments from adjacent branches meet end on (centre) a gap remains. Isaacs Reef crest (Block IRC1), x 10. PLATE 50 BOSENCE, Coralline algal reefs 560 PALAEONTOLOGY, VOLUME 27 The initial colonization of L. congestion is in the form of crusts (PI. 51, fig. 6). In lower reef wall positions, a laminar secondary framework of superimposed crusts may develop (e.g. slab IRCW1, text-fig. 5). The crusts may have grown in competition with Homotrema and/or Tenarea and this results in alternating layers of these organisms and L. congestion in vertical section (PI. 51, fig. 6; text-fig. 8Dii). Similarly, crusts of Lithophyllum may be found within laminar and columnar growths of Porolithon (text-fig. 7). Branches arise from crust perithallial tissue to form branching frameworks (PI. 50, fig. 5). Steneck and Adey (1976) give details of branch morphology which varies in different reef niches. Branch sections are frequently oval with minimum diameters of 1 mm and maximum diameters of 23 mm with a mean of 9- 1 mm. Branches are constructed by a dominant central, zoned medullary hypothallus with elongate rectangular cells arranged in a grid of filaments and cell rows. The outer cortex of the branch is thin, with smaller square cells (PI. 51, fig. 1). Branches may divide by apical or lateral division, but commonly fuse together (PI. 50, figs. 5, 6) which increases the strength of the framework. Branch fusion occurs when filaments approach obliquely or laterally. However, when filaments from adjacent branches approach end on there is no fusion (PI. 50, fig. 6) possibly due to a thick apical cuticle (cf. Cabioch 1972). This framework is encrusted by Homotrema , vermetids, Tenarea sp. and monostromatic crustose corallines (PI. 51, fig. 1) which all add to the branching framework. Less common encrusters are serpulids and bryozoans. Sediment, which is soon cemented, is added to the cavities between branches (PI. 50, fig. 5) and is presumably brought in by waves. Many outer surfaces are bored by dendritic algal borings (c. 10 p,m diameter) which may also mark growth discontinuities within branches (PI. 52, fig. 1 ). Surrounding the borings are darkened areas of altered algal tissue which under the S.E.M. are seen to be cells infilled with micrite cement. This close association suggests that micritization of the thallus may be controlled by the algae or organic decay in and around borings (cf. Bathurst 1971, p. 388). Sponge boring is ubiquitous and may occur before or after colonization by the encrusters mentioned above. Porolithon framework. Three species of Porolithon occur in the St. Croix reefs: P. antillarium , P. fragilis , and P. pachydermum (Adey and Vassar 1975). The latter species is the main reef builder and dominates in high reef crest environments (over 20 cm above m.l.w.s. tides). Here it outcompetes L. congestion as Porolithon is particularly well adapted to resist desiccation and high levels of illumination (Steneck and Adey 1976; Littler 1973). On Isaacs Reef P. pachydermum is common both live on the upper reef crest (text-fig. 4d) and as a dense framework throughout the reef interior where it forms 10% of area in cross-section (text-fig. 5). The abundance decreases with depth into the reef as bioerosion increases and cemented sediment becomes dominant. In Shark Reef Porolithon is not EXPLANATION OF PLATE 51 Fig. 1. Thin-section micrograph showing elongate zoned medullary cells and smaller, squarer cortical cells in outer part of branch of Lithophyllum congestion. Encrusting on coralline branches are Homotrema ntbrum intergrown with monostromatic coralline crusts (arrowed). Isaacs Reef crest (IRC1), x 27. Fig. 2. Thin-section micrograph showing detail of lacy intergrowth of Lithothamnium ruptile (note multipored conceptacle, mid left) and Homotrema. Block IRC3 (text-fig. 5), x 35. Fig. 3. Thin-section micrograph of part of Lithothamnium (L)-Tenarea-Homotrema (H)-vermetid framework. Note early colonizers Homotrema and monostromatic crusts (M) on cemented internal sediment (lower right) giving way to Lithothamnium and Homotrema crusts (cf. text-fig. 8 Diii). Block IRW3 (text-fig. 5), x 24. Fig. 4. Mesophyllum (M)-Tenarea- Homotrema vermetid (V) framework in Shark Reef Lip (SRC1; PI. 50, fig. 3; text-fig. 6). Thin-section micrograph, x 10. Top to right of specimen. Fig. 5. Thin-section micrograph of intergrowth of Neogoniolitlion (N) and Homotrema (H) over earlier cemented sediment (upper right). Block SRW2 (text-fig. 6), x 10. Fig. 6. Thin-section micrograph of laminar Lithophyllum-Tenarea-Homotrema-vermetid framework. Sponge- bored reef rock is overgrown in vertical sequence by Homotrema (H), Tenarea (T), L. congestion (L), vermetid gastropod (V) (infilled with aragonite cement), and L. congestion. Block IRW3 (text-fig. 5), x 25. PLATE 51 BOSENCE, Coralline algal reefs 562 PALAEONTOLOGY, VOLUME 27 found living but it occurs in reef sections with a similar distribution to that found in Isaacs Reef. In the reef crest blocks it occurs as 8-4% of the area but Porolithon only makes up 4% of the reef floor blocks. Porolithon frameworks dominate over Lithophyllum frameworks in the inner crest regions whilst Lithophyllum dominates the outer crest and wall locations (cf. text-figs. 5, 6). P. pachydermum occurs as thick crusts with a thin poorly developed hypothallus. The thick perithallus with cells arranged in filaments and with frequent heterocysts and conceptacles, is apparently of limitless growth. The perithallial tissue forms laminar crusts which normally develop into columnar growths (10-1-5 cm high and 1 cm wide) of undifferentiated tissue (PI. 50, fig. 2). The growth of columns can be seen to have been discontinuous with breaks and rehealed surfaces being common. Algal borings (10 /xm) persist throughout the columns suggesting continuous infestation of Porolithon during life (text-fig. 7). A common epiphyte is the articulate coralline Amphiroa (PI. 50, fig. 2; text-fig. 7). Articulate corallines are not usually preserved in situ but here the bases have been overgrown by the live host to become incorporated in the Porolithon tissue. text-fig. 7. Thin-section micrograph and tracing to illustrate structure and succession in primary frame- works (branching to laminar Lithophyllum congestum , then columnar Porolithon pachydermum followed by Lithophyllum and Porolithon crusts), and secondary Lithothamnium and Lithophyllum crusts. Note frequent intergrowths of Homotrema and vermetid gastropods, infilled sponge borings, minor growth breaks and subvertical algal borings in Porolithon. BOSENCE: CARIBBEAN CORALLINE ALGAL REEFS 563 This relationship is not found in any other coralline in the St. Croix section and appears to be host specific. Amphiroa debris is common in the reef sediments. Vermetids and Homotrema are commonly intergrown within the Porolithon framework. Sponge boring is ubiquitous. Secondary frameworks Within the many reef cavities and below overhangs exist associations of encrusting corallines and other epizoans. The term secondary framework is used to describe these cryptic crustose associations because they overgrow or infill cavities in primary frameworks or cemented sediment and they make up a minor part of the two reef sections (0-7% and 7-7% of Isaacs and Shark Reef, respectively). The major taxa involved in the construction of secondary frameworks as seen in thin section are: Lithothamnium ruptile , Mesophyllum syntrophicum, Lithophyllum congestion (crustose forms), Neogoniolithon sp., Tenarea sp., Vermetid gastropods, and Homotrematid foraminifers. Adey (1975) also lists Archaeolithothamnium dimotum as a cryptic coralline but this has not been seen in these sections. table 3. Frequency of pairing (total 77) of taxa in encrusting associations in thin section. Lithophyllum Lithothamnium Mesophyllum congestum ruptile Neogoniolithon syntrophicum vermetid Homotrema Tenarea 6 3 3 5 8 19 Homotrema 5 6 3 4 9 vermetid Mesophyllum 1 2 1 2 syntrophicum 0 0 0 Neogoniolithon Lithothamnium 0 0 ruptile 0 Of the 7 encrusting taxa only some (15 pairs out of a possible 21 pairings) are found together in these frameworks. The crusts normally occur as alternating layers of 3 or 4 taxa in an association. Table 3 shows the frequency of occurrence of pairs (substrate and overgrowth) of encrusting organisms in secondary frameworks. The coralline Tenarea is abundant and ubiquitous (accounting for 63% of pairings) whilst Homotrema and vermetids are respectively less and less common but still ubiquitous. The corallines Neogoniolithon sp., M. syntrophicum , L. congestion , and Lithothamnium ruptile commonly overgrow or are overgrown by Tenarea , Homotrema , and vermetids but do not occur together in secondary crustose frameworks. This pattern of association permits a classification into five principal secondary frameworks in which Tenarea , Homotrema , and vermetids may occur alone or together with one of four other diagnostic coralline taxa, viz. Tenarea- Homo trema-vermehd framework; Lithothamnium- Tenarea- Homo trema-vevmeixd framework; Mesophyllum-Tenarea-Homotrema-ve rmetid framework; Neogoniolithon-Tenarea-Homotrema-ve rmetid framework; Lithophyllum-Tenarea-Homotrema-vermetid framework. All these frameworks are constructed by laminar to foliaceous crusts of corallines intergrown with Homotrema and vermetids. The competitive intergrowth of these corallines can be seen today in close up underwater photographs (text-fig. 4e) from overhangs and reef cavities. In vertical thin sections an interfingering growth of adjacent encrusters indicates competitive growth (PI. 51, figs. 2-6). Cavities within these frameworks may be common where thin, foliaceous crustose growth occurs ( Lithothamnium-Tenarea-Homotrema-vermetid framework) or rare where thick laminar crusts are 564 PALAEONTOLOGY, VOLUME 27 involved ( Neogoniolithon-Tenarea-Homotrema-vermetid framework). Similarly, bioerosion is commoner in the latter, denser frameworks than in the former, more open frameworks. Cavities are either open, infilled with sediment, or filled with fans of aragonite cement (PI. 51, fig. 6). The secondary frameworks listed above are found in different environments on and within the reefs. The Tenarea dominated framework (PI. 50, figs. 3, 4) is commonest on the underside of Shark Reef crest (SRC1, text-fig. 6; PI. 50, fig. 3) and in what is interpreted as a fossil reef crest on block SRW2 (text-fig. 6). In both situations it overgrows a branching head of Lithophyllum congestion. Occasional thin crusts of this association (but with Homotrema dominating) are found over growth breaks in the primary frameworks. The Lithothamnium-Tenarea-Homotrema-vermetid framework is found in both reef sections in similar successional positions. Most occurrences follow erosion or overgrow fresh Lithophyllum congestion or P. pachydermum frameworks. Other occurrences are where it alternates with the Lithophyllum-Tenarea-Homotrema-vennetid framework. Normally this association occurs in reef cavities (i.e. infilling Echinometra borings: SRC2, text-fig. 6). However, with sections deeper within the reef it is not possible to reconstruct the surrounding niche due to later boring and reef growth. The Neogoniolithon- dominated association is found in the lower reef front wall of Shark Reef (SRW 2, text-fig. 6) and the Mesophyl lum-dommated association occurs in cryptic habitats on both reefs (PI. 50, fig. 3; PI. 51, fig. 3). This association occurs with the large Tenarea-Homotrema- vermetid framework on the underside of the Shark Reef crest. It is also found on blocks on the floor of reef cavities (text-fig. 5, IRF1; text-fig. 6, SRF1 ). The main occurrence of the Lit hophyllum-Tenarea- Homo trema-vermetid framework is in the recently exposed areas on Isaacs Reef crest (text-fig. 5, IRW2-IRW4; PI. 51, fig. 6). This occurrence represents the first stage in the development of the L. congestion branching framework and has been described above. Discussion The occurrence of coralline frameworks in the reefs is shown above to be closely tied to the specialized ecological requirements of the different coralline algae. Therefore, the presence of preserved frameworks in the reef sections (text-figs. 5, 6) can be used to reconstruct ancient reef morphology and environments. In both reef sections there is no evidence of any major erosion surfaces or breaks in growth. Therefore, reef growth is considered to be more or less continuous. Isaacs Reef. The presence of preserved remnants of branching Lithophyllum and Porolithon frameworks in the interior of blocks at the bottom of the Isaacs Reef section (IRF1, text-fig. 5) indicate the former (c. 2,200 b.p.) presence of a high-intertidal reef forming in high energy conditions. Nothing is preserved of the shape or position of the reef crest. The base of the present-day ridge crest blocks (particularly IRC3, 4, 6, 7, text-fig. 5) records the change from an exposed mid-intertidal Lithophyllum reef to a high-intertidal Porolithon reef. The downward-curving Porolithon growth laminae in block I RC3 suggest a former outer edge to this reef crest. A high intertidal Porolithin reef crest has continued through to the present day with Lithophyllum on the front of the seaward edge. Shark Reef. Extensive drilling by Adey (1975) showed that Shark Reef was founded on heads of Acroporci palmata and Millepora. The present section penetrates to the top of the MUlepora colonies. Overgrowing the Millepora are alternating frameworks of Lithophyllum and Porolithon (blocks SRF1, 2, text-fig. 6). The inner part of block SRW2 (text-fig. 6) shows a branching Lithophyllum framework overgrown by 2-6 cm thick convex-outwards secondary frameworks of both Tenarea and Mesophyllum-Tenarea-Homotrema-vermedd type. These frameworks are only found together today on the outer crest and lip of the sheltered Shark Reef (block SRC1, text-fig. 6). Therefore, this structure may be interpreted as an ancient reef lip developing as a response to quieter hydraulic conditions. This sheltered reef crest is succeeded within a few centimetres by Lithophyllum then Porolithon and then nearly 50 cm of well-preserved branching L. congestion. This represents a return to an exposed intertidal reef crest, (bock SRW 1 , text-fig. 6). The remaining 30 cm of reef crest blocks record a high intertidal Porolithon crest with occasional intergrowths of branching Lithophyllum. The upper reef surface today is not concordant with this growth framework. Extensive Echinometra, BOSENCE: CARIBBEAN CORALLINE ALGAL REEFS 565 sipunculid, Lithotrya, and algal borings are eroding this surface and no live Porolithon has been recorded. These sheltered conditions are also recorded by the thick Tenarea-Homotrema-ve rmetid and Mesophyllum-Tenarea-Homotrema-\e rmetid secondary frameworks on the outer and under surface of the lip. Adey (1975) has suggested that the sheltered conditions are due to the build up of an A. palmata barrier reef across the mouth of Boiler Bay (text-fig. 4a) in the last 500 years. ECOLOGICAL SUCCESSION The change, through overgrowth, of one association of corallines to another is recorded in vertical sections in the reef slabs. Large-scale changes in frameworks have been discussed in the preceding sections. The reef frameworks in thin section illustrate ecological succession on a smaller scale. This is shown where fresh (broken) substrates are colonized or where new individuals overgrow and replace previous corallines. In this investigation, 350 overgrowths were recorded and the nature of the junction (bored, broken, simple) was observed. Overgrowths by the same species were not recorded. The successions were analysed by the probability of one taxon being overgrown by another. The sequences were studied within primary frameworks, overgrowths to primary frameworks, and secondary frameworks in the reef crests, walls, and interiors (text-fig. 8). Succession within primary frameworks In primary frameworks the commonest growth sequences, with or without a time break (which is indicated by boring into the substrate), is from laminar to branching Lithophyllum followed by an overgrowth of first laminar then columnar Porolithon. Columnar Porolithin is most commonly overgrown by Lithophyllum crusts (text-figs. 7, 8a). These observations confirm the large-scale association of these two frameworks and, because of the differing tolerances to exposure of these taxa, reflect fluctuating conditions of emergence. This could result from breakage of portions of reef in storms (there is evidence of this happening on the present-day reefs) or fluctuating sea level. Adey and Vassar (1975) measure accretion rates of 1 -5 mm/yr which means that successions without a time break represent changes measured in years or tens of years. This short time span suggests that growth or breakage of reef crest sections is the most likely cause of local alteration of conditions of exposure. Primary framework encrusters The frequencies of overgrowths by encrusters on the primary reef frameworks are illustrated in text- fig. 8b and c. Breaks within Porolithon thalli are colonized mainly by Tenarea and Homotrema which are also the commonest encrusters on the preserved outer surfaces. None of the growth breaks within Lithophyllum was found to be colonized by other taxa. This may reflect the rapid growth rate of this coralline (up to 8 mm/yr, Steneck and Adey 1976). The commonest encrusters, mostly over dead bored branches of Lithophyllum , are Homotrema , Tenarea , and monostromatic crusts. Secondary frameworks An analysis of secondary frameworks identified five associations of encrusters. Within these associations there were alternations of overgrowth resulting in the accreting frameworks. This analysis records the sequence of overgrowths within these frameworks and between one association and another. Results of recording successions from the reef crests and interior cavities from both reefs were similar and have been combined for discussion (text-fig. 8d). Overgrowths in the Shark Reef wall were markedly different and are discussed separately (text-fig. 8e). Reef crest and wall and interior cavities are characterized by a large number of combinations (36 out of a possible 72 overgrowths) of a relatively small number of encrusters (8 corallines + Homotrema). Text-fig. 8d therefore only shows the commonest recorded sequences. Bored and broken surfaces and reef rock are most commonly overgrown by Homotrema , Tenarea , and monostromatic crusts (PI. 51 , figs. 3, 6). The pattern of subsequent colonization is again dominated by associations of Lithothamnium ruptile or laminar Lithophyllum or M. syntrophicum, all with Tenarea and Homotrema. The 566 PALAEONTOLOGY, VOLUME 27 succession of one of these associations by another is only occasionally found and is most commonly an alternation of laminar Lithophyllum and Lithothamnium frameworks (text-fig. 8Dii, Diii). The Shark Reef wall overgrowth sequences are shown in text-fig. 8e. They differ from the previous overgrowths by the addition of large colonies of Millepora and crusts of Neogoniolithon (PI. 51, fig. 5) and Porolithon. Lithophyllum and Mesophyllum are lacking and Lithothamnium is less frequent. Discussion The present-day succession of corallines on the St. Croix reefs has been studied by Adey and Vassar (1975). Plates placed in well-lit areas in 1-2 m of water in Boiler Bay were studied over a year. At or before 20-60 days Leptoporolithon spp. dominate. Tenarea increases in importance up to 100 days and then Neogoniolithon increases at the expense of the early successional species. If grazing is heavy on plates Leptoporolithon and Tenarea may increase. In shady reef cavities Tenarea , Lithothamnium ruptile , Hydrolithon horgesenii occur. Adey and Vassar (1975) did not record the occurrence of Homotrema. KEY in Porolithon sp. -VtYt Mesophyllum > Lithothamnium syntrophicum ruptile monostromatic L crusts + t + Millepora Lithophyllum congestum Neogoniolithon Homotrema growth break cemented s e di ment text-fig. 8. Ecological succession in coralline algal ridge frameworks. Arrows and numbers give direction and probability of the particular overgrowth illustrated; data from 352 observations of thin sections, a, succession of growth forms in Lithophyllum and Porolithon primary frameworks. B, frequency of encrusters within and on Porolithon frameworks, c, frequency of encrusters on Lithophyllum frameworks. Di-iii, succession within secondary frameworks in reef crests, interior cavities, and wall of Isaacs Reef. Ei, ii, succession in secondary frameworks within Shark Reef wall cavities. BOSENCE: CARIBBEAN CORALLINE ALGAL REEFS 567 A comparison of these successional patterns on plates with the vertical sequences described above (text-fig. 8) indicates many similarities, including the pioneer colonization by Tenarea and monostromatic crusts ( cf . Leptoporolithon), and the later occurrence of Lithothamnium ruptile and Tenarea in cryptic environments. However, I additionally record later colonizations by M. syntrophicum and Lithophyllum congestion on both Isaacs and Shark reefs. Neogoniolithon occurs mainly in the reef wall of Shark Reef which corresponds to the positions of the settlement plates in Boiler Bay (Adey and Vassar 1975). Additional information here comes from the presence of Lithothamnium ruptile overgrowing fresh, branching Lithophyllum congestion in exposed areas. No Lithophyllum or Porolithon settled on the Boiler Bay plates but these appeared after 6 months on plates on the more exposed Fancy Algal ridge (Adey and Vassar 1975). The large degree of similarity between the results from these two investigations provides valuable evidence that ecological succession can be established with considerable accuracy from fossil material. In a study of secondary frameworks from Recent and Pleistocene reefs in Barbados, Martindale (1976) noted no colonization succession in plates or vertical sections of crusts. However, Martindale (1976) does record a larger-scale successional change in fossil crusts over coral colonies. These represent the change from open, well-lit coralline associations ( Lithophyllum , Porolithon, and Neogoniolithon) to cryptic associations of corallines ( Mesophyllum , Lithothamnium , Archaeo- lithothamnium), bryozoans, serpulids, and foraminifers. He interprets these trends as recording the change with burial as the Acroproa palmata thickets build up. In reviewing ecological succession, Connel and Slatyer (1977) propose three process models to explain succession: (1) a facilitation model where successional colonizers modify the environment and make it suitable for later species, (2) a tolerance model where first arrivals are fast growing, rapid dispersers and later animals are better able to utilize resources, and (3) an inhibition model where no replacements occur until damage creates an opening which is then occupied by longer lived genera. The St. Croix data does not fit the facilitation model as colonizers in each case are encrusting a coralline substrate. Similarly, the inhibition model is inappropriate as Adey and Vassar (1975) have shown that succession returns to the pioneers after disturbance by heavy grazing. The available data best fits the tolerance model, for which Connel and Slatyer could find little supportive field evidence. Later organisms which overgrow the pioneers, are larger, more robust and longer lived than the thin, early colonizing monostromatic crusts and Tenarea. Grimes (1977) also suggests that vegetational successions are characterized by the replacement of small rapid dispersers, which are efficient at acquiring new space, by larger more durable and massive forms. BIOEROSION In both reef sections, both primary and secondary frameworks are bored to a variable degree (text-figs. 5, 6; PI. 52). The main characteristics of these borings are similar to those described from Caribbean (Scoffin et al. 1980) and Bermudan (Ginsburg and Schroeder 1973; Scoffin and Garrett 1974) patch reefs with the exception of bivalve borings. Echinoidea Borings made by Echinometra lucunter are common today on the upper surfaces of both reefs (text- figs. 5, 6; PI. 50, figs. 1, 3, 5; see also Abbott et al. 1974). Juveniles occupy spaces between coralline branches and as they grow specimens can be seen to have enlarged their holes by removing adjacent corallines. Borings are commonly up to 3-4 cm in diameter but may occasionally reach 6 cm. They penetrate down from the reef crest and sideways into the front wall of the reefs (text-figs. 5, 6) with sinuous J and U shapes in vertical section. Margins of occupied borings are clean, sharp, and truncated frameworks or cemented reef sediments. Stomach contents of Echinometra from Shark Reef contain 5-10% carbonate material by volume which is composed almost entirely of coarse sand-sized fragments of articulate corallines. Very occasional fragments of Homotrema , vermetids, crustose corallines, and reef rock are also found. Abbott et ah (1974) found that Echinometra grazed indiscriminately on epiphytes in Boiler Bay and record a positive correlation between occurrence of 568 PALAEONTOLOGY, VOLUME 27 taxa in gut contents with abundance of epiphytes on reefs. Grazing traces are common on both reef surfaces. Secondary crustose frameworks occur in both present-day and ancient Echinometra borings (text-fig. 6, slab SRC2; PI. 50, fig. 5). Margins of borings may also contain smaller borings of sponges and algae. Diadema antillarium is also a common grazer on the under surfaces of Shark Reef. Possible Diadema faecal pellets (cf. Hunter 1 977) are found in the coarse fractions of inter-reef sediments (see below). Lithotrya Straight, subvertical borings by this endolithic barnacle are common on both reef crests (PI. 52, fig. 3). They possess a distinctive oval cross-section (maximum diameters 5 and 10 mm; see Ahr and Stanton 1973). Margins to borings are sharp, smooth, and cut through both coralline framework and cemented reef sediment. Sinuosity is only seen where borers have to make a deviation around previous cavities. Borings penetrate to a maximum depth of 10 cm. Sipunculoidea Sipunculid worm borings are common on the upper reef surfaces (text-fig. 6). They are characterized by smooth-walled, sinuous borings which are circular in cross-section (maximum diameter 9 mm), and reach lengths of 4 cm. The borings typically have a rounded blind end within the reef. Clionidae Chambers produced by clionid sponges are ubiquitous in reef slabs (text-figs. 5, 6; PI. 52, figs. 2-4). Empty chambers account for 9 and 13-6% of the preserved blocks from Isaacs and Shark reefs, respectively. Both interior and exterior reef surfaces are bored, together with surfaces or previous larger borings. S.E.M. study reveals the characteristic sculpted surfaces of chambers and the excavated chips (PI. 52, figs. 2-5). The chips form a significant proportion of the fine-grained intra- reef sediments (see below). Algae Endolithic algae are common as borers, particularly within coralline thalli. Their nature and occurrence is described above under Lithophyllum congestion and Porolithon primary frameworks. EXPLANATION OF PLATE 52 Fig. 1 . Thin-section micrograph illustrating bored growth break in Lithophyllum congestion. Isaacs Reef crest (IRC1), x 36. Fig. 2. S.E.M. micrograph of clionid sponge boring with excavated (scalloped) chip and other skeletal debris. Block SRC5 (text-fig. 6), x 436. Fig. 3. Lithotrya (L) borings in reef fragment from Isaacs Reef Crest. Note characteristic smooth oval cross- section. Outer margin of specimen well-bored by clionid sponge(s), x 1. Fig. 4. S.E.M. micrograph illustrating clionid sponge-bored coralline infilled with micrite peloids and skeletal debris with later endolithic algae. Block IRC4 (text-fig. 5), x 458. Fig. 5. S.E.M. micrograph of fresh clionid sponge boring in micritized coralline. Note characteristic pitted surface where chips have been removed. Block IRC4 (text-fig. 5), x 24. Fig. 6. Thin-section micrograph of cemented internal sediment illustrating abundant micrite and spar-cementing peloids, grains of Homotrema , and articulate coralline (lower right). Block SRF1 (text-fig. 6), x 100. PLATE 52 BOSENCE, Coralline algal reefs 570 PALAEONTOLOGY, VOLUME 27 REEF SEDIMENTS Inter-reef sediments The algal reefs are surrounded by coral and coralline algal pavements covered with patches of bioclastic sediment. The sediments are commonly wave-rippled during normal wave conditions and support sparse and patchy populations of Penici/lus , Halimeda, and Thalassia. The size distribution and composition of these sediments is shown in text-fig. 9. The sediments from Isaacs Reef are very coarse sands which are moderately well to poorly sorted. They possess a near symmetric to coarsely skewed frequency distribution, the coarse tail being composed of a small number of large coral, mollusc, or coralline fragments. Constituent composition varies with grain size. Crustose corallines, foraminifers, and intraclasts (mainly cemented internal sediments from reefs) decrease, and corals increase in abundance with decreasing grain size. From a visual impression it appears that the sediment grain composition reflects the abundance of grain producers in the environment (see text- fig. 2), the only exception being Halimeda plates which appear over-represented in the sediments. Clionid sponge chips occur in the finest grades (S.E.M. examination) but the majority are probably too small to be deposited in these wave-swept sand patches. ISAACS REEF forereef SHARK REEF back reef INTRA- REEF INTER-REEF INTRA-REEF INTER-REEF INTER-REEF text-fig. 9. Texture and composition of sediments from Isaacs and Shark Reefs. Size-frequency histograms illustrate typical size distributions; point counts of inter-reef sediments were taken from peels of samples combined from each size class of all field samples; intra-reef sediments point-counted from six thin sections from each reef. BOSENCE: CARIBBEAN CORALLINE ALGAL REEFS 571 Shark Reef sediments have been divided into those from the fore and back reef areas for presentation (text-figs. 3, 9). Both areas have very coarse sand to granule-sized sediments which vary from very well to very poor sorting, the best-sorted sediments being in sand patches of the fore reef region which are being continually transported by wave currents. Small numbers of gravel-sized bioclastic grains result in grain size distributions that are coarsely to strongly skewed. Sediment composition contrasts with that from Isaacs Reef in that Halimeda grains are about twice as abundant at the expense of coral grains. This is matched by an increase in cover of Halimeda around the patch reefs compared with Isaacs Reef. An almost pure Halimeda gravel forms the beach sediment behind the reefs. Also contrasting with Isaacs Reef, is an increase in abundance of coralline grains with decreasing grain size. Intraclasts, some of which resemble Diadema faecal pellets (Hunter 1977), are more abundant in the Shark Reef coarse-grained classes. The dominant 2 modal class does not correspond with a greatly increased abundance of any particular grain type and is presumably due to hydraulic sorting. Intra-reef sediment Large areas of both Isaacs Reef (55%) and Shark Reef (37%) are made up of cemented internal sediment (Table 1). Cemented sediments from various parts of the reef all have very similar characteristics (text-fig. 9; PI. 50, figs. 1, 3, 5; PI. 52, figs. 4, 6). They are all subangular to very angular, fine to medium or medium-grained sands with a packstone texture. The dominant constituents being a brown micrite and spar cements (PI. 52, fig. 6). Peloids are locally common (PI. 52, fig. 4). Skeletal grains make up most of the remainder of these apparently grain-supported sediments. Foraminifers (mainly Homotrema ) dominate (PI. 52, fig. 6) and coralline fragments are the next most abundant. Traces of echinoid, bryozoan, serpulid, and spicule fragments also occur. Discussion In a simplistic view of patch reef sedimentation, a direct correlation might be expected between standing crops of carbonate secreting organisms and their abundance in inter- and intra-reef sediments. Scoffin et al. (1980) have investigated in detail the relationship between carbonate production and inter-reef sediments in a Barbados fringing reef. They find that sediments reflect most closely the primary frameworks of the reef. They also note that the sands are very well mixed in sediment patches across the Bellair reef. No carbonate production figures are available for Isaacs and Shark Reefs but a rough comparison can be made of abundance of grain types in inter-reef sediments with a visual impression of standing crops of producers. The only exception to this is the over- representation of Halimeda plates. This is well known from other regions (Garrett et al. 1970) and is accounted for by the rapid growth rate of this plant (Neumann and Land 1975). Evidence that bioerosion is a major source of sediment production from the reefs has been presented above. The removal of reef frameworks by sponges would be expected to result in a large proportion of coral and coralline algal grains in the silt size class of reef sediments. However, this size grade is not present as these sediments are continously wave washed and silts are probably carried out into deeper quiet waters. This was found to be the case in similar reefs by Moore and Shedd (1977). The increased abundance of primary framework grains in the 1 to classes in Shark Reef (corals and coralline algae) and Isaac Reef (corals) may be accounted for by bioerosion as both Diadema and parrot fish produce grains of this size (Schoffin et al. 1980). The higher energy conditions at Isaac Reef exclude these bioeroders from the coralline ridges which may explain the decrease in abundance of corallines in the 1 to classes at this location. Mechanical abrasion of coralline branches and crusts could account for the greater abundance in the coarser-size classes. Intra-reef sediments are significantly different in both texture and composition from inter-reef sediments in both localities. The consistent finer grain size of the internal reef sediments (cf. Garrett et al. 1970) may partly reflect the exclusion of larger particles from entering internal cavities and partly the high intra-reef production of silt-sized sponge chips (PI. 52, figs. 2-4). Grains which are well represented both within and outside the reefs are derived from organisms constructing or living on the reefs: coralline algae, molluscs, foraminifers, and echinoids, with the addition of Millepora in 572 PALAEONTOLOGY, VOLUME 27 Shark Reef. The remaining coral and Halimeda grains are only common in the inter-reef areas. The exclusion of Halimeda and coral fragments from the internal cavities may be explained by their occurrence in the lower sea floors around the reefs and suggests that waves do not carry these grains up into the reef interstices. However, with burial of the reefs they may be expected to be deposited above the internal sediments derived from the reefs. The earlier investigation of coralline patch reefs by Ginsburg and Schroeder (1973) allows some comparisons. External reef sediments were not studied but internal reef sediments contrast with those from St. Croix reefs. The textures of the internal sediments of the Bermuda reefs varied from silt wackestones through to coarse grainstones. The sand-sized material appears to be largely derived from the reef framework and reef encrusters with grain abundance decreasing in the following order: crustose and branching corallines, Homotrema , molluscs and other foraminifers, Halimeda , echinoid, serpulid, and ostracod. This compares well with the internal sediment grains (all sizes) identified in the Cruzan reefs (text-fig. 9). However, the silt-sized debris was found to be quite different in the Bermudan reefs being composed of fragile and encrusting reef organisms mixed with planktonic organisms. This population was not seen in either of the reefs investigated here, in thin section or under S.E.M. Comparisons of inter-reef sediments with other Caribbean examples (mainly coral reefs) show that they are texturally similar (Milliman 1967, 1969; Scoffin el al. 1980). These earlier studies have also found coarsely skewed (i.e. winnowed) coarse sands and gravels forming near-reef sediments. Compositionally it might be expected that the coralline algal reef sediments would differ from coral reef sediments. However, this is not the case (see Milliman 1973 for review). Other Caribbean reef sediments are dominated by either coral, coralline, and Halimeda grains and often show greater proportions of coralline grains than is found in the St. Croix coralline reef sediments. This apparent anomaly might be explained firstly by the abundance of coral stands around the Cruzan algal reefs (text-figs. 2, 3) and secondly by the fact that most of the erosion of the coralline framework is considered to be in the form of silt-sized sponge chips which are not deposited in the inter-reef sediments. COMPARISONS AND CONCLUSIONS Comparisons Previous studies of shallow coralline reefs have been undertaken in the Caribbean ( Boyd et al. 1 963), the Mediterranean (Thornton et al. 1978), and Bermuda (Ginsburg and Schroeder 1973). Each study describes subcircular patch reefs (cup reefs), which may coalesce to form a lobate ridge. In vertical section, preferential growth of the corallines is visible on the outer edge which may result in a raised rim or crest and an overhanging lip. The upper surface of the patch reefs is generally dish-shaped, with less coralline growth and greater bioerosion in the central depressed area. With the exception of Ginsburg and Schroeder (1963) none of the previous authors have described the nature of the coralline reef framework. The corallines vary with location and sample sites on the reefs: Boyd et al (1963) collected samples from the reef sides and reported Neogoniolithon ( Goniolithon ) solubile, Archaeolithothamnium episporum , Lithothamnium sejunctum , Epilithon membranaceum. Thornton et al. (1978) reported a framework of thin crusts of N. notarisii with some LithophyUum indicating a similarity to the well-known Mediterranean ‘trottoirs’ (Peres 1967; Laborel 1961). The most detailed study is by Ginsburg and Schroeder who describe frameworks of crustose corallines, Millepora , and vermetids. The Bermudan reefs are similar in many respects to those from St. Croix but differ in their greater height (8-12 m) and lower porosity. Coralline frameworks are constructed by laminar and columnar Neogoniolithon sp. with lesser Mesophyllum syntrophicum. They do not show the diversity of taxa or morphologies described in this paper. The complex sequences of reef growth, destruction, internal sedimentation, and cementation characteristic of patch reef formation (Schroeder and Zankl 1974; Scoffin and Garrett 1974) are common to the two examples. Some similarities are also apparent with intertidal ledges built by crustose corallines in the Caribbean (Gessner 1970), on the Brazilian coast (Kempf and Laborel 1968), and in the BOSENCE: CARIBBEAN CORALLINE ALGAL REEFS 573 Mediterranean (Peres 1967; Laborel 1961). A comparison with the Pacific algal ridge systems (Emery etal. 1954) and deep-water coralline build-ups from the Mediterranean (Laborel 1961; Laubier 1966) is made in a recent review on coralline reef frameworks (Bosence 19836). Conclusions 1 . The internal structure of two recent coralline reefs has been studied and nearly all the coralline algae living on the surface can be identified in thin sections from the reef interior. The occurrence of these preserved corallines can be used for palaeoenvironmental reconstruction. 2. Two main primary framebuilding coralline algae (L. congestum and Porolithon pachydermum ) which today construct the intertidal algal ridge were abundant in the past in both reef sections. 3. Secondary reef frameworks occur in five associations of corallines (with Homotrema and vermetid gastropods) and each association is characteristic of different sub-environments on the reef and can thus be used in palaeoenvironmental reconstruction of the reef surface. 4. The occurrence of preserved primary and secondary frameworks in the slabbed reef sections permits a generalized reconstruction of past reef morphology and environment. 5. Isaacs Reef has alternated from a high- ( Porolithon dominated) to mid-intertidal ( Lithophyllum dominated) ridge during the last c. 2,200 years. 6. Shark Reef has had a more varied history alternating from an exposed intertidal reef to two periods of sheltered conditions during which a thick subtidal lip of secondary framework grew and the upper reef surface became reduced by bioerosion. 7. A predictable ecological succession of taxa and growth forms, and of subsidiary encrusters, is found in the primary frameworks. 8. Vertical sections through secondary frameworks illustrate predictable palaeoecological successions which equate well with previous settlement-plate experiments. Small fast-growing and presumed rapid dispersers are replaced by more massive, framework-building associations. 9. The major form of erosion on the reefs is bioerosion which is carried out mainly by clionid sponges and echinoids. 10. Inter-reef sediments are well- to poorly sorted coarse sands and gravels with symmetric to coarsely skewed distributions. Size ranges reflect the locally derived reef material and continual movement and winnowing by wave currents. 1 1 . Compositionally, the inter-reef sediments reflect the composition of the reef frameworks, but are relatively depleted in coralline debris. Surprisingly, the sediments have a similar composition to Caribbean coral patch reefs. The relatively small amount of coralline material in the sediments is thought to be due to sponge erosion releasing silt-sized debris which is deposited in quieter waters away from the reefs. 12. Internal reef sediments are well-cemented and reflect the composition of the reef frameworks. Acknowledgements. The assistance, advice, and previous work by Walter Adey and Bob Steneck (Smithsonian Institution) were invaluable in planning this project and I thank them also for reading an earlier version of this paper. Staff and students from the West Indies Laboratory, St. Croix, provided logistical support for fieldwork which made the study possible. Finance was provided by Goldsmiths’ College Research Committee and is gratefully acknowledged. I thank Tom Easter, Owen Green, John Maddicks, and Doreen Norman at Goldsmiths' for technical and secretarial help. REFERENCES Abbott, d. p., ogden, j. c. and abbott, i. a. 1974. Studies on the activity pattern, behaviour and food of the echinoid Echinometra lucunter on beachrock and algal reefs at St. Croix, U.S. Virgin Islands. West Indies Lab., spec. Publ. 4, 1-111. adey, w. h. 1 975. The algal ridges of St. Croix: their structure and Holocene development. A toll. res. Bull. 187, 1-67. - townsend, r. a. and boykins, w. t. 1982. The crustose coralline algae of the Hawaiian Archipelago. Smith. Contrib. mar. Sci. 15, 1-74. and vassar, J. m. 1975. Colonisation, succession and growth rates of tropical crustose coralline algae (Rhodophyta, Cryptonemiales). Phycologia, 14, 55-69. 574 PALAEONTOLOGY, VOLUME 27 ahr, w. m. and standon, r. j. 1973. The sedimentological and palaeoecologic significance of Lithotrya , a rock boring barnacle. Jl sediment. Petrol. 43, 20-23. bathurst, R. G. c. 1971 . Carbonate sediments and their diagenesis. Elsevier, Amsterdam. Pp. 620. bosence, d. w. j. 1983a. Coralline algae from the Miocene of Malta. Palaeontology , 26, 147-173. 19836. Coralline algal reef frameworks. Jl geol. Soc. London , 140, 365-376. — and pedley, h. m. 1982. Sedimentology and palaeoecology of a Miocene coralline algal biostrome from the Maltese Islands. Palaeoclimatol., Palaeoecol ., Palaeogeogr. 37, 9-43. boyd, d. w., kornicker, l. s. and rezak, r. 1963. Coralline algal micro-atolls near Cozumel Island, Mexico. Contrib. Geol. Dept. Wyoming Univ. 2, 105-108. cabioch, j. 1972. Etude sur les Corallinacees,II: morphogenese; consequences systematiqueset phylogenetiques. Cah. Biol. mar. 12, 137-288. connel, j. h. and slatyer, r. o. 1977. Mechanisms of succession in natural communities and their roles in community stability and organisation. Amer. Nat. Ill, 11 19-1144. emery, t. o., tracey, j. i. and ladd, h. s. 1954. Geology of Bikini and nearby atolls. I. Geology. Prof. Pap. U.S. Geol. Surv. 260A, 1-265. garrett, p., smith, d. l., wilson, a. o. and patriquin, d. 1970. Physiography, ecology and sediments of two Bermuda patch reefs. J. Geol. 79, 647-668. GESSNER, F. 1970. Lithothamnium— Terrassen im karibischen Meer. bit. rev. ges. Hydrobiol. 55, 757-762. ginsburg, R. n. and schroeder, ). h. 1973. Growth and submarine fossilization of algal cup reefs, Bermuda. Sedimentology , 20, 575-614. grimes, j. p. 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. Amer. Nat. Ill, 1169-1194. hunter, i. g. 1977. Sediment production by Diadema antillarium on a Barbados fringing reef. Proc. 3rd Int. Coral Reef Symp. Miami , 2, 105-109. rempf, m. and laborel, j. 1968. Formations de vermets et d’Algues Calcaires sur les Cotes du Bresil. Rec. Trav. Stn. Mar. Endoume, 43, 9-23. laborel, j. 1961. Le concretionnment, algal ‘Coralligene’ et son importance geomorphologique en Mediterranee. Ibid. 23, 37-60. laubier, l. 1966. Le Coralligene des Alberes. Monographic biocenotique. Ann. Inst. Oceanogr. 43, 137-316. littler, m. 1973. The population and community structure of Hawaiian fringing-reef crustose Corallinaceae (Rhodophyta, Cryptonemiales). J. exp. Mar. Biol. Ecol. 11, 103-120. martindale, w. 1976. Calcareous encrusting organisms of the Recent and Pleistocene reefs of Barbados, W.I. Unpublished Ph.D. Thesis. Univ. of Edinburgh, 156 pp. milliman, j. 1967. Carbonate sedimentation on Hogsty reef, a Bahamian atoll. Jl sediment. Petrol. 37, 658-676. — 1969. Carbonate sedimentation on four south-western Caribbean atolls. Trans. Gulf Coast Ass. Geol. Soc. 19, 195-206. — 1973. Caribbean coral reefs. In: jones, o. a. and endean, r. (eds.) Biology and geology of coral reefs. Academic Press. Pp. 410. moore, c. h. and shedd, w. w. 1977. Effective rates of sponge bioerosion as a function of carbonate production. Proc. 3rd. Int. Cora! Reef Symp. Miami , 2, 499-505. neumann, a. c. and land, L. s. 1975. Lime mud deposition and calcareous algae in the Bight of Abaco, Bahamas: a budget. Jl sediment. Petrol. 45, 763-786. peres, J. M. 1967. The Mediterranean benthos. Oceanogr. Mar. Biol. ann. Rev. 5, 449-533. schroeder, j. h. and zankl, h. 1974. Dynamic reef formation: A sedimentological concept based on studies of Recent Bermuda and Bahama reefs. Proc. 2nd Int. Cora! Reef Symp. Brisbane , 3, 413-428. scoffin, t. p. and garrett, p. 1974. Processes in the formation and preservation of internal structure in Bermuda patch reefs. Ibid. 2, 429-448. — STEARN, C. W„ BOUCHER, D., FRYDL, P., HAWKINS, C. M., HUNTER, I. G. and MACGEACHY, J. K. 1980. Calcium carbonate budget of a fringing reef on the west coast of Barbados. Pt. 2: Erosion, sediments and internal structure. Bull. mar. Sci. 30,475-508. steneck, r. s. and adey, w. h. 1976. The role of environment in control of morphology in Lithophyllum congestion, a Caribbean algal ridge builder. Botanica mar. 59, 197-215. thornton, s. E., pilkey, o. H. and lynts, G. w. 1978. A lagoonal crustose coralline algal micro-ridge: Balivet el Bibane, Tunisia. Jl sediment. Petrol. 48, 743-750. d. w. j. bosence Department of Geology University of London Goldsmiths’ College Rachel McMillan Building Creek Road, Deptford, London SE8 3BU Laboratory, St. Croix, U.S. Virgin Is. Typescript received 5 April 1983 Revised typescript received I December 1983 Contribution No. 93 from the West Indies THE POSTC RAN I A L SKELETON OF THE UPPER TRIASSIC SPHENODONTID PLANOC EPHALOSAU RUS ROBINSON AE by n. c. fraser and g. m. walkden Abstract. The postcranial skeleton of the Triassic sphenodontid, Planocephalosaurus robinsonae, is described from dissociated remains recovered from the type locality at Cromhall Quarry, South Gloucestershire. A full reconstruction is outlined and its relationships within the Sphenodontidae are briefly discussed. A lower Jurassic eosuchian, Gephyrosaurus bridensis , is shown to share a number of characteristics with P. robinsonae , and Gephyrosaurus is consequently considered to be either a member of the Sphenodontidae or an offshoot from the stem Sphenodontidae. Abundant dissociated sphenodontid and archosaurian reptile remains are known from the Triassic fissure deposits of Cromhall Quarry, South Gloucestershire (Robinson 1973; Fraser 1982; Fraser and Walkden 1983). The skull of the most common of the sphenodontids, P. robinsonae , was recently described by Fraser (1982) and this paper deals with the postcranial skeleton of the same species. Whilst a number of similar-sized reptiles are represented in the deposits the much greater preponderance of Planocephalosaurus elements (text-fig. 1) aids in their separation from the remaining material. However, because the Cromhall sphenodontids have similar postcranial structures, it is still possible to confuse some elements with juveniles of the larger Clevosaurus hudsoni, particularly in the more poorly preserved fossils. To avoid such difficulties only elements from a single site, where Clevosaurus is rare, have been considered. At this site (fissure four, text-fig. 1) archosaurs constitute the major percentage of the non -Planocephalosaurus material and are readily distinguished from sphenodontid elements. Preservation of the bone is generally excellent although few bones are absolutely complete (Pis. 53 and 54 illustrate the typical nature of the material). The numbers of bones recovered which are more than half complete are shown in the Appendix. In addition, hundreds of smaller, yet still quite readily identifiable, fragments have been sorted from the residue and examined. In order that complete bones could be illustrated, most of the reconstructions have been based on more than one specimen, but the major part of any reconstruction is represented by a single specimen which is the one referred to in the legend. Although most of the skeleton of Planocephalosaurus is represented some of the more fragile elements are either incomplete or not known at all. Despite this, a reliable reconstruction has been made which shows Planocephalosaurus as a lizard-like animal (text-fig. 2) with a lightly built skeleton indicating agility and swift action in prey capture and predator avoidance. From a study of its dentition (Fraser and Walkden 1983), Planocephalosaurus was considered to have been primarily insectivorous, although possibly capable of taking newly hatched specimens of small sphenodontids if the opportunity arose. AXIAL SKELETON The vertebrae are generally quite well preserved although the articulation facets for both the ribs and the adjacent vertebrae are commonly a little eroded and the neural spines are usually incomplete. Amongst the vertebrae can be recognized the usual cervical, dorsal, sacral, and caudal elements. [Palaeontology, Vol. 27, Part 3, 1984, pp. 575-595, pis. 53-54.| 576 PALAEONTOLOGY, VOLUME 27 but because the material is completely dissociated it is not possible to determine the exact number of vertebrae in each region. For the purposes of the reconstruction it has been assumed that there are twenty-five presacrals, two sacrals, and between thirty and thirty-six caudals corresponding to the distribution in the only extant sphenodontid, Sphenodon. The centrum is of the notochordal amphicoelous type throughout. There is a rudimentary zygosphenic articulation (text-fig. 9 a: PI. 53, fig. 7) with the development of a zygosphene and zygantrum. The element that was tentatively designated as the epipterygoid of Planocephalosawus (Fraser 1982) is now known to be one of a pair of elements that met in the midline dorsal to the neural canal and together formed the atlas neural arch (text-fig. 3). A ventral process on each element articulated with the odontoid process medioventrally and with the atlas intercentrum ventrally. There is a postzygapophysial articulation with the axis, but the anterior process bears no facets and connective tissue probably attached it to the skull. The axis prezygapophysis takes the form of a simple circular facet that is directed dorsolaterally and overlapped by the atlas neural arch. The odontoid process is formed by the fusion of the atlas and axis centra (text-fig. 4); a faint suture on the dorsal surface is the sole remaining evidence of their separate centres of ossification. The atlas intercentrum has not been positively identified but that of the axis has been fused to the centrum and bears a rib parapophysis on each side. The diapophyses for the axial ribs lie on the centrum. As previously stated it is difficult to assess the precise number of cervical vertebrae; Sphenodon has eight, but the Jurassic sphenodonlids such as Homeosaurus and Kallimodon have just seven. In the reconstruction (text-fig. 2) Planocephalosawus has been shown with eight. Excluding the atlas and axis, the cervicals are typically short with widely spaced zygapophyses angled at approximately 40° (text-fig. 5; PI. 53, fig. 3). In the first one or two elements the parapophyses are situated on the edge of the intercentrum and separate diapophyses occur along the centrum/neural arch boundary, slightly posterior to the level of the parapophyses. The corresponding cervical ribs display separate capitula and tuberculae. The diapophyses on the more posterior cervical vertebrae show a tendency to form an elliptical-shaped facet that is elongated in a posterodorsal-anteroventral direction. These ’00 Site 1 a b c d e f g h ’00 Site 3 abode f g h text-fig. 1 . The abundance of the predominant reptile genera at each of six fossiliferous fissure deposits at Cromhall Quarry, □--spheno- dontid, □ - non-sphenodontid, a — Planocephalosawus, b — Clevo- saurus, c — a pseudosuchian, d— a small sphenodontid, e— a primitive crocodile, f — Sigmala, g— Kue line os aur us, h— a pseudosuchian. FRASER AND W A L K DEN: UPPER TRIASSIC SPHENODONTID 577 oz (N 6 578 PALAEONTOLOGY, VOLUME 27 diapophyses articulated with enlarged tuberculae on the ribs but posteriorly the small capitulum soon becomes reduced until it fails to articulate with the parapophysis and both are subsequently lost. Hoffstetter and Gasc (1969) believe that in Sphenodon the parapophyses migrate dorsally to meet the diapophyses thereby forming elliptical-shaped synapophyses. For the posterior rib facets of Planoceophalosaunis to be considered as true synapophyses, the diapophyses need to have enlarged at the same time as the parapophyses migrated dorsally; however, the evidence would suggest that the diapophyses enlarged to the exclusion of the parapophyses without any fusion of the two facets. In this way Planocephalosaurus apparently differs from Sphenodon in the formation of the elliptical- shaped dorsal rib facets. Where the exact transition between cervical and dorsal vertebrae takes place is unknown, but the dorsal vertebrae are generally longer, with the elliptical facets for the rib articulation situated more dorsally on the neural arch (text-fig. 6). Flowever, these rib facets become progressively smaller in the posterior members of the dorsal series (text-fig. 7). In comparison with the cervical vertebrae the zygosphenic articulation on the dorsals is slightly more pronounced. The two sacral vertebrae (PI. 53, figs. 9, 10) have not been recovered in the fused condition. However, a clear distinction can be made between them (text-fig. 8). In both instances the very stout ribs are fused to the vertebrae with no trace of a suture, but in the first sacral these ribs are directed slightly ventrally, whereas in the second sacral each rib extends laterally and also bifurcates distally. a b text-fig. 3. Planocephalosaurus robinsonae. Atlas text-fig. 4. Planocephalosaurus robinsonae. Re- neural arch, AUP No. 11136. a , lateral view and b, construction of the atlas/axis complex from AUP mesial view. See list of Abbreviations used in Text- No. 11137. a, ventral and 6, lateral aspects, figures, pp. 594-5. text-fig. 5. Planocephalosaurus robinsonae. Re- text-fig. 6. Planocephalosaurus robinsonae. Recon- construction of a cervical vertebra from AUP struction of a mid-dorsal vertebra from AUP No. 11139. No. 11138. a, anterior and b, lateral aspects. a , anterior and b , lateral aspects. FRASER AND WALKDEN: UPPER TRIASSIC SPHENODONTID 579 text-fig. 7. Planocephalosaurus robinsonae. Reconstruction of a posterior dorsal vertebra from AUP No. 1 1096. a, anterior and b, lateral views. An anterior process extends both towards the rib of the first sacral and laterally to an articulation with the ilium. The posterior process bears no articulation facets and was presumably solely for muscle attachment. Such bifurcation of the second sacral ribs is also observed in the three Jurassic sphenodontid genera: Homeosaurus, Kallimodon , and Sapheosaurus (Hoffstetter and Gasc 1969), as well as in Clevosaurus (Swinton 1939). The anterior caudal vertebrae are approximately equal in length to the posterior dorsals and they bear dorsoventrally compressed ribs projecting at right angles to the spinal cord (text-fig. 9c/; PI. 53, fig. 1 2). These ribs are fused to the neural arch and possess shallow ventral grooves. The neural spines s.r a b text-fig. 8. Planocephalosaurus robinsonae. Re- constructions of the sacral vertebrae, a , anterior view of the first sacral from AUP No. 1 1097. The second sacral in b , anterior view and c, dorsal view from AUP Nos. 1 1098 and 1 1099. a text-fig. 9. Planocephalosaurus robinsonae. Recon- structions of caudal vertebrae, a, posterior view of an anterior caudal from AUP No. 11100 and b , lateral view of a mid-caudal from AUP No. 1 1 101. 580 PALAEONTOLOGY, VOLUME 27 are less elongated than in the dorsal series. Posteriorly the vertebrae become more elongate whilst the transverse processes become progressively shorter, more circular in cross-section, and directed posterolaterally rather than laterally (text-fig. 9b). They are eventually lost altogether (PI. 53, figs. 1 3, 1 4). In the same way the neural spines become smaller and the zygapophyses converge until they form nothing more than rudimentary contact points. Fracture planes are present and these appear in the anterior members of the caudal series where the vertebrae are relatively short and still bear small transverse processes (text-fig. 9b): at an estimate between caudal 5 and caudal 10. With the exception of the fused atlas and axis there are small anteroventral and posteroventral facets on all the centra of the vertebral column which testify to the presence of intercentra throughout. Probably as a result of their small size and delicate nature none has been positively identified from the cervical region and only a few have been recovered representing the dorsal region. These dorsal intercentra are simple crescent-shaped bands of bone (text-fig. 10c), but the caudal intercentra take the form of chevron bones which possess a triangular fossa to allow for the passage of the caudal nerves and blood-vessels (text-fig. 10r/). Below this fossa a medial ventrally directed process separated the muscle blocks on either side of the tail. The anterior chevron bones have a basal transverse bar at the point of attachment to the vertebral column, but in the posterior chevrons this bar is reduced so that the chevron is borne on two separate pedicels (text-fig. 10(7; PI. 53, fig. 18). As already mentioned, the ribs were typically of the holocephalous type, the capitulum having been lost and the tuberculum expanded. However, in the first two or three cervical vertebrae there are small parapophyses and correspondingly the first three or four cervical ribs also possessed rudimentary capilula (text-fig. llu). These may not necessarily have articulated with the parapophyses since ligaments could have completed the attachment where the capitula were reduced to mere protruberances. The posterior cervical ribs possess a short shaft that is expanded distally (text-fig. 1 1(7; PI. 53, fig. 17) whilst in the dorsal ribs the shaft is generally narrow and longer — particularly in the anterior dorsals (text-fig. 1 lc). PECTORAL GIRDLE In total, five elements contribute to the pectoral girdle: one interclavicle, two clavicles, and two scapulocoracoids. There is no suture visible separating the scapula from the coracoid. The interclavicle is a T-shaped bone with a slender anterior crossbar that supported the clavicles (text-fig. 1 2a; PI. 53, fig. 1 6). This crossbar curves dorsally at its distal extremities. The facets for the clavicles are situated on the anteroventral edge of the bone and take the form of confluent grooves allowing the paired clavicles to meet at the midline. The clavicles are slender rod-like elements that curve dorsoventrally from their articulation with the interclavicle (text-fig. \2b, c, d). A long slender depression on the posteroventral surface represents the facet for the interclavicle and there is a small notch at the distal end where it makes contact with the scapulocoracoid. However, the posterolateral border of the clavicle possibly abutted against a cartilagenous zone of the scapulocoracoid for a short distance ventral to this notch. EXPLANATION OF PLATE 53 Figs. 1-19. Planocephalosaurus robinsonae. 1, AUP No. 1 1093, atlas/axis complex, right lateral view, x8. 2, AUP No. 11123, atlas/axis complex, anterior view, x 9. 3, AUP No. 1 1094, cervical vertebra, anterior view, x 9. 4, AUP No. 1 1094, cervical vertebra in anterior, left lateral, and posterior views, x 6-5. 8, AUP No. 1 1096, posterior dorsal vertebra, anterior view, x 10. 9, AUP No. 1 1097, 1st sacral vertebra, anterior view, x 8-5. 10, AUP No. 1 1098, 2nd sacral vertebra, anterior view, x 7. 11, AUP No. 1 1098, 2nd sacral verebra, dorsal view, x 7. 12, AUP No. 11124, anterior caudal vertebra, anterior view, x 8. 13, AUP No. 11125, mid- caudal vertebra, left lateral aspect, x 10. 14, AUP No. 11126, caudal vertebra, left lateral aspect, x 7. 15, AUP No. 11110, right scapulocoracoid, posterolateral view, x 5-5. 16, AUP No. 11108, interclavicle, ventral aspect, x 8. 1 7, AUP No. 1 1 127 and 1 1 106, cervical ribs, x 6. 18, AUP No. 1 1 104, chevron bone, x8. 19, AUP No. 11109, right clavicle, x 6-5. PLATE 53 FRASER and WALKDEN, Upper Triassic sphenodontid 582 PALAEONTOLOGY, VOLUME 27 20mm 10 mm c text-fig. 10. Planocephalosaurus robinsonae . Intercentra, a, anterior chevron bone, AUP No. 11 103, in anterior view, b, posterior chevron, AUP No. 11104, anterior view, c, dorsal intercentrum, AUP No. 11102, dorsal view. The scapulocoracoid is generally rather poorly preserved, but there are one or two almost complete specimens (text-fig. 13; PI. 53, fig. 15). The glenoid fossa is the most robust part of the bone and consequently is more frequently preserved. It bears well-developed buttresses to support the proximal head of the humerus. Anterior to the glenoid is the supracoracoid foramen which carried the supracoracoid nerve and associated blood-vessels. Along the posterior margin of the bone, immediately dorsal to the glenoid, is a small tubercle to which the triceps tendon was attached. The element extends dorsally and there is a single fenestration entering into the anterior margin of the scapula blade which, as in the Lacertilia, probably related to the origin of the limb musculature (Romer 1956). text-fig. 1 1. Planocephalosaurus robinsonae. Ribs, a, anterior cervical, AUP No. 11105. b , posterior cervical, AUP No. 1 1 106. c, anterior dorsal, AUP No. 1 1 107. FRASER AND WALKDEN: UPPER TRIASSIC SPHENODONTID 583 text-fig. 12. Planocephalosaurus robinsonae. Dermal elements of the pectoral girdle, a, an interclavicle, AUP No. 11108, in ventral aspect. Right clavicle, AUP No. 1 1 109, in b , anterior view, c, posterior view, and d, dorsal view. text-fig. 13. Planocephalosaurus robinsonae. Reconstruction of a right scapulocoracoid from AUP Nos. 11110 and 11 111. a, lateral and b , mesial aspects. 584 PALAEONTOLOGY, VOLUME 27 a b c d 20mm text-fig. 14. Planocephalosaurus robinsonae. Partial reconstruction of the humerus from AUP Nos. 11112 and 1 1 1 13. a, anterior, b, ventral, c, posterior, and d, dorsal aspects. FORELIMB The humerus (PI. 54, figs. 1 -4) is a slender element bearing expanded and compressed proximal and distal heads with an axial twist of the shaft so that the planes of the two heads are approaching 90° to each other (text-fig. 14). The proximal head is flattened anteroposteriorly with a ridge on the anteroventral edge marking the insertion of the latissimus dorsi muscle (text-fig. 14a). A similar ridge on the posteroventral edge was for the insertion of the deltopectoralis muscle (text-fig. 146). The distal head is dorsoventrally compressed with the entepicondyle expanded slightly more than the ectepicondyle. The entepicondyle is perforated by a foramen (text-fig. 14c) which opens into a deep depression on the ventral surface (text-fig. 146). The ectepicondyle foramen (text-fig. 14r/), which allows for the passage of the radial nerve and blood-vessels, generally appears more as a groove than a foramen since the bone bridging across the canal is thin and membranous and thus prone to fragmentation. The articular surfaces on both proximal and distal heads have not been preserved in any of the specimens recovered. text-fig. 15. Planocephalosaurus robinsonae. a, proximal head of an ulna, AUP No. 11114. 6, lateral aspect of a left ischium, AUP No. 11115. FRASER AND WALKDEN: UPPER TRIASSIC SPHENODONTID 585 text-fig. 16. Planocephalosaurus robinsonae. Composite reconstruc- tion of a left pelvic girdle in lateral aspect. The epipodials are not as well represented being more slender and consequently rather more vulnerable to breakage. Only the proximal end of the ulna is completely known (text-fig. 15n). The expanded head is flattened anteroposteriorly and in all probability bore an olecranon epiphysis which is missing in all the recognizable specimens. There is a shallow depression on the posterior surface of the head outlining an area for muscle attachment. The shaft is narrow and circular in cross-section. PELVIC GIRDLE All three elements of the pelvic girdle are well represented in the deposits and a complete reconstruction is possible (text-fig. 16). The ilium (PI. 54, fig. 5) consists of a posterodorsally directed iliac blade which medially bears articular surfaces for the two sacral ribs (text-fig. 1 7 a). The bone expands ventrally to form the major part of the acetabulum which is bounded dorsally by a well-defined supracetabular buttress (text-fig. 176). On the anterior edge of the element, just dorsal to the buttress, is a small tuberosity for the attachment of the iliotibialis muscle. There are broad ventral contacts with the pubis and ischium and in addition an anterior process sheathed the anterior edge of the pubis thereby lending rigidity to the structure of the girdle. .b ant. text-fig. 17. Planocephalosaurus robinsonae. Reconstruction of left ilium from AUP No. 11116. a, mesial and b, lateral aspects. 1-0 mm 586 PALAEONTOLOGY, VOLUME 27 text-fig. 18. Planocephalosaurus robinsonae. Left pubis, AUP No. 11117. a , lateral and b, mesial aspects. The pubis (text-fig. 18o; PI. 54, fig. 6) bears a dorsal facet for the ilium and a shorter posterior facet for the ischium. An obturator foramen is situated just anterior to the latter facet. The ventral plate of the bone is emarginated posteriorly by the thyroid fenestra which separates the pubis from the ischium ventrally. The ischium (text-fig. 15 b\ PI. 54, fig. 7) formed the posterior half of the puboischiadic plate. The anterior edge has a short facet for the pubis and ventral to this there is a concave margin marking the posterior boundary of the thyroid fenestra. The posterior margin of the bone is extended backwards into a prominent tubercle for the attachment of ligaments and tendons of the tail musculature. HINDLIMB The femur (PI. 54, figs. 8, 9) is a long slender bone with a sigmoid flexure along the length of the shaft (text-fig. 1 9a, b). The proximal expansion bears a well-developed internal trochanter situated ventral to the articulation head. Lying on the anterior surface, positioned between the internal trochanter and the head, is an area for the attachment of the puboischiofemoralis internus. Unfortunately, the epiphyses of the element are missing in all instances and thus other details of muscle attachment are unknown. EXPLANATION OF PLATE 54 Figs. 1 I 5. Planocephalosaurus robinsonae. 1-4, AUP No. 11112, humerus in anterior, ventral, posterior, and dorsal aspects, x 5-5. 5, AUP No. 1 1 132, right ilium, lateral aspect, x5-5. 6, AUP No. 1 1 1 17, left pubis, lateral view, x 5-5. 7, AUP No. 1 1 1 1 5, left ischium, lateral view, x5-5. 8, AUP No. 11118, femur, x4. 9, AUP No. 1 1 128, femur, x 4-5. 10, AUP No. 1 1 1 19, fibia, x 4. 1 1, AUP No. 1 1 129, two tibias, x 4. 12, AUP No. 1 1 120, astragalocalcaneum, x9-5. 13, AUP No. 1 1 121, tarsometatarsal, x 10. 14, AUP No. 1 1 130, phalanges, dorsal, and planar views, x 10. 15, AUP No. 1 1 131, ungual phalanges, x9-5. PLATE 54 FRASER and WALKDEN, Upper Triassic sphenodontid 588 PALAEONTOLOGY, VOLUME 27 text-fig. 19. Planocephalosaurus robinsonae. Reconstruction of a femur from AUP No. 11118. a , anterior and b , posterior aspects. Reconstruction of a tibia from AUP No. 11119 , c, anterior and d , posterior aspects. Of the two epipodials only the tibia (text-fig. 19c, d\ PI. 54, figs. 10, 11) has been confidently identified, but again the articulation surfaces themselves have been poorly preserved. It is a long slender bone that is concave towards the fibula. The posterior surface of the proximal head bears a slight ridge and rugosity for insertion of the iliofibularis musculature. MANUS AND PES A variety of small carpals, tarsals, and metapodials have been recovered from the residues at all the sites, but these have proved difficult to separate into distinct forms and it is likely that the structures of the manus and pes are quite uniform in all the Triassic sphenodontids; varying only in size. The manus and pes of Planocephalosaurus are described from elements recovered solely from site four, but the following comments also serve as an outline for the generalized sphenodontid structure. The small bones of the carpus are difficult to determine since many of their distinguishing characteristics are obscured by erosion and polishing. The astragalus and calcaneum are fused into an astragalocalcaneum with no trace of sutures (text-fig. 20a; PI. 54, fig. 12). It is a relatively flat bone- bearing articulation facets on its dorsomedial surface for the tibia and fibula, and a well-defined concavity for the fourth distal tarsal. The distal tarsals themselves are unknown. With the exception of the fifth metatarsal, the metapodials are all similar, varying only in size and slight details of the proximal head. Each metapodial has a long slender shaft with the proximal head slightly expanded and usually bearing small tuberosities for the attachment of the digital extensor and flexor muscles. The fifth metatarsal (text-fig. 20c, cl), which should more accurately be termed the fifth tarsometatarsal, since it is a fusion of the fifth distal tarsal with the fifth metatarsal, is easily recognized as a small robust bone that is clearly ‘hooked’ and very similar to that observed in Sphenodon. The proximal head has a broad facet for the fourth distal tarsal, and tuberosities on the FRASER AND WALKDEN: UPPER TRIASSIC SPHENODONTID 589 plantar surface were for the attachment of flexor muscles such as the gastrocnemius. Robinson (1975) discussed the role of the hooked fifth metatarsal in the functioning of the hind limb and concluded that it was of major importance in allowing for the opposition of the first digit to the fifth, and thereby increasing the gripping powers of the foot. It also had a role to play in bringing the crus and pes musculature to bear in the locomotor effect of the hind limb. The phalanges (PI. 54, fig. 14) bear deeply concave proximal articulation surfaces whilst the distal head is convex. There is some variation in the shape of the shaft— some have an almost circular cross- section whereas others exhibit a degree of dorsoventral compression and also on occasions possess a shallow ventral ridge. The latter were probably the most proximal in position (Evans 1981). A number of ungual phalanges are known which are mediolaterally compressed (text-fig. 20c; PI. 54, fig. 15). These phalanges possess medial and lateral grooves which may have borne ducts supply- ing secretory glands. As a consequence of the material being completely dissociated the phalangeal formula is unknown, but it may have been the same as Sphenodon , namely: Hand 2, 3, 4, 5, 3 Foot 2, 3, 4, 5, 4 RECONSTRUCTION OF THE SKELETON There is a complete absence of articulated material from which direct measurements of Plano- cephalosaurus could be taken. Thus to deduce the relative proportions of the body the mean sizes of the available completely ossified elements must be calculated. There are, however, only a few complete limb bones which do not provide satisfactory sample sizes from which to calculate means. To rectify this deficiency the full lengths of a number of incomplete bones were extrapolated by direct text-fig. 20. Planocephalosaurus robinsonae. Reconstruction of a left astragalocalcaneum from AUP No. 11120. a, anterior and b, posterior views. Right tarsometatarsal, AUP No. 11121, in c, plantar and d, dorsal views. e, Ungual phalanx, AUP No. 11122, in lateral view. 590 PALAEONTOLOGY, VOLUME 27 comparison with intact representatives of each of the four relevant limb elements, and the following mean lengths were obtained: Forelimb: Humerus 1 1 mm ] _ „ t t, n > 20 mm Ulna 9 mm 1 Hindlimb: Femur 16mm) -r , • n > 29 mm Tibia 13 mm I With respect to the vertebrae, their numbers more than their individual sizes govern the relative proportions of the axial skeleton. With dissociated material it is very difficult to estimate the exact numbers of vertebrae in the column, but the relative abundance of each of the four vertebral types within the deposits (Table 1 ) suggests that it is perfectly acceptable to reconstruct Planocephalosaurus with the same vertebral count as Sphenodon. table L Relative abundances of the four major vertebral types of Planocephalosaurus robinsonae expressed as a percentage of the total vertebral count from two different strata at Site 4. The percentages for Sphenodon are obtained from data given by Hoffstetter and Gasc (1969) where Sphenodon is assumed to possess twenty-five presacrals, two sacrals, and between twenty-nine and thirty-six caudals. P. robinsonae. Level k, site 4 P. robinsonae. Level m, site 4 P. robinsonae. Total numbers at site 4 Sphenodon punctatus Per cent cervical vertebrae 13-3 12-8 12-5 12-7-14-3 Per cent dorsal vertebrae 25-7 26-5 26-2 27-0-30-4 Per cent sacral vertebrae 2-8 3-2 2-9 3-2 3-6 Per cent caudal vertebrae Total number of 58-7 57-7 58-0 57-2-51-8 vertebrae in the sample 218 189 646 The full reconstruction shows Planocephalosaurus having a snout-vent length of approximately 7-5 cm with an additional 7 0-8-5 cm of tail. The forelimb/hindlimb ratio is 69-6%, but this disparity is in common with other eosuchians and is not as great as that generally seen in bipedal reptiles such as Malerisaurus , Saltoposuchus , Bcisiliscus , etc. (Ewer 1965; Chatterjee 1980). In addition, the vertebral numbers suggest that the trunk of Planocephalosaurus was not reduced in length; the combined tibia and femur length being approximately 45% that of the trunk. In bipeds, on the other hand, the latter ratio is much higher: 75% for Malerisaurus and 100% in Basiliscus (Chatterjee 1980). The small limb disparity in quadrupedal eosuchians, such as Planocephalosaurus , may permit better acceleration and the ability to change direction quickly. This, coupled with opposable first and fifth digits would have enabled Planocephalosaurus to negotiate quite rough terrain at speed in its attempts to avoid danger and capture elusive prey. DISCUSSION Following Evans (1980), the family Sphenodontidae is considered to lie within the infraclass Eosuchia. The following discussion concentrates on the affinities of Planocephalosaurus with a second eosuchian, Gephyrosaurus, and assesses the possibility of including the latter within the Sphenodontidae. The rudimentary zygosphenic articulation of Planocephalosaurus (text-fig. 9a) is a character not reported amongst other fossil eosuchians with the exception of Saurosternon and Gephyrosaurus. FRASER AND W A LK DEN: UPPER TRIASSIC SPHENODONTID 591 However, Evans (1981) points out the difficulties of observing such a feature in articulated skeletons and suggests that it might be more widespread than reported. Together with Gephyrosaurus, Homeosaurus , and Sphenodon strong fracture planes occur in the caudal vertebrae of Plano- cephalosaurus. These are absent in many other eosuchian genera. Evans ( 1981 ) suggests this may be the result of the functional importance of the tail in other groups (e.g. for swimming or as a counterbalance) and may not have any bearing on ancestral relationships. In a similar fashion to Sphenodon the scapula and coracoid of Planocephalosaurus have fused into a single unit and the same is also true of Gephyrosaurus. But unlike Sphenodon , other sphenodontids, and Gephyrosaurus , Planocephalosaurus does possess a fenestrated scapulocoracoid. Compared with the lacertilians this fenestration is rudimentary as only one fenestra occurs in the scapula region of the bone compared to two in Iguana (text-fig. 21). Iguana also possesses two fenestra in the coracoid section of the bone whilst in Planocephalosaurus this region is unfenestrated. Evans ( 1981 ) believed that such fenestration is a uniquely lacertilian character, but the present evidence would suggest otherwise and further support is provided by Carroll (1975) who reported a similar condition in Saurosternon. Thus with regard to the pectoral girdle Planocephalosaurus would appear quite advanced, but with respect to the humerus it conforms to the pattern observed in Sphenodon and Gephyrosaurus , possessing both ent- and ectepicondylar foramina. text-fig. 21. The scapulocoracoid ossification of A, Sphenodon , B, Plano- cephalosaurus, and c. Iguana. The pelvic girdle of Planocephalosaurus is very similar to that of Sphenodon and Homeosaurus having a puboischiadic plate perforated by a well-developed thyroid fenestra. This condition is also seen in other advanced eosuchians such as Kuehneosaurus and Gephyrosaurus. The iliac blade of Planocephalosaurus is not as elongated as that of Sphenodon and resembles more closely that of Gephyrosaurus. Thus the postcranial skeleton of Planocephalosaurus does not depart noticeably from the general sphenodontid structure. The main difference is in the structure of the scapulocoracoid where that of Planocephalosaurus exhibits fenestration, but this is atypical of the family. In all other respects, including the cranial morphology (Fraser 1 982), Planocephalosaurus is a typical sphenodontid. At the same time it can be said that there are a number of similarities between Planocephalosaurus and Gephyrosaurus. However, Gephyrosaurus was assigned by Evans (1980) to a new family within the Eosuchia, the Gephyrosauridae. There is therefore reason to believe that there are some affinities between the Sphenodontidae and Gephyrosauridae and a brief resume of cranial morphology would seem to strengthen the argument. Evans (1980) commented on the fusion of both frontals and parietals in Gephyrosaurus and considered this to be unusual within the Eosuchia; however, Planocephalosaurus also shows this characteristic. The sphenodontids generally do not possess a lachrymal and whilst this element is present in Gephyrosaurus it is quite rudimentary. The incomplete lower temporal bar of Gephyro- saurus is a characteristic that also occurs in some members of the Sphenodontidae— including Planocephalosaurus and Clevosaurus (Robinson 1973). The quadriradiate shape of the squamosal is a feature shared with the sphenodontids and Evans herself (1980) noted the similarity of the quadrate-quadratojugal arrangement but concluded that it must be a result of convergence. 592 PALAEONTOLOGY, VOLUME 27 Turning to the palate, the possession of an enlarged row of teeth on the palatine of Gephyrosaurus is another characteristic of the sphenodontids. Whilst there is a general tendency in Clevosaurus and Sphenodon to show a reduction in the number of small teeth scattered across the palatal elements, Planocephalosaurus, in common with Gephyrosaurus , retains a number of small palatal teeth. The posterior process of the dentary of Sphenodon meets the articular complex and braces the lower jaw (text-fig. 22). Such a process is not seen in other eosuchian genera such as Youngina , Tanystropheus , and Kuehneosaurus , but it is known in other sphenodontids such as Planocephalo- saurus and Clevosaurus and is also seen in Gephyrosaurus. Evans (1980) notes the overall similarity of the Gephyrosaurus lower jaw to that of the sphenodontids, including the lack of a splenial, but again concludes that this must be the result of convergence. The number of characteristics shared by Gephyrosaurus and the Sphenodontidae suggest a close relationship between the two (Table 2), the only obvious difference so far noted being the existence of a rudimentary lachrymal in Gephyrosaurus , and it is quite easy to see how this element could have been lost in the sphenodontids. However, one major difference does exist and that is the attachment of the marginal dentition. Whereas Gephyrosaurus has a pleurodont attachment the sphenodontids typically display an acrodont dentition. This difference does not necessarily rule out a close relationship between the two, however, since within the Lacertilia both acrodont and pleurodont forms are recognized. table 2. A comparison of sphenodontids and some other eosuchians. Sphenodon Clevosaurus Planocephalosaurus Homoeosaurus Kuehneosaurus Tanystropheus Gephyrosaurus Rhynchosaurs Youngina Prolacerta Macrocnemus Palaeagama Askeptosaurus Fusion of the frontals o o X o o o X o o 0 0 o o Fusion of the parietals X o X o o X X X o o o o o Lachrymal small or absent X X X X o X X o o o X - o Lower temporal arcade incomplete 0 x/o x/o o X X X o 0 X X - o Enlarged palatine tooth row X X X X o o X o 0 o o - o Dentary with pronounced posterior process X X X X o o X o o X o - o Splenial absent X X X - o X X o 0 o o - o Acrodont dentition X X X X o o o o o o o o o Zy gosphene / zy gantrum X X X - o o X o o o o o o Caudal fracture planes X X X X o X X o o o o 0 o Most presacrals with single headed ribs X X X X o o X o o X o o o Vertebrae amphicoelous and notochordal X X X X o o X o X X X X X Scapulocoracoid a single bone X X X X o 0 X X o X 0 o o Scapulo coracoid fenestrated o o X 0 o o o o o o o 0 o Flumerus with two distal foramina X X X X o X X o X o o X o Thyroid fenestra X X X X X X X o o o X o X Flooked fifth tarsometatarsal X X X X o X X X o X X o o x — character present. o character absent. FRASER AND WALKDEN: UPPER TRIASSIC SPHENODONTID 593 text-fig. 22. Lateral aspects of the dentaries of a, Clevosaurus, b, Plano- cephalosaurus, C, Sphenodon , D, Gephyrosaurus , E, Tanystropheus , and F, Iguana. (c after Robinson (1976), d after Evans (1980), e after Wild (1980), and f after Roiner (1956)). It might be postulated that intermediate forms would exist between sphenodontids and their eosuchian ancestors in which the marginal dentition displayed some degree of pleurodonty. The evidence presented suggests that Gephyrosaurus may be such an intermediate form. If it is not considered to be a ‘true’ sphenodontid then it probably represents an early offshoot from the stem Sphenodontidae. It is also postulated that accompanying this trend towards a firmer anchorage of the marginal dentition, there is a tendency within the Sphenodontidae for a reduction in tooth numbers. Thus it is likely that within the Sphenodontidae and their ancestors there is a spectrum of forms ranging from small, relatively delicate individuals with numerous pleurodont teeth to more robust species possessing firm acrodont teeth with a marked decrease in their absolute numbers (Table 3). Such table 3. Tooth arrangement and insertion in three Triassic eosuchians indicating a probable trend towards acrodonty and a reduction of numbers in the Sphenodontidae. Gephyrosaurus bridensis Planocephalosaurus robinsonae Clevosaurus hudsoni Palatal Dentition Numerous teeth scattered across palatines, pterygoids, and vomers. Ordered into an enlarged tooth row on the palatine Numerous teeth on palatines, pterygoids, and vomers. Pre- dominantly arranged in rows with two enlarged tooth rows on the palatine Reduction in palatal denti- tion. Teeth arranged in two rows on the pterygoids. A single enlarged tooth row on the palatine. Occasionally a few vomerine teeth Insertion of marginal dentition Number of functional marginal teeth in the mature individual: Pleurodont Acrodont Acrodont (a) premaxilla 8-10 4 2-3 ( b ) maxilla 35 40 12-14 5-6 (c) dentary 30 40 13-14 5-6 Juvenile teelh worn away to the bone anteriorly 594 PALAEONTOLOGY, VOLUME 27 dentitional modifications are obviously associated with altered dietary habit, with the skull also becoming generally more robust and consequently capable of withstanding the greater stresses imposed upon it by more demanding diets. Another species that occurs in the Cromhall fauna which is expected to substantiate the evolutionary trends outlined above, is presently being described by D. I. Whiteside (in prep.) from abundant remains occurring in Triassic fissure deposits at Tytherington Quarry (ST 660 890). Acknowledgements. We thank the N.E.R.C. for continued financial support of the work on late Triassic reptile- bearing palaeokarstic phenomena. We are grateful to Mr. Hodges of the Amey Roadstone Corporation Ltd. who allowed us to work in Cromhall Quarry at all times and we also thank the British Museum for providing access to their collections of Clevosaurus and Kuehneosaurus material. APPENDIX Total number of each postcranial element of Planocephalosaurus robinsonae recovered from fissure four, Cromhall Quarry. Element at.ar. at/ax. ce.v. d.v. 1st s. 2nd s. caud. ch. int.c. ce.r. d.v. cl. Numbers 17 12 69 169 8 1 1 377 147 87 61 150 13 Element int.cl. sc. hum. rad. il. isch. pu. fern. tib. 5th met. ast. Numbers 9 12 36 14 27 21 13 26 23 26 3 Abbreviations at.ar. atlas arch int.c. intercentrum il. ilium at. /ax. atlas/axis ce.v. cervical rib isch. ischium ce.v. cervical vertebra d.r. dorsal rib pu. pubis d.v. dorsal vertebra cl. clavicle fern. femur 1st s. 1 st sacral vertebra int.cl. interclavicle tib. tibia 2nd s. 2nd sacral vertebra sc. scapulocoracoid 5th met. 5th tarsometatarsal caud. caudal vertebra hum. humerus ast. astragalocalcaneum ch. chevron bones rad. radius ABBREVIATIONS USED IN TEXT-FIGURES acet. acetabulum f.t. tibia facet a.i. axis intercentrum f.4 d.t. facet for 4th distal tarsal ant.pu.p. anterior pubis process gl.f. glenoid fossa a.pr. anterior process il. ilium cap. capitulum il.ant.f. anterior facet for ilium cl.f. clavicle facet ill'. facet for ilium cor.fo. coracoid foramen il.fib. insertion for iliofibularis muscle c.r. caudal rib int.f. facet for interclavicle d.pop. diapophysis isch. ischium ect.fo. ectepicondylar foramen isch.f. facet for ischium ent.fo. entepicondylar foramen l.p.t. lateral plantar tubercle f.c. facet for centrum m.p.t. median plantar tubercle f.f. fibula facet n.f. nutrient foramen f.p. fracture plane n.s. neural spine FRASER AND W A L K DEN: UPPER TRIASSIC SPHENODONTID 595 Abbreviations used in text-figs, (cont.) ob.fo. obturator foramen r.f. rib facet od.f. odontoid facet s.acet.b. supracetabular buttress od.p. odontoid process sc. scapula blade o.p. outer process s.r.l sacral rib 1 p.pop. parapophysis s.r.2 sacral rib 2 p.f. posterior facet s.r. 1 f. facet for 1st sacral rib prox. proximal head s.r.2f. facet for 2nd sacral rib pr.zyg. prezygapophysis thy. fen. thyroid fenestra pu. pubis tr.tb. tubercle for triceps attachment pu.f. facet for pubis tub. tuberculum p-zyg- postzygapophysis v.p. ventral process REFERENCES Carroll, R. L. 1975. Permo-Triassic ‘Lizards’ from the Karroo. Palaeont. afr. 18, 71-87. chatterjee, s. 1980. Malerisaurus, a new eosuchian reptile from the Late Triassic of India. Phil. Trans. R. Soc. B, 291, 163-200. ewer, R. F. 1965. The anatomy of the thecodont reptile Euparkeria capensis Broom. Ibid. 248, 379-435. Evans, s. e. 1980. The skull of a new eosuchian reptile from the Lower Jurassic of South Wales. J. Linn. Soc. (zoo/.), 70, 203-264. — 1981. The postcranial skeleton of the Lower Jurassic eosuchian Gephyrosaurus bridensis. Ibid. 73, 81-116. fraser, n. c. 1982. A new Rhynchocephalian from the British Upper Trias. Palaeontology , 25, 709-725. — and walkden, G. m. 1983. The Ecology of a Late Triassic reptile assemblage from Gloucestershire, England. Palaeogeogr., Palaeoclimatol., Palaeoecol. 42, 341-365. hoffstetter, r. and gasc, j-p. 1969. Vertebrae and ribs of modern reptiles. In gans, c. et al. (eds. ). Biology of the Reptilia , 1, 201-310. Academic Press, London. robinson, p. l. 1973. A problematic reptile from the British Upper Trias. J. geol. Soc. Lond. 129, 457-479. 1975. The functions of the hooked fifth metatarsal in lepidosaurian reptiles. Colloques int. Cent. natn. Recli. scient. Paris. No. 218. Problemes actuels de Paleontologie — evolution des vertebres, 461-483. — 1976. How Sphenodon and Uromastix grow their teeth and use them. In bellairs, a. d’A. and cox, c. b. (eds.). Morphology and Biology of Reptiles, 43-64. Academic Press, London. romer, a. s. 1956. The Osteology of the Reptiles. University of Chicago Press, Chicago. swinton, w. E. 1939. A new Triassic Rhynchocephalian from Gloucestershire. Ann. Mag. nat. Hist., Ser. II, 4, 591 -594. wild, r. 1980. Die Triasfauna der Tessiner Kalkalpen, XXIV. Neue funde von Tanystropheus (Reptilia, Squamata). Schweiz, palaeont. Abh. 102, 1-43. N. C. FRASER G. M. WALKDEN Department of Geology and Mineralogy Typescript received 3 May 1983 Marischal College Revised typescript received 2 September 1983 Aberdeen AB9 IAS OSTEOLOGY OF THE PALAEOGENE TELEOST ESOX TIE MAXI by MARK V. H. WILSON Abstract. The Palaeocene pike Esox tiemani combines many typically esocid features such as an elongate body, depressible and canine teeth, elongate snout and jaws, almost straight preopercle, small second and third hypurals separated by a gap, large ethmoid process on the ectopterygoid, and anteriorly lobed scales with other features in which it is intermediate between Recent umbrids and esocids. Branchiostegal rays are almost equally divided between the ceratohyal and epihyal, while in Recent umbrids they are more numerous on the ceratohyal and in Recent esocids they are more numerous on the epihyal. The opercle of E. tiemani has a prominent dorsolateral flange that probably covered part of the insertion of the levator operculi muscle. The flange is larger in umbrids and virtually absent in Recent esocids, in which much of the insertion for the levator operculi is on the dorsolateral surface of the opercle. W e now know that the pikes (genus Esox, family Esocidae) have lived in North America apparently continuously since the Palaeocene, sixty million years ago. Fossils of Esox have been reported from the Pleistocene of the Yukon and Ontario (Crossman and Harington 1970) and of Florida (Cavender et al. 1970), the Miocene of Oregon (Cavender et al. 1970), the Oligocene of Montana (Cavender 1977), the Eocene of Ellesmere Island (Estes and Hutchinson 1980) and Colorado (Wilson 1981 ), and the Palaeocene of Alberta and Saskatchewan (Wilson 1980). The Palaeocene esocid E. tiemani Wilson (1980), from the Paskapoo Formation of Alberta, is known from several articulated specimens including the holotype, a complete fish, and from numerous disarticulated bony elements at several sites within the same formation. The osteology of E. tiemani is of special interest because it is the oldest-known esocid, and yet clearly possesses many of the unique specializations of the living pikes. The oldest Eurasian esocoids, which are also of Palaeocene age, belong to the Palaeoesocidae (Sytchevskaya 1976, 1982) and represent rather different skeletal adaptations. The purpose of the this paper is to present a detailed account of the osteology of E. tiemani , based on additional preparations of the holotype specimen, in addition to information obtained from the numerous disarticulated bones collected from the type locality and other sites in the Paskapoo Formation. In addition, a skeletal reconstruction and comments on the phylogenetic relationships of the species are presented. MATERIALS AND METHODS The holotype specimen (UAVP 15002) was briefly described by Wilson (1980). The specimen is complete except for the distal portions of the dorsal fin and the dorsal lobe of the caudal fin. The skull was preserved in part and counterpart and, since the original description, has been prepared by transfer methods. The right side of the skull in the counterpart (text-fig. 1) was embedded in bioplastic and the opposite (left) side prepared (text-fig. 2). This face is referred to in the present paper as UAVP 15002B. The other material which was part of the original collection (UAVP 15005, a skull, and 15006, I 5070, 1 507 1 , and 1 5072, four small partial fish) has been supplemented by additional disarticulated bones collected at the type locality in 1 979 and prepared more recently. These consist of two dentaries (UAVP 17685 and 17686), an angular (UAVP 17670), a parasphenoid (UAVP 17678), an opercle [Palaeontology, Vol. 27, Part 3, 1984, pp. 597-608.) 598 PALAEONTOLOGY, VOLUME 27 (UAVP 1 7676), and a group of scales possibly representing a coprolite (UAVP 17677). In addition to these, a great many Eso.x fossils continue to be recovered from other sites in the Paskapoo Formation. The most notable of these is a series of partial fish, primarily skulls, from the Lovettville Creek site (Wilson 1980, fig. 1, site 4), with catalogue numbers UAVP 15024, 15027-15031, and 17259. Description of the skeleton of E. tiemani involved a detailed comparison of the skeletal elements of the fossil species with skeletons of Recent esocoids, primarily E. Iucius, E. masquinongy, Novumbra hubbsi , Dallia pectoralis, Umbra krameri , U. pygmaea , and U. limi. Osteological features of the Umbridae were summarized by Wilson and Veilleux (1982). Those of esocids have not recently been described in detail, but the descriptions of Sytchevskaya (1976) were supplemented by observations on Recent skeletal material in the University of Alberta Museum of Zoology. For catalogue numbers of the Recent osteological material available for comparative purposes in this study, see Wilson and Veilleux (1982). In almost all features the fossil species was found to be much more similar to the Recent esocids than to the Recent umbrids. Where information on particular parts of the skeleton was lacking in the fossils from the type locality, the reconstruction was prepared by using additional data, first from other fossil specimens from the Paskapoo Formation, and secondly from Recent esocids. The following is a list of the abbreviations used in the figures: AA angulo-articular HS haemal spine PP pelvic plate BH basihyal (glossohyal) HU hypural PR pterotic BR branchiostegal IO infraorbital PS parasphenoid CH ceratohyal LA lachrymal PT post-temporal CL clei thrum LE lateral ethmoid PU preural centrum CT canine or fixed teeth MS mesopterygoid QU quadrate DE dentary MT metapterygoid SC supracleithrum DT depressible teeth MX maxilla SM supramaxilla EC ectopterygoid OP opercle SN supraneural EH epihyal PA palatine SO subopercle ES extrascapular PC postcleithrum uc ural centrum EU epural PE proethmoid UH urohyal FR frontal PH parhypural UN uroneural HH hypohyal PM premaxilla VO vomer HM hyomandibula PO preopercle OSTEOLOGY The skull roof of E. tiemani closely resembles that of the Recent species E. Indus and E. masquinongy. Frontals are elongate and narrow anteriorly, rounded posteriorly, and have a prominent supraorbital sensory canal, enclosed in bone, as in the Recent species (text-fig. 1). Nasals are not preserved in any specimen. UAVP 1 5005 has a pair of small supraorbitals. Parietals are separated by the supraoccipital and extend laterally to the pterotics. A canal-bearing extrascapular is present in the holotype (text-fig. 1). Proethmoids are elongate and tapered posteriorly, broader and diverging anteriorly. Lateral ethmoids are seen in UAVP 1 5002B where they have a conical shape, convex anterolaterally (text-fig. 2a). Pterotics, also visible in this specimen, are elongate anteroposteriorly as in E. Iucius. Infraorbitals and lachrymal are only poorly preserved in UAVP 15002B, where they appear to be similar to those in Recent Esox (text-fig. 2). The parasphenoid is preserved in UAVP 1 5002B and 17678. It is narrow and elongate, as in Recent Eso.x. The vomer is broad and truncate anteriorly (text-fig. 1 ), narrower and tapered posteriorly, and possesses depressible teeth along its ventral surface. There are no fixed or canine teeth on the vomer such as are found at the anterior end of the vomer in E. masquinongy (Cavender et at. 1970). Premaxilla, maxilla, and supramaxilla are preserved best in UAVP 15002B (text-fig. 2). The premaxilla bears a series of small depressible teeth along its oral margin. Anteriorly the bone is WILSON: PALAEOCENE PIKE ESOX 599 triangular and posteriorly a narrow extension underlies the anterior portion of the maxilla. The maxilla is markedly curved medially at its anterior end, where it articulates with the palatine. Posteriorly it is narrow and gently curved. A single elongate supramaxilla extends just beyond the posterior end of the maxilla, as in Recent Esox. The dentary and angulo-articular are preserved in UAVP 15002, 15002B, 15005, 17670, 17685, and 1 7686. Dentaries are among the most common Esox fossils at other localities. The mandible is more elongate than in other early Tertiary Esox described by Sytchevskaya (1976), having an ‘articular angle’ of about 50 degrees (text-tigs. 1, 2). The dentary is slender anteriorly and bears the mandibular sensory canal in a bone-enclosed tube near its ventral margin. Fixed or canine teeth are borne posteriorly and depressible teeth anteriorly. About ten canine teeth are present per ramus, from a point ventral to the anterior end of the maxilla, to a point just anterior to the coronoid process. The largest canine teeth are at the mid-point of the series. For a fish of comparable size, the canine row appears to have slightly more teeth and to extend slightly further anteriorly than in E. Indus. Palatines are like those of E. Indus', elongate bones with numerous depressible teeth which grade in size from largest anteromedially to smallest posteriorly and along the posterolateral margin (text- figs. 1 , 2). The largest anterior teeth (text-fig. 2) have the characteristic truncated bases of depressible teeth (Wilson, 1980, fig. 2h, i), unlike anterior palatine teeth of E. masquinongy which are of the canine type (Cavender et al. 1970). The ectopterygoid is robust and angled, with a prominent ethmoid process as in Recent Esox (text- fig. 2). The inesopterygoid is small, the metapterygoid is large, and the quadrate robust with a prominent anterodorsal strut that supports the ectopterygoid, as in Recent Esox. The symplectic is nearly straight, as in Esox , Novumbra, and Dallia , but not Umbra (Wilson and Veilleux 1982). text-fig. 1. Skull of Esox tiemani holotype, UAVP 15002, in dorsolateral view. 600 PALAEONTOLOGY, VOLUME 27 The shapes of the hyoid arch bones can be seen in UAVP 15002B (text-fig. 2) and 15005, where they are very similar to the corresponding bones in E. Indus. The hyomandibula has a long, posteroventrally directed opercular arm, a prominent laterally directed preopercular strut, a thin anteroventral flange lying against the metapterygoid, and a shaft directed slightly anteroventrally, forming a right angle with the opercular arm and about one and a half times as long as the latter. The epihyal is elongate, tapered posteriorly, and with a gently curved ventral margin as in Recent esocids but not umbrids. The ceratohyal is also elongate and hourglass shaped, very similar to that of Recent Esox. Ventral hypohyals are small conical bones with slightly projecting ventral tips (text-fig. 2). Branchiostegals are acinaciform and in the holotype number eleven on the right side and twelve on the left, where six attach to the medial and posteroventral surfaces of the ceratohyal, and six to the ventrolateral surface of the epihyal (text-fig. 2). The total number of branchiostegals is low for known Esocidae (Crossman 1960, Sytchevskaya 1976), agreeing only with some specimens of E. americanus, but differing in the distribution of branchiostegals between the epihyal and ceratohyal. Also in the holotype (text-fig. 2), the urohyal is seen to be a long, slender bone, tapered anteriorly, somewhat expanded dorsoventrally at its posterior end, and slightly wider than deep at its anterior end. Branchial-arch bones are not preserved, except for the dermal tooth-plate of the basihyal (glossohyal). In UAVP 15002B and in 15005 it is seen to be thin and broad, tapering gradually from its truncated anterior end, as in Recent esocids. The preopercular is like that of Recent Esox but appears stouter. It is slighty bent at the angle (text- fig. 2), with a bone-enclosed sensory canal running the entire length of the bone and an anteromedial flange somewhat larger than the one in Recent Esox and reminiscent of that in Novumbra. Details of the interopercle are not visible but the subopercle (text-fig. 2) is only slightly curved, has roughly parallel dorsal and ventral margins, and has a prominent vertically directed articular process at its anterodorsal corner, as in Recent Esox. It differs from subopercles of the latter in being somewhat less elongate for its depth. One of the most striking differences between E. tiemani and all other esocids, including Eurasian fossil forms (Sytchevskaya 1976), is the shape of the opercle (text-figs. 1, 2). Like those of Recent Esox the opercle has vertical anterior and posterior margins, but the ventral and posteroventral margins are more rounded, the articular process at the anterodorsal corner is more slender, and in overall proportions the bone is deeper relative to its length than in the modern species. In addition, the dorsal margin near the articular process bears a thin, plate-like dorsal extension which is continuous with the lateral surface of the body of the opercle, except for a shallow groove. On the medial surface a ridge, corresponding to the dorsal margin of the bone in other esocids, extends from the articular process to the posterolateral corner of the bone, and is separated from the dorsal extension by a narrow fossa, seen best in UAVP 1 5029 where it is filled with sediment. The fossa extends about two- thirds the distance from the articular process to the posterodorsal corner. Few significant differences are seen between the pectoral girdle of E. tiemani and modern esocids. The cleithrum has a greatly elongated ventral arm, seen in UAVP 1 507 1 , and a slender postcleithrum extends posteroventrally from the angle, which is located just behind the posterior end of the subopercle. The supracleithrum, visible in the holotype (text-fig. 1) and in UAVP 15006 (text-fig. 3), bears a canal along its posterodorsal border. The post-temporal is forked, but whereas in modern esocids this bone is thin and plate-like, it appears in the holotype to be more strut-like (text-fig. 2). Pectoral rays number fourteen on the right side of the holotype. Pelvic plates are virtually the same as those of modern esocids and umbrids other than Dallia (Wilson and Veilleux 1982): elongate pubic processes anteriorly, flanked medially by thin pubic plates, with rounded iliac plates posteriorly upon which the fin rays are borne in an anterolateral to posteromedial oblique row. Pelvic rays consist of a splint and approximately eleven rays, as counted in UAVP 15070 (text-fig. 4). The vertebral column consists of nineteen caudal and approximately forty precaudal vertebrae, including ural centra. The first caudal vertebra, taken to be the first vertebra with an expanded haemal arch, is above the middle of the anal pterygiophore series, and below the middle of the dorsal pterygiophore series. Precaudal vertebrae bear slender ribs which do not appear to have autogenous WILSON: PALAEOCENE PIKE ESOX 601 text-fig. 2. Skull of Esox tiemani holotype, UAVP 15002B, in ventrolateral view, a, ammonium-chloride- coated cast of the specimen, b. Drawing of the same specimen. 602 PALAEONTOLOGY, VOLUME 27 text-fig. 3. Trunk and posterior portion of skull of Esox tiemani, UAVP 15006, a specimen with estimated standard length of 215 mm. parapophyses. Neural spines are also slender, but neural arches are expanded somewhat, especially on anterior vertebrae. Epineurals are present on all precaudal and the first few caudal vertebrae. Epipleurals or epihaemals are present on the last few precaudal and the first few caudal vertebrae. A series of slender, S-shaped supraneurals, seen best in UAVP 1 5006 (text-fig. 3) and 1 5070 (text-fig. 4), extends from just behind the pectoral girdle to just in front of the dorsal fin origin. The dorsal and anal fins are situated far posteriorly (text-fig. 5) as in other esocoids. Dorsal rays number approximately fifteen or sixteen, preceded by several shorter unbranched rays, and are supported by sixteen long, slender pterygiophores (proximal radials) in the holotype and by fifteen in UAVP 1 5070, counting the anteriormost-forked pterygiophore as one. The anal fin originates about the length of two vertebrae more posteriorly than the dorsal fin (text-figs. 4, 5), and consists of twelve text-fig. 4. Trunk and anterior caudal region of UAVP 15070, estimated standard length 150 mm; ventrolateral portion of skull and portions of trunk of UAVP 15071, estimated standard length 70 mm; and portions of the trunk and caudal region of UAVP 15072, estimated standard length 70 mm. WILSON: PALAEOCENE PIKE ESOX 603 text-fig. 5. Complete specimen of Esox tiemani, holotype, UAVP 15002, ammonium-chloride-dusted cast of original specimen. principal rays, preceded by several shorter unbranched rays. Anal pterygiophores number thirteen in UAVP 1 5070 and fifteen in the holotype. The margin of the anal fin is rounded, as in other esocoids. Ossified middle and distal radials cannot be distinguished in either the holotype or in UAVP 1 5070, so it is possible that they remain unossified as they do in some umbrids (Wilson and Veilleux 1982). The caudal skeleton is preserved well in the holotype (text-fig. 6), in which it has two ural centra, two lower and four upper hypurals, a single uroneural, and three very slender epurals. The most posterior neural arch and spine is that on the second preural centrum. A distinct gap separates hypurals 2 and 3, which are less broad dorsoventrally than hypurals I and 4, as in Novumbra and other species of Esox (Wilson and Veilleux 1982). The first hypural, the parhypural, and the last few haemal spines bear anteriorly directed processes near their bases. The caudal fin consists of nineteen principal rays, with nine branched rays above and eight below the mid-line. Dorsal procurrent rays number approximately ten, and are decidedly S-shaped and anteroposteriorly expanded near the middle of the series. Ventral procurrent rays number eleven, and are more uniform in shape, except that the anteriormost procurrent ray seems larger and oriented more anteroposteriorly than its neighbours. The articulated specimens display only the exposed portions of their scales. These are circular to slightly elongate, with fine, closely spaced, concentric circuli. No trace of cardioid scales (Scott and Crossman 1973) was seen on any specimen. Extrapolating from the presence of eight scale lengths within the length of three vertebrae on UAVP 15070, the species has a total of approximately 144 lateral-line scales, making it comparable to E. masquinongy in this respect. Entire scales are preserved in UAVP 17677, in which they can be seen to have the typical features of modern species of Esox : focus about two-thirds the length of the scale from the anterior margin, scale broadly oval to almost rectangular, anterior field deeply cleft by one to three (usually two) radii which produce a lobed anterior edge to the scale. Scales like these have also been reported from the Eocene of Colorado (Wilson 1981 ). RECONSTRUCTION AND COMPARISONS E. tiemani is reconstructed in lateral view in text-figs. 7 and 8. The skull is restored with an elongate lower jaw projecting anterior to the snout, and with the quadrate lower jaw articulation ventral to the posterior margin of the orbit. These features are essentially as preserved in the holotype. The lachrymal, although not well preserved in any specimen, is reconstructed as extending anteriorly 604 PALAEONTOLOGY, VOLUME 27 text-fig. 6. Drawing of caudal region of Esox tiemani , holotype, UAVP 15002. on to the snout as in Recent Esox (text-fig. 7). Sytchevskaya (1976) has used the ratio of snout length to postorbital head length to distinguish certain species of Esox. The ratio for E. tiemani is approximately IT, somewhat lower than values for E. masquinongy, higher than values for E. americanus , E. borealis, and E. lepidotus, and about the same as values for E. lucius , E. reicherti, and E. niger. The number of branchiostegals is lower than for Recent esocids. Some specimens of E. americanus and rarely E. lucius have a similar total number. However, in the great majority of specimens of all Recent esocids (Crossman 1960) the number of branchiostegals on the epihyal exceeds that on the ceratohyal by two or three, whereas in E. tiemani the branchiostegals are distributed almost equally between the two bones. The situation is reversed in Recent umbrids (Wilson and Veilleux 1982), where more branchiostegals are found on the ceratohyal. The presence, in the fossil, of more branchiostegals on the left ceratohyal than on the right is the usual situation for modern esocids (Crossman 1960), in which the left branchiostegal membrane usually overlaps the right. WILSON: PALAEOCENE PIKE ESOX 605 The opercular region is shorter anteroposteriorly than in Recent esocids, and the prominent anterodorsal flange (text-fig. 9b) on the opercle distinguishes the species from all other esocids. In young specimens of E. Indus the dorsal opercular margin is in two sections, a lateral ridge extending posteriorly and slightly ventrally from the articular process across about two-thirds of the width of the bone, and the posterodorsal margin, which runs from the posterodorsal corner of the bone, anteriorly and slightly ventrally along the dorsomedial edge to a point about one-third the width of the bone from the articular process. Where the two ridges overlap, in the middle third of the bone, there is a shallow groove (text-fig. 9d). The levator operculi muscle (Winterbottom 1974) inserts along the lateral ridge and on the lateral face of the opercle posterodorsal to it, as well as on the dorsal edge and dorsomedial surface of the bone. The situation in young specimens of E. masquinongy is similar (text-fig. 9c), except that the lateral ridge forms a flange, like that in E. tiemani but much smaller in extent. As well the overlap between the ridges and the groove between them is scarcely present. The opercle of the Palaeocene to Eocene palaeoesocid Boltyschia has what might be interpreted as an incipient or remnant anterodorsal flange, and a ridge on the medial surface of the bone running posteriorly from the articular process (Sytchevskaya 1976, pi. i, figs. 3-4). In most umbrids, the opercle is not as rectangular as it is in esocids or even Boltyschia , and all three extant genera have a plate-like extension dorsal to a line joining the articular process with the posterodorsal corner of the bone. Opercles of Novumbra are most similar to those of Boltyschia. In Novumbra (text-fig. 9a) there is a medial ridge running posteriorly from the articular process, and the dorsal flange is broadest anteriorly. The levator operculi inserts entirely on the medial surface of the opercle, dorsal to the ridge. text-fig. 8. Reconstruction of Esox tiemani in lateral view, intermuscular bones and scales omitted. 606 PALAEONTOLOGY, VOLUME 27 A text-fig. 9. Comparison of opercles in lateral view (upper drawings) and medial view (lower drawings). A, Novumbra hubbsi, UAMZ 3713. b, Esox tiemani , UAVP 15029. c, E. masquinongy , UAMZ 3744. D, E. Indus , UAMZ4876. The articular process of the opercle is also longer in E. tiemani than in Recent species. In overall shape the opercle of E. tiemani is somewhat intermediate between those of Boltyschia and Novumbra on the one hand, and extant species of Esox on the other, because it is more rectangular than the former but less rectangular and less elongated than the latter. One can visualize a transformation series from a Novumbra- like opercle, through Boltyschia , E. tiemani, and E. masquinongy, to the condition in E. lucius (text-fig. 9). The majority of the dorsal margin of the opercle of Recent esocids is thus probably homologous with the medial ridge of umbrids. The dorsal opercular flange of umbrids is the homologue of the anterodorsal and lateral ridge in esocids. The opercle of the Oligocene species E. dispar differs further from these in having an elongate yet rounded opercle (Sytchevskaya 1976). In overall body form (text-fig. 8) E. tiemani is a long, slender fish. Postcranially the skeleton is comparable to that of Recent North American esocids with the exception of meristic differences. A meristic comparison of relevant features of E. tiemani with other fossil and Recent species is given in Table I . DISCUSSION Esox tiemani is anatomically the best-known fossil esocid, thanks mostly to the excellent preservation of the holotype. There is no doubt that it represents a distinct species because of the mosaic of meristic differences between the fossil and other known esocids (Table 1 ). In one of these, the branchiostegal ray distribution, the fossil is intermediate between Recent umbrids and esocids. In addition to the meristic characters, there is a unique combination of proportional and shape features of the skull which serve to distinguish this species from others. The most striking of these is the shape of the opercle, as described above. Others include the somewhat stouter, slightly bent preopercle with well- developed anteromedial flange, and the elongate lower jaw with perhaps a greater ratio of fixed or canine teeth to depressible teeth along the oral margin of the dentary. WILSON: PALAEOCENE PIKE ESOX 607 table 1. Comparison of meristic features of Esox tiemani with other Recent and fossil species of Esox. Data for other esocids from Sytchevskaya ( 1976). BR — branchiostegal rays, P— pectoral fin rays, V— pelvic fin rays, D — total dorsal fin rays, A — total anal fin rays, CS — supplementary caudal rays, TV — total vertebrae, CV — caudal vertebrae. BR P V D A CS TV CV E. lucius 11-20 1 1 18 8-13 17-26 14-21 20 56-65 21-22 E. masquinongy 16-19 14 11 21-24 20-22 27 63-67 21-22 E. reicherti 13-14 12 9-12 17-22 15-18 27 64-65 21 -22 E. americanus americanus 11-16 13-17 8-10 15-19 14- 17 ? 44-51 19 E. a. vermiculatus 9-14 14 45 9-10 17-21 15 19 26 42-49 18 E. niger 12-16 15 10-11 20-21 16-19 22-28 49-54 20-21 E. borealis 13-15 12- 13 9-11 18- 19 ? 7 56-58 21-23 E. papyraceus ? 13? 10? 17 15 13 48 20 E. lepidotus 14 15 13 19-21 19 17 18 52-60 18 E. dispar ? 13 1 1 19 ? 7 61 18 E. tiemani 11-12 14 11 19 15 21 59 19 The question of the phylogenetic relationships of the Palaeocene species is more difficult to answer, because the position of the esocids within the fossil and Recent esocoid families is poorly understood, as are the relationships among esocid species. Nelson (1972) proposed a division of Recent Esox species between two subgenera: Esox , includ- ing E. lucius, E. reicherti, and E. masquinongy ; Kenoza , including E. americanus and E. niger. Unfortunately, the features of the cephalic sensory canals used by Nelson to separate the subgenera are not clearly visible on the available specimens of E. tiemani. Nelson did, however, suggest that the number of total vertebrae seen in Kenoza (42-54) was primitive for the genus as a whole, and that the greater number seen in the subgenus Esox (56-67) was a derived condition. Nelson’s hypothesis would suggest that the vertebral number in E. tiemani (59) is indicative of a closer relationship to the subgenus Esox. On the other hand, the low branchiostegal number and the distribution of branchiostegals between ceratohyal and epihyal, as discussed above, together with the opercular structure, are all features in which E. tiemani is intermediate between Recent umbrids and esocids. These features therefore indicate that the Recent subgenera of Esox are more closely related to each other than either is to E. tiemani. The evidence bearing on this question is admittedly slim, and a more reliable estimate of relationships must await further study. With respect to the question of the position of the esocids within the Esocoidei, E. tiemani demonstrates that a large number of fundamental esocoid features (including posteriorly situated dorsal and anal fin, S-shaped supraneurals, and toothless maxillae) and esocid features (depressible and canine teeth, elongate snout, jaws, and trunk, elongate vomer and palatines, ethmoid process on ectopterygoid, nearly straight preopercle, squared opercle, small second and third hypurals separated by a gap, and anteriorly lobed scales) are geologically much older than was previously known. The significance of this is that in the absence of E. tiemani, the fossil record of esocoids appeared to show a tendency for the older esocoids ( Boltyschia , Palaeoesox) to be anatomically more like umbrids than like modern esocids (Sytchevskaya 1 976). Eurasian fossil esocids, similarly, tended to have such features as less elongate jaws and fewer vertebrae compared with some modern esocids. Whether these features are indeed primitive for esocids remains to be seen. But the apparent correlation with greater geologic age must be rejected in view of the structure of E. tiemani , which suggests that such esocid features as elongate bodies and jaws and depressible teeth were completely evolved before features such as branchiostegal ray numbers and opercular structure, in which E. tiemani is still intermediate between Recent umbrids and esocids. 608 PALAEONTOLOGY, VOLUME 27 Acknowledgements. Specimens used in this study were prepared by Allan Lindoe. Drawings were prepared by Diane Hollingdale. Robert Williams discussed several aspects of esocid anatomy. This research was supported by Natural Sciences and Engineering Research Council of Canada Grant A9180. REFERENCES CA vender, T. M. 1977. A new Tertiary fish fauna from southwestern Montana. Geological Society of America , Abstracts , 9, 715. — lundberg, j. G. and wilson, r. l. 1970. Two new fossil records of the genus Esox (Teleostei, Salmoniformes) in North America. Northwest Science, 44, 176-183. crossman, e. j. 1960. Variation in number and asymmetry in branchiostegal rays in the family Esocidae. Canadian Jour. Zoo/. 38, 363-375. and buss, k. 1965. Hybridization in the Family Esocidae. Jour. Fisheries Research Board Canada , 22, 1261-1292. and harington, c. R. 1970. Pleistocene pike, Esox Indus, and Esox sp., from the Yukon Territory and Ontario. Canadian Jour. Earth Sciences, 7, 1 130-1138. estes, R. and hutchinson, j. H. 1980. Eocene lower vertebrates from Ellesmere Island, Canadian Arctic Archipelago. Palaeogeogr., Palaeoclimat., Palaeoecol., 30, 325-347. nelson, g. J. 1972. Cephalic sensory canals, pitlines, and the classification of esocoid fishes, with notes on galaxiids and other teleosts. American Museum Novitates, 2492, 1 49. scott, w. B. and crossman, e. j. 1973. Freshwater fishes of Canada. Fisheries Research Board Canada, Bulletin, 184, 966 pp. sytchevskaya, E. c. 1976. The fossil esocoid fishes of the USSR and Mongolia. Trudy Paleontologicheskogo Instituta, Akad. Nauk SSSR, 156, 1-116. 1982. A new data on the Palaeogene freshwater fish fauna of N. Eurasia. 4th European Ichthyological Congress, Hamburg, Abstracts, 294. wilson, m. v. h. 1980. Oldest known Esox (Pisces: Esocidae), part of a new Paleocene teleost fauna from western Canada. Canadian Jour. Earth Sciences, 17, 307-312. — 1981. Eocene freshwater fishes from the Coalmont Formation, Colorado. Jour. Paleont. 55, 671-674. and veilleux, p. 1982. Comparative osteology and relationships of the Umbridae (Pisces: Salmoniformes). Zool. Jour. Linn. Soc. 76, 321-352. winterbottom, r. 1974. A descriptive synonymy of the striated muscles of the Teleostei. Proc. Acad. Nat. Sci. Philadelphia, 125, 225-317. Typescript received 23 June 1983 Revised typescript received 15 November 1983 MARK V. H. WILSON Department of Zoology University of Alberta Edmonton, Alberta T6G 2E9 Canada A NEW FRESHWATER LIMULOID FROM THE MIDDLE TRIASSSC OF NEW SOUTH WALES by J. W. PICKETT Abstract. Dubbolimulus peetae gen. et sp. nov. is described from freshwater strata of middle Triassic age near Dubbo, New South Wales, and is referred to the new family Dubbolimulidae. Previous reports of xiphosurans from Australia are reviewed; Pincombella belmontensis Chapman, 1932 and Hemiaspis tunnecliffei Chapman, 1932 are shown to be respectively an insect and a trilobite. Reports of merostomes from Australia are contained in only eight publications. Three of these (McCoy 1899; Gill 1951; Caster and Kjellesvig-Waering 1953) deal with eurypterids, and are not further mentioned in this article. A fourth (Quilty 1972) reports an unnamed Cambrian aglaspid from Tasmania. The remaining four (Chapman 1932; Riek 1955, 1968/9; Riek and Gill 1971 ) are concerned with xiphosurans, and are summarized briefly. AUSTRALIAN XIPHOSURANS Chapman (1932) established the taxa P. belmontensis from the insect beds (?Boolaroo Sub-group, late Permian) near Belmont, New South Wales, and H. tunnecliffei from the late Silurian Dargile Formation at Studley Park, Kew, Victoria. Much better preserved than either of these is the magnificent specimen of the rather anomalous Austrolimulus fletcheri from the Beacon Hill shale lens in the middle Triassic Hawkesbury Sandstone of the Sydney district (Riek 1955, 19686). The youngest xiphosuran so far known from the Australian continent is Victalimulus mcqueeni Riek and Gill, 1971 from the early Cretaceous fish bed (Strzelecki Group) near Koonwarra in Victoria (Waldman 1971 ). Three of these papers are based on careful examination of available material. On the other hand, neither of the two species described by Chapman (1932) is xiphosuran. The holotype specimen of P. belmontensis (PI. 56, fig. 2) is clearly the wingless and, for the most part, legless body of a hemipteran with a small scutellum and distinct pronotal lobes. Riek (1968n) lists sixty-two species of Hemiptera from the horizon yielding P. belmontensis. These are based on generally incomplete animals, either wings, heads, or thoraxes; since so few are known in all aspects of their morphology, no attempt is made here to reconcile P. belmontensis with any of the described forms. Chapman’s figures (1932, pi. 14, figs. 1 -3) are difficult to interpret in relation to the specimen. He appears to have taken some irregularity in the matrix to be the anterior border of his figs. 2 and 3 (‘Anterior border . . . marked out by a definite rust-stained impression’) and the real anterior margin to mark the edge of ‘opercular plates’. Chapman considered that he was examining the specimen in ventral aspect, whereas it represents the back of the insect from which the wings have been torn. This, together with its distorted position, may have led him to interpret the only preserved, jointed leg as a dorsally attached telson. Chapman’s fig. 2 was reproduced by Stormer (1955, fig. 13.6), who, for reasons now apparent, could not place Pincombella within any of the defined subdivisions of the suborder Limulina, listing it (p. P23) under ‘Superfamily and Family uncertain’. For similar reasons it was referred to ‘incertae sedis’ by Bergstrom (1975). The true affinities of the other species described by Chapman, H. tunnecliffei , are not im- mediately so clear. It is certain that Chapman’s reconstruction (1932, pi. 14, fig. 5) bears no resemblance to the actual specimen. The broad anterior border with radiate markings is a conchoidal fracture lying outside the true area of the specimen, which is marked by dark colouration; there is no suggestion of pleural spines, rather the lateral profile is fairly straight; there is no hint of a telson. The | Palaeontology, Vol. 27, Part 3, 1984, pp. 609-621, pis. 55-56. | 610 PALAEONTOLOGY, VOLUME 27 specimen (MUGD 1201, not 1801 asquoted by Chapman, refigured here as PI. 56, fig. 5) consists for the greater part of the thoracic area of a trilobite with relatively wide axis (13-5 mm) and narrow pleurae (6-3 mm). The pygidium is entirely missing and the fragmentary cephalon shows no detail at all. It is probably a poorly preserved homalonotid trilobite (pers. com. K. S. W. Campbell and P. A. Jell). One such species, Trimerus harrisoni (McCoy, 1876), has been described from the Dargile Formation in the Melbourne area. Chapman’s damaged specimen has been compared alongside the holotype of T. harrisoni , NMV P7503. The axis of Chapman’s specimen is more clearly demarcated than in the holotype of T. harrisoni , but the latter has suffered very little distortion compared with the former. The clearer delineation of the axis of'//’ tunnecliffei is taken to be a result of the distortion. It is highly probable that the name H. tunnecliffei is a junior synonym of Homalonotus harrisoni McCoy (reassigned to Trimerus by Gill 1949). SYSTEMATIC PALAEONTOLOGY Repositories. The repositories of specimens quoted are denoted by the following abbreviations: MM, Geological and Mining Museum, Sydney; MUGD, Department of Geology, University of Melbourne; NMV, National Museum of Victoria, Melbourne. Superfamily limuloidea Zittel, 1885 1885 Limulidae Zittel, p. 643. 1944 Limulacea Zittel; Raymond, p. 504. 1952 Limulacea Zittel; Stormer, p. 636. 1955 Limulacea Zittel; Stormer, p. P21. 1975 Limulacea Zittel; Bergstrom, p. 303. Discussion. The suffix of the superfamilial name has been changed in conformity with Recommenda- tion 29a of ICZN. Stormer (1952, 1955) included three families in the Limuloidea: Paleolimulidae Raymond, 1944; Mesolimulidae Stormer, 1952; and Limulidae Zittel, 1885. The family Limulidae was redefined by Riek and Gill (1971) to include those genera previously referred to the Mesolimulidae, thus comprising the genera Limulus , Tachypleus , Carcinoscorpius , Mesolimulus, Psammolimulus , and their new genus Victalimulus . They excluded the genus Limulitella Stormer, 1952, placing it with the Paleolimulidae and regarding it as a probable synonym of Paleolimulus. The type species of Limulitella , Limulus bronni Schimper, 1850, is in need of re-examination to establish the status of the genus. A fourth family name, Austrolimulidae, was established by Riek (1955) for the highly individual Austrolimulus; no other genus has yet been referred to this family. Via Boada and Villalta (1966) established a new family Heterolimulidae for their genus Heterolimulus from the Triassic of Spain. Bergstrom (1975) summarized the classification up to this point. More recently Romero and Via Boada (1977) described the genus Tarracolimulus from the same horizon as Heterolimulus , and included the latter without comment in the family Limulidae, a conclusion with which I agree. Dubbolimulus gen. nov. has features which differentiate it from each family (although, if Limulitella is set aside because of its doubtful status, two of the families contain but a single genus), so that it would be possible to establish a separate family for it. Dubbolimulus can be separated from the Austolimulidae by the absence of exaggerated genal spines and of 'free posterior segments’ on the opisthosoma. It is distinguished from the Paleolimulidae by the fact that the ophthalmic ridges do not meet in front of the cardiac lobe, by the absence of distinct annulation of the axis of the opisthosoma, and by the absence of movable lateral spines on the opisthosoma. This last feature represents a point of similarity with the Austrolimulidae; Austrolimulus is the only other limuloid genus which lacks movable spines on the opisthosoma. A characteristic peculiar to the Austrolimulidae is the fact that the anterior margin of the opisthosoma, at its junction with the prosoma, extends well beyond the area bounded by the ophthalmic lobes (Riek 19686, fig. 1), whereas there is a clear correlation between these two features in other genera. Features which distinguish Dubbolimulus from genera of the Limulidae are: (a) absence of movable PICKETT: MIDDLE TRIASSIC LIMULOID 61 1 spines, ( b ) apparent absence of free lobes on the anterior part of the opisthosoma, (c) the greatly unequal prosoma and opisthosoma, and (cl) genal angles which are distant from the lateral margins of the opisthosoma. Setting aside (c) and (d) as being possibly of lesser significance phylogenetically, the other two characters could be either features which no antecedent of Dubbolimulus ever possessed, or ones which have been secondarily lost. Movable spines are unknown in Palaeozoic Limulina until the Permian Paleolimulus avitus , in which both the movable spines (‘stylets’) and the free lobe are specifically mentioned by both Dunbar (1923) and Raymond (1944). According to Ivanov (1933, fig. 54) the movable spines originate as the distal tips of the somites of the opisthosoma, which separate and move posteriorly during development of the embryo, being already present in the first larval stage. In view of the complexity of such a development it seems reasonable to regard all those forms in which it is expressed as belonging to a single lineage. The same applies to the sharing of somite VI, between the prosoma and opisthosoma, which appears to be connected with the production of the free lobes of the opisthosoma. These two features are fairly general among Limuloidea. Bergstrom (1975, p. 295) regards the sharing of somites VI and VII between prosoma and opisthosoma as distinctive of the suborder Limulina. While the gross characters of Dubbolimulus link it to genera of Limulidae and Paleolimulidae, it is considered likely that the apparent absence of the free lobe on the opisthosoma is due rather to a masking of the characteristic by general simplification of morphological features of the opisthosoma, and that the absence of movable spines is another expression of the same trend. Phylogenetically, this means that Dubbolimulus is an offshoot of the main lineage of Limuloidea, rather than an independent derivative of a lineage closer to either Belinuroidea or Euproopoidea. This conclusion notwithstanding, Dubbolimulus cannot be assigned to either of the subfamilies of Limulidae, which are distinguished on the basis of characters entirely lacking in D. peetae. Although it shares the absence of movable spines with Austrolimulus and is of approximately the same age, the extraordinary genal spines and the ‘free posterior segments’ of the opisthosoma of A. fletcheri preclude consideration of any close relationship between Austrolimulus and Dubbolimulus. It is therefore appropriate to establish a new family for the latter. Raymond’s original diagnosis of the superfamily (1944, p. 504) included movable lateral spines as a criterion. Their absence in Dubbolimulus and Austrolimulus excludes this as a suprafamilial character. An alternative phylogeny for xiphosurans has been presented by Fisher (1981, 1982), based on cladistic principles. This focuses on the nature of the prosoma/opisthosoma articulation, and varies from earlier suggestions (e.g. Bergstrom 1975) in deriving the Limuloidea and Euproopoidea independently from the Belinuroidea. Both the younger groups are characterized by fusion of the opisthosomal segments, and Fisher uses the declination of the occipital band of the prosoma to determine polarity of various conditions of it with respect to the primitive (unfused) condition of the opisthosoma. Fisher (1977) has argued in favour of a ventral enrolment in Euproops , affording protection of the underparts. The articulation in Limulus , however, serves to achieve strong flexuring in the dorsal direction (Richter 1964), associated with righting strategies; the remarkable ophthalmic spines of Euproops absolutely preclude any similar dorsal flexuring, even had the structure of the occipital band allowed it. It seems to me that the morphological differences in the occipital band discussed by Fisher (1981) reflect these differences in function, which are in all probability generally characteristic of each superfamily. The free segments reported at the rear of the opisthosoma in Paleolimulus (one segment) and Austrolimulus (three segments) support a derivation of Limuloidea from Belinuroidea. However, there appears to be some doubt about the nature of these structures. The most recent reconstruction of Paleolimulus by Fisher (1981, fig. 3b) excludes such a segment, and Raymond ( 1 944, p. 50) seems to have had some reservation concerning it. I have examined the holotype of A. fletcheri with specific reference to this point. Certainly there are indistinct features suggesting transverse structures at the rear of the opisthosoma, but the ‘segments’ are definitely fused and were incapable of any independent movement. More significantly, they are confined to the area of the opisthosomal doublure, and thus had no appendages corresponding to them. Separation of the counterparts of the holotype has resulted in a transverse fracture of the exoskeleton anterior to these ‘segments’, so that 612 PALAEONTOLOGY, VOLUME 27 the posteriormost part of the opisthosoma is held on the dorsal counterpart. Consequently, although only dorsal features can be observed for most of the skeleton, the rear of the opisthosoma and the telson can only be observed in ventral view. While the possibility remains that the poorly expressed transverse features represent true segmentation, the absence of any trace of appendages tends to belie such an interpretation. In the absence of material, other than the holotype specimen, with which to check this feature, it seems wiser not to regard the transverse structures as reflecting a true segmentation, and it should not be considered a primitive character in either Paleolimulus or Austrolimulus. A comprehensive phylogeny of xiphosurans is presented by Fisher ( 1 982, text-fig. 1 ). The nature of the text of his article implies that this phylogeny was arrived at by cladistic analysis. However, no discussion of the parameters used in achieving the result is presented, which makes it difficult to compare Dubbolimulus on the same basis. It is difficult to understand his close placement of A. fietcheri with Psammolimulus gottingensis and Limulitella bronni. Austrolimulus , with its reduced opisthosoma, exaggerated genal spines, absence of movable spines, and absence of posterolateral facets, is the most aberrant of all limuloids known so far; none of these features is characteristic of either of the other two genera on this branch of Fisher’s proposed phylogeny. I regard both Austrolimulus and Dubbolimulus as being derived independently from the mainstream of Limuloidea. Some of the remarks made by Fisher (1981) relating to the function of certain features of the limuloid carapace are of interest in the present context: in particular, (a) reduction of prosomal spines, reducing drag during swimming, ( b ) the significance of the free lobes in maintaining channels for respiratory currents during shallow burial, and (c) the function of movable spines as sensors against the substrate. All three of these features are reduced in both Austrolimulus and Dubbolimulus and, in the light of Fisher’s observations, allow comment on the autecology of these genera. Fisher (1981, p. 57) predicts an association between development of free lobes and the importance of burrowing and burial in the activities of limuloids. This is an attractive suggestion which, coupled with the reduction in drag during swimming because of the absence of spines, suggests that swimming heavily outweighed burrowing in the relative significance of the activities of Dubbolimulus. Further support for this suggestion comes from Fisher’s (1975) analysis of degree of vaulting of the prosoma in Limulus (strongly vaulted) and Mesolimulus (much flatter), in which he concludes that the two shapes are better suited to burrowing and swimming respectively. As the prosoma of Dubbolimulus was probably much flatter than that of Limulus , this strengthens the suggestion that burrowing was not an important activity of D. peetae. The movable lateral spines, which occur in all limuloid genera except Austrolimulus and Dubbolimulus , serve to transmit information from the substrate, allowing precise orientation. The extended genal spines of Austrolimulus , coupled with the long telson, provide a refined mechanism of maintaining orientation (in both longitudinal and lateral senses) probably more critical than that provided by the movable spines; this could account for the obsolescence of movable spines in that genus. Dubbolimulus, however, does not show any features which could take over this function of the movable spines, so their absence in that genus remains unexplained. Family dubbolimulidae fam. nov. Type genus. Dubbolimulus gen. nov. Diagnosis. Limuloidea without dorsal or lateral spines, except for the genal angles of the prosoma and the posterolateral terminations of the opisthosoma adjacent to the telson joint; opisthosoma fused, much smaller than the prosoma; genal angles distant from lateral margins of opisthosoma. Genus Dubbolimulus gen. nov. Type species. D. peetae sp. nov. Diagnosis. Prosoma semicircular, smooth; posterior margin lying near 90° to axis and bearing posteromarginal facet; opisthosoma with smoothly curved lateral margin without lateral spines; PICKETT: MIDDLE TRIASSIC LIMULOID 613 scarcely wider than the distance between the ophthalmic ridges of the prosoma; free lobes not apparent on dorsal surface of opisthosoma. Derivation of name. From the type locality near Dubbo, New South Wales, Australia. Dubbolimulus peetae sp. nov. Plate 55; Plate 56, figs. 3, 4 Diagnosis. As for genus. Material. Holotype and only specimen preserved as counterparts MM F27693 (dorsal) and MM F27694 (ventral). The specimen is preserved in a red-brown, iron-rich, slightly micaceous shale with common plant remains, most conspicuously Dicroidium odontopteroides var. moltenense Retallack. The specimen was recovered south of Western Plains Zoo, Dubbo, at approximate grid reference 151004 (yards), Dubbo 1:250,000 geological sheet (SI/55-4), 148° 37' E, 32° 19' S (text-fig. 1 ). Plant fossils are abundant at the locality, including many species illustrated in a recent paper by Holmes (1982) which deals with the flora of a locality 15 km to the south. This flora was considered middle Triassic by Holmes. text-fig. 1. Map of the district around Dubbo, New South Wales, showing the type locality of Dubbolimulus peetae gen. et sp. nov. arrowed. 614 PALAEONTOLOGY, VOLUME 27 Description. The specimen consists of the flattened prosoma and opisthosoma, still in juxtaposition. The telson is lacking. Traces of some of the appendages have been impressed through the carapace. Separation has occurred in the dorsal exoskeleton, as there is little difference between dorsal and ventral counterparts (PI. 55, figs. I, 2); traces of the appendages are clearer in the ventral counterpart. The specimen may have been lying in a slight depression before burial, as compression has depressed the inner parts of the prosoma below both the prosomal margin and the opisthosoma. The outline of the flattened prosoma approximates closely to a semicircle. It is 27-8 mm wide immediately in front of the genal angles and 14 0 mm long. The dorsal surface is marked by ophthalmic ridges which begin near the posterior margin at a point midway between axis and lateral margin. The compound eyes begin c. 10 mm anterior of this point, and are probably just under 2 0 mm long. The visual surface of the eye is not adequately preserved, but its position is marked by an outward inflection in the ophthalmic ridge. In front of the eye the ridge curves gently inward and runs forward again to a point 6-5 mm from the posterior margin and 5-5 mm from the axis, where it is inflected axially. The ophthalmic ridge is not continuous across the front of the prosoma, but terminates at a point 3-5 mm from the axis and 3 0 mm from the anterior margin. The left ophthalmic ridge is not as well preserved. The cardiac lobe is trapezoidal, 4 0 mm wide posteriorly, 2-3 mm wide at the front, and 7 0 mm long. The axial furrows are now the highest points on the flattened specimen, presumably corresponding to a greater amount of cuticular material beneath (apodemes, ventral exoskeleton, appendages). The axis is depressed and is marked by a crack (without separation) which reaches to within 3 0 mm of the anterior margin. It is not possible to count the number of apodemes, though there were certainly not less than four. The genal angles are produced into short (1-5 mm), blunt, genal spines which project posterolaterally. There may have been a narrow border (c. 0-2 mm wide) similar to that in Limulus , but details of this are obscured by flattening, possible concentric fracturing near the margin, and the fact that the doublure, apparently markedly angulated like that of Limulus, is impressed through the dorsal skeleton. The trace of this angulation disappears 3 0 mm from the posterior margin, suggesting that the angulation fades out, as in Limulus. In Limulus the angulation defines a triangular widening of the doublure in front of the mouth; this preoral area is present in Dubbolimulus, but is semicircular and clearly impressed through the carapace. There is a slight indication of a pair of median ocelli 7 0 mm from the anterior margin, where an increase in height of the ridge produced by the longitudinal axial crack suggests the original presence of a short, axially aligned, ocellar ridge. Alternatively, a median structure represented by an anteriorly convex curve 1 -5 mm wide and 2T mm from the anterior margin may represent the trace of an ocellar tubercle forming the anterior edge of the interophthalmic region (PI. 56, fig. 4, arrows). The posterior margin of the prosoma is rather straight, deflected posteriorly only at the genal angles. Between the posterior margin and the rounded anterior portion of the prosoma are triangular areas, here termed pos- teromarginal facets (after Meischner 1962, p. 185, ‘hintere Randfacette’); they are defined by the inner edge of the genal spine, the posterior edge of the cardiac lobe, and the posterior end of the ophthalmic ridge. The anterior margins of the posteromarginal facets are formed by an angulation along which the dorsal surface of the prosoma is deflected ventrally. Crushing of the facets during compression has obscured details of morphology along most of the posterior margin of the prosoma. The opisthosoma is conspicuously smaller than the prosoma, measuring 12-0 mm across its greatest width near the junction with the prosoma, and 7-5 mm from the anterior margin to the lobes on either side of the telson attachment; this makes it about equal in size to the area within the ophthalmic ridge of the prosoma. The lateral margins run parallel for 2 0 mm before turning axially, so that the opisthosoma tapers markedly to a minimum width of 4-2 mm, T5 mm from the posterior margin, which has a shallow re-entrant c. 1 -0 mm deep for the telson attachment. The lateral margins bear no trace of either fixed or movable spines. As on the prosoma, the cardiac lobe is defined by the most raised part of the specimen, bounding a trapezoidal area 4 0 mm wide anteriorly and 2-5 mm wide at the rear. These ridges show traces of probably six apodemes. The area between the ridges is rather depressed. The lateral margins show a narrow border not more than 0-5 mm wide, which may correspond to the ventral doublure. If this is the case, the doublure is relatively narrower than in Limulus, possibly reflecting the absence of lateral spines. There is no indication of anterolateral free lobes on the opisthosoma; rather, the EXPLANATION OF PLATE 55 Figs. 1, 2. Dubbolimulus peetae gen. et sp. nov. Holotype specimen. Ballimore Formation, middle Triassic, Dubbo, New South Wales, x 4. 1, MM F27693, upper counterpart, showing the impression of the doublure, the posterolateral facets, and suggestions of apodemes on the opisthosoma. 2, MM F27694, lower counter- part, low angle lighting showing the appendages. PLATE 55 PICKETT, Dubbolimulus 616 PALAEONTOLOGY, VOLUME 27 doublure continues around the anterolateral corner, whereas in Limulus, with its well-developed free lobes, the morphology is quite complex in this area. Some prosomal appendages have been impressed through the carapace (PL 55, fig. 2). Traces of five appendages are evident on the right side. Assuming that appendage I (the chelicera) would not be apparent and that the first ridge represents appendage II (by analogy with Limulus ), a group of four legs (appendages II-V) is succeeded by a gap (in which there are two smaller, unidentified impressions) and another long, laterally directed leg, corresponding to the posteriorly directed appendage VI of Limulus. None of the legs projects beyond the rim of the carapace, but legs V and VI almost reach it. No joints in the legs are discernible. They are almost straight, with a slight anterior curve. On the left there are less regular impressions of possibly five legs reaching almost to the margin, but except for appendage VI these are less easily identifiable. A possible reconstruction of D. peetae gen. et sp. nov. is shown in text-fig. 2. Discussion. The specimen has been much flattened during compaction of the sediment. This is most obviously expressed as (a) the impression of legs, doublure, and apodemes through the carapace, ( b ) the presence of an axial crack and some damage in the region of the left ophthalmic ridge, and (c) wrinkles on the outer part of the prosoma. This latter point particularly may reflect the fact that the exoskeleton was not strongly mineralized. The wide angle between the rear margin of the prosoma and the lateral edges of the opisthosoma is an unusual feature in limuloids, and was originally attributed to anterolateral movement of the genal angles during compaction. To examine the effects of compression on limuloids, four juvenile specimens of Limulus polyphemus were embedded in clay and subjected to considerable pressure applied by means of screw clamps. Some lateral spread occurred in all cases, the least amount occurring in the specimen illustrated in PI. 56, fig. 1, in which the maximum width of the prosoma increased by 1-4%. The length of the prosoma, however, was reduced by over 10%. In no case was there any change which suggested that the condition observed in Dubbolimulus could have resulted from the spreading of an original configuration of genal angles similar to that in Limulus. The prosoma outside the ophthalmic ridges has suffered the greatest damage, being compressed into a number of concertina-like folds roughly concentric with the margin. It is apparent that the dorsal spines greatly reinforce the exoskeleton (thicker at these points), as the areas around them have suffered little crushing. Probably for this reason there is no development of a longitudinal axial crack as in the holotype of D. peetae , since young L. polyphemus have six axial spines, three on the prosoma and three on the opisthosoma. Insufficient pressure was achieved to impress the appendages through the dorsal exoskeleton. By contrast, a specimen flattened between numerous sheets of paper in a hydraulic press suffered considerable anterolateral movement of the genal angles, and the spines on the ophthalmic ridges were flattened sideways. EXPLANATION OF PLATE 56 Fig. 1. Limulus polyphemus. Epoxy resin cast of juvenile specimen compressed in deformation experiment. Note the concentric folds of the exoskeleton outside the ophthalmic ridges, and the proximity of the genal spines to the lateral margins of the opisthosoma, x 0-95. Fig. 2. Pincombella belmontensis Chapman. NMV PI 3646, latex cast of holotype, Boolaroo Sub-group, late Permian, Belmont, New South Wales, x 7-5. Figs. 3, 4. Dubbolimulus peetae gen. et sp. nov. MM F27693, holotype, Baltimore Formation, middle Triassic, Dubbo, New South Wales. 3, upper counterpart, x 1. 4, detail of upper counterpart showing median area of prosoma in ocellar region, x 14; the arcuate structure is the rear edge of the doublure; arrows indicate the levels of structures which may be interpreted as ocelli. Fig. 5. Trimerus harrisoni (McCoy). MUGD 1201, latex cast of ‘ Hemiaspis ’ tunneclijfei Chapman, Dargile Formation, late Silurian, Studley Park, Melbourne, Victoria. The area in shadow near the top of the figure lies outside the real margin of the specimen, but is what was interpreted as a wide border with radiate markings by Chapman (1932, pi. 14, fig. 5). Thoracic pleurae are visible on the right, but only the posterior pleurae are preserved on the left, x 1 . PLATE 56 PICKETT, Australian xiphosurans 618 PALAEONTOLOGY, VOLUME 27 It is apparent that the hydraulic conditions of compression under geological circumstances hinder lateral spreading of strongly profiled organisms, as established by Walton (1936) and Rex and Chaloner (1983) for plant compressions. I thus conclude that the outline of Dubbolimulus as preserved is very close to the original. Having determined this, there is some support for the undisturbed rear margin of the prosoma from other aspects of the morphology. If the prosoma was moderately arched (as it must have been to create room for the appendages, their musculature and the digestive organs), the possibility emerges, with an outline like that of Dubbolimulus , that the arched posterior margins beyond the edges of the opisthosoma left a wide gape, exposing the underparts of the prosoma. Such a morphology would be out of character with the general style of limuloids, which are close to their substrate on all sides so that their ventral surface is well protected. The rear margin of the prosoma of Dubbolimulus bears a triangular area on either side of the interophthalmic region which is interpreted as a ventrally deflected posteromarginal facet, thus largely closing the gape at the rear of the prosoma. This structure occurs in Psammolimulus (Meischner, 1962) and Victalimulus, though not described in the latter. It is also present in the oldest member of the group, Paleolimulus , being clearly indicated in Dunbar’s (1923) figures, in ‘ Limulus ’ woodwardi Watson, 1909, and in Mesolimulus walchi , in which it has a somewhat exaggerated development. PICKETT: MIDDLE TRIASSIC LIMULOID 619 Ivanov (1933, fig. 63) figured a specimen of M. walchi and remarked (1933, p. 295) that, on the underside of the prosoma, there is an obvious furrow corresponding to the pleural part of the anterior border of the sixth segment. He equated this with a ridge in the same position on the dorsal surface of other specimens. This structure may be one taken up by Stormer as a primitive feature (1952, p. 630: 'A furrow along the inner margin of the genal angles may also be a primitive character’). It is possible that the angulation at the anterior border of the posteromarginal facet corresponds to the boundary between segments V and VI but, as it appears to have a function in protecting the ventral surface of the prosoma, this is not necessarily so. Habitat. Dubbolimulus was recovered from strata which are undoubtedly of freshwater origin. No marine Triassic rocks are known in New South Wales, the nearest being 800 km distant near Gympie in Queensland (Fleming 1966; Runnegar 1969). The Ballimore Formation is the basal unit of the Great Artesian Basin in the Dubbo area, and lies unconformably on folded strata of Ordovician to Devonian age. The oldest marine strata in the New South Wales portion of the Great Artesian Basin are early Cretaceous (Aptian) in age (Scheibnerova 1976; Morgan 1980); the marine succession in this basin is the result of a north to south transgression. To the east there was an area of high land, just as there is today. Thus the locality was probably as remote from the ocean in Triassic times as it is at present, and any possibility of marine influence can be ruled out. The general area abounds in plant fossils (e.g. Holmes 1982), indicating a limnic situation, and the specimen of Dubbolimulus was recovered while collecting plant fossils. Riek and Gill (1971) suggested that their specimen of Victalimulus , also from freshwater strata, may have migrated there for breeding purposes. In view of the distance of the type locality from any areas of marine sedimentation, such a possibility seems remote for Dubbolimulus. There has been a certain reluctance on the part of palaeontologists to attribute a freshwater habitat to fossil limuloids (Riek and Gill 1971; Holland et al. 1975). None the less, in addition to the species described by those authors ( V. mcqueeni and C aster olimulus kletti ), there are others which have been recovered from non-marine strata: A.fletcheri (Riek, 1955), Psammolimulus gottingensis (Meischner, 1962). On the other hand, species of undoubted marine origin are M. walchi (Stormer 1952; Barthel 1974), Heter olimulus gadeai and Tarrac olimulus rieki (Via Boada et at. 1977), ‘ Limulus ’ woodwardi (Watson, 1909) and ‘L.’ syriacus (Woodward, 1879). Species referred to Limulitella appear to belong to both groups, e.g. L. bronni (Bill, 1914) and L. liasokeuperinus (Braun, 1860) appear to be from non-marine strata, whereas L. vicensis (Bleicher, 1 897) is marine. Colleagues have been strongly influenced by the marine nature of modern limulids, and by the rare reports of some species in fresh waters. However, Annandale (1909, p. 295) claimed that Carcinoscorpius rotundicauda is ’mainly if not entirely estuarine’ and indicated that it occurs almost 1 50 km from the mouth of the Hooghly River. The remoteness of Dubbolimulus from any marine areas suggests strongly that its whole life cycle was spent in fresh water. Riek and Gill ( 1971, p. 207) have suggested that Victalimulus migrated to fresh water for breeding. If one accepts a migration between marine and limnic environments for V. mcqueeni , it seems more likely, in view of the basically marine nature of extant forms, that the breeding phase would have been marine, as, for instance, in many Recent terrestrial crabs. Acknowledgements . I am indebted to Mrs. Judie Peet for making the specimen of D. peetae available. Colleagues Peter Jell and Edgar Riek took part in valuable discussions and read a draft of the manuscript. The article is published with the permission of the Secretary, New South Wales Department of Mineral Resources. REFERENCES annandale, N. 1909. The habits of Indian king crabs. Rec. Indian Mas. 3, 294-295. barthel, k. w. 1974. Limulus : a living fossil. Horseshoe crabs aid interpretation of an Upper Jurassic environ- ment (Solnhofen). Naturwissenschaften , 61, 428-433. Bergstrom, j. 1975. Functional morphology and evolution of xiphosurids. Fossils Strata , 4, 291-305, 1 pi. 620 PALAEONTOLOGY, VOLUME 27 bill, p. c. 1914. Ueber Crustaceen aus dem Voltziensandstein des Elsasses. Mitt. geol. Landesanst. Els.-Loth. 8, 327-330. bleicher, m. 1897. Sur la decouverte d’une nouvelle espece de Limule dans les marnes irisees de Lorraine. Bull. Soc. Sci. Nancy , Ser. 2, 14, 116-126. braun, c. F. 1860. Die Thiere in den Pflanzenschiefern der Gegend von Bayreuth. caster, K. E. and kjellesvig-waering, E. N. 1953. Melbournopterus , a new Silurian eurypterid from Australia. J. Paleont. 27, 153-156. chapman, f. 1932. Two new Australian fossil king-crabs. Proc. R. Soc. Viet, n.s., 44, 100-102. dunbar, c. a. 1923. Kansas Permian insects. Part 2. Paleolimulus , a new genus of Paleozoic Xiphosura, with notes on other genera. Am. J. Sci., Ser. 5, 5, 443-454. fisher, d. c. 1975. Swimming and burrowing in Limulus and Mesolimulus. Fossils Strata, 4, 281-290. — 1977. Functional significance of spines in the Pennsylvanian horseshoe crab Euproops danae. Paleobiol. 3, 175-195. 1981 . The role of functional analysis in phylogenetic inference: examples from the history of the Xiphosura. Am. Zool. 21, 47-62. — 1982. Phylogenetic and macroevolutionary patterns within the Xiphosurida. Proc. Third N. Am. Paleont. Corn. 1, 175-180. Fleming, p. J. G. 1966. Eotriassic marine bivalves from the Maryborough Basin, southeast Queensland. Pubis geol. Surv. Qd , 333 (Palaeont. Paps 8), 17-29. gill, e. d. 1949. Palaeozoology and taxonomy of some Australian homalonotid trilobites. Proc. R. Soc. Viet. n.s., 61, 61-73. — 1951. Eurypterids — scorpions of the sea. Victorian Nat. 68, 128-133. Holland, f. d., jr., erickson, j. m. and o’brien, d. e. 1975. C aster olimulus , a new Late Cretaceous generic link in limulid lineage. Bull. Am. Paleont. 67 (287), 235-249, 2 pis. holmes, w. b. k. 1982. The Middle Triassic flora from Benolong near Dubbo, western New South Wales. Alcheringa , 6, 1-33. ivanov, p. p. (iwanoff, p. p.) 1933. Die embryonale Entwicklung von Limulus moluccanus. Zool. Jb., Abt. Anatomie und Ontogenie , 56, 163-348. mccoy, F. 1876. Prodromus of the palaeontology of Victoria. Decade III, 1-40, pis. 21-30. Government Printer, Melbourne, and Trubner and Co., London. 1899. Note on a new Australian Pterygotus. Geol. Mag. N.s., dec. 4, 6, 193-194. meischner, K.-D. 1962. Neue Funde von Psammolimulus gottingensis (Merostomata, Xiphosura) aus dem mittleren Buntsandstein von Gottingen. Palaeont. Z. (LI. Schmidt Festband), 185-193, pis. 19, 20. Morgan, r. p. 1980. Palynostratigraphy of the Australian Early and Middle Cretaceous. Mem. geol. Surv. N.S. W. Palaeont. no. 18, 1-153, 38 pis. quilty, p. g. 1972. A Middle Cambrian xiphosuran (?) from Western Tasmania. Pap. Proc. R. Soc. Tasm. 106, 21-23. Raymond, p. E. 1944. Late Paleozoic Xiphosurans. Bull. Mus. comp. Zool. Harv. 94, 475-508. rex, G. m. and chaloner, w. g. 1983. The experimental formation of plant compression fossils. Palaeontology, 26, 231-252, pis. 30-33. richter, i.-E. 1964. Bewegungsstudien an Limulus. Verb. dt. zool Ges., Miinchen 1963, Zool. Anzeiger Suppl. 27, 491-497. riek, e. f. 1955. A new xiphosuran from the Triassic sediments at Brookvale, New South Wales. Rec. Aust. Mus. 23, 281-282. — 1968a. Undescribed fossil insects from the Upper Permian of Belmont, New South Wales. Ibid. 27, 303- 310. — 1968A A re-examination of two arthropod species from the Triassic of Brookvale, New South Wales. Ibid. 27, 313-321. — and gill, e. d. 1 97 1 . A new xiphosuran genus from Lower Cretaceous freshwater sediments at Koonwarra, Victoria, Australia. Palaeontology, 14, 206-210, pi. 33. romero, A. and via boada, l. 1977. Tarracolimulus rieki, nov. gen., nov. sp., nuevo limulido del Triasico de Montral-Alcover (Tarragona). Cuad. Geol. iber. 4, 239-246. runnegar, b. 1969. A lower Triassic ammonoid fauna from southeast Queensland. J. Paleont. 43, 818-828, pis. 103, 104. scheibnerova, V. 1976. Cretaceous foraminifera of the Great Australian Basin. Mem. geol. Surv. N.S.W. Palaeont. no. 17, 1 -265, 77 pis. SCHIMPER, w. ph. 1850. Palaeontologia alsatica. Mem. Soc. Hist. nat. Strasbourg, 4, 5-7. PICKETT: MIDDLE TRIASSIC LIMULOID 621 ST0RMER, L. 1952. Phylogeny and taxonomy of fossil horse-shoe crabs. J. Paleont. 26, 630-639. — 1955. Merostomata. In moore, r. c. (ed.). Treatise on Invertebrate Paleontology. P. Arthropoda 2. P4- P41. Geological Society of America and University of Kansas Press, New York, and Lawrence, Kansas. via boada, l. and villalta, j. f. 1966. Heterolimulus gadeai, nov. gen., nov. sp., representant d’une nouvelle famille de Limulaces dans le Trias d’Espagne. C.r. somm. Seanc. Soc. geol. Fr ., Ser. 7, 8, 57-59. [Spanish translation in Acta geol. hispan. 1, 9-11] — and esteban cerda, m. 1977. Paleontologia y paleoecologia de los yacimientos fossiliferos del Muschelkalk superior entre Alcover y Mont-Ral (Montanas de Prades, Provincia de Tarragona). Cuad. Geol. iber. 4, 247-256. waldman, m. 1971. Fish from the freshwater Lower Cretaceous of Victoria, Australia, with comments on the palaeoenvironment. Spec. Pap. Palaeont. 9, 1-124, pis. 1-18. walton, J. 1936. On the factors which influence the external form of fossil plants, with descriptions of the foliage of some species of the Palaeozoic equisetalean genus Annularia Sternberg. Phil. Trans. R. Soc. B226 (535), 219-237. watson, d. m. s. 1909. Limulus woodwardi , sp. nov. from the Lower Oolite of England. Geol. Mag. n.s., dec. 5, 6, 14-16. woodward, H. 1879. On the occurrence of a fossil king-crab ( Limulus syriacus ) in the Cretaceous formation of Lebanon. Q. Jl geol. Soc. Load. 35, 554-555. zittel, K. A. 1885. Handbuch der palceontologie. I. Palaozoologie. II. MoUusca und arthropoda. R. Oldenbourg, Miinchen und Leipzig, 893 pp. j. W. PICKETT Geological and Mining Museum 36 George Street Typescript first received 3 August 1983 Sydney, N.S.W. 2000 Revised typescript received 10 November 1983 Australia 4* RAMSEYOC RINUS AND RISTNA CRINUS FROM THE ORDOVICIAN OF BRITAIN by STEPHEN K. DONOVAN Abstract. Ramseyocrinus Bates, 1968, has hitherto been included in the family Eustenocrinidae Ulrich, 1925, but it differs from all other members of this family in having only four radials, an anal X which is supported by two radials (rather than two or three superradials, or a brachianal and superradial), a cup which is about as wide as high, and a column which is tetrameric proximally and tetragonal holomeric distally. A new family, Ramseyocrinidae, is erected for this genus. Of the true eustenocrinids, Ristnacrinus Opik, 1934 is recognized from the British Ordovician for the first time, based upon the occurrence of its distinctive columnals with synarthrial articulating ridges. Ristnacrinus columnals are known from four localities of Cautleyan-Rawtheyan (?Hirnantian) age and comparison with similar columnals from the Swedish Boda Limestone (Ashgill) suggests that more than one species may be present in Britain. Only about thirty crinoid species have been described from the British Ordovician (Ramsbottom 1961; Bates 1965, 1968; Brower 1974; Donovan 1983). However, dissociated skeletal elements of crinoids, especially columnals, are common particularly in the Caradoc and Ashgill. Only highly distinctive columnals are of practical use in taxonomy, as homeomorphs between distantly related taxa are probably common ( Broadhead and Strimple 1 977). However, it is possible to make a generic identification of at least some columnals (Donovan 1 983). One of the most distinctive of all columnals found in the Ordovician is that of the dististele of Ristnacrinus Opik, 1 934, which has been recognized by a number of subsequent authors (Chauvel and Le Menn 1972, 1979; Chauvel el al. 1975; Briskeby 1981; Wright 1983). The radial plates of the dorsal crinoid cup are defined as those ossicles which are radial in position. Basal plates have an interradial orientation. Some inadunate crinoids possess two radial plates in some or all rays of the cup. Such a compound radial is composed of an inferradial (supported by the basal plates) and a superradial (supported by the associated inferradial). Ristnacrinus belongs to the family Eustenocrinidae, whose members possess five compound radials. Re-examination of a second British Ordovician crinoid which was thought to belong to this family, Ramseyocrinus cambriensis (Hicks) (Bates 1968), reveals that, despite similarities, it is certainly not a eustenocrinid. This species is therefore included here in a new crinoid family, the Ramseyocrinidae. Terminology used in this paper follows Moore et al. (1968), Ubaghs (1978), and Webster (1974). SYSTEMATIC PALAEONTOLOGY Class crinoidea J. S. Miller, 1821 Subclass inadunata Wachsmuth and Springer, 1881 Order disparida Moore and Laudon, 1943 Family ramseyocrinidae nov. Diagnosis. Monocyclic crinoids with a low, cylindrical cup, as wide as the proximal column. Four radials, with an anal X supported by two radials. Basals concealed or absent. Four arms, isotomously branched. Anal X bears at least three further plates in the anal series. Proximal stem quadripartite; distal stem holomeric, tetragonal. (Palaeontology, Vol. 27, Part 3, 1984, pp. 623-634.) 624 PALAEONTOLOGY, VOLUME 27 Discussion. Bates (1968) assigned R. cambriensis to the Eustenocrinidae but it differs from other members of this family by having only four non-compound radials, an anal X which is supported by two radials (rather than two or three superradials, or a brachianal and superradial) and a column which is tetrameric proximally while retaining tetragonal symmetry throughout. Differences of the cup plate arrangements are illustrated by text-fig. 1. In Eustenocrinus five basals are apparent and are offset from the radials, all of which are compound (text-fig. 1a). The anal series is supported by the superradials of the B-, C-, and D-rays. The first circlet of primibrachs are fixed to the cup. Only four arms are present. Ristnacrinus has basals which are either hidden or absent (text-fig. 1b). The five radials are all compound, each superradial supporting an arm. Ossicles of the anal series are small and are supported by the C- and D-ray superradials, to the left of the C-ray brachial. Peniculocrinus (text-fig. 1e) is similar to Eustenocrinus but the anal series is supported by a large brachianal in the C-ray (i.e. five arms are present) and the D-ray superradial. The first two circlets of brachials are fixed. In order to contrast these eustenocrinid genera with Ramseyocrinus it is first necessary to define the plates of the cup in the latter. Two alternative interpretations are shown (text-fig. lc, d). Text-fig. Id is based mainly upon the interpretation of Bates (1968, p. 407), although his inferradianal is labelled as an anal X. In this plating scheme basals are three in number, the large basal in the A- and E-rays apparently being fused from two smaller plates. Radials are not offset from basals but are directly supported by them. This is unsatisfactory because the ‘basals’ are radial in orientation. The four arms are directly supported by the radials. The anal X is supported by the two smaller basals. The interpretation preferred in this paper is shown in text-fig. lc. Basals are either hidden or absent. The cup is composed of four radials, surmounted by a circlet of four fixed brachials (the ramseyocrinid Entrochus primus does not have this circlet of fixed brachials; R. J. Prokop, pers. comm.). The radials in the A- and E-rays are fused. The four arms arise directly from the radials. The anal X is supported by the two smaller radials. This interpretation (text-fig. lc) is preferred because of the obvious similarities of the cup plating to that of crinoids such as Eustenocrinus (although Ramseyocrinus does not have split radials, merely radials surmounted by fixed brachials). Radials may be defined as the most proximal plates in each ray (Moore et al. 1952, p. 608). It is apparent that the most proximal plates in each ray of Ramseyocrinus are those called basals by Bates (1968) (text-fig. Id). It is therefore correct to regard his ‘basals’ as radials. Moore et al. (1978, p. T554) propose that both basals and split radials are present, i.e. the most proximal circlet of free brachials in text-fig. lc are incorporated in the cup as superradials, the fixed brachials (this paper) are inferradials and the lowest plate circlet are basals. The first two plates of the anal series are regarded as a compound radial. Similar arguments to those stated above can be applied to show that the lowest plate circlet in the cup is not composed of basals. The lowest free brachial circlet (text-fig. lc) does not appear to be fixed. If present, and it is emphasized that this cannot be proved, the basals must be extremely small and are probably concealed by the lobes of the stem. Many articulate crinoids have a hidden circlet of infrabasal plates (cryptodicyclic) (Rasmussen 1978). However, these infrabasals are concealed by relatively broad, circular, or pentagonal columnals. The lobes of the proxistele in R. cambriensis are narrow and separated. The anal tube can be recognized in two specimens of R. cambriensis (NMW 29.308.G296 and Manchester Museum LI 2360). It is also seen in Entrochus primus (R. J. Prokop, pers. comm.). In both species the tube is formed of a column of about four plates. If this new interpretation is now compared with Eustenocrinus , Ristnacrinus , and Peniculocrinus , a number of differences are apparent. Each of the eustenocrinid genera has five compound radials, whereas Ramseyocrinus has only four radials, none of which is compound. The cups ofeustenocrinids are tall, sometimes incorporating brachials. The cup of Ramseyocrinus is about as wide as high, including a circlet of fixed brachials. The anal series of Ramseyocrinus is supported by two radials, whereas that in eustenocrinids is supported by three superradials (Eustenocrinus), two superradials ( Ristnacrinus ), or a superradial and a fixed brachianal (Peniculocrinus). This last point is particularly DONOVAN: ORDOVICIAN CRINOIDS 625 BrA Brachianal Basals,brachials Radials, superradials Inferradials Fixed brachials Anal plates X Anal X- plate 9 Basals hidden or absent text-fig. 1. Cup plating diagrams for: a, Eustenocrinus (after Ulrich 1925; Moore 1962); b, Ristnacrinus (after Opik 1934); c, d, Ramsey ocrinus. Two interpretations of the cup are shown, both based on NMW 29.308.G296 and G318 (counterparts). In c the basals are hidden or absent, radials and a circlet of fixed brachials are present, and the anal X is supported by two radials (this paper). In d basals and radials are present, two basals supporting the anal X (based on the description of Bates 1968). e, Peniculocrinus (after Moore 1962). 626 PALAEONTOLOGY, VOLUME 27 important, although it is apparent that minimal reduction of the superradials in the C- and D-rays of Ristnacrinus could lead to the anal X being supported by two inferradials. The stem of Ramseyocrimis is also radically different from that of eustenocrinids. The unusual column of Ristnacrinus is discussed in detail below but it is sufficient to note here that most columnals are holomeric (i.e. composed of a single ossicle) and circular, with a central synarthrial ridge. Peniculocrinus has a subpentagonal column (Moore et at. 1978, p. T554), with what appear to be lenticular columnals (cp. Moore et al. 1978, text-fig. 347. 2e, p. T555, with Taylor 1983, text-fig. 41, p. 61). The stem of Eustenocrinus is circular and apparently pentameric (Ulrich 1925, p. 99). The column of Ramseyocrimis differs from all these by showing fourfold symmetry (text-figs. 2, 3). The proxistele of R. cambriensis is tetralobate and quadripartite. Distally the column is square but holomeric. text-fig. 2. Ramseyocrimis cambriensis { Hicks). Columns and columnals. a, reconstruction of the articular facet of a proximal columnal (Donovan 1983, text-fig. 2c). b, NMW 29.308.G220, dissociated mere, c, d, NMW 29.308.G296 and G3 1 8 (counterparts), proxistele beneath cup. e, f, BM(NH) E3, proxistele immediately beneath cup (e) and slightly more distal (f, in which the columnal marked by a dot is 36 mm below the base of the cup). G, tentative reconstruction of the articular facet of a distal columnal. Lumen shape, size, and orientation conjectural. All camera lucida drawings of latex casts, except a and g. DONOVAN: ORDOVICIAN CRINOIDS 627 Genus ramseyocrinus Bates, 1968 Type species. Dendrocrinus cambriensis Hicks, 1873, designated by Bates (1968). Diagnosis. As for the family Ramseyocrinidae, with the first circlet of brachials fixed. Discussion. The first primibrach circlet of the only other Ramseyocrinid known, Entrochus primus , is free above the radials (R. J. Prokop, pers. comm.). Ramseyocrinus cambriensis (Hicks), 1873 Text-figs, lc, d, 2, 3 1873 Dendrocrinus cambriensis Hicks, p. 51, pi. 4, figs. 1 7—20. 1960 Iocrinus? cambriensis (Hicks) Ramsbottom, pp. 5, 6, pi. 3, figs. 9-11. 1968 Ramseyocrinus cambriensis (Hicks) Bates, pp. 406-409, pi. 76, figs. 1-5. Diagnosis. Radials fused in the A- and E-rays; arms branched isotomously at least four times. Material. Ramsbottom (1961, p. 5) recognized Sedgwick Museum (SM) A16739 as a syntype (Hicks’s fig. 17; also a gutta-percha impression in the Geological Survey Museum collection, 1GS GSM 82819) and noted IGS GSM 59428 as a gutta-percha impression of the specimen in Hicks’s fig. 20. His description is based mainly on British Museum (Natural History) specimen BM(NH) E3. Bates (1968, p. 407) discovered the counterpart mould of one of Hicks’s syntypes (now numbered SM A 16739a, b) but based his description on National Museum of Wales (NMW) 29.308.G296 and G318 (counterpart moulds). This description is based on these specimens and also NMW 29.308.G220 and G470(a) (not counterparts), and University College of Wales, Aberystwyth, specimen UCW 21467. Horizon and Localities. Porth Gain Beds, Ogof Hen Formation (Lower Arenig), Ramsey Island, Dyfed, NGR SM 708 252 (Pringle 1930). Specimens are also known from a quarry on Lleithyr Farm, Dyfed, near to the ruins and by the side of the road to Carnedd, NGR SM 748 248 (Jenkins 1979). Description. Stem (text-figs. 2, 3a, c) wide proximally, tetrameric, quadrilobate, lobes almost parallel-sided beneath cup, becoming less pronounced distally. Columnals tend to be irregular in height, particularly near the top of the stem (text-fig. 2c-e); diameter of columnals also varies slightly but no more than three orders of columnals are ever present, i.e. two orders of internodal. Articulation synostosial. Lumen tetragonal with inwardly curved sides, angles corresponding to meric sutures (text-fig. 2a). Meres lozenge-like (text-fig. 2b). Distal stem square in section and holomeric; heteromorphic N212 (Webster 1974) but the sequence of internodals is sometimes incomplete. Articulation apparently synostosial (text-fig. 2g). The stem terminates in a planar, coiled holdfast (text-fig. 3a). Basals hidden or absent (text-figs, lc, d, 3b). Four radials, those in the A- and E-rays fused to form a single large plate (the line of fusion apparent as a groove on NMW 29.308.G318, whereas boundaries between separate plates are clearly evident). Four fixed brachials, two supported by the large radial and one by each of the small radials. Anal X lower than the radials and supported by the two small radials. Anal series of at least four plates of similar morphology. Tegmen not seen. Four arms (text-fig. 3b) supported directly by the four radials. Arms branch isotomously at least four times. At least ten primibrachs, elliptical in section, wider than high, with a narrow, shallow, ventral food groove. Maximum of thirteen secundibrachs, although there may be as few as four (Bates 1968, p. 407); the most proximal three are low, becoming about as wide as high more distally. Tertibrachs higher than wide, ten at most, quadribrachs tall, thin, about twelve, pentibrachs narrow, tall, at least ?five. Discussion. Only four crinoid genera are known which have a tetrameric column: Ramseyocrinus , Colpodecrinus (Sprinkle and Kolata 1982; Donovan 1983), Tetragonocrinus (Yeltysheva 1964), and Entrochus Barrande (previously only known from dissociated columnals but to be fully described from more complete material by Dr. R. J. Prokop). R. cambriensis is also the oldest crinoid so far discovered in Britain. The proximal stem is well known from several well-preserved specimens (text-figs. 2a-f, 3b, c). These are all tetralobate and tetrameric, with a random intercalation of internodals often apparent (text-fig. 2c-f). 628 PALAEONTOLOGY, VOLUME 27 The columnal morphology has been deduced from NMW 29.308.G220 (text-figs. 2b, 3c). Details of the articular facet are not shown by specimens of ‘whole’ animals (none is preserved complete with the distal stem) but this specimen is partly disarticulated, enabling the morphology of individual meres to be determined. From this a reconstruction of the facet has been made (text-fig. 2a). The latera of the meres are sub-parallel and it is deduced that the stem fragment is from some distance beneath the cup. The minimal contact between meres of the same columnal did not favour preservation of whole columnals but instead the stem first disarticulated along its meric sutures, producing an effect analogous to peeling the skin from a banana. The separate stem lobes then disarticulated into individual meres. text-fig. 3. Ramseyocrinus cambriensis (Hicks), a, NMW 29.308.G470(a), dististele with planar, coiled holdfast, x 4. b, NMW 29.308.G3 1 8, cup with proximal column and arms, x 3- 1 . c, NMW 29.308. G220, partially disarticulated stem with a dissociated mere in the bottom left corner, x 3-5. All latex casts whitened with ammonium chloride. DONOVAN: ORDOVICIAN CRINOIDS 629 The description of the distal stem is based on NMW 29.308.G470(a) (text-fig. 3a) from Ramsey Island. It cannot be proved that this is R. cambriensis but the assumption is based on the absence of evidence for other pelmatozoans at the type locality of the Ogif Hen Formation and the tetragonal symmetry of the specimen. If the assumption is correct, there must have been a change from tetramerism to holomerism distally. The planar coil of the pluricolumnal is suggestive of an attachment structure and is unlike the primitive holdfasts of certain other early crinoids such as Echmatocrinus (Sprinkle 1973) and Aethocrinus (Ubaghs 1969). A suggested reconstruction of the articular facet of a coluinnal of the dististele is shown in text-fig. 2o. Bates (1968, p. 408) considers R. cambriensis to be the most primitive eustenocrinid, possibly closely related to the ancestral stock of the disparids. This may be so but it is unlikely that Ramseyocrinus is a direct ancestor of later monocyclic inadunates. It has many unique features not found in later disparids (apart from Entrochus ), particularly the tetragonal symmetry of the stem and the presence of four arms and radials. Crinoid columnals with fourfold symmetry are not common in the British Ordovician until the Ashgill, when they are more likely to be derived from camerates such as Xenocrinus or Colpodecrinus. A final reason for doubting the close relationship of Ramseyocrinus to eustenocrinids is stratigraphic. R. cambriensis is limited to the Lower Arenig of South Wales, whereas the earliest Ristnacrinus columnals are Upper Llandeilo (see below) and both Eustenocrinus and Peniculocrinus are of Trenton age (mid Caradoc). Family eustenocrinidae Ulrich, 1925 Diagnosis. Monocyclic crinoids with compound radials in all five rays. Basals may or may not be seen. Anal X supported by the B-, C-, and D-ray superradials ( Eustenocrinus ), by the D- and C-ray superradials ( Ristnacrinus ), or by the D-ray superradial and a brachianal in the C-ray ( Peniculo- crinus). Cup conical. Arms isotomously branched. Column round to sub-pentagonal in section. Genus ristnacrinus Opik, 1934 Type species. Ristnacrinus marinus Opik, 1934, by original designation. Diagnosis of stem. Stem circular in section; latera convex; proxistele heteromorphic, Nl, with symplexially articulated columnals. Articular facet of a distal columnal has a central, synarthrial ridge flanked by two depressed ligament areas which are surrounded by a marginal rim; latera convex. Discussion. Ristnacrinus , Caleidocrinus multiramus , and an undescribed columnal species from the Boda Limestone of Osmundsberget, Sweden, are the only Ordovician crinoids known to have columnals with synarthrial articulation. The latter is elliptical and low (possibly myelodactylid) and the columnals of C. multiramus have less convex latera and are proportionally taller than those of Ristnacrinus. The type species of Ristnacrinus , R. marinus , is founded on a particularly complete specimen, three cups and fragmentary columnals and pluricolumnals from the Caradocian Dj (Johvi) Stage of Estonia (Opik 1934), which is approximately equivalent to the late Harnagian to early Soudleyan of Britain (Williams et al. 1972). The near-complete specimen lacks a distal termination to the column but retains about 35 cm of the stem. This is divided into a short proxistele and a much longer dististele. The proxistele of fifteen columnals is heteromorphic, N I (Webster 1974), with alternating nodals and internodals with convex and concave latera, respectively. These have a radial symplexial articulation (Opik 1934, p. 4). The dististele is highly flexible, homeomorphic, and composed of columnals with a synarthrial articulating ridge on each articular facet. Dissociated columnals of this type are easily recognized as belonging to Ristnacrinus. Opik ( 1 934, p. 1 ) notes that similar columnals are also found below (Kukruse Limestone and Oil Shale, Stages C2 and C3, Upper Llandeilo to early Harnagian) and above (Keila Limestone, D2 Stage, late Soudleyan) the Johvi Stage. 630 PALAEONTOLOGY, VOLUME 27 D. K. Wright (1983) has noted columnals from the Longvillian and Woolstonian of Snowdonia, North Wales, which are almost certainly Ristnacrinus (his Cyclocyclopa D). The species R. cirrifer Le Menn is based upon dissociated, cirriferous columnals from the Upper Ordovician (Ashgill) of Coat-Carrec, Argol, south-east of Brest, Brittany, France (Chauvel and Le Menn 1972). Ristnacrinus cf. cirrifer is also known from the Ashgill of Aragon, Spain (Chauvel et al. 1975; Chauvel and Le Menn 1979). Briskeby (1981) described columnals of R. ?marinus from the Upper Ordovician of Hadeland, north of Oslo, Norway. Finally, columnals of Ristnacrinus sp. have been found below the Swedish Kullsberg Limestone (Upper Llandeilo) at Kullsberg, Dalarma, and in the Ashgill Boda Limestone (C. R. C. Paul, pers. comm.). Synarthrial articulation has evolved three, or perhaps four, times in crinoid columns, in inadunates (ristnacrinids, including C. multiramus , and myelodactylids), camerates (platycrinitids), and articulates (bourgueticrinids). The British Ashgill Ristnacrinus columnals are similar to those illustrated by Opik ( 1 934, pi. 1 (labelled pi. 2), fig. 2). One of Opik’s columnals appears to bear some sort of process, possibly a cirrus. If so, this would seem to invalidate R. cirrifer Le Menn. Indeed, columnals of this genus show such limited morphological variation, despite their distinctive appearance, that it is here considered expedient to regard them all as Ristnacrinus sp. text-fig. 4. Scanning electron micrographs of Ristnacrinus sp. from the Ashgill of Britain. a~c, BM(NH) E69201. A, c, articular facets at each end of the pluricolumnal; b, lateral view, d, BM(NH) E69202, articular facet. E, f, BM(NH) E69207. e, lateral view; f, articular facet, g, BM(NH) E69199, articular facet. Scale bars represent 1 mm. DONOVAN. ORDOVICIAN CRINOIDS 631 Ristnacrinus sp. Text-figs. 4, 5 Material. Eleven British Museum (Natural History) specimens: BM(NH) E69197 to E69207. Horizons and Localities. These columnals come from four localities. (u) BM(NH) E69205. At the top of the quarry in the wood, behind the school, Corwen, Clwyd. NGR SJ 071 433. Rawtheyan siltstones. Collected by Mr. B. Cullen. ( b ) BM(NH) E69206. Loose at the 4-39-70 to +4T00 m level, west side of Sholeshook railway cutting (Paul 1973, fig. 5, loc. I/2W), near Haverfordwest, Dyfed. NGR SM 968 171. Sholeshook Limestone. Cautleyan (Price 1973, 1980). (c) BM(NH) E69207. Old factory wall at Ffrydan, near Bala, Gwynedd. NGR SH 922 367. Rhiwlas Lime- stone. Rawtheyan Zone 5. (d) BM(NH) E69197 to E69204. Keisley Limestone (Upper Ashgill), Keisley, near Appleby, Westmorland. E69202 came from Keisley Bank, above the west quarry, NGR NY 7140 2390 (these are the lowest exposed beds of the Keisley Limestone). E69204 is from the interbedded mudstones and nodular lime- stones in the north face of Keisley west quarry, NGR NY 7130 2385 (lowest horizons in the quarry). E69198 was in situ in the younger limestone horizons of the east face of this quarry. E69197 and E69199 to E69201 come from the east face of the west quarry, where they were washed out from a muddy horizon about 10 m from the top of the section. Finally, E69203 was part of the infill of a trench in the path to the quarry, NGR NY 7135 2380 (A. D. Wright 1982). British Ashgill Boda Limestone text-fig. 5. KD/LD and KD/KH plots for Ristnacrinus sp. from the British Ashgill {left) and the Boda Limestone (right), where KD = columnal diameter, LD = lumen diameter, and KH = columnal height. Lines of best fit determined by Bartlett’s method (Fryer 1966). 632 PALAEONTOLOGY, VOLUME 27 Description. Columnal circular to slightly elliptical in outline (cp. text-fig. 4d and 4g with 4f). Lumen circular, central. Axial canal planar-sided. Articular facet traversed by a central, synarthrial articulating ridge which sometimes has a shallow longitudinal groove. In elliptical specimens the articulating ridge lies close to, but not on, the shortest radius (text-fig. 4f). Synarthrial ridge flanked by two large, depressed, semicircular ligament areas. Circumference of the articular facet bordered by a raised rim. Columnals low or barrel-shaped with a convex latus (text-fig. 4b, f). None of the British Ashgill specimens shows an appreciable divergence of the articulating ridges. Columns homeomorphic, although some columnals are apparently wedge-shaped (text- fig. 4b) possibly due to adjacent columnals not resting parallel to each other on their fulcral ridges. Discussion. Crinoid columnals may be examined by bivariate analysis of suitable parameters (Jeffords and Miller 1968; Roux 1977, 1978; Le Menn 1981). Ristnacrinus sp., from the British Ashgill and the Swedish Boda Limestone (Ashgill), have been examined using KD/LD and KD/KH plots (text- fig. 5), where KD = columnals diameter, LD = lumen diameter, and KH = columnal height. The Swedish specimens came from a single horizon and give a very close correlation to lines of ‘best fit’. It is probable that only a single species of Ristnacrinus is present at this horizon. These columnals seem to differ from the British specimens by showing fulcral ridge divergences of up to about 90°. The British columnals, however, despite having a good KD/LD relationship, generate an apparently random distribution of points on the KD/KH graph. This may be because the columnal height of a single species is highly variable, or it is possibly due to more than one species being represented. As specimens come from four localities, which span the interval Cautleyan to Rawtheyan (Hirnantian?), the latter is most probable. However, it is unlikely that these species could be readily separated without analysing large collections from each of the four localities. These are not readily available, the entire known Ristnacrinus fauna from the British Ashgill having been used to produce text-fig. 5. Acknowledgements. I wish to thank Dr. C. R. C. Paul for critically reading an early draft and loaning specimens of Ristnacrinus columnals from the Boda Limestone, Mr. M. Birtle for help in the field at Keisley, Mr. B. Cullen for generously giving me the specimen from Corwen (now BM(NH) E69205), and Mr. C. J. Veltkamp for taking the scanning electron micrographs in text-fig. 4. This work was carried out during the tenure of Natural Environment Research Council research studentship GT4/80/GS/55. Latex casts of Entrochus primus were kindly supplied by Dr. Rudolf J. Prokop of the National Museum, Prague, Czechoslovakia. REFERENCES bates, D. E. B. 1965. A new Ordovician crinoid from Dolgellau, North Wales. Palaeontology , 8, 355-357, pi. 45. 1968. On ‘ Dendrocrinus' cambriensis Hicks, the earliest known crinoid. Ibid. 11, 406-409, pi. 76. briskeby, p. 1981. Klassifikasjon av krinoidstilker fra den over-ordoviciske Kalvsjoformasjonen pa Hadeland. Unpublished thesis, Oslo, Norway. broadhead, t. w. and strimple, h. l. 1977. Permian platycrinitid crinoids from Arctic North America. Can. J. Earth Sci. 14, 1166-1175, I pi. brower, J. c. 1974. Upper Ordovician xenocrinids (Crinoidea, Camerata) from Scotland. Paleont. Contr. Univ. Kans. 67, 1-25, 3 pis. chauvel, J. and le menn, j. 1972. Echinodermes de l’Ordovicien Superieur de Coat-Carrec, Argol (Finistere). Bull. Soc. geol. mineralog. Bretagne , ser. C, 4, 39-61, 3 pis. 1979. Sur quelques echinodermes (Cystoides et Crinoides) de f Ashgill d’Aragon (Espagne). Geobios, 12, 549-587, 3 pis. — melendez, b. and le menn, j. 1975. Les Echinodermes (Cystoides et Crinoides) de l’Ordovicien superieur de Luesma (Sud de FAragon, Espagne). Estudios geol. Inst. Invest, geol. Lucas Mallada, 31, 351-364, 3 pis. donovan, s. k. 1983. Tetrameric crinoid columnals from the Ordovician of Wales. Palaeontology, 26, 845-849. fryer, H. c. 1966. Concepts and methods of experimental statistics. Allyn and Bacon, Rockleigh, N.J. hicks, h. 1873. On the Tremadoc rocks in the neighbourhood of St. David’s, South Wales, and their fossil contents. Q. Jl geol. Soc. Lond. 29, 39-52, pis. 3-5. jeffords, R. M. and miller, T. H. 1968. Ontogenetic development in late Pennsylvanian crinoid columnals and pluricolumnals. Paleont. Contr. Univ. Kans., Echinodermata Art. 10, 1-14, 4 pis. DONOVAN: ORDOVICIAN CRINOIDS 633 jenkins, c. J. 1979. Stratigraphy and graptolite biostratigraphy of the Llanvirn Series’ type area, St. David’s, Dyfed, Wales. Unpublished Ph.D. thesis, Cambridge. lemenn, j. 1981. Les Crinoides. In morzadec, p., Paris, f. and racheboeuf, p. (eds.). Le Devonien inferieur de la tranchee de la Lezais. Mem. Soc. geol. mineralog. Bretagne , 24, 261-273, pis. 32, 33. miller, j. s. 1821. A natural history of the Crinoidea or lily-shaped animals , with observation on the genera Asteria, Euryale, Comatula , and Marsupites. Bryan and Co., Bristol. moore, r. c. 1962. Ray structures of some inadunate crinoids. Paleont. Contr. Univ. Kans., Echinodermata Art. 5, 1-47, 4 pis. — jeffords, R. M. and miller, T. H. 1968. Morphological features of crinoid columns. Paleont. Contr. Univ. Kans., Echinodermata Art. 8, 1-30, 4 pis. — lalicker, c. G. and fischer, a. G. 1952. Invertebrate fossils. McGraw-Hill, New York, Toronto and London. — lane, n. G., strimple, H. l. and sprinkle, j. 1978. Order Disparida Moore and Laudon, 1943. In moore, r. c. and teichert, c. (eds.). Treatise on invertebrate paleontology , part T. Echinodermata 2 (2), T520-T564. Geological Society of America and University of Kansas Press. — and laudon, l. r. 1943. Evolution and classification of Palaeozoic crinoids. Geol. Soc. Am. Spec. Pap. 46, 153 pp., 14 pis. OPIK, a. a. 1934. Ristnacrinus, a new Ordovician crinoid from Estonia. Tartu Ulik. Geol.-Inst. Toim. 40, 7 pp., 2 pis. PAUL, c. R. C. 1973. British Ordovician cystoids, part 1. Palaeontogr. Soc. [ Monogr .], 1-64, pis. 1-11. price, d. 1973. The age and stratigraphy of the Sholeshook Limestone of southwest Wales. Geol. J. 8, 225-246. — 1980. A revised age and correlation for the topmost Sholeshook Limestone Formation (Ashgill) of South Wales. Geol. Mag. 117, 485-489. pringle, j. 1930. The geology of Ramsay Island (Pembrokeshire). Proc. geol. Ass. 41, 1-31, pis. 1-3. ramsbottom, w. H. c. 1961. A monograph of the British Ordovician Crinoidea. Palaeontogr. Soc. [Monogr.], 1-37, pis. 1-8. rasmussen, h. w. 1978. Articulata. In moore, r. c. and teichert, c. (eds.). Treatise on invertebrate paleontology, part T. Echinodermata 2 (5), T813-T928. Geological Society of America and University of Kansas Press. roux, m. 1977. Les Bourgueticrinina du Golfe de Gascoigne. Bull. Mus. natn. Hist. nat. Paris, 3rd ser. 426 ( Zool . 296), 25-83, 10 pis. — 1978. Ontogenese, variability et evolution morphofonctionnelle du pedoncle et du calice chez les Millercrinidae (Echinodermes, Crinoides). Geobios, 11, 213-241, 2 pis. sprinkle, J. 1973. Morphology and evolution of blastozoan echinoderms. Spec. Publ. Harvard University Mus. Comp. Zool., Cambridge, Mass. — and kolata, d. r. 1982. ‘Rhomb-bearing’ camerate. In sprinkle, j. (ed.). Echinoderm faunas from the Bromide Formation (Middle Ordovician) of Oklahoma. Paleont. Contr. Univ. Kans. Monogr. 1, 206- 211, 1 pi. taylor, p. d. 1983. Ailsacrinus gen. nov., an aberrant millericrinid from the Middle Jurassic of Britain. Bud. Br. Mus. nat. Hist. Geol. 37, 37-77 . ubaghs, G. 1969. Aethocrinus moorei Ubaghs, n. gen., n. sp., le plus ancien crinoide dicyclique connu. Paleont. Contr. Univ. Kans., Pap. 38, 1-25, 3 pis. — 1978. Skeletal morphology of fossil crinoids. In moore, r. c. and teichert, c. (eds.). Treatise on invertebrate paleontology, part T. Echinodermata 2 (I), T58-T216. Geological Society of America and University of Kansas Press. ulrich, e. o. 1925 (for 1924). New classification of the ‘Heterocrinidae’. In foerste, a. f. (ed.). Upper Ordo- vician faunas of Ontario and Quebec. Mem. geol. Surv. Can. 138, 82-104, pis. 6, 7. wachsmuth, c. and springer, f. 1881. Revision of the Palaeocrinidea, part II. Family Sphaeroidocrinidae, with the subfamilies Platycrinidae, Rhodocrinidae and Actinocrinidae. Proc. Acad. nat. Sci. Pltilad. 175-411, pis. 17-19. Webster, G. d. 1974. Crinoid pluricolumnal noditaxis patterns. J. Paleont. 48, 1283 1288. WILLIAMS, A., STRACHAN, I., BASSETT, D. A., DEAN, W. T., INGHAM, J. K., WRIGHT, A. D. and WHITTINGTON, H. B. 1972. A correlation of Ordovician rocks in the British Isles. Spec. Rep. geol. Soc. Lond. 3, 74 pp. wright, a. d. 1982. The Ordovician -Silurian boundary at Keisley, Northern England. In bruton, d. l. and williams, s. H. (eds.). Abstracts for meetings 20, 21 & 23 August 1982, IV International Symposium on the Ordovician System. Paleont. Contr. Univ. Oslo, 280, 60. 634 PALAEONTOLOGY, VOLUME 27 wright, D. K. 1983. Crinoid ossicles from Upper Ordovician benthic marine assemblages from Snowdonia, North Wales. Palaeontology , 26, 585-603, pi. 65. yeltysheva, R. s. 1964. Stebli Ordovikskikh morskikh liliy Pribaltiki (Nizhniy Ordovik). Voprosy Paleont. 4, 59-82, 4 pis. [In Russian.] STEPHEN K. DONOVAN Department of Geology Trinity College Typescript received 6 September 1983 Dublin 7 Revised typescript received 14 November 1983 Ireland RECONSTRUCTION OF THE JAWS AND BRAINCASE IN THE DEVONIAN PLACODERM FISH BOTH RIO LEPIS by G. C. YOUNG Abstract. New material of the antiarch Bothriolepis , from the Gogo Formation (early Upper Devonian, Canning Basin, Western Australia), provides morphological details of the visceral jaw elements, which were not previously known in antiarchs. The palatoquadrate lacks a high orbital process, and was attached to the ventral part only of the suborbital (mental) plate. This shows that the ethmoidal region of the braincase must have been considerably deeper than previously thought. Detailed descriptions are given of the dermal elements in the jaws (suborbital and infragnathal) and cheek (submarginal, prelateral, mfraprelateral). On the evidence of the palatoquadrate the mental plate of antiarchs is homologized with the suborbital in other placoderms. The absence of supragnathals may be secondary, and the differentiation of the infragnathal into biting and non-biting divisions probably evolved independently in antiarchs and euarthrodires. Reassembly shows that the submarginal and infraprelateral plates in Bothriolepis fitted closely against the anterior ventrolateral to close the operculum. A new restoration of the endocranium is presented, based on the identification of a posterior postorbital process and cucullaris fossa in Asterolepis. It is suggested that in Bothriolepis the palatoquadrate had an amphistylic connection to subocular and subnasal shelves, that the lateral pit was bounded posteriorly by the anterior postorbital process to form a mandibular muscle fossa, and that the preorbital recess housed the rhinocapsular bone. Comparison with Yunnanolepis indicates that the preorbital depression in this form contained a discrete rostral capsule with lateral nasal openings, and that the ‘orbital fenestra’ in those antiarchs with a preorbital depression is equivalent to the suborbital fenestra of Bothriolepis. Certain characters defining the relationships of antiarchs to other placoderm groups are summarized in cladogram form; prelaterals, infraprelaterals, a long obstantic margin, and prominent posterolateral corners on the skull are proposed as synapomorphies of antiarchs and euarthrodires. Fusion of the quadrate to the postsuborbital is a possible additional synapomorphy of actinolepids and phlyctaenioids. The remarkable preservation of fishes from the Frasnian (early Upper Devonian) Gogo Formation in the Canning Basin of north-western Australia is now well known through the publications of Miles (1971, 1977), Gardiner and Bartram (1977), Miles and Young (1977), Miles and Dennis (1979), and Dennis and Miles (1979a, b\ 1980-1983). A preliminary faunal list and brief comments on the occurrence of this diverse fish fauna was presented by Gardiner and Miles (1975). The new information on the structure of the placoderm Bothriolepis presented here is based on material from a large collection of Gogo fish specimens made in 1972 by the Bureau of Mineral Resources and the Geological Survey of Western Australia. Bothriolepis is a well-known member of the highly specialized placoderm order Antiarcha, which had a cosmopolitan distribution during the Late Devonian, typically in non-marine sediments. Over fifty named species have been referred to the genus (e.g. Denison 1978). However, the occurrence of Bothriolepis in strictly marine calcareous sediments, as is the case with the Gogo fish fauna, is very uncommon. The mode of preservation in the Gogo material (whole fish in calcareous nodules) has permitted preparation using acetic acid techniques as developed by Toombs (1948) and Toombs and Rixon (1959), to reveal in intricate detail the skeletal morphology of this form. This new information largely confirms, but also supplements and enlarges upon, the comprehensive previous accounts of the morphology of Bothriolepis (see Stensio 1931, 1948). Many prepared specimens of Bothriolepis from the Gogo Formation are held in the British Museum (Natural History) and will be described in a forthcoming account as a new species by Dr. (Palaeontology, Vol. 27, Part 3, 1984, pp. 635-661, pis. 57-59-1 636 PALAEONTOLOGY, VOLUME 27 R. S. Miles. The new information on jaw and braincase structure described here is based on a single specimen in which delicate and extremely fragile surface (perichondral) ossifications of various cartilaginous elements were recognized during preparation. These structures have not previously been identified in Bothriolepis , in which the braincase and gill arches were predominantly cartilaginous. The material described below is housed in the Commonwealth Palaeontological Collection (prefix CPC) in the Bureau of Mineral Resources, Geology and Geophysics, Canberra. The terms ‘interperichondral space’, ‘closed margin’, and ‘open margin’ are used in the description of perichondrally ossified elements as defined by Miles and Young (1977, p. 145). The following terms for dermal bones of the cheek region are substituted for those previously applied to antiarchs: submarginal for extralateral, and suborbital for mental plate. Evidence for establishing these homologies is presented below. MATERIAL The specimen (CPC 25205) on which this study is based includes the three major cheek bones (submarginals, infraprelaterals, and the left prelateral), upper and lower dermal jaw elements from both sides, a possible extramandibular plate, both palatoquadrates, and the left Meckel’s cartilage, the latter elements attached to their respective dermal bones. The nuchal plate is missing from the skull-roof, but the complete anterior region of the skull together with the well-preserved and articulated ventral trunk armour has enabled the space containing the orobranchial cavity to be delineated by direct reassembly of the specimen. DESCRIPTION 1 . Dermal elements of the jaws and cheek. Because of their excellent preservation these bones provide several new morphological details not previously covered in Stensio’s comprehensive descriptions (1931, 1948). The suborbital (mental) plate, which is the upper dermal jaw element, is shown in external view in PI. 58, fig. 1 . In shape it corresponds well with this bone in B. canadensis , but the lateral notch is more pronounced, with a longer posterior process (p.pr, text-fig. 2), and the posterodorsal corner is less marked, as in B. cellulosa or Grossilepis (Stensio 1948, fig. 35). The external surface is ornamented in its dorsal part, with a broad strip of smooth bone along the denticulated ventral margin. Stensio regarded these as distinct ‘tooth’ and ‘sensory canal’ components of a compound bone, but this seems unlikely. The shallow groove (gr.ul, text-figs. 2, 3) separating these two regions probably housed soft tissue forming an upper lip. The infraorbital sensory groove (ifc) crosses the bone in a similar fashion to that of other antiarchs. On the inner surface of the plate is a distinct ridge (r.pq), as previously noted in B. canadensis (Stensio 1931, 1948). Posteriorly it forms an inconspicuous thickening, but anteriorly it is well developed to support the palatoquadrate (text-fig. 3d, e), with its concave upper surface forming a shallow groove pierced by several foramina (f.vasc, text-fig. 2b). In both specimens the anterior end of this ridge is clearly seen in front of the broken perichondral lining of the palatoquadrate (PI. 59). The groove-like upper surface of the ridge may correspond to the ‘neurovascular groove’ in Holonema (Miles 1971, p. 135), although at least some of its contained foramina appear to open externally into the groove (gr.ul) delineating the biting portion of the bone on the outer surface. In its relationship to the palatoquadrate this ridge might be thought to correspond to the suborbital crista on the suborbital plate in brachythoracid euarthrodires (e.g. Miles and Westoll 1968, fig. 13), but the possibility that the adductor mandibulae musculature may have passed above it (see below) argues against this comparison. A final structure of note is the rounded inner projection just beneath the anterodorsal corner of the plate (pr.dm, text-figs. 2, 3, 5). This forms a narrow flattened area on the anterior (mesial) margin, facing towards the symphysial plane with the suborbital in position against the skull. It may have been an attachment site for an anterior ligamentous connection between the suborbital plates of each side. YOUNG: BRAINCASE OF BOTH RIOLEPIS 637 The infragnathal (PI. 58, figs. 2, 3) is, as previously noted (Stensio 1931, 1948), made up of a posterior non-biting and an anterior biting division, as in some euarthrodires. Some new details can be added to previous descriptions of this bone. The presence of a denticulate biting margin on the anterior division (Stensio 1931, p. 64) is confirmed. The posterior division is broader than the anterior, with fairly straight lateral and posteromesial margins, and a convex anteromesial margin. text-fig. 1. (See list of Abbreviations used in Text-figures, p. 661.) Bothriolepis sp. CPC 25205. Gogo Formation, Canning Basin, Western Australia, a, left prelateral and infraprelateral plates reassembled against the anterior end of the left submarginal plate, visceral view, b, the same, in lateral view, c, d, right infraprelateral in lateral and anterior views respectively, e, possible right extramandibular plate, external view. F, reassembly of the left submarginal, prelateral, and infraprelateral plates against the anterior ventrolateral plate of the trunk-shield, ventral view. 638 PALAEONTOLOGY, VOLUME 27 text-fig. 2. Bothriolepis sp. CPC 25205. Gogo Formation, Canning Basin, Western Australia. Right sub- orbital plate with palatoquadrate attached, in ventral (a) and dorsal views (b). There is a prominent posterolateral corner (pic, text-fig. 5d). The dorsal surface is partly divided into mesial and lateral parts (mf.df) by a posterior continuation of the biting margin of the anterior division (dent). In these respects the bone closely resembles the infragnathal of Grossilepis (Stensio 1948, fig. 36c-e). The ventral surface of the bone (PI. 58, fig. 3) is concave over its whole length, where it was affixed to Meckel’s cartilage (see below). Anteriorly this concavity forms a deep groove, giving way on the posterior non-biting division to a shallow depression flanked by low crests along the lateral and anteromesial margins. The posteromesial margin (pmm, text-fig. 5d) has a bevelled edge with a somewhat cancellous texture, and lacks a crest. Just inside this margin in both examples is a conspicuous groove (PI. 58, fig. 3) for a nerve or vessel positioned between the cartilage and the dermal bone. This may correspond to the groove on the posterior division of the infragnathal in coccosteids (e.g. Dennis and Miles 1979a, fig. 14), where, however, it runs on the mesial surface of the bone. EXPLANATION OF PLATE 57 Bothriolepis sp. CPC 25205. Gogo Formation, early Late Devonian, Canning Basin, Western Australia. Fig. 1. Incomplete skull-roof in visceral view, with right submarginal plate attached, x2. Figs. 2, 3. Left submarginal and prelateral plates in lateral and mesial views respectively, x4. PLATE 57 YOUNG, Bothriolepis from Gogo 640 PALAEONTOLOGY, VOLUME 27 Stensio (1948, p. 96) suggested that the anterior and posterior divisions of the infragnathal in Bothriolepis lay respectively lateral and mesial to Meckel’s cartilage, this being the same arrangement as in brachythoracid euarthrodires. It is clear from this new material that the infragnathal was positioned mainly dorsal to the cartilage element, but Stensio’s interpretation would still seem to be correct. On the anterior biting division the outer ventral crest delimiting the groove for the cartilage is much more pronounced (PI. 58, fig. 3), and the reverse applies on the posterior division. In addition, the anterior and posterior divisions are twisted with respect to each other about the long axis of the bone, to accentuate this special morphological relation between dermal and cartilage elements. Of the dermal cheek plates the prelateral is known only from the left side. In life it was closely attached to the submarginal (extralateral) plate (PI. 57, figs. 2, 3). It is essentially a triangular bone, as in B. cellulosa (e.g. Stensio 1948, fig. 31), and lacks the short dorsal margin seen in B. canadensis. However, it is proportionately higher and shorter in CPC 25205 than in the prelateral of B. cellulosa, although plate proportions are known to be somewhat variable in B. canadensis. Externally the bone is crossed by a sensory groove between its dorsal and ventral corners (psoc, text-fig. 1), presumably equivalent to the postsuborbital sensory canal of euarthrodires. The crest on the visceral surface beneath the sensory groove, previously known in B. cellulosa, is shown in this specimen to represent the anterior margin of the area in contact with the submarginal plate. The submarginal plate of the right side is shown in position against the skull-roof in PI. 57, fig. 1. It makes two moveable attachments with the skull, the anterior connection in the form of a distinct anterodorsal process (ad , , text-fig. 1), as in other species (e.g. Stensio 1948, fig. 105). On both sides the ventral margin of the submarginal is thickened about half-way along its length and displays a flattened articular surface of spongy bone (av, text-figs. 1a, 9b) in a corresponding position to the ventral longitudinal crista in B. canadensis (Stensio 1948, fig. 105c). Reassembly of the specimen shows that this surface abutted against a dorsally facing bevelled edge along the lateral margin of the subcephalic division of the anterior ventrolateral plate, immediately lateral to its anterolateral corner (text-fig. If). This would have effected a seal to the branchial chamber when the operculum was closed. The infraprelateral plates, which are poorly known in other species, are excellently preserved in CPC 25205. As figured by Stensio in B. canadensis (e.g. 1 948, fig. 34), the ornamented part comprising the ventral division of this bone lies against the ventral border of both the submarginal and prelateral plates, and carries a continuation of the sensory groove from the latter plate. The unornamented lateral part of the plate (apl, text-fig. lc, d) fitted inside the prelateral and submarginal plates in a loose attachment. The three plates from the left side are shown reassembled in text-fig. 1a, b. Also of interest is the way the infraprelateral fitted along the anterior margin of the anterior ventrolateral plate, immediately inside its anterolateral corner. Here a slight notch is developed, and a narrow overlap area on the posteromesial edge of the infraprelateral fitted loosely inside the anterior ventrolateral. Direct reassembly of the specimen suggests a small gap around the anterolateral corner of the anterior ventrolateral (text-fig. If), and it seems probable that the infraprelateral was carried in a flexible region of skin forming the floor of the branchial chamber. An associated thin dermal plate (text-fig. 1e) may be a right extramandibular, since it resembles this element as figured in B. canadensis by Stensio (1948, fig. 1 10), and carries a groove along its presumed anterior margin which may have been a sensory canal or pit-line. EXPLANATION OF PLATE 58 Bothriolepis sp. CPC 25205. Gogo Formation, early Late Devonian, Canning Basin, Western Australia. Fig. 1. External view of right suborbital plate with palatoquadrate attached, x 6. Fig. 2. Right infragnathal in dorsal view, x 6. Fig. 3. Left infragnathal in ventral view, x 6. PLATE 58 YOUNG, Bothriolepis from Gogo 642 PALAEONTOLOGY, VOLUME 27 text-fig. 3. Bothriolepis sp. CPC 25205. Gogo Formation, Canning Basin, Western Australia. Restoration of the left suborbital plate and palatoquadrate. a, internal view (cf. PI. 59, figs. 1, 2). b-e, sections at the levels indicated (dermal bone cross-hatched, cartilage stippled, articular surfaces with regular stipple). 2. Palatoquadrate. This element is similarly preserved on the inner side of each suborbital (mental) plate (PI. 59, figs. 1, 2). Contrary to the reconstruction of Stensio (1969, fig. 135), the palatoquadrate is long and low, and attaches mainly to the more ventral parts of the plate. Its closed dorsal margin lies just above the inner ridge (r.pq, text-fig. 3) in the central part of the plate. Posteriorly it curves slightly upward and over the lateral notch in the plate (PI. 58, fig. 1). The autopalatine part of the palatoquadrate was relatively deep, its lateral (labial ) perichondral surface extending and attached to the biting margin of the plate (text-fig. 3d). Exactly the same situation is seen in both plates, so any displacements between visceral and dermal elements can be discounted. The mesial (lingual) perichondral lining of this region is incomplete, but shows the autopalatine to have decreased in thickness ventrally. It is unlikely that the palatoquadrate would have projected past the biting margin of the suborbital, and it is assumed that it terminated here as a thin ventral border (text-figs. 3a, 5b). A similar situation is met with in Holonema and Romundina (Miles 1971, fig. 34; 0rvig 1975, pi. 2, fig. 5). The anteromesial extent of the palatoquadrate is not shown, but it is assumed that it formed an anterolateral connection with the braincase, as in other forms. The broken anterior edge of the outer perichondral lining curves away from the visceral surface of the suborbital at this level, and suggests an inward flexure of the palatoquadrate to form the orbital connection with the braincase. On both YOUNG: BRAINCASE OF BOTH RIOLEPIS 643 sides the anterodorsal edge of the perichondral lining is similarly preserved as a distinct notch (PI. 59, figs. 1, 2; n, text-fig. 2b), which may have been the posterior border of an articular area (art, text- fig. 3a). However, this needs confirmation on better material, as it is not certain that this perichondral edge is a natural margin. Just behind the notch the dorsal margin of the palatoquadrate forms a laterally directed shelf (r.epq, text-fig. 3d), which extends posteriorly until it touches the suborbital plate at its dorsal margin (text-fig. 3c). At least on the right palatoquadrate this shelf has an open lateral margin (text-fig. 2b). This shelf delimits dorsally the space described below as an extension of the adductor fossa. As such it corresponds in position to the extra-palatoquadrate ridge in acanthodians or elasmobranchs (e.g. Miles 1973; Young 1982). The metapterygoid region of the palatoquadrate is less deep but wider than the autopalatine, with a conspicuous dorsal flexure forming a ventral embayment, clearly the fossa for the adductor mandibulae musculature (f.am, text-fig. 3). Just in front of this fossa the palatoquadrate is separated from the suborbital ventrally, but is in contact dorsally. The inner ridge of the suborbital plate is reduced here to a low thickening, above which the concave inner surface of the suborbital and the concave outer (labial) surface of the palatoquadrate enclose a space in open communication ventrally with, and probably forming a dorsal extension of, the adductor fossa (f.am, text-fig. 3c). The dorsal-most opening through the perichondral lining, high on the lingual face above the adductor fossa, has slightly thickened rims showing it to be a natural opening representing another articular surface in contact with the endocranium (art.pb, text-fig. 3). The homologization of these various connections with the braincase amongst different placoderms is still uncertain. In Romundina , Ctenurella, and Buchanosteus there are three or more connections between the anterior portion of the palatoquadrate and the subocular shelf. The posterior two in Buchanosteus , by their close association with the groove for the efferent pseudobranchial artery, may be assumed to correspond, topographically at least, to the palatobasal connection in other fishes (Young 1979, p. 336). It is not clear how these correspond to the articular areas in Romundina , but in relation to the groove for the efferent pseudobranchial artery on the endocranium the posterior area labelled by 0rvig (1975, pi. 2, fig. 5) may correspond to the anterior part of the palatobasal connection in Buchanosteus , and another articular area may have been developed in the non-preserved part of the palatoquadrate immediately behind. This latter connection would have had an anterodorsal position relative to the adductor fossa, as with this posterior articular area in Bothriolepis (art.pb). However, in Holonema an articular area in much the same position was interpreted by Miles ( 1971 ) as for the orbital connection. In Bothriolepis a more anterior double orbital articulation has been restored after Romundina (art, text-fig. 3a), but it should be noted that in at least one euarthrodire ( Dicksonosteus ; see Goujet 1975) this was reduced to a single connection. In Holonema there were apparently no articular areas on the autopalatine corresponding to those of Romundina (Miles 1971, fig. 57). At its lateral end the palatoquadrate of Bothriolepis is constricted, broader than high, and protrudes through the lateral notch in the suborbital plate (text-fig. 2), so that it lies above the posterior process (text-fig. 3b). The most posterior part of the ventral perichondral surface in both examples is inflected downwards, and this is interpreted as the edge of the condyle for the mandibular joint (cd.art). The more completely preserved palatoquadrate from the right side indicates that the condyle partly straddled and was in contact with the end of the posterior process of the sub- orbital plate. In previous reconstructions of the palatoquadrate in Bothriolepis (Stensio 1948, fig. 7; 1969, figs. 41, 42, 135; Miles 1971, fig. 1 12; Denison 1978, fig. 3) it has been depicted as a high short element somewhat similar to that of ptyctodontids, with a prominent orbital process extending dorsally to articulate with the endocranium beneath the orbital cavity. The closed dorsal margin of the perichondral lining in this new material shows this reconstruction to be incorrect. The palato- quadrate was low and long, without an orbital process, and occupied a position inside the ventral part only of the suborbital plate. 3. Meckel's cartilage. The posterior portion of this element was identified attached to the non-biting posterior division of the left infragnathal, from which it was subsequently detached (PI. 59, figs. 3, 4). It is a fairly flat broad element which was fixed to the ventral surface of the posterior 644 PALAEONTOLOGY, VOLUME 27 text-fig. 4. Bolhriolepis sp. CPC 25205. Gogo Formation, Canning Basin, Western Australia. Posterior part of left Meckel’s cartilage (cf. PI. 59, figs. 3, 4). a, dorsal view, showing flat surface affixed to posterior non-biting division of the infragnathal. b, ventral view, c, section through level indicated by arrows. division of the infragnathal as oriented in text-fig. 4a. A dorsal view of the detached element (PI. 59, fig. 3) shows clearly the area in contact with the dermal bone as a largely unossified region bounded anteriorly by a thickened perichondral margin (a.m, text-fig. 4a). As preserved, the area of contact corresponds closely in shape to the flat central region bounded by ridges on the ventral surface of the posterior non-biting division of the infragnathal (PI. 58, fig. 3). There can be little doubt that a long narrow process of the cartilage (lost during preparation) extended to the extremity of the biting division of the infragnathal, in its ventral groove. The expanded posterior part of the cartilage projected anteriorly from beneath the infragnathal (a.sh, text-figs. 4a, 5d). This, together with the dorsolateral face of the non-biting division, probably formed the ventral area of attachment for the adductor mandibulae musculature. The ventral surface of the preserved portion of the cartilage is completely ossified (text-fig. 4b, c), except for a deep embayment in its posterior margin (art.v), which has a slightly everted rim and was probably a cartilaginous articular surface. The function of such an articulation is uncertain, however, as it would have faced ventrally beneath the mandibular joint. It might correspond to the facet on the articular bone of ptyctodontids which may have received an element of the hyoid arch (Stensio 1969, fig. 152; Miles and Young 1977, fig. 24). A similar facet is present also in brachythoracids, and was referred to by Stensio (1963) as the ‘supraglenoid area’ (see Miles and Dennis 1979, p. 52). The mandibular articulation is not preserved in CPC 25205, but its likely position can be inferred from a consideration of the relationship between the upper and lower jaw elements (see below). Again this element is much shorter and broader than restored by Stensio (1948, 1969), and there is evidence that the mandibular joint was located adjacent to the posterior end of the infragnathal, and not some distance from it. EXPLANATION OF PLATE 59 Bothriolepis sp. CPC 25205. Gogo Formation, early Late Devonian, Canning Basin, Western Australia. Figs. I, 2. Internal views of right and left suborbital plates with palatoquadrates attached, x 6. Figs. 3, 4. Posterior part of left Meckel’s cartilage in dorsal and ventral views respectively, x 6. PLATE 59 YOUNG, Bothriolepis from Gogo 646 PALAEONTOLOGY, VOLUME 27 RESTORATION 1. Jaws. A restoration of upper and lower dermal and visceral jaw elements in biting position is presented in text-fig. 5. The following constraints were applied in this restoration. ( a ) Reassembly shows that the flat mesial edges of the suborbital plates do not fit against each other, indicating an intervening ligamentous or cartilaginous connection, as previously suggested for Bothriolepis (e.g. Stensio 1948, 1969). As preserved (text-fig. 6) these plates lay symmetrically about the midline, but by fitting together the skull-roof and trunk armour and laying in the upper elements in the available space (see below) it became evident that they had been subject to post-mortem rotation. The configuration of the dorsal (anterior) margins of the suborbital plates in relation to the rostral margin of the skull-roof indicates a slight separation between left and right elements, as Stensio concluded from a study of B. canadensis (1948, fig. 34). ( b ) By direct experimentation with upper and lower dermal elements their biting margins were oriented in presumed biting position, with the denticle rows subparallel. The biting margin of the infragnathal has much stronger curvature, and it is clear that it could not have closed outside the upper biting margin. It also seems unlikely that it bit inside this margin, because of obstruction by the palatoquadrate. It is possible, however, to place the jaw elements with good alignment between upper and lower denticle rows such that they occluded against each other. There is little freedom to adjust the two elements with respect to each other without the biting edges losing their alignment, so this is assumed to be a reasonable approximation of the correct biting position. Camera lucida drawings were prepared of each element in biting view. text-fig. 5. Restoration of jaw cartilages in Bothriolepis , after CPC 25205. a, dorsal view of a left suborbital and palatoquadrate, oriented as in b. b, ventral (occlusal) view of left suborbital and palatoquadrate. c, ventral view of right infragnathal and Meckel’s cartilage, oriented as in d. d, dorsal (occlusal) view of right infragnathal and Meckel’s cartilage. YOUNG: BRAINCASE OF BOTH RIOLEP/S 647 (c) The likely position of the mandibular joint provides a third constraint on jaw position. As noted above, the posterior end of the palatoquadrate protrudes laterally to the posterior process of the suborbital plate. From the shape of the broken perichondral margins it is probable that the articular facet on the quadrate was immediately adjacent to the end of this dermal process, and the corresponding surface on the articular was immediately adjacent to the posteromesial margin of the posterior division of the infragnathal. To bring these surfaces together in the reconstruction the posterior division of the infragnathal needed a position as far lateral as possible, and it was necessary to assume some mesial flexure of the quadrate inside the posterior process of the suborbital plate. The restored position of the mandibular joint was determined both by superimposing outline drawings of the two elements, and by direct reassembly of the specimens. It should be noted that in this new restoration (text-fig. 5) the palatoquadrates are separated anteriorly, and do not form a symphysis (cf. Stensio 1948, 1969). This is a primitive condition for placoderms at least (e.g. euarthrodires, palaeacanthaspids, gemuendinids, petalichthyids), and probably also for gnathostomes (acantho- dians, osteichthyans). The lower jaws of each side are also widely separated, but were possibly connected anteriorly by a median basimandibular, as restored by Stensio (1969, fig. 135a). However, there is no preserved evidence in this material for the existence of this element. The above procedure permits the reconstruction of upper and lower jaw elements with respect to each other, but gives no indication of their relationship to the skull-roof and endocranium. However, by reassembling the skull-roof against the trunk-shield the space available for the orobranchial cavity could be delimited. The submarginal and prelateral plates close in the gill chamber laterally, the former plate fitting closely against the anterior ventrolateral to effect an adequate seal with the operculum closed, as in other placoderms (e.g. Young 1980, fig. 18). As previously determined (Stensio 1931, 1948) the mouth must have opened through that semicircular space delimited by the fairly straight leading ventral edge of the trunk-shield and the strongly curved rostral margin of the text-fig. 6. Botliriolepis sp. CPC 25205. Gogo Formation, Canning Basin, Western Australia. Specimen in ventral view, partly prepared, to show preserved position of jaw elements, x 3. 648 PALAEONTOLOGY, VOLUME 27 text-fig. 7. Botlniolepis sp. CPC 25205. Gogo Formation, Canning Basin, Western Australia. Right lateral view showing skull-roof reassembled against the trunk-shield. The dashed line from the leading ventral edge of the trunk-shield to the rostral margin of the skull-roof includes the assumed mouth position. skull-roof. A lateral view of the reassembled Gogo specimen (text-fig. 7) confirms that this rostral margin and its lateral extremities lie in a slightly curved plane forming the anterior continuation of the flat ventral surface of the trunk-shield. The lack of angularities or projections in the overall conformation of the skull and trunk armours suggests that protruding jaws would have been most unlikely. As in previous restorations therefore (e.g. Stensio 1948, 1969), it is assumed that the mouth was a transverse opening situated behind and slightly below the rostral margin of the skull-roof. By suitably elevating the posterior part of the trunk-shield in CPC 25205 with the skull attached, it was possible to lay in the upper and lower jaw elements in their approximate positions on the flat surface supporting the specimen. At this angle the curved dorsal margins of the suborbital plates conform fairly well with the concave rostral margin of the skull-roof. The limited space available anterior to the anterior ventrolaterals shows that the upper jaws must have been carried adjacent to the rostral margin in a flap of skin. In this position it is clear that the mandibular joint must have been located approximately inside the prelateral plate, with the preserved dorsal articular surface (art.pb, text-fig. 3a) facing toward the lateral pit, but some distance below it. The palatoquadrate lies anteromesial to the articular process of the submarginal plate, the slope of its upper surface in the quadrate region conforming fairly well to the slope of the articular process. It is reasonable to suggest therefore that the epihyal fitted between these two structures. If the epihyal articulated against the anterior postorbital process of the endocranium, as it does in some other placoderms, then this process must have terminated approximately adjacent to the prelateral crista, and slightly above the anterior articular area for the submarginal. This assumes, of course, that the epihyal was neither attached to nor incorporated in the submarginal plate and its articular process. YOUNG: BRAINCASE OF BOTH RIOLEPIS 649 2. Endocranium. The form of the palatoquadrate, as described above, demonstrates that the endocranium of Bothriolepis must have been much deeper than previously supposed, at least in its anterior parts, in order to effect an articulation with the palatoquadrate. The previous restoration of the endocranium by Stensio (1948, figs. 6, 7; 1969, figs. 41, 42), largely followed by other workers (Miles 1971, fig. 1 12f; Denison 1978, fig. 3a), was based on the well-developed impressions for the dorsal surface of the endocranium on the visceral skull-roof surface in the region behind and lateral to the orbital fenestra (the ‘otico-occipital depression’ of Stensio). Anteriorly little information is provided by the skull-roof, and here Stensio relied on the assumption that the palatoquadrate was a short, deep element, with a prominent dorsal process which formed an orbital connection with the endocranium in a similar position, relative to the orbits, to this connection in other forms. To develop a new restoration of the braincase, based on the morphology of the palatoquadrate as described above, it is first necessary to consider the various endocranial processes in placoderms and their relations to visceral arch elements and associated muscles. I have previously suggested (Young 1979, 1980) that there are grounds for homologizing most of the endocranial processes and associated fossae amongst the various major groups of placoderms, on the assumption that these structures were developed in different ways according to differences in the arrangement of the main muscles controlling movement of the cheek and jaws against the skull-roof and endocranium. By using these previously proposed homologies (text-fig. 8), and assuming constancy of morphological relations and function of corresponding endocranial structures in Bothriolepis , a new restoration of the endocranium has been prepared (text-fig. 9). The major changes in proportion resulting from dorsal migration of the orbits and nares, and the development of a prominent rostrum, appear to have had less effect on the posterior parts of the endocranium, where homologies to corresponding structures in other forms are fairly clear. These may be considered first, to establish a framework for interpreting the ethmoid and orbital regions. The craniospinal process (pr.csp, text-fig. 8) in Bothriolepis supports the dermal neck-joint, and can be assumed homologous to this process in other placoderms as identified by Young (1980, fig. 24). The same process in Stensio’s restoration (1969, figs. 41, 42) was termed the ‘supravagal process’. The supravagal process as redefined by Young (1980, p. 56) was apparently either extensively reduced or absent in Bothriolepis , as was the posterior postorbital process. However, there is evidence that in text-fig. 8. Endocranial processes in placoderms. a, ventral view of skull-roof and endocranium in Buchanosteus (after Young 1979, fig. 2). b, Asterolepis , skull-roof in ventral view (after Stensio 1969, fig. 138c). c, Bothriolepis , skull-roof in ventral view (after CPC 25205). Not to scale. 650 PALAEONTOLOGY, VOLUME 27 some antiarchs the latter process was retained. A deep pit lying lateral to the paramarginal crista and opening posteriorly is known in Asterolepis (e.g. Stensio 1 948, fig. 1 7). This pit is bounded laterally by a crest of bone passing on to the postmarginal plate, which can be readily homologized with the inframarginal crista in euarthrodires such as Coccosteus (e.g. Miles and Westoll 1968). This homology is supported by the presence of a postmarginal sensory groove on the postmarginal plate in some primitive antiarchs (e.g. Zhang Miman 1980, pi. 1). In Buchanosteus the inframarginal crista (cr.im, text-fig. 8a) forms an extension to the posterior postorbital process, and marks the anterior boundary of the cucullaris fossa. Comparison with the visceral skull-roof surface in Asterolepis (e.g. Stensio 1969, fig. 138c) indicates that this lateral pit, which faces posterolaterally toward the obstantic margin, also represents the cucullaris fossa (fo.cu, text-fig. 8b). Although there is no evidence that the posterior postorbital process was retained in Bothriolepis, a small process corresponding in position to that of Asterolepis is shown in the restoration (pr.ppo, text-fig. 9), to facilitate homologization of surrounding structures. However, there is good evidence of the shape of the dorsal aspect of the anterior postorbital process, which in Bothriolepis and other antiarchs must have occupied an anterolateral extension of the otico-occipital depression (pr.apo, text-fig. 8b, c). In most other placoderms this process carried one or two subterminal articular surfaces for visceral arch elements (palatoquadrate, epihyal, opercular cartilage). In Asterolepis the depression for this process lies adjacent to a dermal thickening which supported the connection and articulation of the submarginal plate against the skull-roof ( a1SM , a2SM , text-fig. 8). In Bothriolepis there is a similar arrangement, although a more complex (apparently dermal) moveable articulation is developed between the articular process of the submarginal plate and the skull-roof. Whether the thickening supporting these articulations is entirely dermal in origin, or in fact represents the ossified terminal part of the anterior postorbital process, depends on the likely position of the epihyal element and its relationship to the submarginal plate. This is further considered below. In the restorations the anterior postorbital process is shown extending as far forward as the transverse lateral groove (tig, text-fig. 9). In Buchanosteus the subocular shelf formed a floor to the orbital cavity, and was continuous anteriorly with the subnasal shelf (sns, text-fig. 8a) which extended beneath the separately ossified rostral capsule (Young 1979). A similar arrangement is seen in Romundina , where the rostral capsule and nasal openings are dorsally placed (0rvig 1975), and also in Brindabellaspis , where the two elements of the braincase show incipient fusion (Young 1980). The distribution of this character indicates that the separation of the endocranium into rhinocapsular and postethmo-occipital components was probably a primitive placoderm feature. Again, in both Buchanosteus and Romundina the palatoquadrate articulated against the lateral edge of the subocular shelf, or was closely held to it by ligaments, and this is also likely to be a primitive placoderm feature. In Bothriolepis , Stensio (1969, figs. 41, 42) restored a high orbital process on the palatoquadrate, which articulated against the braincase in a dorsal position beneath the orbital fenestra, with the autopalatines of each side forming an anterior symphysis. However, as shown above, the palato- quadrate was low and broad, and it is clear that there must have been considerable anteroventral extension of the braincase to receive the palatoquadrate articulation. By comparison with Buchanosteus , and especially Romundina (in which the nares are dorsally placed), it can reasonably be assumed that the palatoquadrate retained its normal connection with subocular and subnasal shelves, even though the orbits and nasal openings had migrated dorsally to a mid-line position. In the restorations, therefore, three articulations between the palatoquadrate and subocular and subnasal shelves are shown: an anterior double articulation with an ectethmoid process, corresponding to the orbital connection known in Kujdcmowiaspis , Dicksonosteus , Ctenurella , Buchanosteus, and probably Romundina (Stensio 1963; Goujet 1975; 0rvig 1975; Miles and Young 1977; Young 1979); and a posterior single articulation, corresponding topographically to the palatobasal articulation, which is also known in Buchanosteus and probably Romundina (Young 1979, fig. 12; cf 0rvig 1975, pi. I ). To what extent the subocular and subnasal shelves of Bothriolepis were in contact with the visceral surface of the skull-roof is uncertain, but the poriferous area and associated ridges beneath the premedian plate (e.g. Stensio 1 948, fig. 1 5) suggest attachment in this region (the endocranial ‘rostral YOUNG: BRAINCASE OF BOTH RIOLEPIS 651 process’ of Stensio’s restoration). The anterolateral corners of this rostral process in B. canadensis approximate in position to the ectethmoid processes restored here (pr.ect, text-fig. 9a). As such they are appropriately placed to have received the anterolateral ends of each palatoquadrate, with the jaws in their correct position. apo a2 text-fig. 9. Restoration of endocranium, mandibular and hyoid arch elements in Bothrio- lepis, Dermal skull-roof, and cheek plates of one side, shown in outline, a, dorsal view, with right palatoquadrate in approximate position (modified in part after Stensio 1969, fig. 41). B, ventral view, with epihyal (unknown) and palatoquadrate in approximate positions. Position of right ceratohyal (unknown) and right Meckel’s cartilage based on assumed position of palatoquadrate and epihyal. 652 PALAEONTOLOGY, VOLUME 27 Clearly there must have been muscles of mandibular derivation controlling movement of the palatoquadrate against the braincase, and in other forms (Young 1979, 1980) an embayment in the lateral endocranial wall, bounded posteriorly by the anterior postorbital process, has been proposed as the likely site of insertion of the m. levator palatoquadrati (fo.md, text-fig. 8a). In Bothriolepis the thickening on the lateral plate determined above as adjacent to or part of the anterior postorbital process, forms the posterior boundary of the lateral pit, which in this Gogo specimen (PI. 57, fig. 1 ) is a distinct irregular depression readily interpreted as a site for muscle insertion. An embayment in the endocranial wall in this position, corresponding to the mandibular muscle fossa identified in Buchanosteus (Young 1979, fig. 2), is shown on the restorations (fo.md, text-fig. 9). It is possible, however, that this fossa was partly closed in ventrally by the subocular shelf. The presence of extensive subocular and subnasal shelves in Bothriolepis, as is required to effect a connection with the palatoquadrate, leads to a new interpretation of the preorbital recess in this form. This structure, a backwardly opening cavity beneath the posterior part of the premedian plate, has been interpreted by Stensio (1948, 1969) as a cartilage-lined space completely occupied by the nasal sacs (text-fig. 10a). The recess is floored by a lamina of bone projecting posteroventrally from the visceral surface of the premedian plate, and enclosed in cartilage under Stensio’s restoration. The shape of the recess amongst different species of Bothriolepis is well known (e.g. Stensio 1948, figs. 13-15; Miles 1968, figs. 7, 43, 58), and there can be no doubt that the recess was either filled or text-fig. 10. Paramedian sections through the preorbital region of the head in Bothriolepis. a. after Stensio (1969, fig. 135b). b, new interpretation showing suggested position of rhinocapsular bone. YOUNG: BRAINCASE OF BOTH RIOLEP1S 653 lined with cartilage. In view of the position of the preorbital recess in relation to the remainder of the braincase as restored here, I suggest that this structure was in fact the rhinocapsular bone of the endocranium (rh.cap, text-figs. 9a, 10b). Disregarding changes in proportion, it can be noted that this structure occupies the same position in relation to the subocular and subnasal shelves of Kujdanowiaspis , Buchanosteus , or Romundina , as does the rhinocapsular bone in these forms (Stensio 1963; 0rvig 1975; Young 1979). The bony lamina forming the floor of the recess, and enclosed in cartilage under Stensio’s restoration (an anomalous relation if this lamina is of dermal origin), may be reinterpreted by comparison with Brindabellaspis (Young 1980, fig. 4). I suggest that it is a thickened remnant of a double perichondral layer representing contiguous surface ossifications of the two separate cartilages comprising the primitive braincase in placoderms. In Brindabellaspis this lamina is completely enclosed within the cartilage of the fused endocranium, and can be shown to have the correct morphological relations to surrounding nerves and vessels for it to represent the interface between rhinocapsular and postethmo-occipital portions of the endocranium. A similar interpreta- tion for Bothriolepis presents no difficulty in relation to the optic nerve, which in other placoderms actually emanates between these two cartilages. Furthermore, the rhinocapsular bone thus delimited in Bothriolepis is attached to the rostral and pineal plates as in the other primitive placoderms mentioned above. The restoration shows posterolateral extensions of the rhinocapsular bone partly enclosing the orbits laterally, as do the antorbital processes in Brindabellaspis (Young 1980, fig. 8). DISCUSSION 1 . Dermal bones of the jaws and cheek. It is now evident that the so-called mental plate of antiarchs, a paired canal-bearing dermal bone forming the upper biting margin of the mouth, is homologous with the suborbital plate of other placoderms. This is confirmed by the fact that the palatoquadrate is attached to its inner margin. In view of the form of the palatoquadrate as demonstrated by this new material, any suggestion that the lateral plate incorporates homologues of the suborbital plate in whole or in part (e.g. Stensio 1948, p. 200) may be discounted. On the other hand, the lower dermal jaw-bone in Bothriolepis includes differentiated blade and biting portions, as does the infragnathal in euarthrodires, and might therefore be considered the homologue of this bone. If so, it could be suggested that the absence of any supragnathal elements in Bothriolepis (and other antiarchs) is a secondary condition. This is consistent with assumed phylogenetic relationships of the antiarchs. However, there is other evidence indicating that the differentiated lower jaw-bone in antiarchs may have arisen within the group, in which case this argument would not apply. This is more fully considered below. Whether the antiarchs may have lost two pairs of supragnathals is an interesting question, since Miles and Young (1977) proposed this as a synapomorphy of euarthrodires. Further information on jaw structure in yunnanolepids may illuminate this point. In advanced euarthrodires the quadrate part of the palatoquadrate is fused inside the postsuborbital plate, although in primitive euarthrodires and some other placoderms this plate is not readily recognized. In Bothriolepis both the prelateral and infraprelateral plates are canal-bearing bones lying adjacent to the mandibular joint, and one or both may represent a postsuborbital element which has either lost its close connection to the quadrate, or never had such a connection (see below). Behind the suborbital-postsuborbital unit the operculum in placoderms is covered by a large dermal bone, the submarginal plate. In several groups this plate articulates with the endocranium through a cartilage fused to its inner surface. Again there is little doubt that the bone in antiarchs previously termed the extralateral plate is homologous to the submarginal (e.g. Miles 1971; Young 1980). In some asterolepidoids (e.g. Nilsson 1941) the plate carries a groove on its inner surface similar to the groove for the epihyal in phlyctaenioids (e.g. Goujet 1972, 1975), and in yunnanolepids the submarginal lies adjacent to a notch in the lateral skull-roof margin resembling the notch in a corresponding position in euarthrodires (Zhang Miman 1980; Young 1980, p. 53). However, the development of a dermal articulation between the submarginal and the lateral plate in Bothriolepis is unique amongst placoderms, although an analogous articulation is seen between the suborbital and 654 PALAEONTOLOGY, VOLUME 27 postorbital plates in Buchanosteus (Young 1979). It is worth noting that this articulation in Bothriolepis is supported by a thickened ridge of bone beneath the lateral plate, interpreted above as being adjacent to, or part of, the anterior postorbital process. In euarthrodires the submarginal has an equivalent position, being connected to the endocranium through an articulation between the epihyal and the end of the anterior postorbital process. 2. The preorbital recess and preorbital depression in antiarchs. It has been suggested above that the preorbital recess of Bothriolepis housed the rhinocapsular bone of the endocranium, which was incompletely fused to the postethmo-occipital bone. In placoderms generally, the occurrence of a discrete rostral capsule in several distantly related groups indicates that this is a primitive placoderm feature (Young 1979, p. 341; 1980). It is of interest therefore that in the primitive yunnanolepid antiarchs from south China, and some other forms, the preorbital recess is not developed (Zhang Guorui 1978; Zhang Miman 1980). Instead, there is a broad dorsal depression in front of the orbital fenestra, recently interpreted by Janvier and Pan (1982) as the primitive condition for antiarchs, which was modified to form the recess of Bothriolepis and other more advanced forms by posterodorsal growth of the anterior border of the depression. This being so, one could suggest that the rostral capsule occupied the preorbital depression in yunnanolepids, and because it retained its primitive relation to the main portion of the endocranium as a discretely ossified unit, it is commonly lost and has largely gone unrecognized in these forms. In fact, remains of the rostral plate are known only in one specimen of Yunnanolepis parvus , as described by Zhang Miman (1980), who has, however, restored the nasal openings in a posterior position behind the preorbital depression and within the orbital fenestra. Consistent with this interpretation is the suggestion by Janvier and Pan (1982) that the preorbital depression in Yunnanolepiformes is homologous to the depression described by 0rvig (1975) on the ‘median prerostral plate’ of Romundina. There are thus two alternative interpretations of the preorbital depression in primitive antiarchs. Under my interpretation of the preorbital recess in Bothriolepis , the preorbital depression is no more than a cavity which contained a discretely ossified rostral capsule in a position somewhat similar to that proposed by Zhang Guorui (1978). Under this interpretation the nasal capsules have a wholly dorsal position relative to the premedian plate. The alternative interpretation of Janvier and Pan (1982) proposes that the preorbital depression lies largely in front of and above the nasal capsules, and may be homologous to the depression in the dermal bone surface described by 0rvig (1975) as lying in front of the rostral capsule in Romundina. Under this interpretation the nasal capsules are ventrally situated relative to the premedian plate, perhaps extending forward on either side of the subpremedian ridge, as proposed by Zhang Miman (1980, p. 186). In support of the second interpretation might be cited the fact that in Y. parvus the floor of the preorbital depression is ornamented, as is the depression on the prerostral plate in Romundina. But this raises a difficulty with Janvier and Pan’s explanation of the fate of the depression in more advanced antiarchs, which requires that the ornamented (dermal) floor of the depression must have sunk into the endocranium to become enclosed in cartilage as the floor of the preorbital recess in bothriolepids. Similarly, the dermal ornament on the floor of the preorbital depression would appear to contradict my suggestion that the depression contained the rostral capsule, since under this interpretation the floor of the depression (like the floor of the preorbital recess in Bothriolepis ) must be of perichondral derivation. Since the preorbital depression in Sinolepis and Microbrachius is entirely lacking in ornament (Liu and Pan 1958; Hemmings 1978), the nature of the dermal tubercles in the preorbital depression of yunnanolepids is of particular interest. My observations on specimens in the Institute of Vertebrate Palaeontology and Paleoanthropology, Beijing, confirmed that in Y. chii the preorbital depression is smooth in the region of the premedian plate, but tuberculate in more lateral parts of the depression. The situation is less clear in Y. parvus because of its small size, but tubercles are again present in the lateral parts of the depression, and may extend towards the mid-line. The observation that these tubercles are much finer in the depression than on the surrounding bone surface indicates an explanation of their occurrence consistent with the interpretation of the preorbital recess in Bothriolepis presented above. I suggest that these fine tubercles surrounded the nasal openings, just as YOUNG: BRAINCASE OF BOTH RIOLEPIS 655 very similar fine ornament is observed in the notch of the nasal opening in Romundina (0rvig 1975), and Buchanosteus (e.g. Miles 1971, fig. 105). In Yunnanolepis the preorbital depression has groove-like lateral extensions, and I suggest that the nares opened laterally into these grooves. Any contact between the floor of the depression and the rhinocapsular bone of the endocranium would thus have been restricted to a narrow region of the mid-line. Under such circumstances dermal bone may grade imperceptibly into unornamented bone of perichondral derivation, as is the case in the rostral capsule of Buchanosteus (Miles 1971, p. 186). It should be noted that this interpretation is at variance with the conclusion of Zhang Guorui (1978, p. 1 54) that the nares in Yunnanolepis may have opened anteriorly, as in Remigolepis, and not laterally as in Bothriolepis , but there is also a phylogenetic argument supporting my view (see below). A final point relates to the position of the nasal capsules in yunnanolepids and whether they may have been partly or wholly contained beneath the premedian plate on either side of the subpremedian ridge as proposed by Zhang Miman (1980). In Y. chii the so-called ‘orbital fenestra’ is a constricted oval-shaped opening facing anteriorly and somewhat dorsally into the preorbital depression. By comparison with the morphology of the orbital cavity as now known in some detail in other placoderms (e.g. 0rvig 1975; Young 1979, 1980) it can be suggested that the following nerves and vessels (all paired) must have gained access to the orbital cavity through this opening: optic (II), oculomotor (III), trochlearis (IV), profundus (V), and abducens (VI) nerves, the ophthalmica magna artery and possibly a branch of the orbital artery, and presumably an orbital or orbitonasal vein. In view of the small size of the ‘orbital fenestra’ in Yunnanolepis it seems inconceivable that the nasal capsules themselves could have been situated beneath it — this would require a similar position for the telencephalon of the brain, and a tortuous dorsal path for the optic nerves to reach the orbits, and of extensive nasal tubes to reach the nares. Considerations of space make it equally unlikely that the nasal capsules were positioned above the fenestra, through which the olfactory nerves passed in addition to those nerves and vessels just mentioned. In my opinion the only interpretation of the known structure of yunnanolepids and other antiarchs which is consistent with detailed endocranial morphology as known in other placoderms, is as follows: that the ‘orbital fenestra’ in various antiarchs with a preorbital depression is misnamed, being strictly homologous to the suborbital fenestra of Bothriolepis ; and that the suborbital fenestra of antiarchs generally corresponds, at least in its dorsal parts, to the anterior fenestra of the endocranial cavity in forms like Buchanosteus (Young 1979, fig. 8), or Romundina (0rvig 1975, pi. 2, fig. 2). As such, the equivalents of the crista supraethmoidalis in these forms, in antiarchs would have attached to the anterior edge of the postpineal plate, thereby marking the anterior limit of attachment of the postethmo-occipital bone of the endocranium to the dermal skull-roof. It is possible that the eye-stalk as known in other placoderms was secondarily lost in antiarchs. Vessels such as the orbital artery, which in other forms passed up through the suborbital shelf to reach the eyeball (e.g. Young 1980, fig. 10), must have entered the orbit through the ventral part of the suborbital fenestra. 3. Phylogenetic implications. The adoption of cladistic techniques in analysing phylogenetic relationships within the placoderms (e.g. Miles and Young 1977; Dennis and Miles 1979-1983; Miles and Dennis 1979; Young 1979, 1980, 1981b; Janvier and Pan 1982; Lelievre et al. 1982; Long 1983) has during the last few years generated a number of new ideas regarding placoderm evolution. These are testable in the sense that they make predictions about the morphology of various groups which can be checked as new information becomes available. An important aspect in this procedure is that any new morphological observation or interpretation of any taxon within the group under study is relevant to all other observations and interpretations of other taxa within the group. This results from the unifying character imparted to phylogenetic hypotheses by the use of cladistic techniques. With regard to antiarch phylogeny, an outline cladogram was presented by Young (1981a), and a more detailed proposal along similar lines was put forward by Janvier and Pan (1982). Interrelationships of antiarchs have been most recently discussed by Long (1983) and Young (in press), and relationships of antiarchs by Goujet (in press). In the context of these proposals, and more general schemes of placoderm interrelationships (e.g. Denison 1975, 1978; Miles and Young 1977; Young 1980), several ideas and observations developed above regarding the morphology of the head in Bothriolepis have 656 PALAEONTOLOGY, VOLUME 27 wider phylogenetic implications. These are the number of supragnathal elements in the upper jaw, the differentiation of the infragnathal into distinct blade and biting portions, the dermal articulation between the submarginal plate and the skull-roof, the relation between the quadrate and the postsuborbital plate, and the position of the nares in antiarchs. The phylogenetic arguments on each are briefly presented in points a-e below, and summarized in the cladogram of text-fig. 12. (a) Supragnathals. The suggestion by Miles and Young (1977) that two pairs of supragnathals may be a synapomorphy of euarthrodires predicts that other placoderm groups will have one or none of these elements. There are no supragnathals in Bothriolepis, but a single pair has recently been reported in phyllolepids (Long, in press), which would mean that this was also the primitive condition in antiarchs, under the scheme of placoderm interrelationships proposed by Miles and Young (1977). Alternatively, the absence of supragnathals in Bothriolepis can be seen as consistent with Goujet’s (in press) view of antiarch relationships. (i b ) Infragnathal. Similarity between the infragnathal of Bothriolepis and that of some euarthro- dires has been mentioned above. In a series of papers on brachythoracid euarthrodires from Gogo (Dennis and Miles 1979a, b, 1980-1983; Miles and Dennis 1979), the differentiation of the infragnathal into distinct posterior blade and anterior biting regions has been used as a synapomorphy of various brachythoracids (see also Young 19816, fig. 17). This was based on the evidence that in Holonema the infragnathal is not differentiated into two portions (Miles 1971 ), that in Phlyctaenius the few known infragnathal remains (assumed by Heintz 1933 to be incompletely preserved) show only a biting region, and that various isolated denticulate bones from the early Devonian of Utah, also lacking a blade portion (Denison 1958, fig. 101), have been attributed to actinolepid euarthrodires. In view of the specialized nature of the dentition in Holonema it is possible that the blade on the infragnathal has been secondarily lost, and the evidence relating to Phlyctaenius is equivocal (e.g. Heintz 1933; Miles 1969). However, it seems reasonable on available evidence to attribute the isolated infragnathals of Denison (1958) to the associated actinolepids, and it is noteworthy that the corresponding element in phyllolepids resembles these examples in general form (Long, in press). Taking account of the differentiated infragnathal in Bothriolepis , therefore, there are three alternative interpretations of the history of this element worthy of consideration, under an assumption that antiarchs and euarthrodires are sister groups: (i) The differentiated infragnathal was present in the common ancestor of antiarchs and euarthrodires, and was secondarily lost in more primitive members (actinolepids, some phlyctaenioids) of the latter group. This alternative seems unlikely, in view of the form of the phyllolepid element mentioned above, and is unparsimonious in requiring two reversals in evolution. (ii) Euarthrodires are paraphyletic, the absence of the blade on the infragnathal in some phlyctaenioids (e.g. Holonema , ? Phlyctaenius) is secondary, and the differentiated infra- gnathal was inherited from the common ancestor of antiarchs and phlyctaenioids, which are sister groups. Evidence against this proposal is evidence supporting euarthrodiran monophyly. Of three characters proposed as euarthrodiran synapomorphies by Miles and Young (1977, p. 134), only one can now be sustained (see Young 1979, p. 347; 19816, p. 261; Young and Gorter 1981). This is the possession of two pairs of supragnathals, but even this is not firmly established for actinolepids. An anterior supragnathal is preserved in situ in Kujdanowiaspis (Stensio 1963, pi. 62), and disarticulated elements which may be actinolepid posterior supragnathals have been described by Denison (1958, 1960; see also Miles 1969, p. 145). However, better evidence is required to confirm that there is a posterior supragnathal on the autopalatine in actinolepids. (Also of relevance in this connection are the position of the supragnathals in phyllolepids— on the palatoquadrate or the braincase— and the number of supragnathals in Wuttagoonaspis.) Another possible euarthrodiran synapomorphy is pro- posed below (point d). (iii) Actinolepids and some phlyctaenioids are primitive in possessing an infragnathal which lacks YOUNG: BRAINCASE OF BOTH RIOLEPIS 657 differentiated blade and biting portions. Resemblances between these and the infragnathal of phyllolepids are symplesiomorphies. The differentiated infragnathal of brachythoracids and antiarchs arose independently in these two groups. Assuming that the Euarthrodira is a monophyletic taxon, this hypothesis is to be preferred on the grounds of parsimony. If, however, actinolepids are shown not to possess two pairs of supragnathals, then alternative (ii) would emerge as the preferred hypothesis. It also follows that the structure of the infragnathal cannot be used as evidence against Goujet’s (in press) alternative hypothesis of antiarch relationships. (c) Submarginal articulation. As noted above, the development of a dermal articulation between the submarginal plate and the skull-roof is a condition so far known only in Botliriolepis , and it thus has the status of an autapomorphy. However, further information on the presence or absence of this structure in other antiarchs thought to be closely related to Botliriolepis (e.g. Dianolepis , Wudinolepis, Microbrachius , Hyrcanaspis) should clarify relationships within the Bothriolepidoidei. (d) Quadrate and postsuborbital plate. It is well known from the works of Stensio (e.g. 1963, 1969) that the palatoquadrate in many advanced brachythoracids is represented only by autopalatine and quadrate ossifications. However, the notion that the intervening region may have been reduced to a ligamentous connection (e.g. Miles 1969, p. 144; 1971, p. 194; 0rvig 1975, p. 65) is no longer held (Miles and Dennis 1979, p. 49). On the evidence of Holonema (Miles 1971), Dicksonosteus (Goujet 1975), and Buchanosteus (Young 1979) it is clear nevertheless that there was a phyletic reduction and loss of perichondral ossification of the metapterygoid region within the Phlyctaenioidei. In phlyctaenioids, and presumably in euarthrodires generally, the quadrate is closely connected to the postsuborbital plate of the cheek. However, in the palatoquadrate of the palaeacanthaspid Romundina , 0rvig(1975, p. 65) has noted that the metapterygoid isperichondrally ossified on all sides at its posterior end, and neither the quadrate nor the postsuborbital plate is known. In view of the condition of the palatoquadrate in other gnathostomes, I am not convinced by 0rvig’s suggestion that in Romundina the quadrate was a separate unit connected only by ligaments to the metapterygoid. Perhaps the mandibular joint had a high position on the non-preserved mesial surface of the palatoquadrate (see 0rvig 1975, pi. 2, fig. 5). Whatever the correct interpretation for Romundina , the apparently anomalous condition of the quadrate in relation to its dermal bone cover. O Bothriolepis O Pterichthyodes -• Asterolepis Yunnanotepis O Bothriolepis Asterolepis Remigolepis • anterior nares O lateral nares ▲ long obstantic margin as a derived feature □ short obstantic margin as a derived feature Yunnanolepis text-fig. 11. Alternative cladograms for some antiarchs, under an assumption that Yunnanolepis had anterior nares. In a and b the Asterolepidoidei (bracketed) are assumed monophyletic on the evidence of the short obstantic margin as a derived feature (synapomorphy). This feature is assumed primitive in c and d, making the asterolepidoids a paraphyletic group. 658 PALAEONTOLOGY, VOLUME 27 a b c d phlyctaenioids actinolepids antiarchs Wuttagoonaspis Phyllolepis text-fig. 12. Interrelationships of some placoderms. Pro- posed synapomorphies, as discussed here, are: a, sliding dermal neck joint, posterior median ventral plate (if absent in petalichthyids), one pair of supragnathals. b , posterior lateral plate and pectoral fenestra in trunk-shield, c, long obstantic margin with prominent posterolateral corners on skull-roof, elongate rather than subovate submarginal plate, dermal prelateral and infraprelateral plates (or their homo- logues) in the cheek unit, d , second pair of supragnathals, quadrate fused to postsuborbital plate. together with the fact that there was also no close association between the quadrate and an overlying dermal bone in Bothriolepis , points to the possibility that the postsuborbital/quadrate connection may be another synapomorphy defining euarthrodires. In addition, if the prelaterals and infraprelaterals of antiarchs are homologous with the postsuborbitals and infrapostsuborbitals of euarthrodires (Miles 1971; Denison 1978), then the possession of these bones may be proposed as another synapomorphy uniting these two groups. (e) Position of the nares in antiarchs. To account for the presence of fine ornament in the preorbital depression I suggested above that the nares in Yunnanolepis opened laterally. This was also the opinion of Zhang Miman (1980), although in other respects (position of nasal capsules, etc.) our interpretations of the nasal region differ considerably. Conversely Zhang Guorui (1978) proposed that in Yunnanolepis the nares opened anteriorly as in Remigolepis (e.g. Stensio 1948, fig. 16) and Aster olepis (e.g. Lyarskaya 1981, fig. 67). It is of significance, however, that in Pterichthyodes and probably Gerdalepis (Gross 1941; Stensio 1948; Hemmings 1978) the rostral plate is developed like that of Bothriolepis , with lateral notches for the nares bounded anteriorly by a prenasal division of the plate. In a phylogenetic context this character distribution can be interpreted in several ways, depending on whether the asterolepidoid antiarchs are regarded as monophyletic or paraphyletic. Young and Gorter (1981) suggested that the short obstantic margin facing posteriorly with posterolaterally extended postmarginal plates was an asterolepidoid synapomorphy, and that the group was monophyletic. If the nares opened anteriorly in Yunnanolepis , this would require either that the condition developed independently of that seen in Aster olepis and Remigolepis (text-fig. 11a), or that it was the primitive antiarch condition, which was separately lost in Bothriolepis and Pterich- thyodes (text-fig. 1 1 b). Both alternatives are less parsimonious than the assumption that in Yunnano- lepis the nares were laterally placed, this being the primitive condition for antiarchs generally. On the other hand, if the asterolepidoid antiarchs were assumed to be paraphyletic, the interpretation of Yunnanolepis as having anteriorly opening nares could be explained either as the primitive condition for antiarchs, which was lost in the common ancestor of Pterichthyodes and Bothriolepis (text-fig. 11c), or as a unique specialization of Asterolepis, Remigolepis, and Y unnano- lepis (text-fig. 1 Id). However, under neither interpretation can the similarities in the obstantic margin and configuration of the postmarginal plate in Bothriolepis and Yunnanolepis be adequately explained. This must be the derived condition under this interpretation (since the other state of this character, as developed in Asterolepis and Remigolepis, is assumed to be symplesiomorphic) and in both schemes is independently acquired (text-fig. 1 1c, d). On the grounds of parsimony therefore it YOUNG: BRAINCASE OF BOTH RIOLEPIS 659 can be concluded that laterally opening nares, as in Bothriolepis , was the primitive antiarch condition, and that the anterior position of the nasal openings in Asterolepis and Remigolepis is a synapomorphy of these genera. This is consistent with the observation that in other placoderms with ventral nares these are also directed more or less laterally (e.g. Stensio 1963, figs. 10, 17). It can be predicted therefore that better material of Yunnanolepis will demonstrate that this form also had the primitive arrangement of lateral nares. Acknowledgements. Messrs. R. W. Brown, W. Peters, and P. W. Davis photographed the material, and G. J. Butterworth provided assistance with drafting. I am indebted to Professor Pan Jiang, Dr. Zhang Guorui, and Dr. Liu Shifan for advice, helpful discussion, and hospitality during a visit to China sponsored by the Australian Academy of Science and Academia Sinica. Dr. P. J. Jones kindly read through the manuscript. Published with the permission of the Director, Bureau of Mineral Resources, Geology and Geophysics, Canberra. REFERENCES denison, r. h. 1958. Early Devonian fishes from Utah. 3, Arthrodira. Fieldiana, Geol. 11, 461-551. 1960. Fishes of the Devonian Holland Quarry Shale of Ohio. Ibid. 555-613. — 1975. Evolution and classification of placoderm fishes. Breviora , 432, I -24. — 1978. Placodermi Handbook of Paleoichthyology, volume 2. h.-p. schultze fed. ). Gustav Fisher Verlag, Stuttgart, 128 pp. dennis, k. d. and miles, R. s. 1979a. A second eubrachythoracid arthrodire from Gogo, Western Australia. Zool. J! Linn. Soc. 67, 1 -29. — 19796. Eubrachythoracid arthrodires with tubular rostral plates from Gogo, Western Australia. Ibid. 297-328. — 1980. New durophagous arthrodires from Gogo, Western Australia. Ibid. 69, 43-85. — 1981. A pachyosteomorph arthrodire from Gogo, Western Australia. Ibid. 73, 213-258. — 1982. A eubrachythoracid arthrodire with a snub-nose from Gogo, Western Australia. Ibid. 75, 153-166. — 1983. Further eubrachythoracid arthrodires from Gogo, Western Australia. Ibid. 77, 145-173. gardiner, b. G. and bartram, A. w. h. 1977. The homologies of ventral cranial fissures in osteichthyans. In Andrews, s. m., miles, R. s. and walker, a. d. (eds.). Problems in vertebrate evolution 227-245. Academic Press, London. — and miles, r. s. 1975. Devonian fishes of the Gogo Formation, Western Australia. Colloques int. Cent. Natn. Red i. Scient. 218, 73-79. goujet, d. 1972. Nouvelles observations sur la joue d 'Arctolepis (Eastman) et d’autres Dolichothoraci. Ann Is Paleont. 58, 3-11. — 1975. Dicksonosteus , un nouvel arthrodire du Devonien du Spitsberg. Remarques sur le squelette visceral des Dolichothoraci. Colloques int. Cent. Natn. Rech. Scient. 218, 81-99. — (in press). Placoderm interrelationships: a new interpretation, with a short review of placoderm classifications. Proc. Linn. Soc. New South Wales, 107. gross, w. 1941. Neue Beobachtungen an Gerdalepis rhenana (Beyrich). Palaeontographica, A 93, 193-214. heintz, a. 1933. Some remarks about the structure of Phlyctaenaspis acadica Whiteaves. Norsk geol. Tidsskr. 14, 127-44. hemmings, s. K. 1978. The Old Red Sandstone antiarchs of Scotland: Pterichthyodes and Microbrachius. Palaeontogr. Soc. ( Monogr .), 131, 1-64. janvier, p. and pan jiang 1982. Hyrcanaspis bliecki n.g. n.sp., a new primitive euantiarch (Antiarcha, Placodermi) from the Middle Devonian of northeastern Iran, with a discussion on antiarch phylogeny. N. Jb. Geol. Paldont. Abh. 164, 364-392. lelievre, h., janvier, p. and goujet, d. 1982. Les vertebres Devoniens de l’lran central. IV: Arthrodires et ptyctodontes. Geobios , 14, 677 -709. Liu, t.-s. and p’an, k. 1958. Devonian fishes from the Wutung Series near Nanking, China. Palaeont. Sin. 141, 1-41. [In Chinese and English.] long, j. a. 1983. New bothriolepid fish from the Late Devonian of Victoria, Australia. Palaeontology , 26, 295-320. — (in press). New phyllolepids from Victoria, and the relationships of the group. Proc. Linn. Soc. New South Wales , 107. 660 PALAEONTOLOGY, VOLUME 27 lyarskaya, l. A. 1981. Baltic Devonian Placodermi Asterolepididae. Zinatne, Riga, 152 pp. [In Russian with English abstract ] miles, r. s. 1968. The Old Red Sandstone antiarchs of Scotland. Family Bothriolepididae. Palaeontogr. Soc. ( Monogr .), 122, 1-130. 1969. Features of placoderm diversification and the evolution of the arthrodire feeding mechanism. Trans. R. Soc. Edinb. 68, 123-70. 1971. The Holonematidae (placoderm fishes), a review based on new specimens of Holonema from the Upper Devonian of Western Australia. Phil. Trans. R. Soc. B 263, 101-234. 1973. An actinolepid arthrodire from the lower Devonian Peel Sound formation. Prince of Wales Island. Palaeontographica, A 143, 109-18. — 1977. Dipnoan (lungfish) skulls and the relationships of the group: a study based on new species from the Devonian of Australia. Zoo/. Jl Linn. Soc. 61, 1-328. — and dennis, k. 1979. A primitive eubrachythoracid arthrodire from Gogo, Western Australia. Ibid. 66, 31-62. — and westoll, t. s. 1968. The placoderm fish Coccosteus cuspidatus Miller ex Agassiz from the Middle Old Red Sandstone of Scotland. Part I. Descriptive morphology. Trans. R. Soc. Edinb. 67, 373-476. and young, G. c. 1977. Placoderm interrelationships reconsidered in the light of new ptyctodontids from Gogo, Western Australia. In Andrews, s. m., miles, r. s. and walker, a. d. (eds.). Problems in vertebrate evolution. 123-198. Academic Press, London. nilsson, t. 1941. The Downtonian and Devonian vertebrates of Spitsbergen. 7. Order Antiarchi. Skr. Svalbard Ishavet, 82, 1-54. orvig, t. 1975. Description, with special reference to the dermal skeleton, of a new radotinid arthrodire from the Gedinnian of Arctic Canada. Colloques int. Cent. Natn. Rech. Scient. 218, 41-71. stensio, e. a. 1931. Upper Devonian vertebrates from East Greenland, collected by the Danish Greenland expedition in 1929 and 1930. Meddr. Gronland , 86, 1-212. — 1949. On the Placodermi of the Upper Devonian of East Greenland. 2. Antiarchi: subfamily Bothrio- lepinae. With an attempt at a revision of the previously described species of that family. Ibid. 139 ( Palaeozool . Groenland. 2), 1-622. — 1963. Anatomical studies on the arthrodiran head. Part I. Preface, geological and geographical distribution, the organisation of the head in Dolichothoraci, Coccosteomorphi and Pachyosteomorphi. Taxonomic appendix. K. Svenska VetenskAkad. Handl. 9, 1-419. — 1969. Elasmobranchiomorphi Placodermata Arthrodires. In piveteau, j. (ed.). Traite de Paleontologie, 4, 71-692. Masson, Paris. toombs, h. a. 1948. The use of acetic acid in the development of vertebrate fossils. Museums Jl, 48, 54-55. — and rixon, a. e. 1959. The use of acids in the preparation of vertebrate fossils. Curator, 2, 304-312. young, G. c. 1979. New information on the structure and relationships of Buchanosteus (Placodermi, Euarthrodira) from the Early Devonian of New South Wales. Zool. Jl Linn. Soc. 66, 309-352. 1980. A new Early Devonian placoderm from New South Wales, Australia, with a discussion of placo- derm phylogeny. Palaeontographica, A 167, 10-76. 1981a. Biogeography of Devonian vertebrates. Alcheringa, 5, 225-243. — 19816. New Early Devonian brachythoracids (placoderm fishes) from the Taemas-Wee Jasper region of New South Wales. Ibid. 5, 245-271. 1982. Devonian sharks from south-eastern Australia and Antarctica. Palaeontology , 25, 817-843. — (in press). Comments on the phylogeny and biogeography of antiarchs (Devonian placoderm fishes), and the use of fossils in biogeography. Proc. Linn. Soc. New South Wales, 107. — and gorter, j. d. 1981. A new fish fauna of Middle Devonian age from the Taemas/Wee Jasper region of New South Wales. Bull. Bur. Min. Res. Geol. Geoph. 209, 83-147. zhang guorui 1978. The antiarchs from the Early Devonian of Yunnan. Vertebr. PalAsiat. 16, 147-86. [In Chinese with English summary.] zhang miman 1980. Preliminary note on a Lower Devonian antiarch from Yunnan, China. Ibid. 18, 179-190. [In Chinese and English.] G. c. YOUNG Division of Continental Geology Bureau of Mineral Resources Geology and Geophysics PO Box 378, Canberra City, ACT Australia, 2601 Typescript received 28 September 1983 YOUNG: BRAINCASE OF BOTHRIOLEPIS 661 ABBREVIATIONS USED IN TEXT-FIGURES 1 olfactory nerve ifc infraorbital sensory groove 2 approximate position of optic nerve LA lateral plate ADL anterior dorsolateral plate laf articular structure for dermal neck-joint AVL anterior ventrolateral plate lat.p lateral pit a,SM anterior articular area for submarginal leg main lateral lme sensory groove plate Mk Meckel’s cartilage a,SM posterior articular area for submarginal mf dorsomesial face of non-biting division of plate infragnathal ad anterior biting division of infragnathal Nu nuchal plate ad, anterodorsal process of submarginal plate n notch, possibly representing part of an ad 2 posterior articular area for connection articular area with skull na nasal opening a.m thickened perichondral margin of area in n.cav nasal cavity contact with dermal bone oa.AVL overlap area for anterior ventrolateral apl unornamented lateral division of infra- oa.PrL overlap areas for prelateral prelateral plate ood otico-occipital depression a.pr anterior process of Meckel’s cartilage P pineal plate art articular areas for orbital connection PP postpineal plate with braincase PM postmarginal plate art.md articular surface for mandibular joint PNu paranuchal plate art. op articular area for opercular cartilage or PrL prelateral plate epihyal PrM premedian plate art.pb articular area for palatobasal connection Py infraprelateral plate with braincase p.br brachial process art.v ventral articular area, possibly for hyoid pd posterior non-biting division of infrag- arch element nathal a.sh anterior shelf of Meckel’s cartilage pe pars pedalis of brachial process aup autopalatine part of palatoquadrate per perichondral bone av ventral articular area of submarginal p.etho postethmo-occipital section of endo- plate cranium cd.art articular condyle for mandibular joint pin pineal foramen ch ceratohyal pic posterolateral corner of infragnathal cir semicircular pit-line groove pmm posteromesial margin of posterior division cri prelateral crista of infragnathal cr2 postlateral crista pnt articular structure for dermal neck-joint cr.im inframarginal crista p.pr posterior process of suborbital plate cr.pm paramarginal crista pq palatoquadrate cr.pto postorbital crista pr.ant antorbital process csl central sensory line pr.apo anterior postorbital process cv cranial cavity pr.csp craniospinal process d.end endolymphatic duct pr.dm dorsomesial process of suborbital plate dent denticulate biting margin pr.ect ectethmoid process df dorsolateral face of non-biting division of pr-gl glenoid process infragnathal pr.ppo posterior postorbital process eh epihyal psoc postsuborbital sensory groove f.am adductor fossa R rostral plate f.ax axillary foramen r.epq extrapalatoquadrate ridge fe.orb orbital fenestra rh.cap rhinocapsular section of endocranium fe.sorb suborbital fenestra r.pq dermal ridge supporting palatoquadrate fo.cu cucullaris fossa SM submarginal plate fo.md mandibular muscle fossa SO suborbital plate fp funnel pit sns subnasal shelf f.vasc foramina, probably vascular soa subobstantic area gr groove, possibly for sensory pit-line spio foramina for spino-occipital nerves gr.ul groove for upper lip tig transverse lateral groove NOTES FOR AUTHORS The journal Palaeontology is devoted to the publication of papers on all aspects of palaeontology. Review articles are particularly welcome, and short papers can often be published rapidly. A high standard of illustration is a feature of the journal. Four parts are published each year and are sent free to all members of the Association. Typescripts should conform in style to those already published in this journal, and should be sent to Dr. D. E. G. Briggs, Department of Geology, Goldsmiths’ College, University of London, Creek Road, London SE8 3BU, England, who will supply detailed instructions for authors on request (these were published in Palaeontology 1977, 20, pp. 921-929). Special Papers in Palaeontology is a series of substantial separate works conforming to the style of Palaeontology. SPECIAL PAPERS IN PALAEONTOLOGY In addition to publishing Palaeontology the Association also publishes Special Papers in Palaeontology. Members may subscribe to this by writing to the Membership Treasurer: the subscription rate for 1984 is £30 (U.S. $53) for Institutional Members, and £15 (U.S. $26.50) for Ordinary and Student Members. A single copy of each Special Paper is available to Ordinary and Student Members only , for their personal use, at a discount of 25% below the listed prices. Non-members may obtain copies, but at the listed prices, from Marston Book Services, P.O. Box 87, Oxford 0X4 1LB, England. RECENT PALAEONTOLOGICAL ASSOCIATION PUBLICATIONS Special Papers in Palaeontology Numbers 119 are still in print and are available (post free) together with those listed below: 20. (for 1977): Fossil Priapulid Worms, by s. c. morris. 155 pp. 99 text-figs ., 30 plates. Price £16 (U.S. $28). 21. (for 1978): Devonian Ammonoids from the Appalachians and their bearing on International Zonation and Correlation, by m. r. house. 70pp., 12 text-figs., 10 plates. Price £12 (U.S. $21). 22. (for 1978, published 1979): Curation of Palaeontological Collections. A joint Colloquium of the Palaeontological Association and Geological Curators Group. Edited by m. g. bassett. 279 pp., 53 text-figs. Price £25 (U.S. $44). 23. (for 1979): The Devonian System. A Palaeontological Association International Symposium. Edited by m. r. house, c. t. scrutton and m. g. bassett. 353 pp., 102 text-figs., 1 plate. Price £30 (U.S. $52.50). 24. (for 1980): Dmoflagellate Cysts and Acritarchs from the Eocene of Southern England, by s. p. bujak, c. downie, g. l. eaton and g. l. williams. 100 pp., 24 text-figs., 22 plates. Price £15 (U.S. $26.50). 25. (for 1980): Stereom Microstructure of the Echinoid Test, by a. b. smith. 81 pp., 20 text-figs., 23 plates. Price £15 (U.S. $26.50). 26. (for 1981): The Fine Structure of Graptolite Periderm, by p. r. crowther. 119 pp., 37 text-figs., 20 plates. Price £25 (U.S. $44). 27. (for 1981): Late Devonian Acritarchs from the Carnarvon Basin, Western Australia, by g. playford and r. s. dring. 78 pp., 10 text-figs., 19 plates. Price £15 (U.S. $26.50). 28. (for 1982): The Mammal Fauna of the Early Middle Pleistocene cavern infill site of Westbury-sub-Mendip, Somerset, by m. j. bishop. 108 pp., 47 text-figs., 6 plates. Price £25 (U.S. $44). 29. (for 1982): Fossil Cichlid Fish of Africa, by j. a. h. vancouvering. 103 pp., 35 text-figs., 10 plates. Price £30 (U.S. $52.50). 30. (for 1983): Trilobites and other early Arthropods. Edited by d. e. g. briggs and p. d. lane. 276 pp., 64 text-figs., 38 plates. Price £40 (U.S. $70). Field Guides to Fossils 1. (1983): Fossil Plants of the London Clay, by m. e. collinson. 121 pp., 242 text-figs. Price £7-95 (U.S. $14). Other Publications 1982. Atlas of the Burgess Shale. Edited by s. c. morris. 31 pp., 24 plates. Price £20 (U.S. $44). © The Palaeontological Association , 1984 Palaeontology VOLUME 27 PART 3 CONTENTS Classification of the Echinodermata ANDREW B. SMITH 431 A muscle attachment proposal for septal function in Mesozoic ammonites R. A. HENDERSON 461 Principal floras of Palaeozoic marine calcareous algae BORIS CHUVASHOV and ROBERT RIDING 487 Palaeoecology of marginal marine sedimentary cycles in the Albian Bear River Formation of south-western Wyoming FRANZ T. FURSICH and ERLE G. KAUFFMAN 501 Sclerochronology and carbonate production in some Upper Jurassic reef corals OMER E. A li 537 Construction and preservation of two modern coralline algal reefs, St. Croix, Caribbean DANIEL W. J. BOSENCE 549 The postcranial skeleton of the Upper Triassic sphenodontid Planocephalosaurus robinsonae N. C. FRASER and G. M. WALKDEN 575 Osteology of the Palaeocene teleost Esox tiemani MARK V. H. WILSON 597 A new freshwater limuloid from the middle Triassic of New South Wales J. W. PICKETT 609 Ramseyocrinus and Ristnacrinus from the Ordovician of Britain STEPHEN K. DONOVAN 623 Reconstruction of the jaws and braincase in the Devonian placoderm fish Bothriolepis G. c. young 635 Printed in Great Britain at the University Press , Oxford hr David Stanford , Printer to the University issn 0031-0239 Published by The Palaeontological Association London Price £19-50 THE PALAEONTOLOGICAL ASSOCIATION The Association was founded in 1957 to promote research in palaeontology and its allied sciences. COUNCIL 1984-1985 President'. Professor C. Downie, Department of Geology, University of Sheffield, Sheffield SI 3JD Vice-Presidents : Dr. J. C. W. Cope, Department of Geology, University College, Swansea SA2 8PP Dr. R. Riding, Department of Geology, University College, Cardiff CF1 1XL Treasurer. Dr. M. Romano, Department of Geology, University of Sheffield, Sheffield SI 3JD Membership Treasurer. Dr. A. T. Thomas, Department of Geological Sciences, University of Aston, Birmingham B4 7ET Institutional Membership Treasurer. Dr. A. R. Lord, Department of Geology, University College, London WC1E 6BT Secretary. Dr. P. W. Skelton, Department of Earth Sciences, Open University, Milton Keynes MK7 6AA Circular Reporter: Dr. D. J. Siveter, Department of Geology, University of Hull, Hull HU6 7RX Marketing Manager: Dr. R. J. Aldridge, Department of Geology, University of Nottingham, Nottingham NG7 2RD Editors Dr. D. E. G. Briggs, Department of Geology, Goldsmiths’ College, London SE8 3BU Dr. P. R. Crowther, Leicestershire Museums Service, Leicester LEI 6TD Professor L. B. Halstead, Department of Geology, University of Reading, Reading RG6 2AB Dr. R. Harland, British Geological Survey, Keyworth, Nottingham NG12 5GG Dr. T. J. Palmer, Department of Geology, University College of Wales, Aberystwyth SY23 2AX Other Members Dr. E. N. K. Clarkson, Edinburgh Dr. C. R. C. Paul, Liverpool Dr. D. Edwards, Cardiff Dr. A. B. Smith, London Dr. P. D. Lane, Keele Professor T. N. Taylor, Columbus Dr. A. W. Owen, Dundee Overseas Representatives Australia: Professor B. D. Webby, Department of Geology, The University, Sydney, N.S.W., 2006 Canada: Dr. B. S. Norford, Institute of Sedimentary and Petroleum Geology, 3303-33rd Street NW., Calgary, Alberta Japan : Dr. I. Hayami. University Museum, University of Tokyo, Hongo 7-3-1, Bunkyo-Ku, Tokyo New Zealand: Dr. G. R. Stevens. New Zealand Geological Survey. P.O. Box 30368. Lower Hutt U.S. A.: Dr. R. Cuffey, Department of Geology, Pennsylvania State University, Pennsylvania Professor A. J. Rowell, Department of Geology, University of Kansas, Lawrence, Kansas 66045 Professor N. M. Savage, Department of Geology, University of Oregon, Eugene, Oregon 97403 South America: Dr. O. A. Reig, Departamento de Ecologia, Universidad Simon Bolivar, Caracas 108. Venezuela MEMBERSHIP Membership is open to individuals and institutions on payment of the appropriate annual subscription. Rates for 1984 are: Institutional membership . . . £38-50 (U.S. $67.50) Ordinary membership . . . £18 00 (U.S. $31 ) Student membership . . £1 1-50 (U.S. $20) Retired membership . . . £9-00 (U.S. $16) There is no admission fee. Correspondence concerned with Institutional Membership should be addressed to Dr. A. R. Lord, Department of Geology, University College, Gower Street, London WC1E 6BT, England. Student members are persons receiving full-time instruction at educational institutions recognized by the Council. On first applying for membership, an application form should be obtained from the Membership Treasurer, Dr. A. T. Thomas, Department of Geological Sciences, University of Aston, Gosta Green, Birmingham B4 7ET. Subscriptions cover one calendar year and are due each January; they should be sent to the Membership Treasurer. All members who join for 1984 will receive Palaeontology , Volume 27, Parts 1 -4. All back numbers are still in print and may be ordered from Marston Book Services, P.O. Box 87, Oxford OX4 1LB, England, at £ 19-50 (U.S. $34.50) per part (post free). Cover: The Coralline Crag (Pliocene) bryozoan Cribrilina sp. showing a group of ten feeding zooids each with an ovicell and paired adventitious avicularia on either side of the orifice This specimen was figured as Lepralia punctata Hassall in G. Busk’s Palaeontographical Society monograph of Crag Poly:oa( 1859, pi. 4, fig. I ). It is reillustrated here by means of a new technique, scanning electron microscopy of the uncoated specimen using back-scattered electrons, x 75. REVIEW OF THE DISTRIBUTION OF THE COMMONER ANIMALS IN LOWER SILURIAN MARINE BENTHIC COMMUNITIES H \ 4Pp \\ by l. r. m. cocks and w. s. mckerrow xk. % j Abstract. The distribution of the commoner species, most of which are brachiopods, in thirty large collections from the late Llandovery of the Welsh Borderland gives extra data on the previously published Lingula , Eocoelia, Pentamerus, Stricklandia, and Clorinda communities. The constituents of the communities were not usually interdependent, but lived together in comparable habitats with similar external parameters. The depths at which the communities lived are reviewed and it is concluded that the total depth range of the community spectrum was probably less than 200 m. Studies of brachiopod-dominated associations of lower Silurian age from the Welsh Borderland began over twenty years ago, and these associations were subsequently grouped into animal communities (Ziegler 1965; Cocks 1967; Ziegler et al. 1968a) which were distributed stratigraphically into organized patterns subparallel to the inferred shoreline (Ziegler et al. 19686). It is now opportune to review some aspects of these studies, which came before most scientific work on Palaeozoic communities of other ages; to present more data on the variation within the described communities; and to discuss some aspects of the ecology of the individual constituents. Exceptional preservation, such as that in the Cambrian Burgess Shale or the Carboniferous Mazon Creek beds, sometimes enables a palaeontologist to glimpse the whole range of flora and fauna to be found on the sea floor, and to compare it with that living today. However, it has long been realized that the average preserved fossil collection from the vast majority of ordinary localities represents only a small proportion of the original biota, both in biomass and diversity, even assuming that the fossils have not been carried dead into the area of deposition. Thus some palaeontologists have questioned the use of the word ‘community’ when describing repetitive associations of fossils such as those from the Welsh Borderland Llandovery rocks. However, we are unrepentant since, as can be seen by comparison with modern-day situations, such associations assuredly reflect the original communities of which they formed part, and we feel that the shorthand terminology of referring to these preserved associations as ‘communities’ is justified. We continue to define our communities in a relatively broad way (Cocks and McKerrow in McKerrow 1978), unlike the narrower community groupings of Boucot (1975). Upper Llandovery communities The Llandovery is an exceptionally good period to study clastic level-bottom animal communities for several reasons: first, there is an excellent framework for accurate correlation (Cocks 1971) using both graptolites and shelly fossils, in particular by using the evolution of selected brachiopods such as Stricklandia (Williams 1951 and later authors) and Eocoelia (Ziegler 1966); secondly, because there was a very widespread distribution of a single faunal province due to the relative nearness of several land masses at that time (Cocks and Lortey 1982, fig. 5): the same communities are very widely distributed across North America, Europe, and parts of Asia; and thirdly, because of the substantial erosion during the glacioeustatic regression at the Ordovician-Silurian boundary, the edges of the Llandovery shelves were relatively well marked (like the continental shelves of the present day following the Pleistocene glaciation), and can be recognized using sedimentological criteria as well as by the distribution of benthic communities. | Palaeontology, Vol. 27, Part 4, 1984, pp. 663-670.) 664 PALAEONTOLOGY, VOLUME 27 All the communities discussed in this paper lived in the open shelf sea and on clastic bottoms. We assume that the salinity of all the environments was normal marine, with the exception of some of the near-shore Lingula community. Whether or not all of the constituents of the Lingula community, such as the large bivalves, some tentaculitids and brachiopods such as Stegerhynchus decetnplicatus, were able to thrive in reduced salinities is more doubtful. There are no carbonate deposits included in the samples described, although the same named communities are known from carbonate sediments, for example in North America and Estonia; but other associations are also found in those places, including some associated with bioherms and shallow-water micrites, which are outside the scope of this review. We have reidentified the specimens in 30 of the 94 collections originally made by Ziegler and Cocks (Ziegler et al. 1968 6, Appendix 1) from the late Llandovery of the Welsh Borderland. The real numbers in each collection are greater than the totals shown in text-figs. 1 and 2 because only the most commonly occurring brachiopod or molluscan valve was counted; for example, in Collection 1 3 there were 104 pedicle valves and 85 brachial valves present of Pentamerus oblongus , but only the 104 were included in the collection total shown of 220. Thus the 30 collections represent well over 10,000 specimens. The percentages present of all of the different species in the thirty collections were calculated, and the total numbers of species are shown in text-fig. la broken down into brachiopods and other phyla. The average number of brachiopod species in each community increases from 4 in the Lingula community to 20 in the Clorinda community, but by contrast the average number of species other than brachiopods does not vary so much, although there is a general progressive increase in total diversity from the Lingula to the Clorinda communities. The total proportions of brachiopods are also tabulated (text-fig. 1 b) and, although the brachiopods vary from 46% to 97% of the total in any one collection (average 72%), there is no general pattern of relative brachiopod dominance in any one community, and the average remains surprisingly constant between the communities. Collection number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Locality (Ziegler et al. 1968. p.780-1) 24 32 34 35 44 11 4 46 43 1 37 63 59 45 52 53 13 16 12 73 40 68 70 69 54 60 49 58 55 56 Total fossils 373 52 185 119 348 267 324 662 204 273 351 1044 220 300 122 122 126 141 91 522 185 293 330 409 211 163 333 167 360 381 1 brachiopods 4 3 4 7 6 9 10 12 9 8 12 15 9 14 4 7 13 7 5 12 11 7 12 13 9 16 22 17 20 23 ies i non"drachiopods 12 7 7 8 14 11 16 18 12 14 16 15 6 11 8 6 2 9 8 15 10 8 25 11 11 9 18 10 9 13 / Total 16 10 11 15 20 20 26 30 21 22 28 30 15 25 12 13 15 16 13 27 21 15 37 24 20 25 40 27 29 36 Community LINGULA EOCOELI A PENTAMERUS STRICKLANDIA CLORINDA a v. brachs 4 av. brachs. 9 av brachs. 10 av. brachs 10 av. brachs. 20 av. non- br. 9 av. non-br. 13 av non -br 9 av non-b r. 13 av. non-b 12 13 22 19 23 32 text-fig. 1 . Numbers of brachiopod and other species in thirty collections from the late Llandovery of the Welsh Borderland, and the average numbers of species in the various animal communities described by Ziegler et al. ( 1 968c/). Middle row, proportions of brachiopods (shaded) to species of other phyla in the same collections. Bottom row, percentages of tabulate corals in the same collections. COCKS AND McKERROW: SILURIAN MARINE BENTHIC COMMUNITIES 665 To construct text-fig. 2, only genera that occur as 10% or more in more than one collection were selected for inclusion. The total distributions of these genera in all thirty collections are shown, so that the relative abundance and community ranges may be seen at a glance. Superimposed on the community structure are some distributions caused by local clumping. To eliminate these random effects we do not show forms that occur as over 10% in only one collection, and these are the brachiopods Dolerorthis (10% in Collection 1 1), Isorthis (1 1% in Collection 9), Salopina (10% in Collection 9), Brachyprion arenacea (16% in Collection 12), and ‘ Meristina furcata (12% in Collection 13), and the bryozoan Hallopora (16% in Collection 16). From text-fig. 2 it can be seen that the community range of each taxon varied widely both in selectivity and abundance, from Pentamerus , which is rare outside its named community but very abundant within it, through Atrypa , which occurs in the Eocoelia to Clorinda communities but with a weak indication of an abundance maximum in the Pentamerus and Stricklandia communities, to streptelasmatid corals, which (apart from their absence in the Lingula and part of the Eocoelia communities) are widespread throughout the rest of the community spectrum with little indication of an abundance maximum. Although the brachiopod diversity increased steadily up to the Clorinda community, the largest brachiopods are to be found in the Pentamerus and Stricklandia communities; not just the eponymous pentamerides but the bigger strophomenides such as Leptostrophia and the larger species of Leptaena and also the larger specimens of atrypids. We have no quantitative data on the biomass involved, but from our collecting we would estimate that these mid-shelf communities appear to have been the areas representing the optimal conditions for the growth of large brachiopods, in contrast to the higher-energy shallower environments on the one hand and the relatively plankton-starved deeper water on the other. Further aspects of the size and feeding efficiency of some Silurian brachiopods were considered by Fiirsich and Hurst (1974). The Llandovery communities are made up of an agglomeration of animals, each of which had a distinct but specific tolerance to a range of marine bottom conditions, but which do not appear to have had any marked degree of mutual interdependence. Of course there are exceptions, for example, some of the bryozoans and cornulitids needed larger shells for their attachment, but most of the individual brachiopods, trilobites, molluscs, corals, and other larger benthos would have been unaware of, and independent from, their neighbours. The normal relationships between inter- dependent members of the same community today are either as successive members of a food chain or as providing shelter or anchorage. In the Llandovery most food chains were not long; the vast majority of the fauna were suspension feeders or deposit feeders. The numbers of predators and scavengers was probably small, and confined to echinoderms, cephalopods, a small minority of arthropods such as eurypterids (not phyllocarids, contra Watkins 1979, p. 249), and, perhaps, some soft-bodied worms. The gastropods present were all archaeogastropods which were almost certainly algal grazers or to a lesser extent deposit feeders; there are no confirmed predatory gastropods of Silurian age. Thus the distribution of the sedentary benthos must have been controlled partly by the availability of food and partly by physical factors (Fiirsich and Hurst 1974). When two or more species persistently occur together, particularly in substantial numbers, then this indicates that the external parameters would have been suitable for them all, but this does not mean that they need have been biologically interdependent. Relationships with depth and sediment type Since the work of Ziegler (1965) there has been dispute as to the extent to which the distribution of Llandovery communities is {a) depth related and ( b ) dependent on sediment type. Let us examine these in turn. Critics such as Watkins (1979, p. 250) have pointed to the discrepancy between the interpretation of Ziegler et al. (19686, fig. 13), which shows a smooth gradient across the late Llandovery shelf and communities, and the subsequent sedimentological work of Bridges (1975, fig. I I a), which shows a more varied gradient, and in particular an emergent Longmynd spur in Shropshire at that time. In fact a situation such as the latter was explicitly envisaged by Ziegler et al. (19686, caption to fig. 12). Their fig. 13 was inevitably generalized, but an analysis of the sediments 666 PALAEONTOLOGY, VOLUME 27 and faunas round the Longmynd shows that they do yield progressively more off-shore ecogroups in each section as the transgression proceeded. The best proof that the Llandovery communities were depth-related comes from the studies of eustatic changes in Silurian sea levels (McKerrow 1979; Johnson 1980; Johnson et al. 1981; Colville and Johnson 1982). The mere fact that these studies show that the communities shifted seawards or shorewards synchronously on the various different palaeocontinents indicates conclusively that the water depth and the communities are directly linked, and that the sequence of ecogroups is Collection number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Locality (Ziegler et al. 1968. p.780-1) 24 32 34 35 44 11 4 46 43 1 37 63 59 45 52 53 13 16 12 73 40 68 70 69 54 60 49 58 55 56 Lithology SI SI SA SA S/S SA C/S M SI S/S SI SA M/SI S/S SI SI SI SI S/S S/S S/S SA S/S SA SI M/SI M M M M/SI Community LINGULA EOCOELIA PENTAMERUS STRICKLANDIA CLORINDA LINGULA BIVALVES (all) STEGERHYNCHUS EOCOELIA "TENTACULITES" LEPTOSTROPHIA PENTAMERUS/ PENTAMEROIDES STREPTELASMATID CORALS ATR YPA/PROTATRYPA STRICKLANDIA/ COSTISTRICKLANDI A CLORINDA COOLINIA GLASSIA EOPLECTODONTA * _* *_ * 20°3 ^ 1 ▲ a. V//Y//a777\ Total fossils 373 52 185 119 348 267 324 662 204 273 351 1044 220 300 122 122 126 141 91 522 185 293 330 409 211 163 333 167 360 381 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 | text-fig. 2. Commoner taxa in thirty collections from the late Llandovery of the Welsh Borderland (only taxa occurring as more than 10% in two or more collections are included). The collections are attributed to the various animal communities described in Ziegler et al. (1968a). COCKS AND McKERROW: SILURIAN MARINE BENTHIC COMMUNITIES 667 everywhere the same during late Llandovery time. Of course depth is not directly linked to distance from shore— the width of the community bands seen in the central and eastern United States can be over 200 km per community, a tenfold increase over many of the band widths seen in the Welsh Borderland. The actual depths involved have been debated. Ziegler (1965) calculated from the displacement of communities by lava flows in Pembrokeshire and Gloucestershire that the depth ranges were in Tens of feet rather than hundreds of feet’, but more rigorous field work in the same areas left these conclusions doubtful. Hancock et al. (1974) postulated depths of up to 1500 m for later Silurian communities, although this view was subsequently modified by Hurst (1976) after Shabica and Boucot (1976) had pointed out that the cephalopods were not imploded in the deepest assemblages, indicating that the maximum depth range was probably less than 600 m. We incline to a shallower figure. The distinctive porous coenosteoid structure of heliolitid corals is also found in Recent scleractinian corals (B. R. Rosen pers. comm.), but amongst which it is confined almost exclusively to zooxanthellates, i.e. those corals that are symbiotic with dinoflagellate algae ( Rosen 1981). Zooxanthellate corals do not live in depths greater than 240 m and the vast majority of genera live in water shallower than 100 m (Rosen 1977). In the Silurian, heliolitids are most common in carbonate bioherm environments, but they are a subsidiary element of the tabulate coral fauna (text-fig. lc) in clastic environments, being recorded from the Eocoelia , Pentamerus , and Stricklandia communities up to Collection 23 of text-fig. 1, 2. This suggests that the Stricklandia Community inhabited depths of less than about 200 m, and probably even less than 100 m, leaving only the Clorinda Community as a candidate for deeper water in Llandovery times. Offshore of the Clorinda community the diversity and abundance of shelly benthos drops rapidly (the Marginal Clorinda Community of Cocks and Rickards 1 969). Although it is dangerous to compare the relative widths of community band distribution, such a comparison in the Welsh Borderland indicates that the Clorinda community is unlikely to have occupied substantially more space and width on the sea-floor than the Stricklandia community, and thus a total depth range of not more than 250 m and perhaps less than 1 50 m seems the most likely. If it was greater then the community shifts caused by eustatic changes in sea-level would indicate that the real figures for the rising or lowering of the water would have been improbably high; for example, if the Clorinda community had really occupied a depth band of from 200 m to 600 m, then a 400 m rise in sea-level (which could have been caused by a combination of eustacy and local tectonics) would have been needed to change from the Stricklandia to the Marginal Clorinda communities, a shift seen in many other places as well as the Welsh Borderland. It would seem that this is less probable than the more modest changes needed if it were postulated that the total Lingula to Clorinda community depth range was less than 200 m. Brenchley and Cocks ( 1 982, p. 807) also concluded that the depth spectrum of the latest Ordovician communities found in the Oslo region, Norway, was unlikely to have been much more than 100 m; this figure was based both on an analysis of the sedimentological structures present and also on estimates of the depth differences likely to have been involved in the contemporary glacio-eustatic fall in sea-level. In normal non-glacial littoral and sublittoral environments today the clastic sediments found can be of any grain size from cobbles to mud, but towards the deeper parts of the shelf (in areas not subject to major tectonic activity) the coarser fractions are progressively eliminated such that only silt and mud are common at the outer shelf margin. On text-fig. 2 we have indicated the sediment type for each of the Llandovery collections and these range from conglomeratic sandstone (C/S) through sandstone (SA), silty sandstone (S/S), siltstone (SI), muddy siltstone (M/SI) to mudstone (M). It can be seen that, although there is a higher proportion of mudstones in the Clorinda community collections, nevertheless there is a very poor correlation between sediment type and individual communities, an obvious example being the Eocoelia community which has been found in the widest variety of sediments from conglomeratic sandstones to mudstones. The same applies to individual species and genera, e.g. Stegerhynclms (text-fig. 2). Of course this does not mean that all benthic animal communities are or were substrate independent, but merely that the dominant shelly benthos of Llandovery age, such as brachiopods, corals, and tentaculitids, were mostly epifaunal, and. 668 PALAEONTOLOGY, VOLUME 27 assuming that a suitable spat attachment surface had been found, the individuals were tolerant of a wide range of substrates. Infaunal forms and burrowers would have been more likely to have been sediment specific, but these were not common in Llandovery time, and even lingulids have been found in a wide range of sediment types and grain size. Trilobites, on the other hand, appear to have been more directly linked to particular substrates in some cases, but no trilobite is recorded at more than 5% in the thirty collections ( Warburgella in Collection 30 and Phacops s.l. in Collection 23), and only Encrinurus and Dalmanites are known from many of the shallower-water localities. Some trace fossils are also considered to have been related to bathymetry (Seilacher 1967). It is difficult to assess the effect of varied turbulence on the animals present. The greater the water turbulence the greater the food supply, but the greater the strength needed for pedicle attachments, holdfasts, etc., which was probably a direct factor in the distribution of some of the more common taxa. Most strophomenides, for example, thrived best in lower energy environments, and the large pentamerides were clearly less vulnerable to wave damage in the middle part of the shelf, particularly since their pedicles were no longer functional as adults. The effects of turbidity are better known. Most brachiopods can clean any excess sediment from their lophophores, unlike most bivalves, whose gills become clogged fairly rapidly (Steele-Petrovic 1975). Corals can also survive after fairly turbid episodes, although they will be killed if the sediment covers the polyps completely. Comparison with communities of other ages It is instructive to compare the Silurian palaeoecological regime with those both earlier and later. In the Cambrian, although most communities are dominated by trilobites and are thus related to sediment type, brachiopods appear to have chiefly occupied a single, relatively shallow-water community niche, with the middle to deeper shelf inhabited mostly by other phyla. During the Ordovician there was a gradual spatial expansion of brachiopods, crinoids, and corals. Lockley’s review (1983) suggests in the text that Ordovician communities were very sediment specific; however, the actual data that he presents (1983, text-fig. 6) only loosely bears out his assertion in that just 5 out of 24 communities are found in a single sediment type, and the remaining 19 occur in two or more sediment types, and, moreover, Lockley’s sediment categories are very broad. It seems more probable that, at least in post-Arenig times, the distributions of Ordovician brachiopods were also depth-related in a general way. Certainly by late Ashgill times, the associations were distributed in a regular order across the shelf, as can be demonstrated in the well-preserved regressive sequence seen around Oslo, Norway (Brenchley and Cocks 1982); and because the shelf sediments there are so similar throughout (mostly varieties of lime-mud), some other factor more directly related to water depth must be invoked to account for the community distribution. After the Llandovery there were changes in the community structure, partly due to extinctions of some of the dominant forms, which resulted in changes in the relative abundances of certain families and genera; Calef and Hancock (1974), Hurst (1975), and Watkins (1979) have described these later Silurian communities. During the Devonian, bivalve autecology underwent a dramatic change to exploit many more infaunal as well as semi-faunal and epifaunal habitats. From that time onwards molluscan-dominated benthic communities became much more common that hitherto, and the infauna much more dominant than the epifauna. Some Silurian ecogroups, such as the shallow-water lingulid-rhynchonellid associations, persisted into Carboniferous or even later times, but the complete spectrum of brachiopod-dominated communities across the entire shelf is seldom fully developed after Devonian times. However, the autecology of individual brachiopod genera and species continued to evolve so that they could live in every-increasing absolute water depths, culminating in the abyssal forms known living today; although even now most brachiopods live in water shallower than 200 m. The time from the later Ordovician to the Devonian was that in which the brachiopods were the dominant forms of shelly benthos across the widest variety of habitats, and thus this was the period when brachiopods can be used as indices for the whole spectrum of level bottom clastic communities. It is these communities that were more directly depth related, in contrast to some others in earlier and later times. COCKS AND McKERROW: SILURIAN MARINE BENTHIC COMMUNITIES 669 Acknowledgements. We thank N. J. Morris, B. R. Rosen, and other colleagues for discussion. This paper formed a contribution to the Symposium on Autecology of Silurian Organisms at Glasgow, September 1983 (IGCP Project 53 — Ecostratigraphy). REFERENCES boucot, a. j. 1975. Evolution and extinction rate controls. Elsevier, Amsterdam, 427 pp. brenchley, p. j. and cocks, L. R. m. 1982. Ecological associations in a regressive sequence: the latest Ordovician of the Oslo-Asker district, Norway. Palaeontology , 25, 783-815, pis. 85, 86. bridges, p. h. 1975. The transgression of a hard substrate shelf: the Llandovery (Lower Silurian) of the Welsh Borderland. J. sedim. Petrol. 45, 79-94. calef, c. e. and Hancock, n. j. 1974. Wenlock and Ludlow marine communities of Wales and the Welsh Borderland. Palaeontology , 17, 779-810. cocks, l. r. m. 1967. Depth patterns in Silurian marine communities. Marine Geol. 5, 379-382. 1971. Facies relationships in the European Lower Silurian. Mem. Bur. Rech. geol. minier. 73, 223-227. — and fortey, r. a. 1982. Faunal evidence for oceanic separations in the Palaeozoic of Britain. Jl geol. Soc. Lond. 139, 465-478. and rickards, R. b. 1969. Five boreholes in Shropshire and the relationships of shelly and graptolitic facies in the Lower Silurian. Q. Jl geol. Soc. Lond. 124 [for 1968], 213-238, pis. 9-11. colville, v. r. and Johnson, m. E. 1982. Correlation of sea level curves for the Lower Silurian of the Bruce Peninsula and Lake Timiskaming (Ontario). Can. J. Earth Sci. 19, 962-974. fursich, f. t. and hurst, j. m. 1974. Environmental factors determining the distribution of brachiopods. Palaeontology, 17, 879-900. Hancock, n. j., hurst, j. m. and fursich, f. t. 1974. The depths inhabited by Silurian brachiopod communities. Jl geol. Soc. Lond. 130, 151-156. hurst, j. m. 1975. Wenlock carbonate, level bottom, brachiopod-dominated communities from Wales and the Welsh Borderland. Palaeogeogr. Palaeoclimatol. Palaeoecol. 17, 227-255. — 1976. The depths inhabited by Silurian brachiopod communities. Geology , 4, 709-710. Johnson, M. E. 1980. Palaeocological structure in Early Silurian platform seas of the North American midcontinent. Palaeogeogr. Palaeoclimatol. Palaeoecol. 30, 191-215. — cocks, l. r. m. and copper, p. 1981. Late Ordovician -Early Silurian fluctuations in sea level from eastern Anticosti Island, Quebec. Lethaia , 14, 73-82. lockley, M. G. 1983. A review of brachiopod dominated palaeocommunities from the type Ordovician. Palaeontology, 26, 1 1 1 145. mckerrow, w. s. (Ed.) 1978. The ecology of fossils. Duckworth, London and Massachusetts Institute of Technology Press, 384 pp. — 1979. Ordovician and Silurian changes in sea level. Jl geol. Soc. Lond. 136, 137-145. ROSEN, b. r. 1977. The depth distribution of Recent hermatypic corals and its palaeontological significance. Mem. Bur. Rech. geol. minier. 89, 507-517. — 1981 . The tropical high diversity enigma — the coral’s-eye view. In forey, p. l. (ed.). The evolving biosphere, pp. 103 129. British Museum (Natural History) and Cambridge University Press. seilacher, a. 1967. Bathymetry of trace fossils. Marine Geol. 5, 413-428. shabica, s. v. and boucot, a. j. 1976. The depths inhabited by Silurian brachiopod communities. Geology , 4, 132, 187-9. steele-petrovic, h. m. 1975. An explanation for the tolerance of brachiopods and relative mtolerence of filter- feeding bivalves for soft muddy bottoms. J. Paleont. 49, 552-556. watkins, R. 1979. Benthic community organisation in the Ludlow Series of the Welsh Borderland. Bull. Br. Mus. nat. Hist. (Geol.) 31, 175-280. williams, a. 1951. Llandovery brachiopods from Wales with special reference to the Llandovery district. Q. Jl geol. Soc. Lond. 107, 85-136, pis. 3-7. ziegler, A. M. 1965. Silurian marine communities and their environmental significance. Nature , Lond. 207, 270-272. 1966. The Silurian brachiopod Eocoelia hemisphaerica (J. de C. Sowerby) and related species. Palaeontology, 9, 523-543, pis. 83, 84. 670 PALAEONTOLOGY, VOLUME 27 ziegler, a. m., cocks, L. R. m. and bambach, r. k. 1968a. The composition and structure of Lower Silurian marine communities. Lethaia, 1, 1-27. and mckerrow, w. s. 19686. The Llandovery transgression of the Welsh Borderland. Palaeontology , 11, 736-782. L. R. M. COCKS Department of Palaeontology British Museum (Natural History) Cromwell Road, London SW7 5BD W. S. MCKERROW Manuscript received 1 3 July 1983 Department of Geology and Mineralogy Revised manuscript received 23 January 1984 Parks Road, Oxford 0X1 3PR GROWTH ANALYSIS OF SILURIAN ORTHOCONIC NAUTILOIDS by R. A. HEWITT Abstract. Evidence from orthocone septal strength implies approximate depth limits of 200 m for the near- shore dwelling actinocerids, 500 m for large orthocerids, and no more than 1500 m for the small pelagic orthocerids found in 'graptolitic shales'. These estimates refer to the initial depositional depth of fully septate conchs; not the habitat depth of pelagic species, nor the occurrence of fragmented and reworked specimens. The further interpretation of the autecology of the large orthocerids must be based either on the distantly related, coiled genus Nautilus, or their large coleoid descendants. Studies of the growth rate of large orthocerids test their ecological similarity to large predatory coleoids. Cycles of lirae spacing in Geisonocerina, annulation amplitude in Dawsonoceras , and septal spacing in a variety of Silurian orthocones, show annual periodicities implying growth rates of about 1 00 mm per year. Large conchs with a length of TO to T5 m had a protracted growth phase for over fifteen years, followed by at least one year as a slowly growing mature stage. The adolescent increase in body weight and mantle cavity volume is even less than that of Nautilus, suggesting little ecological similarity to coleoids. Although the depth limits (Table 1) and swimming position of orthoconic nautiloids can be estimated from their shell morphology (Westermann 1973, 1977), it is difficult to interpret other aspects of their autecology. Perhaps the main dilemma results from their greater phylogenetic affinity with descendant lower Devonian to Recent coleoids (Bandel et at. 1983), than the primitive Nautilus. Although Nautilus resembles the coleoids in being a voracious predator (Saunders et al. 1978), it is not well equipped for rapid swimming by jet propulsion, diurnal or other rapid changes in depth, and visual capture of prey (Packard 1972, p. 292; Chamberlain 1981; Ward et al. 1981). Hewitt and Watkins (1980) pointed out that the small body size and mantle cavity volume of most pelagic orthocerids are inconsistent with interpretations involving intelligent and highly mobile, squid-like predators (McKerrow 1978; Gould 1983, p. 249). But the 0-5 to 3 0 m long orthocones found in relatively inshore Silurian facies are consistent with this autecological reconstruction. This latter hypothesis was tested and rejected by an esoteric approach based on the view of Packard (1972) and Chamberlain (1981), that the volumetric increase in the mantle cavity during ontogeny was of paramount importance to cephalopods which capture prey, or avoid the cannibalistic attentions of their larger relatives, through efficient jet propulsion. Although Nautilus moves by jet propulsion, it has not developed the rapid escape reactions of coleoids and presumably obtains protection from an external shell and retiring habits. These and other ecological differences between Recent cephalopods are evident from their growth rates. One Sepia species probably grows a 0-5 m long shell and 10 kg body within two years (Packard 1972); but the tiny mesopelagic Spiru/a has a similar growth period. It grows about twenty chambers per year (deduced from Clarke 1970) compared to up to one per day in Sepia (Choe 1963). Cochran et al. (1981) and Ward et al. (1981) imply that Nautilus grows five to sixteen chambers per year, over a period of two to six years required to attain a body weight of about 0-5 kg. The growth rate and population turnover of the larger squids is sometimes even faster than that of Sepia , but there appears to be little reduction in growth rate due to shell formation (Packard 1972). Small cephalopod species, with inferior or redundant jet- propulsion adaptations, have a slow growth rate. If it can be shown that the large Silurian orthocerids had a growth rate which is equal to, or less than, that of Nautilus, it may be reasonably assumed that the ecological diversity of coleoids only developed after they lost the external shell of their ancestors. [Palaeontology, Vol. 27, Part 4, 1984, pp. 671-677.] 672 PALAEONTOLOGY, VOLUME 27 table 1. Strength indices of Carboniferous phragmocones calibrated by the penultimate chamber of Nautilus to estimate implosion depths. Terminology after Westermann (1977, 1982). The isotropic phosphate connecting rings in Goniatites choctawensis Shumard came from the same shale as brown sparry calcite connecting rings of Bactrites quadrilineatus Girty (Mapes 1979, sample Ml). Rayonnoceras (M21-2 and M26) has a perispatium cemented with pyrite and isotropic phosphate, surrounded by inner sparry calcite and brown prismatic layers. These actinocerid rings were calculated as spherical membranes. The 'horny tubes’ of Buckhorn Asphalt orthocerids are composed of bituminous aragonitic laminae. Despite these differences from the Nautilus ‘horny tube’, there is a correlation between connecting ring and septal strength indices. The proportion of nacre in the shell thickness Sw is nac. %. The Sw and septal spacing A of coiled shells refers to the venter, and their ri is half internal whorl height. Genus * = from Westermann (1982) Connecting ring Septa Shell wall h /Tin r H m 10M r Depth m Depth m Ss fim R mm 1000S R A mm nac. /o mm Ravonnoceras 155 7300 21 260 200 133 22-7 5-9 17-5 1067 67? 20 Ravonnoceras 78 2700 2-9 360 264 68 8-7 7-8 4-7 530 64? 9 Nautilus *74 *480 15-4 955 559 210 12-7 16 5 7-3 440 60 14 Nautilus *150 *1160 12-9 800 800 778 33-0 23-6 260 910 75 48 Bactrites 78 456 17-1 1060 610 178 9-9 180 5-4 325 70? 7-65 ' Pseudorthoceras' 27 157 17-1 1060 712 29 1-4 210 1-3 67 41 1-28 Goniatites 16 89 18-2 1129 7 — — 0-8 34 — 0-38 Mitortlioceras 85 355 23-9 1482 1469 117 2-7 43-4 4-6 440 57 2-26 Goniatites 21 64 30-6 1900 - 6 - — — 54 — 0-22 The Nautilus chambers measured by Denton and Gilpin-Brown ( 1 966) increased in volume at four times the exponential rate of the larger chambers of the 1 - 5 m long Ludfordian orthocerid measured by Hewitt and Hurst (1983). Since the latter taxon (‘ Orthoceras ’ ludense J. de C. Sowerby, resembling "O' alienum Hall) increased the shell volume to balance the weight of both the body and posterior aragonite deposits, it is evident that either the body grew slower than in Nautilus or that over twenty to sixty-four chambers were added per year. But the evidence for a cycle of ten orthocerid chambers per year (Hewitt and Hurst 1983), supported here by additional studies, indicates that they were added at a similar rate to Nautilus. This conclusion that large Silurian orthocones had a slower increase in body weight than Nautilus , is related to the controversy over the use of ornamentation in growth-rate studies (Pannella 1972, 1975; Kahn and Pompea 1978; Pompea et at. 1979; Saunders and Ward 1979; Hewitt and Watkins 1980; Hughes 1981; Doguzhaeva 1982). Orthocone morphology cannot be analysed purely as adaptative autecology, physiology, environmental cycles, or life history. It is, however, difficult to explain the major growth cycles of large orthocerids, except by seasonal growth-rate changes. Since the annual growth cycles of Sepia are not sinuous it is preferable to analyse septal spacing and external lirae spacing by autocorrelation analysis. The method is discussed by Davis (1973, pp. 225-226) and the problems of analysis of nautiloid growth increments are reviewed by Hewitt and Hurst (1983). Briefly, the increment lag with the most significant parameter ZL (which should be greater than 1-96) indicates the likely wavelength or repeat distance of a cycle. The parameter Z, is only valid when the lag is less than one-quarter of the analysed series of growth increments and when there is no trend in the data. The latter can be standardized to remove ontogenetic trends, or larger growth cycles. Specimens are in Bristol Museum (BRSMG), Birmingham University (BU), Greene Memorial Museum (GMM), Milwaukee Public Museum (MPM), Field Museum (PE), Redpath Museum (RM), and McMaster University (S). Shell diameter was calculated from circumference. HEWITT: GROWTH OF SILURIAN ORTHOCONIC NAUTILOIDS 673 CYCLIC VARIATIONS OF LIRAE SPACING Kahn and Pompea (1978) admitted only three exceptions to their speculative thesis that orthocones grew eight to sixteen diurnal external transverse lirae (ridges) per synodic monthly chamber. Two were small aragonitic longicones from the Carboniferous Buckhorn Asphalt, i.e. Mitorthoceras ( = ‘ Orthoceras unicamera Smith’) with forty-five lirae overlying each chamber and ‘ Pseudorthoceras knoxense (McChesney)’ with thirty lirae overlying each chamber. They proposed (p. 608) that these were unusual in being restricted to a ‘shallow shelf and inland sea habitat (50-100 m depth)’. The criteria of Westermann (1973), cited by Kahn and Pompea (1978), shows that Mitorthoceras may have had the deepest-known nautiloid habitat (Table 1) and imply that only imploded or reworked nautiloids occur in strata deposited at depths much in excess of 1 500 m. A series of 315, 103 ^m wide lirae, were studied in a 5 mm diameter Mitorthoceras. They averaged 44-8 per chamber in thin section. Asymmetric lirae of alternating amplitude grade into zones of narrow rounded lirae defined by striae. They originated as spherulitic prisms, radiating outwards from the nacreous layer of the shell wall. These ‘cameral cycles’ have thirty-four to seventy-nine increments (mean 55-4). Mural deposits formed over an interval of five to six cameral cycles (Crick 1982), and consist of 219 layers flanked by an outer translucent zone (180 ^m) of seventeen layers, implying about one layer per lira. The siphuncle ‘horny tube’ has about twenty laminae per chamber. A retreat of the apertural mantle formed a high-amplitude lira passing into a translucent band within the outer prismatic layer. Five minor ‘breaks’ in growth were identified from the occurrence of these bands. They are associated with the narrow rounded lirae; but the retreat structure initiated the growth of asymmetric lirae. The number of lirae between breaks is: 87 + , 4, 25, 8, 71, 4, and 1 15 + . The deep-water orthocerid displays gradational cycles of lirae spacing related to the internal layers of the conch, interrupted by mantle retreat events which may result from migration events. The latter should not be confused with unconformities defined by the pattern of the lirae, resulting from wounds or other local damage to the apertural mantle (Saunders et al. 1978, p. 138), and concentrated in mature growth stages. Geisonocerina wortheni (Foerste) attained a diameter of 89 mm and a length of 1-48 m in the dolomites of the Telychian Brandon Bridge Member of Wisconsin (MPM 26360). A 558 mm series of 903 rounded lirae, defined by oblique external striae, was measured between diameters of 14 and 45 mm (MPM 25357 from Old Burlington Quarry, east side of White River). The lirae do not enlarge in proportion to the shell diameter, implying that fourteen were grown per chamber at the posterior end, compared to forty-five anteriorally. The anterior sutures and 180 mm long apex are missing. Unconformities occurring at an interval of 500 lirae increased lirae spacing. There are significant periodicities of eleven (ZL = 4-5), nineteen (ZL = 4-3), and ninety-four (ZL = 2-6) lirae. The latter are related to eight major cycles of 110 + 30 lirae defined by the variation in average lirae spacing within the shorter ‘cameral cycles’. An average of 6-32 festooned, 98 /.un wide microlirae occur between the striae at the posterior end of this specimen, where there are bimodal cycles of twenty-one lirae (ZL = 2-5) and a probable seasonal cycle of 128 lirae. If the poorly preserved microlirae are semi-diurnal they suggest that the periodicity of the lirae varied from 1-75 to 4-5 days. Offshore Gorstian 1G. recticinctum (Blake) had a maximum diameter of 13 mm and a length of 150 mm. They show a similar ornamentation to Mitorthoceras. The external mould from Llangammarch Wells cited by Hewitt and Watkins (1980, p. 107) has cycles of lirae spacing with a lag of 18 0 (ZL = 3-87 from 109 lirae). The 3 mm long cycles have a diameter of 2 to 4 mm. A series of eighty lirae (Watkins sample 214) show cycles of 31-17 lirae (lags 30 to 32 with ZL = 2-5 to 2-6). The 7 mm long cycles have a diameter of 1 -53-0 mm. They are too long to be ‘cameral cycles’ and give a plausible estimate of the number of solar days per synodic month. 674 PALAEONTOLOGY, VOLUME 27 CYCLES OF SEPTAL SPACING Ludlow age longicones from Sardinia show minor cycles of septal spacing. Sphaerorthoceras (Serpagli and Gnoli 1 977, pi. 4, figs. 2, 5, 6) contains six cycles between diameters of 0-7 and 4 0 mm. The mean wavelength/diameter ratio (4-7, range 3-9) differs from the mean periodicity of 9-27 chambers in showing no increase during ontogeny. Arionoceras affine (Meneghini) has a 30 mm long cycle of nine chambers at diameters of 5 to 10 mm (Serpagli and Gnoli 1977, pi. 6, fig. 4); but A. submoniforme (Meneghini) displays four cycles of four chambers (3-4 mm wavelength) between diameters of 1-3 and 3-4 mm (Serpagli and Gnoli 1977, pi. 7, fig. 2). Typical cycles occur in the Waukesha and Racine Dolomites at Lannon, near Milwaukee. A 24 to 48 mm diameter G. wauwatosense (Whitfield) from the upper 0 04 m of the former horizon showed three cycles of eight chambers (mean length 91-2 mm and about ninety lirae). Protokionoceras from a 2 m ‘Lannon Stone’ section showed a 1 74 mm long cycle of nine chambers (diameter 21 to 25 mm); but more breviconic Kionoceras had six cycles of 41 chambers (32 to 51 mm diameter). Kolebaba (1977, fig. 2) illustrated an asymmetric cycle of twenty-five chambers in a Gorstian Vericeras ambigena (Barrande) (diameter of 10 to 4-5 mm, length 21 mm). An 'O' ludense (Hewitt and Hurst 1983, fig. 4) has an asymmetric cycle of twenty-one, a major cycle of thirty-eight, and minor cycles of ten chambers. If the minor cycles are annual, they imply a growth period of ten to twelve years for 50 mm diameter longicones. This average growth rate is consistent with the growth period of ten years implied for 0-75 m long individuals of G. wortheni by lirae cycles. The 1 -5 m long individuals of the two species were probably fully grown within twenty years and had a body weight of at least 1 kg. SIGNIFICANCE OF ORNAMENTATION Dawsonoceras has oblique annulations, ornamented with festooned transverse lirae forming 70 /xm thick flanges. Lirae spacing is reduced over the thickened crests of the higher amplitude annulations. The annulation wavelength, which is also septal spacing, increased in proportion to the shell diameter until the diameter reached 30 mm, but then remained at 7-10 mm. Hughes (1981) found lirae composed of numerous lamellae. If the striae seen between lirae in annulation troughs define semi- diurnal increments, then these lirae formed in three solar or lunar days within a low-amplitude annulation with nine lirae, compared to 4-5 solar days in high-amplitude annulations with seven lirae. The implied monthly production of annulations was tested by assuming that the cycles of annulation amplitude had a seasonal origin. The diameter of this 185 mm long increment increased from 38 to 50 mm. The twenty-one (BU Holcroft collection 56) annulations have an average of 71 lirae. The spacing of 158 lirae (up to fifteen measurements per lira at x 50) and the distance between the base of each lira and a ruler attached to the annulation crests, were used to calculate their diameter and cross-sectional volume (excluding flanges). The diameter lags of six (ZL = 8-7) and twelve (ZL = 6-6) lirae, corresponded to lags of six (ZL = 6-9), twelve (ZL = 2-8), and thirteen (ZL = 3-3) lirae obtained by volume. Their weighted first harmonic was 6T4. Longer periodicities with ZL over 1-96 have a weighted mean of 84-52 lirae; but the most significant lag is ninety lirae (ZL of 4-4 by volume and 3-0 by diameter). A smoothed moving average showed volume growth maxima separated by a 116 mm length of 97-5 lirae and 13-21 annulations. This seasonal cycle has four low amplitude annulations (35 mm length) with a mean of 9-25 lirae, compared to 6-67 lirae in the remaining high amplitude annulations resulting from optimum growth. Similar cycles were seen in eight Dawsonoceras (Table 2). Racine Dolomite specimens show sixteen other cycles of annulation amplitude, with a mean of 9-50 annulations. Of these only the Dudley specimen (56) encrusted by Halysites and the Sussex specimen from a fissure in a bioherm without corals, grew one annulation per month. Adding together long specimens from these and other localities, we arrive at an estimate of 1 32 annulations between diameters of 3-8 and 57-3 mm. Conch HEWITT: GROWTH OF SILURIAN ORTHOCONIC NAUTILOIDS 675 length increased at a single linear growth rate over the last ninety-three annulations, representing 780 mm out of a total length of 975 mm (another has forty-nine expanding annulations in 427 mm). The existence of cyclic variations in ornamentation restricts the scope of functional interpretations of structural elements, camouflage patterns, and devices for reducing drag by increased surface roughness (Chamberlain 1981, p. 299). A depth limit of 500 m, implied by the septal strength index of Indiana Dawsonoceras (Laurel Member, Flower 1962; 8S/R . 1000 = 15) appears excessive for life over epicontinental carbonate facies. Dawsonoceras from the argillaceous Rochester and middle Elton formations display unusually low annulations and narrow lirae. Thus if there was any advantage in the development of annulations, it was likely to be related to the greater value of camouflage over limestone facies than in dark, turbid environments. The apertural end of a large specimen from argillaceous dolomite (RM 2644) shows a well-defined zone of 109 narrow lirae without annulations. This observation and the general tendency of mature nautiloids to show a greatly reduced septal spacing, suggests that the low amplitude phases of the growth cycles represent a less extreme reduction of growth rate than that associated with maturity. In contrast, the vast majority of coleoids grow rapidly, breed once, and then die (Packard 1972). The 1 m long Dawsonoceras grew at a slow and periodically variable rate for about fifteen years, grew even more slowly when mature, and may have lived longer without increasing body weight. The growth and breeding characteristics of coleoids are not a primitive trait. It is not clear whether the cycles resulted from seasonal migrations between two different environments; or local changes in temperature, hydrography, and food supply through the year. The paucity of abrupt changes in growth and apertural unconformities supports the latter view. A sedentary life as bentho-necton would explain variation in growth cycles and ornamentation between localities (Table 2) and the problematical stratigraphic value of Dawsonoceras "species’ defined by gradational variations in these characters. Flower (1942, 1962) reviews some of these difficulties. This interpretation is consistent with the evidence from lirae spacing and septal spacing in less- ornate orthocerids. Together they imply a mode of life combining the buoyancy of Nautilus and the table 2. Dawsonoceras show variation in their lirae/annulation ratio (L/a on left) which is not a function of mean diameter (dia.) or diameter/annulation wavelength (dm. I A). Cycles defined by L/a were one to four times longer than their diameter (Yr/dia.). Specimens are from: Homerian limestones of Dudley (BU Holcroft Collection 33, 56, 76; Ketley Collection 476); Sheinwoodian shales of Rochester Formation at Lockport (RM ‘1 138'); Sheinwoodian Waukesha Dolomite of Lannon (PE 18943, 19017) and Sussex; Racine Dolomite of Milwaukee (including GMM 12965, 12445, and a 26th Street Quarry specimen); Homerian Lockport Dolomite, all members in Hamilton area (BRSMG Cc829; RM 2644, 2648; S699). n = number of annulations in sample. Sample Annulation sample Seasonal cycles n dia. mm range mm dia. /A Lja range Sample dia. mm Yr/dia. Cycle L a L/a Dudley 57 40 22-57 4-8 8-17 5-18 BU 56 47-5 2-4 97 13 7-4 BU 476 26-5 2-5 81 9 90 Rochester 24 31 26 36 6-2 8-83 5- 12 RM 1 138 28-0 T9 84 10 8-5 RM 1138 315 IT 65 7 9-3 Waukesha 43 29 21-38 3-5 9-21 4-16 Sussex 26 3-8 84 13 6-5 Sussex 34 2-2 103 11 9-4 Racine 22 34 12-51 3-6 1 1 84 7 10 12965 310 2-3 8 7-0 12445 41-5 2-2 10 10-5 26th st. 48-8 1-4 7 110 Hamilton 48 39 19-45 4-4 13-00 9-23 RM 2648 40-6 2-2 112 9 12-4 RM 2644 44-6 1-5 74 5 14-8 676 PALAEONTOLOGY, VOLUME 27 swimming position of coleoids with the ponderous feeding behaviour of some carnivorous gastropods and arthropods. In addition, there were small orthocerid species, which probably drifted with the zooplankton rather than migrate like large squids. Acknowledgements. I thank Dr. M. A. Whyte (Sheffield University) for computing data and reading the manuscript. Dr. D. G. Mikulic and J. Kluessendorf (Illinois Geological Survey) showed me museum collections and exposures. Dr. R. H. Flower, Dr. R. H. Mapes, Dr. M. L. K. Curtis (Bristol Museum), Dr. T. H. Clark (Redpath Museum), Dr. P. M. Sheehan (Milwaukee Public Museum), and Dr. I. Strachan (Birmingham University) loaned to me cited material in their charge. Dr. G. E. G. Westermann read the manuscript and provided financial assistance from his NSERC grant. This paper formed a contribution to the Symposium on Autecology of Silurian Organisms at Glasgow, September 1983 (IGCP Project 53— Ecostratigraphy). REFERENCES bandel, k., reitner, j. and sturmer, w. 1983. Coleoids from the Lower Devonian Black Slate (‘Ekinsriick- Schiefer’) of the Hunsruck (West Germany). Neues Jb. Geol. Palaeont. Abh. 165, 297-417. chamberlain, j. a. 1981 . Hydromechanical design of fossil cephalopods. In house, m. r. and senior, j. r. (eds.). The Ammonoidea. Systematics Association Special volume, no. 18, 289-336. Academic Press London. choe, s. 1963. Daily age markings on the shell of cuttlefishes. Nature , Lond. 197, 306, 307. Clarke, M. R. 1970. Growth and development of Spirula spirula. J. mar. biol. Ass. U.K. 50, 54-64. cochran, j. K., rye., D. M. and landman, N. h. 1981. Growth rate and habitat of Nautilus pompdius inferred from radioactive and stable isotope studies. Paleobiology , 7, 469-480. crick, R. E. 1982. The mode and tempo of cameral deposit formation: Evidence of orthoconic nautiloid physiology and ecology. Proc. N. Am. paleont. Convention , Montreal 1982 , 1, 113-118. davis, J. c. 1973. Statistics and data analysis in geology, 550 pp. Wiley, New York. denton, e. j. and gilpin-brown, j. b. 1966. On the buoyancy of the Pearly Nautilus. J. mar. biol. Ass. U.K. 46, 723-759. doguzhaeva, l. 1982. Rhythms of ammonoid shell secretion. Lethaia , 15, 385-394. flower, r. h. 1942. Cephalopods from the Clinton Group of New York. Bull. Am. Paleont. 27, 124-153. — 1962. Notes on the Michelinoceratida. Mem. Inst. Min. Technol. New Mex. 10, 19-58. GOULD, s. J. 1983. Hen's teeth and horse's toes , 413 pp. Norton, London. hewitt, R. a. and hurst, J. M. 1983. Aspects of the ecology of actinocerid cephalopods. Neues Jb. Geol. Palaeont. Abh. 165, 362-377. — and watkins, r. 1980. Cephalopod ecology across a late Silurian shelf tract. Ibid. 160, 96-117. hughes, w. w. 1981. Shell ornamentation in fossil nautiloids: a problem for Earth-Moon dynamics. Abstr. Progm. geol. Soc. Am. 13, 477. kahn, p. g. k. and pompea, s. m. 1978. Nautiloid growth rhythms and dynamical evolution of the Earth-Moon system. Nature, Lond. 275, 606-61 1 . kolebaba, I. 1977. New information on longitudinally sculptured orthoceroids. Cas. Miner. Geol. 22, 125-138. landman, n. h. 1983. Barnacle attachment on live Nautilus: Implications for Nautilus growth rate. Veliger , 26, 124-127. 1984. Ammonoid growth rhythms. Lethaia , 16, 248. mapes, r. h. 1979. Carboniferous and Permian Bactritoidea (Cephalopoda) in North America. Paleont. Contr. Univ. Kans. Art. 64, I -75, pis. 1-41. mckerrow, w. s. 1978. The ecology of fossils, 384 pp. Duckworth, London. Packard, a. 1972. Cephalopods and fish: The limits of convergence. Biol. Rev. 47, 241-307. pannella, G. 1972. Palaeontological evidence on the Earth’s rotational history since the early Precambrian. Astrophs. Space Sci. 16, 212-237. — 1975. Palaeontological clocks and the history of the Earth’s rotation. In rosenberg, g. d. and runcorn, s. K. (eds.). Growth rhythms and the history of the Earth's rotation , 253-284. Wiley, London. pompea, s. m., kahn, p. g. k. and culver, r. b. 1979. Paleoastronomy and nautiloid growth: A perspective. Vistas Astr. 23, 185-205. saunders, w. b. 1984. Nautilus growth and longevity: evidence from marked and recaptured animals. Science, 224, 990-992. — spinoza, c., teichert, c. and banks, R. c. 1978. The jaw apparatus of Recent Nautilus and its palaeonto- logical implications. Palaeontology, 21, 129-141. HEWITT: GROWTH OF SILURIAN ORTHOCONIC NAUTILOIDS 677 saunders, w. b. and ward, p. d. 1979. Nautiloid growth and lunar dynamics. Lethaia , 12, 172. serpagli, e. and gnoli, m. 1977. Upper Silurian cephalopods from southwestern Sardinia. Boll. Soc. paleont. ital. 16, 153-196. ward, p., carlson, b., weekly, m. and brumbaugh, b. 1984. Remote telemetry of daily vertical and horizontal movement of Nautilus in Palau. Nature , Load. 309, 248-250. — greenwald, l. and magnier, y. 1981. The chamber formation cycle in Nautilus macromphalus. Paleobiology , 7, 48 1 -493. westermann, g. e. G. 1973. Strength of concave septa and depth limits of fossil cephalopods. Lethaia , 6, 383-403. - 1977. Form and function of orthoconic cephalopod shells with concave septa. Paleobiology , 3, 300-321. 1982. The connecting rings of Nautilus and Mesozoic ammonoids: implications for ammonoid bathymetry. Lethaia , 15, 373-384. Manuscript received 8 June 1983 Revised manuscript received 28 September 1983 R. A. HEWITT 12 Fairfield Road Eastwood Leigh-on-Sea Essex SS9 5SB Note added in press. Although Landman (1984) has rightly questioned the assumptions of Doguzhaeva (1982), it would be unwise to base interpretations only on our present knowledge of Nautilus. A recent volte-face by Ward appears to have increased the chance that fossil nautiloids formed one or two growth increments per solar day, as a result of vertical migrations avoiding sunlight (Ward et ai 1984). The well documented migrations across the reef front in Palau and the lateral peregrination of one individual, recorded as 16 km in ten days, indicates that Nautilus is an active member of the bentho-necton. Moreover, the life-span of seventeen to twenty years deduced by a questionable extension of growth rates seen in sub-mature Nautilus (Saunders 1984), is inconsistent with the evidence suggesting that a juvenile, barnacle encrusted Nautilus formed over sixteen chambers in 340 days (Landman 1983). But these debates do not seriously alter the above conclusions about Silurian nautiloids. MODE OF LIFE AND AUTECOLOGY OF SILURI AN-DEVONI AN GRAMM YSI I DAE (BIVALVIA) by L. F. MARSH Abstract. The Grammysiidae are a Palaeozoic family of mainly infaunal anomalodesmatan Bivalvia that lived in very shallow marine environments. Since they were edentulous, the problem of shearing of the valves during burrowing was overcome by the cincture (radial furrows and ribs), which folded the ventral margin. This folded venter provided an active saw during burrowing. Species that burrowed in very high-energy environments showed very marked ventral folding, strong concentric ribs on the anterior end, a fairly large anterior end, and usually a disc-shaped, not highly inequilateral, shell of low inflation. Byssally attached forms reduced the anterior end, expanded the posterior end, had a less folded venter, more elongate shell, lost the strong anterior concentric ribbing, and were more inflated. Shell form and ornamentation is related to the environment and mode of life. Bivalves of the Palaeozoic family Grammysiidae (largely Silurian and Devonian) belong to the subclass Anomalodesmata and, like most forms in this subclass, lack dentition. The main external morphological features of the family are shown in text-fig. 1 . The most important characteristic is the radial cincture, composed of ribs and/or furrows which pass from the umbones to the ventral margin. The only inequality in the valves is in the form of the cincture. This differs in the two valves so that a rib on one valve alternates with a furrow on the other to cause the ventral margin to be folded (PI. 60, fig. 3). The function of this is discussed further below. Most Grammysiidae also possess concentric, bifurcating ribs (text-fig. 1), a lunule, and an escutcheon. By comparison with modern bivalves (and particularly the work on modern bivalves by Stanley 1970), it is possible to ascertain the functional significance of the shell morphology in the Palaeozoic Grammysiidae. MORPHOLOGY AND MODE OF LIFE Since all species of the Grammysiidae show the greatest inflation in the dorsal to central part of the shell, they must all have been, at least partly, infaunal. The relationship between position of maximum inflation and mode of life in modern bivalves is shown by Stanley (1970, p. 27 and fig. 8). The Grammysiidae must also have lived close to the sediment surface, in permanent contact with the sediment-water interface, since none of them possessed a pallial sinus. Many species of Grammy- siidae lived in the littoral zone and all species lived in very shallow water. Interpretation of the life environments are readily confirmed by sedimentary structures and lithologies as recorded, for example, by Potter and Price (1965) and Potter (1967, p. 280). Some species, however, lived in lower energy environments (Holland and Lawson 1963) and were not frequently exhumed. Consequently they appear to be more streamlined for permanent infaunal life. The morphology of some species indicates that they became infaunally or semi-infaunally byssally attached. All members of the Grammysiidae lack dentition. This would have presented problems to an active burrowing bivalve, since the valves would have sheared over each other during burrowing if some assistance to burrowing was not available. I consider that this lack of dentition, associated in most cases with active burrowing, led to the development of the radial and concentric ornament in most of the species. The radial cincture (ribs and furrows) combined with concentric ribbing would have provided an active sawing mechanism to aid rapid, repeated burrowing. Stanley (1970, p. 64) recognized the importance of the function of this type of ornament in modern bivalves. It is of note that the species (Palaeontology, Vol. 27, Part 4, 1984, pp. 679-691, pis. 60-61.) 680 PALAEONTOLOGY, VOLUME 27 that lived in the higher energy conditions, and hence were more likely to be exhumed frequently, have better developed ornament, e.g. Grammysia triangulata (Salter) (PI. 61, fig. 1; text-fig. 3k) and G. grammysioides (Salter) (PI. 61 , fig. 2; text-fig. 3 p). The latter species also shows marked bifurcation of the concentric ribbing as an additional burrowing tool, as do G. sp. nov. d. (Marsh 1976, pp. 359-363, pi. 14, figs. 1-9; pis. 15-18; pi. 19, figs. 1-9) (herein PI. 60, figs. 4, 5; text-fig. 3d) and G. cingulata (Hisinger) (PI. 60, figs. 1 -3; text-fig. 3c). The cincture alone folded the venter (PI. 60, fig. 3) and would have provided a sawing mechanism. Most of the species of this family can be shown to have had a strong ligament to help prevent shearing of the valves, which would have occurred owing to the lack of dentition. Scars made by ligament attachment along the hinge line have been illustrated by Bambach (1971, p. 180, fig. 10). RIGHT VALVE POSTERIOR LEFT VALVE text-fig. 1. Main morphological features of Grainmysiidae. EXPLANATION OF PLATE 60 Figs. 1 5 = ancestral grammysiid species. Figs. 13, Grammysia cingulata, Much Wenlock Limestone, Formation, Dudley, BM L49130, right valve, left valve, and ventral view showing folding produced by the alternation of ribs and furrows of the cincture in each valve. Figs. 4, 5, G. sp. nov. d , Much Wenlock Limestone Formation, Dudley, BU 276, 294, right and left valves. Figs. 6, 7, G. sp. nov. o , Wenlock sandstone, Bryn Craig, Usk inlier, IGS 24217, right and left valves. All x 2. PLATE 60 MARSH, Silurian Grammysia text-fig. 2. Relation of burrowing rate to gross shell shape and shell ornamentation in recent bivalves (from Stanley 1970, fig. 25). 3 2 3 0 - DISC o o + BLADE + 0 CD 2 8- 2 4- Z o 2 2 - 2 0 - 18 - 0 6 - 0 4 - 0 2 - \ \ \ +p +, \ Rapid Burrowers Slow Burrowers ® o + SPHERE h+\ b”A, +J \*n \ . \ + ®CZ) CYLINDER • RAPID SHALLOW BURROWERS a BYSSALLY ATTACHED FORMS -4- CONCENTRIC ORNAMENTATION MORE STRONGLY MARKED ON ANTERIOR END 0 0 - r- 0 5 “I 1 1 I I 1 1 10 1-5 2 0 2 5 3 0 3-5 4 0 LENGTH / HEIGHT text-fig. 3. Relation of inferred burrowing rate to gross shell shape and shell ornamentation in Grammysiidae. MARSH: ECOLOGY OF SILURI AN-DEVONIAN GRAMMYSIIDAE 683 The shapes of the shells of Grammysiidae vary considerably. The shape of a particular species is a reflection of its mode of life and adaptation of the animal to its environment, as produced by natural selection. A change in the environment and, therefore, in the mode of life of a species resulted in a change in morphological features such as the shell form. Species evolved owing to selection pressure produced by environmental changes. In areas where environmental changes occurred speciation was much more rapid than in areas where environmental conditions were stable (Raup and Stanley 1971, p. 99). Bretsky (1973, p. 2090) recognized this in stating that \ . . the degree of morphological distinctiveness at any one time in the fossil record for a particular clearly defined ecological grouping probably reflects . . . the degree of difference between local environmental settings’. A particular species of grammysiid thus survived in one area where the environment did not change and yet was replaced by different species in an area where the environment did change. Stratigraphical ranges of any one species thus vary considerably from place to place, being dependent on the environment. Grammysiid bivalves are therefore very good environmental indicators. It is common to find morphologically similar bivalves developed at different horizons owing to convergent evolution in unrelated species living in similar environments. From the generalized, straight-hinged ancestors G. cingulata and G. sp. nov. r/., differently shaped species evolved as adaptations to different environmental conditions. Disc- and triangular-shaped forms, with concentric ribs much more strongly marked on the anterior end, evolved to cope with very high energy conditions and frequent exhumations. Other forms became byssally attached, lived in rather lower energy conditions and developed more elongated shells, and reduced the anterior end. The relationship of shell form to mode of life in living bivalves was shown by Stanley (1970, p. 60, fig. 25) (herein text-fig. 2). Compare text-fig. 3 for the Palaeozoic Grammysiidae, which shows the positions of the type specimens of each of the grammysiid species. The more rapid burrowers have a high height/inflation ratio and high length/height ratio. They are therefore either disc- or blade- shaped shells. Concentric ornamentation more strongly marked on the anterior end also appears to assist burrowing speed. If, however, the shell shape itself allows extremely rapid burrowing, then the additional concentric ornament appears to be unnecessary, e.g. plot 1 of G. extrasulcata (Salter), an elongate shell with very low inflation. From detailed biometric studies of Silurian-Devonian Grammysiidae, two groups which exhibit distinctive morphological features have been recognized. These are interpreted as follows: Group (i). Rapid, shallow burrowers (text-fig. 4). Group (ii). Semi-infaunal, byssally attached species (text-fig. 5). Group (i) The shallow-burrowing Grammysiidae living in very high energy environments (where exhumation from the sediment during life is commonplace and, therefore, the ability to carry out rapid, repeated burrowing throughout life is a necessity) show a combination of the following morphological features (text-fig. 4): (a) A shell that is not highly inequilateral, to aid ease of rocking from side to side during burrowing. (b) A shell shape that is discoidal or triangular, or very elongate and of very low inflation. (c) A hinge line that is either strongly reclining towards the venter or parallel to the venter, not divergent (text-fig. 6, e.g. G. triangulata). (d) The greatest height of the shell at the umbones. ( e ) Low inflation. (/) High height/inflation ratio. (g) A large anterior end, to allow ease of penetration into the sediment as burrowing was initiated (text- fig. 7), e.g. G. triangulata to contrast with the byssally attached G. obliqua. (h) An anterior end furnished with strong concentric ridges (for sawing) that are either absent or are only very weakly marked on the rest of the shell. (/) A well-developed, strongly marked cincture that causes a very marked folding of the entire venter, to act as a sawing mechanism. (j) The valves usually found disarticulated and frequently broken. 684 PALAEONTOLOGY, VOLUME 27 text-fig. 4. Typical morphological features of the shell developed in rapid, shallow burrowing Grammysiidae, e.g. Grammysia triangulata. EXPLANATION OF PLATE 61 Fig. 1, Grammysia triangulata , Whitclillian, Kirkby Moor Flags, Benson Knot, Kendal, Cumbria, a rapid burrower in very high energy conditions, IGS 12487, left valve. Fig. 2, G. grammysioides, derived fossil in Triassic Budleigh Salterton Pebble Bed, SE Devon, specimen derived from Lower Siegenian, Devonian of Brittany, BM LI 5, 821, a right valve, another rapid burrower. Figs. 3-6, G. obliqua , Ludlow, Usk inlier, a byssally attached species showing oblique cincture, very small anterior end, and greatest height of shell at posterior end. Fig. 3, IGS 12468, holotype, right valve. Fig. 4, NMW G482, left valve. Fig. 5, IGS G Sb 4225, dorsal view, showing dorso-posterior gape and scars of attachment of the strong external ligament along the escutcheon. Fig. 6, BM L 5438 (a) antero-ventral view, with anterior end uppermost, showing narrow antero-ventral byssal gape (bg) and folded venter (fv) near posterior end, produced by the alternation of the cincture in each valve. All x 2. PLATE 61 MARSH, Silurian Grammysia 686 PALAEONTOLOGY, VOLUME 27 Group ( ii ) Byssally attached infaunal or semi-infaunal Grammysiidae show a combination of the following morphological features (text-fig. 5): (a) An elongate shell. ( b ) A highly inequilateral shell. (c) A hinge line that is strongly divergent from the venter towards the posterior end (text-fig. 6, e.g. G. obliqua). (d) The greatest height of the shell at the posterior end of the hinge line. ( e ) High inflation. (/) Low height/inflation ratio. SHELL HIGHLY INEQUILATERAL VIEW OF LEFT VALVE Hinge line strongly divergent from venter towards posterior CROSS SECTION TO SHOW INFLATION • HEIGHT DORSAL VENTRAL He ight Inflation = 10 - 11 text-fig. 5. Typical morphological features of the shell developed in semi- infaunal byssally attached Grammysiidae, e.g. Grammysia obliqua. MARSH: ECOLOGY OF SILURIAN DEVONIAN GRAMMYSIIDAE 687 (g) A greatly reduced anterior end (text-fig. 7, e.g. G. obliqua). (/?) The small anterior end marked with fine concentric ridges that are not better marked on the anterior end than they are on the rest of the shell. (/') A reduced, very obliquely marked cincture composed only of furrows, which causes a folding only of the extreme posterior end of the venter. (j) An expanded posterior end. ( k ) An anterior-ventral byssal gape. (/) The valves usually found closed and articulated. G. obliqua McCoy illustrates all the above features (PI. 61, figs. 3-6). SPECIAL ADAPTATIONS Some specialized adaptive morphology has been noticed in certain species. The byssally attached G. obliqua (text-fig. 3 r) evolved from G. cingulata( text-fig. 3c), a moderately active shallow burrower, via a new species G. sp. nov. o. (text-fig. 3o) (Marsh 1976, pp. 405-41 1, pis. 34-36; herein PI. 60, figs. 6, 7) which became byssally attached. From text-fig. 3 it can be seen that both the byssally attached species are much more inflated than active burrowing species like G. triangulata(k). Even G. cingulata (c), the ancestral species, would have had only a modest burrowing speed as indicated by its plot on the graph in text-fig. 3. G. sp. nov. o. (text-fig. 8) shows certain changes in its morphology (compared with G. cingulata) concurrent with the acquisition of an adult byssus. The anterior end is reduced and the cincture is more oblique. Certain of the ancestral features, however, remain, e.g. the straight hinge line. 80 - 75 - 70- 65- 60 - 55 -I G cingulata G sp nov d 80- 75- 70 65 60- 55- 50 45 40 G sp. nov. / G triangulate G t-xtrasulcata text-fig. 6. The angle in degrees between dorso-posterior and ventro-posterior in six grammysiid species. No. of specimens No. of specimens PALAEONTOLOGY, VOLUME 27 Length of beaks from anterior end/Length x 100% G. cingulata G. sp. nov. d G. obliqua 45 50 "T" 55 text-fig. 7. The length of the beaks from the anterior end as a percentage of the total length in six grammysiid species. eg DUDLEY eg CARDIFF eg. USK 8 eg USK TORTWORTH Grammysia Grammysia Gen. nov. Grammysia Grammysia cingulata sp. nov. d sp. nov. sp. nov.o sp. nov. f text-fig. 8. Grammysiid life position and environmental zonation in the upper Wenlock (Homerian). MARSH: ECOLOGY OF SILURIAN DEVONIAN GRAMMYSIIDAE 689 text-fig. 10. Grammysiid life position and environmental zonation in the Ludlow (Whitcliffian) of the Lake District, England 690 PALAEONTOLOGY, VOLUME 27 The umbonal ridge (text-fig. 8) running from the umbo to the postero-ventral angle marks off a much less-inflated dorso-posterior slope to the shell. I believe that the dorso-posterior slope would have been above the sediment in life position and that the umbonal ridge marked the sediment-water interface. I do not believe that the posterior furrow of the cincture marked the sediment-water interface, nor that the whole of the posterior end of the shell protruded from the sediment as Bambach (1971, p. 175, fig. 9) suggested for ‘G. obliqua (although the species he discussed was the very similar, but more highly evolved species G. acadica Billings). The two posterior furrows of the cincture are not in the same position on the two valves and, therefore, if the posterior furrows did represent the sediment-water interface, then this interface would have been at different levels on the two sides of the specimen. The umbonal ridge therefore probably represents the sediment-water interface in life and only the extreme dorsal posterior end protruded from the sediment. It is, however, certainly possible that from time to time some of the sediment was removed from above the bivalve by current, tidal, and wave action, and indeed the nature of the sediments suggest this. With the frequent inwashes of sediment that occurred, G. sp. nov. o must have been able to retain contact with the sediment surface. It seems likely that it did this by changing the angle at which it lived relative to the sediment-water interface. As suggested by Bambach (1971, p. 178) for ‘ G . obliqua , G. sp. nov. o probably normally lived at an angle of about 40° to the sediment-water interface. If it was inundated with sediment, I believe that it could have increased this angle by action of the foot and use of the cincture and concentric ridges, to maintain contact between the siphons and the sediment-water interface (text-fig. 8). It could have done this without damage to the byssus (Marsh 1976, p. 410, fig. 67). Certainly in the high-energy environment in which it lived, G. sp. nov. o must have been able to cope with small influxes of sediment that would have occurred from time to time. At other times erosion took place and this species then returned to its more normal life position. The shell of G. sp. nov. o therefore shows special adaptation to a semi-infaunal byssally attached mode of life in an environment of moderately high energy. The descendent species G. obliqua probably retained a cincture to pull itself down into the sediment from time to time to avoid exhumation (Marsh 1976, pp. 421, 422) and to avoid complete inundation when moderate rates of sedimentation occurred, in a similar way to G. sp. nov. o. General life positions are illustrated in text-figs. 8-11. text-fig. 1 1. Grammysiid life position and environmental zonation in the topmost lower Downton of South Central Wales. MARSH: ECOLOGY OF SILURIAN DEVONIAN GRAMMYSIIDAE 691 Acknowledgements. I acknowledge the receipt of an ILEA Research Assistantship, during the tenure of which much of the research for this paper was carried out. Thanks are expressed to Miss J. E. Bruch (Exploratory Division, British Gas) for production of the originals of some of the figures. I also thank Dr. J. F. Potter (Farnborough College of Technology) and my husband. Dr. R. C. Marsh (South London College) for critically reading the manuscript. This paper formed a contribution to the Symposium on Autecology of Silurian Organisms at Glasgow, September 1983 (IGCP Project 53— Ecostratigraphy). Specimens are housed in the British Museum (Natural History), London (BM), Institute of Geological Sciences, London (IGS), Birmingham University (BU), and the National Museum of Wales, Cardiff (NM W). REFERENCES bambach, r. k. 1971. Adaptations in Grammysia obliqua. Lethaia , 4, 169-183. bretsky, p. w. 1973. Evolutionary patterns in Palaeozoic Bivalvia-documentation and some theoretical consideration. Bull. geol. Soc. Am. 84, 2079-2095. Holland, c. H. and lawson j. d. 1963. Facies patterns in the Ludlovian of Wales and the Welsh Borderland. Lpool Manchr geol. J. 3(2), 269-288. marsh, L. F. 1976 'British Grammysiidae’. Ph.D. thesis (unpubl.), Univ. Leics. 2 vols., i-xxx, 1-512, 46 pis. potter, j. f. 1967 Deformed micaceous deposits of the Llandeilo Region, South Wales. Proc. Geol. Assoc. 78, 277-288. and price j. h. 1965. Comparative sections through rocks of Ludlovian-Downtonian Age in the Llandovery and Llandeilo districts. Ibid. 76, 379-402. raup, D. M. and Stanley, s. M. 1971. Principles of Palaeontology , W. H. Freeman, 388 pp. salter, j. w. 1848. Palaeontological appendix to Phillips, J. The Malvern Hills compared with the Palaeozoic district of Abberley, Woolhope, Mayhill, Tortworth & Usk. Mem. geol. Surv. ( G.B. ), 2, pt. 1. 331-386, 30 pis. Stanley, s. m. 1970. Relation of shell form to life habits in the Bivalvia (Mollusca). Mem. geol. Soc. Am. 125, 1-292, 40 pis. LINDSAY F. MARSH Geology Section Department of Physical Sciences Manusciipt received 21 May 1983 South London College, Knights Hill Revised manuscript received 22 January 1984 London SE27 0TX JANEIA SI L U RICA, A LINK BETWEEN NUCULOIDS AND SOLEMYOIDS (BIVALVIA) by LOUIS LILJEDAHL Abstract. Janeia silurica Liljedahl, 1984 from the Silurian of Gotland has unusual characters in common with the deposit-feeding nuculoids and the systematically controversial solemyoids, generally considered to have a life habit intermediate between deposit- and suspension-feeding but which in fact lives in symbiosis with chemoautotrophic bacteria. The extensive silicified material available is occasionally extremely well-preserved, reflecting soft-part anatomy of the muscles of the foot as well as of the mantle. J. silurica , which has conspicuous traces of the pallial muscles of the mantle (fused margins?), probably had an efficient system of cleaning the mantle cavity, a typically solemyoid feature. However, it was probably a more active burrower than the extant Solemya since it presumably had a larger foot and smaller gills. The gills were thus used for respiration only, as in nuculoids, while the inferred character of the mantle indicates an evolutionary trend towards the solemyoid life habit. In the animal kingdom the suspension-feeding lamellibranch bivalves are regarded as being the most successful exploiters of the almost inexhaustible primary biomass of the oceans, i.e. the plankton supply. The strikingly expanded ctenidia of these bivalves, which are used exclusively for food collecting, occupy most of the mantle cavity and the individual filaments are far more numerous than the protobranch ctenidia of deposit-feeders (nuculoids). Each filament is considerably longer than the short protobranch filament and is thus by far more efficient. In contrast to the lamellibranch ctenidia the protobranch ctenidia are used mainly for respiration. The affinities of Solemya have always been controversial. Gross soft-part morphology points to a relationship with the nuculoids (e.g. Yonge 1941 , 1959) but specialization along different lines makes its systematic position uncertain (e.g. Purchon 1978). The life habit of Solemya is poorly understood, considered by some to be intermediate between the deposit- and suspension-feeding mode of life (e.g. Yonge 1941) and by others as a suspension-feeder (e.g. Stanley 1970). However, recent discovery of procaryotic symbionts in S. velum (Cavanaugh et al. 1 98 1 ) in combination with the fact that the gut of solemyoids is extraordinarily minute, and in some species even completely absent (Reid and Bernard 1980), suggests that these bivalves obtain nutrient mainly through symbiosis with chemoautotrophic bacteria. Morphological comparison of Janeia silurica with nuculoids and solenyoids The silicified Wenlock fauna at the locality Mollbos 1, Gotland, Sweden contains eleven bivalve species, one of which is J. silurica Liljedahl, 1984 (see Liljedahl 1983, 1984 (in press)). This species is represented by more than 500 valves containing extraordinarily well-preserved specimens, some of which exhibit traces of the soft parts. The discovery of J. silurica extended the known stratigraphical range of the solemyoids to the late Early Silurian, the oldest previously known representative being of mid Devonian age (Newell 1 969). Pojeta (1978, p. 231, fig. 4) considered an Ordovician nuculoid as showing solemyoid characters (described by Eichwald 1880, p. 991, pi. 39, fig. 10, as Nucula aedilis) with the expanded anterior part of the shell and the enlarged anterior adductor muscle scar being the most typical features. I agree that the lateral outline of this species is possibly reminiscent of the solemyoid form (but see also the close similarity in shape with the nuculoid? Dystactella subnasuta Hall and Whitfield; see McAlester 1 968, p. 27, pi. 5). However, since dentition and other internal features are unknown, it is necessary to await the discovery of better-preserved material to understand its systematic position. Pojeta’s ( 1 978) [Palaeontology, Vol. 27, Part 4, 1984, pp. 693-698.] 694 PALAEONTOLOGY, VOLUME 27 second example of a nuculoid showing solemyoid features is Ctenodonta nasuta (Hall). It is elongate, has a slightly expanded anterior part but a posterior end which is longer than the anterior one (see Salter 1859, pi. 8, fig. 2 and Pojeta 1971, pi. 4, fig. 10). J. silurica shows certain similarities to the solemyoids combined with a number of primitive features characteristic of extant Nuculoida. The shell of J. silurica is thin, as in the extant Solemya, and elongate with an extended anterior part (text-fig. 1:1, 2, 3). It is not, however, as strictly cylindrical as that of the extant Solemya. The shell of Solemya gapes at both ends while in J. silurica the valve margins fit together except for the dorsal margin, the left valve overlapping the right (text-fig. 1 : 5 and 3). Like the extant Solemya , the ligament of J. silurica consists of a posterior internal/external part and an anterior external part (text-fig. 1 : 1, 4, 6). The ligament construction was probably strong to judge by the unusually large ligament area (text-fig. 2b). text-fig. I . Janeia silurica Liljedahl, 1984. 1, External dorsal view of articulated specimen, anterior to the right. SGU 3426/3427, x 1.8, sample G77-29LJ. 2, External lateral view of articulated specimen, anterior to the right. Note overlapping left valve, same specimen as in 1 , x 1 .8. 3, Internal lateral view of a right valve (holotype). Note conspicuous traces of pallial muscles. SGU 3608, x 2.4, sample G79-79LJ. 4, Internal lateral view of umbonal part of a right valve. Note chondrophore extending above beak. SGU 3318, x 6.1, sample G77-28LJ. 5, Vertical section of an articulated specimen. Note silicified possible folded in ventral margin of the mantle, SGU 3592/3593, x 4.4, sample G79-78LJ. 6, Posterior view of beak and chondrophore, same specimen as 4, x 7.8. All specimens are in the Type Collection of the Geological Survey of Sweden. The material was coated with ammonium chloride prior to photography. B Janeia silurica Solemya togata text-fig. 2. Muscular impressions (densely stippled) and ligament area (loosely stippled), a, Yoldia limatula! : 1 = anterior adductor muscle. 2a, lb = anterior pedal protractor muscles. 3 = anterior pedal retractor muscle. 4 = pedal elevator muscle (dorsomedian muscle). 5 = posterior pedal retractor muscle. 6 = posterior adductor muscle; note conspicuously large posterior pedal retractor muscle (No. 5). After Heath 1937, pi. 10, fig. 83. b, Janeia silurica Liljedahl: 1 = anterior adductor muscle scar. 2a, 2b = anterior pedal protractor muscle scars. 3 = anterior pedal retractor muscle scar. 4 = pedal elevator muscle scar. 5 = posterior pedal retractor muscle scar. 6 = posterior adductor muscle scar, c, Solemya togata Poli: 1 = anterior adductor muscle. 2 = anterior pedal protractor muscle. 3 = anterior pedal retractor muscle. 4 = pedal elevator muscle. 5 = posterior pedal retractor muscle. 6 = posterior adductor muscle. After Pelseneer 1891, pi. 9, fig. 15, ligament extension after Owen 1959, p. 217, fig. 3c. 696 PALAEONTOLOGY, VOLUME 27 The muscular impressions of J. silurica (text-fig. 2b; see also Liljedahl 1984, fig. 15) are deeply incised, their distribution much resembling that of extant nuculoids (text-fig. 2a; cf. Heath 1937), similarities with Solemyci also being evident (text-fig. 2c; cf. Pelseneer 1891). The adductor muscle scars are almost equal in size (cf. reduced posterior adductor muscle in extant Solemyci in Newell 1 969, p. N242). The anterior pedal protractors are absent or minute in Solemya (2 in text-fig. 2c) but well developed in J. silurica (2 a in text-fig. 2b). Scar 2b of J. silurica possibly also indicates a pedal protractor muscle since it has a position similar to the most anterior pedal muscles in nuculoids which function as pedal protractors (2a, 2b in text-fig. 2a). Scar 3 in text-fig. 2b of J. silurica may be homologous with the anterior pedal retractor in nuculoids (3 in text-fig. 2a) while scar 4 in text-fig. 2b of J. silurica probably corresponds to the pedal elevator in both nuculoids and solemyoids (4 in text- fig. 2a, c). The posterior pedal retractor muscle scar of J. silurica (5 in text-fig. 2b) is minute as in solemyoids (5 in text-fig. 2c) in contrast to the extremely large posterior pedal retractor muscle of the nuculoids (5 in text-fig. 2a). Thus the pattern of pedal muscle scars and the greatly extended anterior part of the shell of J. silurica together suggest that the foot was correspondingly enlarged occupying more than half the mantle cavity, and that it protruded from the anteroventral part of the shell. Accordingly it must have had a function similar to that in Solemya (cf. Drew 1900; see also Liljedahl 1984, fig. 15 showing a reconstruction of the foot of J. silurica). The conspicuously deep and broad traces of the pallial muscles of J. silurica , which are less accentuated in the posterior and anterior extremities, indicate that the radial muscles of the mantle edge were unusually strong (text-fig. 1 : 3). Because of their uneven impression these scars also suggest partially fused mantle edges (cf. Solemya in Drew 1900, p. 264, figs. 9, 10, 11, 12) with one anterior and one posterior opening. If these assumptions are true, then this species was probably able to withdraw the ventral margins of the mantle with great force, perhaps as in Solemya (see silicified possible replica of an infolded ventral margin of the mantle in text-fig. 1:5; cf. Drew 1900, pp. 264, 265). The anteriorly extended shell and the distribution and size of the impressions of the pedal muscles of J. silurica indicate that the foot occupied more than half the mantle cavity, thus leaving about a third of the cavity for the ctenidia (cf. ctenidia of solemyoids occupying about half the mantle cavity in Yonge 1941, p. 93). If this assumption holds, then the ctenidia of J. silurica were probably not involved in ingestion (as they are to some extent in Solemya; Yonge 1941, p. 116) but were used for respiration only, as in nuculaceans (Yonge 1941, pp- 115, 143). The shape of the shell with its posterior umbonal slope and with the position of the posterior adductor muscle scar, make a strictly posterior exhalant current, as in Solemya , improbable in J. silurica , the direction presumably being ventroposterior. Thus, unlike Solemya it probably did not have free-swimming ability (cf. Drew 1900). Ecology of extant nuculoids and solemyoids Extant nuculoids live in soft muddy to sandy bottoms, moving about only sparingly (Yonge 1941, pp. 81, 82). They collect food by means of the extended labial palps, or proboscides, which when text-fig. 3. Vertical section of umbonal region of Janeia silurica showing chondrophore of right valve and reconstructed ligament (stippled). LILJEDAHL: SILURIAN BIVALVIA, GOTLAND 697 protruded between the opened valves collect and guide food particles from the sediment to the mouth (Yonge 1941, p. 114). The protobranch ctenidia, the main function of which is respiration, are fairly small, and though they may possibly be involved in feeding they cannot be compared with those of the lamellibranch bivalves in this respect (Yonge 1941, p. 1 15). Extant Solemya thrives in rather firm, sandy mud in which it digs itself down in a Y-shaped burrow (Stanley 1970) where it usually stays for the greater part of its time (Yonge 1941, p. 96). The protobranch ctenidium of this genus has a surface comparable with that of a lamellibranch ctenidium and the labial palps are small, not reaching beyond the mantle margins, i.e. they are not able to collect food outside the shell (Yonge 1959, p. 213). Solemya inhales water anteriorly by the actions of muscles. The water is heavily laden with suspended sediment, and food particles are transferred by the labial palps from the ctenidium to the mouth (Yonge 1959). The gut of Solemya is of such extremely reduced size that it alone cannot possibly provide enough nutrient material (Allen and Sanders 1969, p. 388). However, as mentioned above its paradoxically small gut (absent in some species) is probably compensated by its symbiosis with chemoautothrophic bacteria in sulphide-rich levels of the sediment. Possibly Solemya uses the lower part of its Y-shaped habit as a connection with its symbionts (suggested by L. Jeppsson). Autecology of Janeia silurica The bivalve fauna at Mollbos is dominated by deposit-feeders (90%, Liljedahl 1984) probably reflecting the high silt-clay content of the carbonate sediment (cf. Sanders 1958, 1960). J. silurica comprises approximately 20% of all deposit-feeding bivalves and is the third commonest species. There are facts in favour of niche diversification between the different deposit-feeders at this locality (Liljedahl in press). J. silurica is believed to have inhabited a somewhat deeper level in the sediment than the remaining deposit-feeders (Liljedahl 1984, fig. 34). It may have lived symbiotically with chemoautotrophic bacteria at a sulphide-rich level of the bottom (like S. velum , see Cavanaugh et al. 1981) where it did not have to compete for food with other species (cf. discussion above of extremely reduced size and absence of gut in Solemya). The hypothesis that J. silurica was not as specialized a passive suspension-feeding deposit-feeder as the extant Solemya but instead was an active burrower, is supported by the assumed differences in the size of the foot and ctenidia of this species and Solemya. The absence of a gape in J. silurica would have given this species an advantage while burrowing, in preventing excessive sediment from entering the mantle cavity. The almost stationary Solemya , on the other hand, has no need of completely closed shell margins. Instead the valves gape at each end, which is advantageous since it can inhale and exhale with the valves closed. The assumed robust ligament construction of J. silurica , in combination with the edentulous hinge and deeply incised, large, subequal adductor muscle scars, may have made it possible for this species to close the valves quickly and vigorously as in Solemya (cf. Yonge 1941, p. 115) and with greater force and speed than in nuculoids. These have interlocking hinge teeth which are considered a hindrance to such rapid contraction. The conspicuous impressions of the pallial muscles, as well as the inferred morphology of the foot and accessory muscles, suggest that J. silurica could infold the (fused?) ventral margin of the mantle and simultaneously withdraw the foot, possibly in much the same way as in Solemya (see Yonge 1941, p. 136). Thus, J. silurica probably had a system for removing indigestible matter from the mantle cavity, perhaps more effective than in extant nuculoids (cf. living nuculoids; Yonge 1941, p. 83). Acknowledgements. I am indebted to a number of people for valuable help. Lennart Jeppson, the late Anders Martinsson, Kent Larsson, and Sven Laufeld examined and improved the manuscript. Christin Andreasson assisted with the drawings, Margaret Greenwood-Petersson gave linguistic help, and Sven Stridsberg took the photographs. Most of the work was carried out at the Department of Historical Geology and Palaeontology, Lund and part of it at Allekvia Research Station, Gotland. This paper formed a contribution to the Symposium on Autecology of Silurian Organisms at Glasgow, September 1983 (IGCP project 53 Ecostratigraphy). 698 PALAEONTOLOGY, VOLUME 27 REFERENCES allen, j. a. and sanders, h. l. 1969. Nucinella serrei Lamy (Bivalvia: Protobranchia), a monomyarian solemyid and possible living actinodont. Malacologia , 7, 381-396. CAVANAUGH, C. M., GARDINER, S. L., JONES, M. L., JANNASCH, H. W. and WATERBURY, J. B. 1981. Procaryotic Cells in the hydrothermal vent tube worm Riftia pachyptilia Jones: possible chemoautothrophic symbionts. Science , 213, 340-342. drew, G. a. 1900. Locomotion in Solenomya and its relatives. Anat. Anz. 17, No. 15, 257-266. eichwald, E. 1860. Lethaea rossica ou paleontologie de la Russie. Vol. 7, pt. 2. 681-1657. Stuttgart, E. Schweizerbart. heath, h. 1937. The anatomy of some protobranch mollusks. Mem. Mus. r. Hist. nat. Belg. 2e Ser. 10, 1-26. liljedhl, l. 1983. Two silicified Silurian bivalves from Gotland. Sver. Geol. Unders. C, 799, 51 pp. 1984. Silurian silicified bivalves from Gotland. Ibid. C, 804, 82 pp. — (in press) Ecological aspects of a silicified bivalve fauna from the Silurian of Gotland. Lethaia , 1984. mcalester, a. l. 1968. Type species of nuculoid bivalve genera. Mem. geol. Soc. Am. 105, 143 pp. Newell, N. d. 1969. Systematic descriptions. In moore, r. c. (ed.). Treatise on Invertebrate Palaeontology, Part N, Mollusca 6, Bivalvia 1, 489 pp. Geol. Soc. Am. and University of Kansas Press. owen, G. 1959. The ligament and digestive system in the taxodont bivalves. Proc. malacol. Soc. Lond. 33, 215-223. pelseneer, p. 1891. Contribution a l'etude des lamellibranches. Archs Biol., 11, 147-312. pojeta, j. jr. 1978. The origin and early taxonomic diversification of pelecypods. Phil. Trans. R. Soc. Lond. B, 284, 225-246. purchon, r. d. 1978. An analytical approach to a classification of the Bivalvia. Ibid. 425-436. reid, r. g. b and Bernard, f. r. 1980. Gutless bivalves. Science, 208, 609-610. Salter, j. w. 1 859. Fossils from the base of the Trenton Group. Geol. Survey Canada, Figures and Descriptions of Canadian Organic Remains , decade 1, 47 pp. sanders, H. L. 1958. Benthic studies in Buzzards Bay. 1. Animal-sediment relationships. Limnol. Oceanogr. 3, 245-258. — 1960. Benthic studies in Buzzards Bay. 3. The structure of the soft-bottom community. Ibid. 5, 138-153. Stanley, s. m. 1970. Relation of shell form to life habits of the Bivalvia (Mollusca). Mem. geol. Soc. Am. 125, 296 pp. stasek, c. r. 1972. The molluscan framework. In florkin, m. and scheer, b. t. (eds. ). Chemical zoology, Vol. 7, Mollusca, 1-44. yonge, c. m. 1941. The protobranchiate Mollusca: A functional interpretation of their structure and evolution. Phil. Trans. R. Soc. Lond. B, 230, 79-147. — 1959. The status of the Protobranchia in the bivalve Mollusca. Proc. malacol. Soc. Lond. 33, 210-214. LOUIS liljedahl Department of Historical Geology and Palaeontology Solvegatan 13 Manuscript received 13 July 1983 S-223 62 Lund Revised manuscript received 14 January 1984 Sweden AUTECOLOGY AND DISTRIBUTION OF THE SILURIAN BRACHIOPOD DUBARIA by BRIAN JONES and JOHN M. HURST Abstract. Smooth-shelled atrypoid brachiopods, Dubaria varians (Poulsen, 1943) and Dubaria sp. nov., occur in Llandovery strata of western North Greenland in cryptic habitats associated with biostromes and reefs. This ecological preference probably explains their patchy spatial and temporal distribution. It may explain the stratigraphic distribution of Dubaria since other species in the genus also occur in association with reefs. External morphological features of Dubaria are comparable with the morphological adaptions of a species of Atrypoidea which lived in direct association with reefs in the upper Silurian of Arctic Canada. AuTECOLOGYcan be important to biostratigraphy since the interactions between single species and their environments may ultimately control their stratigraphic distribution. Inherent to the topic is the establishment of the environment in which the species lived, for without that knowledge it is impossible to examine the autecology of the species concerned. In some instances the nature of the environment has been inferred from the fossils themselves, a practice that ultimately leads to circular arguments once the autecology of the species is considered. The Silurian-Devonian brachiopod Dubaria provides an excellent example of how the knowledge of autecology assists in the understanding of the stratigraphic distribution of a genus or a species. Boucot and Perry (1981, p. 212) argued that the low diversity Dubaria Community inhabited a quiet-water environment since the brachiopod shells are virtually always articulated and show no sign of abrasion. Thus, it might be expected that Dubaria would be a common element of Silurian-Devonian faunas for there is ample evidence that quiet-water environments were common in that time period. This is especially true in the light of the comment by Johnson et al. (1978, p. 803) that \ . . Dubaria is more of a biofacies indicator than a biochronological tool . . However, as demonstrated in this paper, Dubaria evidently had particular ecological needs that served to restrict its occurrence to a well-defined ecological niche. This paper utilizes occurrences of Dubaria on a world-wide basis (Table 1 ) and from western North Greenland (Table 2) to outline the ecological factors that controlled the distribution of the genus. DUBARIA IN NORTH GREENLAND Spatial and temporal distribution Dubaria were found at ten localities in North Greenland (text-fig. 1 and Table 2). Although strata between these localities were examined in detail, no further specimens of Dubaria were found. The stratigraphic distribution of Dubaria is similarly patchy; although it has been recorded at five different horizons in the Llandovery of North Greenland, strata between those horizons apparently contain no specimens of the genus (text-fig. 2 and Table 2). This patchy spatial and temporal distribution suggests that other factors such as ecological requirements probably controlled the distribution of Dubaria. Ecological distribution At Kap Schuchert, D. varians and Dubaria sp. nov. occur in the uppermost part of the Alegatsiaq Fjord Formation in direct association with small, localized biostromes that are surrounded by (Palaeontology, Vol. 27, Part 4, 1984, pp. 699 706. | 700 PALAEONTOLOGY, VOLUME 27 text-fig. 1. Distribution of Silurian platform carbonates (Hurst 1980) and deep-water basin clastic sediments (Hurst and Surlyk 1982) on North Greenland. Localities 1 to 1 0 indicate where Dubaria were collected (Table 1). table 1. Recorded occurrences of Dubaria. Note the patchy spatial and temporal ranges of the genus as well as the species. * indicates species for which material has been studied. Species Reference Geographic location Stratigraphic level D. varians This paper NW Greenland Llandovery* D. reel inis Rubel 1970 Estonia Llandovery* D. sp. nov. This paper NW Greenland Llandovery* D. legrinus Ivanovskii and Kulkov 1974 Altai-Sajan Upper Llandovery D. tenera Nikiforova and Modzalev- skaya 1968 Siberia Llandovery D. rongxiensis Wan 1978 Tsuei-shan Yung Tzi Province, China Middle Silurian D. sp. D Johnson et al. 1976 Central Nevada Lower-middle Ludlow D. sp. Johnson et al. 1976 Central Nevada Lower-middle Ludlow D. lentenoisi Termier 1936 North Africa Lower Ludlow* D. megaeroides Johnson et al. 1973 Johnson and Boucot 1970 Central Nevada Upper Pridoli D. sp. Johnson, 1973 Central Nevada Middle Lochkovian D. thesis Johnson, 1975 Smith 1980 Arctic Canada Upper Lochkovian D. sp. Johnson and Boucot 1970 Carnic Alps Budnanian (Ludlow and Pridoli) JONES AND HURST: AUTECOLOGY OF DUBARIA 701 Limestone •1 Sandv turbidites 1 a ± | Black chert i _ Mudstone and Resedimented Dolomite 1 - tJ thin turbidites 1° - ° M conglomerate Biostromes Mudstone Reefs G| Green chert text-fig. 2. (Above) Stratigraphic scheme for Peary Land, North Greenland (from Surlyk et al. 1980 and Surlyk and Hurst 1984). (Below) Stratigraphic scheme for western North Greenland based on Washington Land and Hall Land (Surlyk and Hurst 1984). Asterisks indicate stratigraphic location of Dubaria (Table 2). 702 PALAEONTOLOGY, VOLUME 27 crinoidal rudstones (text-fig. 3). The poorly sorted nature of these rocks and the abundant coral and stromatoporoid colonies, many of which are overturned, suggests accumulation in relatively high-energy, shallow-water carbonate shelf environments. The isolated pockets of skeletal sands, which occur as isolated pockets up to 10 m thick and several tens of metres long, may represent the remains of skeletal sand shoals (Hurst 1980). The Offley Island Formation represents a carbonate shelf, biostromal unit consisting of level-bedded stromatoporoid floatstones and skeletal, crinoidal, and stromatoporoid rudstones (text-fig. 3) associated with small reefs (Hurst 1980). The biostromes probably represent skeletal sand shoals that accumulated in an agitated, shallow-water subtidal carbonate platform environment. An undulatory sea-floor topography is reflected by the complex vertical and lateral relationship of the facies. The rare laminated pellet lime mudstones probably accumulated in the lower energy environments, that existed in the topographic lows between the sand shoals. Dubaria occurs only in the higher energy skeletal sand shoals, never in the low-energy pelletal lime mudstones. Large D. varians from upper Llandovery reefs (text-fig. 3c) also occur in pockets and as skeletal components in debris beds, suggesting that they inhabited high energy environments. Dubaria occurrences in North Greenland are associated with high-energy biostromes and/or reef environments. Contrary to this opinion, Boucot and Perry (1981, p. 212) argued that the low diversity Dubaria Community inhabited a quiet-water environment. This dichotomy may reflect the scale at which environmental factors are viewed. Clearly, on a large scale the environment is a high energy one. However, even modern day, high-energy reefal environments encompass quiet, cryptic habitats that on a local scale are quite different in character. For example, Logan (1977) demonstrated that the modern brachiopods Thecidellina barret ti, Argyrotheca sp., and Platidia sp. inhabit deep recesses in Eden's Rock, a large patch reef off the coast of Grand Cayman Island. Such recesses typically have little light and little water movement (Logan 1977, p. 91). Logan argued that intense predation pressures and severe competition for space forced the brachiopods to inhabit such cryptic habitats. The analogy with Dubaria is striking. table 2. Summary information on Dubaria from west North Greenland. * Geological Survey of Canada locality number. Dv = Dubaria various., D = Dubaria sp. nov. Locality Species Locality on Text-fig. 1 Collector Formation Age Number of Specimens 256354 Dv 10 Mabillard j | Unnamed > Carbonate Llandovery to early 10 256357 Dv 10 Mabillard ! Reefs Wenlock 57 82370 Dv 7 Dawes 1 Unnamed ? 2 82368 Dv 7 Dawes > Carbonate 9 4 82367 Dv 7 Dawes Reefs ? 5 184128 Dv 9 Peel Unnamed Carbonate Reefs Llandovery to early Wenlock 1 211789 Dv 2 Norford Aleqatsiaq Fjord Fm. early Llandovery 2 211765 D 6 Norford Offley Is. Fm. Llandovery to early Wenlock 4 216887 D 2 Hurst Aleqatsiaq Fjord Fm. early Llandovery 1 JONES AND HURST: AUTECQLOGY OF DU BARI A 703 text-fig. 3. a, biostrome of laminar stromatoporoids and bioclastic debris, Offley Island Formation, Hall Land, b, crinoidal rudstone, Offley Island Formation, Hall Land, c, carbonate buildup (x) surrounded by off-reefal sediments (y) in Peary Land. Dubaria various occurs in both reefs and biostromes. See text-figs. I and 2 for location and stratigraphy. 704 PALAEONTOLOGY, VOLUME 27 WORLDWIDE DISTRIBUTION OF DUBARIA Dubaria occurs at many different, widely separated localities throughout the world (Table 1). As on Greenland, the stratigraphic distribution of Dubaria on a world-wide basis is also patchy (Table 1). For example, Dubaria is found in Llandovery strata but only rarely in the Wenlock (Table 1). These patchy spatial and temporal distributions also suggest that ecological factors may have controlled the occurrences of this brachiopod on a wide scale. Unfortunately, this suggestion cannot always be verified because the sedimentological setting of Dubaria is commonly not documented. Atrypopsis reclinis Rubel (= D. reclinis ) is common in the Llandovery strata on the Estonian island of Hiiuma in the northern Baltic. Rubel (1983 written comm.) states that ‘the known distribution of Atrypopsis reclinis in the North Baltic is really restricted to the bioherms or related rocks’. D. lantenoisi, the type species of the genus, occurs in the Ludlow of Morocco (Termier 1936). Termier and Termier (1960, fig. 390) illustrated a series of reefs surrounded by shale and indicated that Dubaria occurred in direct association with the reefs. Other illustrations (Termier and Termier 1960, figs. 391 and 394) show Dubaria associated with algal limestone from the reefs. Johnson ( 1 975, p. 16) inferred that Dubaria from the upper Lochkovian of Bathurst Island was also associated with reefs. Elsewhere the ecological setting of the genus cannot be ascertained from the available data. COMPARISON OF ECOLOGICAL SETTINGS OF DUBARIA AND ATRYPOIDEA Atrypoidea Mitchell and Dun, 1920 which is very common in the Ludlow strata of Arctic Canada, is morphologically similar to Dubaria. Atrypoidea differs, however, in not having dental lamellae in the pedicle valve. Most Atrypoidea appear to have inhabited a soft-substrate in quiet-water, subtidal environment (Jones 1982) and thus occupied a different ecological niche than that proposed for Dubaria. However, one species of Atrypoidea did develop the ability to inhabit reefs and/or the skeletal sands surrounding the bioherms (Jones and Narbonne 1984). In the areas between the reef only A. foxi occurs (Jones 1974). Atrypoidea bioherma has an external morphology directly comparable with that of D. various. Both have a similar overall shell form, both apparently had functional pedicles, and both developed rectangular-shaped deflections of the anterior commissure. text-fig. 4. Pedicle, lateral and anterior views of Dubaria varians , x 1.5, specimens are from sample 256367 (Table 1) and are deposited in the Geologisk Museum, Copenhagen, a-c, MGUH 16097; d-f, MGUH 16098. JONES AND HURST: AUTECOLOGY OF DU BA RIA 705 CONCLUSIONS Dubaria and A. biohenna sp. nov. both inhabited quiet-water, cryptic habitats associated with reefs. The similarity in the external morphology of the Dubaria species and A. biohernia strongly suggests that this general morphological scheme was adapted specifically for such a habitat. In particular, the rectangular form of the anterior deflection of the commissure appears to have been a necessity, presumably in some way assisting the feeding process. Comparison with similar modern reef- dwelling brachiopods suggests that predation or more likely competition from other organisms restricted their occurrence to such cryptic habitats. This may explain why Dubaria apparently does not occur in quiet-water environments of a more open aspect (for example, in quiet-water lagoonal deposits). The result of such spatial restriction is that brachiopods such as Dubaria have a very patchy stratigraphic occurrence. Acknowledgements. The material from Greenland was studied by Jones during a visit to Copenhagen financed by the Central Research Fund from the University of Alberta. Publication is by permission of the Director of the Geological Survey of Greenland. This paper formed a contribution to the Symposium on Autecology of Silurian Organisms at Glasgow, September 1983 (IGCP Project 53— Ecostratigraphy). REFERENCES boucot, a. j. and perry, d. g. 1981. Lower Devonian brachiopod dominated communities of the Cordilleran Region. In gray, j., boucot, a. j. and berry, w. b. n. (eds.). Communities of the past , 185-222. Hutchinson Ross Publishing Company, Stroudsburg. hurst, j. m. 1980. Silurian stratigraphy and facies distribution in Washington Land and Western Hall Land, North Greenland. Bull. Gronlands geol. Unders. 138, I 95. and surlyk, f. 1982. Stratigraphy of the Silurian flysch sequence of North Greenland. Ibid. 145, 125. ivanovskii, a. b. and kulkov, n. p. 1974. Rugosa, brachiopods and stratigraphy of the Silurian of Mountainous Altai and Sajan. Trudy Inst. Geol. Geofiz. sib. Otd. 231, 121. Johnson, j. G. 1973. Mid-Lockhovian brachiopods from the Windmill Limestone of Central Nevada. J. Paleont. 47, 1013-1030. — 1975. Devonian brachiopods from the Quadrithyris Zone (Upper Lochkovian), Canadian Arctic Archipelago. Bull. geol. Surv. Can. 235, 5-56. and boucot, a. j. 1970. Brachiopods and age of the Tor Limestone of Central Nevada. ./. Paleont. 44, 265-269. — and murphy, m. a. 1973. Pridolian and early Gedinnian age brachiopods from the Roberts Mountains Formation of Central Nevada. Univ. Calif. Pubis, geol. Sci. 100, 75. 1976. Wenlockian and Ludlovian age brachiopods from the Roberts Mountains Formation of Central Nevada. Ibid. 115, 102. penrose, n. l. and wise, m. t. 1978. Biostratigraphy, biotopes and biogeography in the Lower Devonian (Upper Lochkovian, Lower Pragian) of Nevada. J. Paleont. 52, 793-806. jones, b. 1982. Paleobiology of the Upper Silurian brachiopod Atrypoidea. Ibid. 56, 912-923. and narbonne, g. m. 1984. Environmental controls on the distribution of Atrypoidea species. Can. ./. Earth Sci. 21, 131-144. logan, a. 1977. Reef-dwelling articulate brachiopods from Grand Cayman, B.W.I. Proc. 3rd. Int. Coral Reef Symposium , 1, 87-93. Nikiforova, o. i. and modzalevskaya, t. l. 1968. Some Llandovery and Wenlockian brachiopods of northwest part of Siberian Platform. Scientific Notes Paleont. and Biostrat., Scient. Res. Inst., Geol. Arctic , Ministry of Geol., USSR, 21, 50-78. rubel, m. r. 1970. Brachiopody Pentamerida i Spiriferida Silura Estonii , 75 pp. Institut Geologii Akademii Nauk Estonskoj SSR, Tallinn. smith, r. e. 1980. Lower Devonian (Lockhovian) biostratigraphy and brachiopod faunas, Canadian Arctic Islands. Bull. geol. Surv. Can. 308, 155. surlyk, f. and hurst, j. m. 1984. The evolution of the early Paleozoic deep-water basin of North Greenland. Bull. geol. Soc. Am. 95, 131-154. and bjerreskov, m. 1980. First age-diagnostic fossils from the central part of the North Greenland foldbelt. Nature, 286, 800-803. 706 PALAEONTOLOGY, VOLUME 27 termier, h. 1936. Etudes geologiques sur le Maroc Central et le Moyen Atlas Septentrional. Notes Mem. Paleont. Maroc Serv. Geol. Div. Mines and Geologic, 33, 1087-1421. and termier, G. 1960. Paleontologie Stratigraphique , 515 pp. Masson & Co. Paris. wan, z. Q. 1978. Atlas of fossils of southwest China. Sichuan Volume, 347, 348. Geological Publishing House, Peking, China. BRIAN JONES Palaeontological Collections Department of Geology University of Alberta Edmonton, Alberta Canada T6G 2E3 JOHN M. HURST Manuscript received 29 June 1983 Revised manuscript received 19 August 1983 British Petroleum Developments Ltd Britannic House Moor Lane London EC2Y 9BU, U.K. A REAPPRAISAL OF THE LOWER CARBONIFEROUS LEPIDOPHYTE ESKDALIA KIDSTON by B. A. THOMAS and S. V. MEYEN Abstract. Two new species of Eskdalia are described from the Lower Carboniferous of Siberia. Eskdalia has been thought to possess true leaf scars produced as a result of leaf abscission. The new specimens similarly have misleading oval areas thought to indicate leaf abscission. However, closer examination and consideration of sedimentary effects upon preservation, shows that leaves are still attached although their preservation is poor. The genus is rediagnosed and compared with other similar lepidophytes. Eskdalia was originally thought to be a fern rachis by Kidston (1883), but was later clearly shown to be a ligulate lycopod shoot (Chaloner 1967; Thomas 1968). Such specimens have very distinct longitudinally elongated oval areas which have been interpreted as leaf scars by comparison with other genera and because they lack maceratable cuticle. Stem cuticle can be prepared, but perforations are present at the sites of the ovals. Small depressions can be seen in the upper margins of many of the ovals and macerations showed them to be the remains of ligule pits. Similar cuticle with perforations and ligule pit tubes have been described many times from the Russian paper coal of the lower Carboniferous of the Moscow basin. They were orginally described as species of Lepidodendron by Eichwald (1860), Auerbach and Trautschold (1860), and Goeppert (1861) or of Bothrodendron by Zeiller (1880, 1882). Zalessky (1915), however, named them as species of a new genus Porodendron and was followed by most Russian authors until they were referred to Eskdalia (Thomas 1968; Meyen 1972, 1976). Meyen (1976), however, also remarked that there is a similarity between Eskdalia and certain other Angaran lepidophytes, notably Tomiodendron varium (Radcz.) Meyen and those described as Lepidodendron (?) cf. planum by Rasskazova (1962) (now Tunguskadendron borkii Meyen and Thomas in press). Also, in comparison with Angarodendron and Ulodendron , Meyen made the suggestion that the perforations seen in such stem cuticles may possibly be the result of cuticle being absent from persistent leaves, rather than the leaves having been shed by means of definite abscission layers located at the levels of the stem cuticle perforations. There are other genera which similarly show cuticles with perforations but no ligule pits, even though they are leafy and probably heterosporous. So suggestions have been made about leaves having no cuticle, ligules being either short or situated on unclear or unpreserved leaves, or such axes being remnants of cones that have lost their sporophylls and sporangia. There is an obvious need for much more information on this topic. It was therefore thought to be highly desirable to reinvestigate Eskdalia and certain similar forms in an attempt to clarify this situation. LOCALITIES AND STRATIGRAPHY OF NEW MATERIAL Two new collections were available for study, each with several specimens. Both are housed in the collections of the Geological Institute of the Academy of Sciences in Moscow. Collection 1 constitutes specimen nos. 3779/81-92 and 4034/2-5. They are preserved on a blue green siltstone from ‘Kyutyungde' locality, the right bank of the Kyutyungde River in Eastern Siberia, 12 km from its mouth and 2 km below the mouth of the Khalomalokh (70° 43' North, 1 23 East). The sediments have been dated by associated marine fauna as Tournaisian (it is difficult to be more precise, but it is not equivalent to that part of the Tournaisian that is referred in W. Europe to the Devonian. Collection 2 Constitutes specimen nos. 4034/1 and 9. They are preserved [Palaeontology, Vol. 27, Part 4, 1984, pp. 707-718, pis. 62-63.| 708 PALAEONTOLOGY, VOLUME 27 in grey shale from the number K-3 borehole (depth 930 m), between the Kempendyai and Namana Rivers, 65 km east of the village of Kempendyai, Eastern Siberia (roughly 62° North, 1 19° 48' East). It is dated as the upper part of the Kurunguryakh suite (Visean) of the Vilyui syncline (Kolodeznikov 1982). The miospores from ‘Kempendyai’ have been described by Pashkevich et al. (1978). DESCRIPTION All the specimens clearly show the longitudinal elongated oval areas typical of Eskdalia and remains of ligule pits can be seen in their upper angles. One minor difference between the new specimens and those described previously by Zalessky (1915), Chaloner ( 1 967), and Thomas ( 1 968 ) is the manner of preservation of the pits. The earlier specimens had tubes of cuticle representing the inner linings of the actual pits. The new Russian specimens have their pits preserved as casts similar to those described by Meyen (1972, 1976) and Thomas and Purdy (1982). Specimens from both collections were examined closely for details. This involved transferring portions of stems using Darrah solution (Darrah 1952), degaging around the edges of the ovals and at the sides of the stems, and macerating stem compressions using Schulze’s solution followed by dilute ammonia solution. Both collections show some variation, but there appear to have been enough differences to treat them initially as new species. The maximum size of the stem fragments is governed by the dimensions of the rock samples, but we have the first evidence of branching within this genus. Two specimens from the borehole each show an equal dichotomy even though these are amongst the smallest fragments of stem referrable to this genus. Length of preserved axis is therefore not the controlling factor in showing whether the plants branched. Leaf abscission producing oval leaf scars is one of the main diagnostic characters of Eskdalia as interpreted by Chaloner (1967) and Thomas (1968). The degaging and transfer techniques, however, have revealed a very suggestive fact. The longitudinally elongated ovals are outlined by a very thin line of compression which continues into the matrix. The extra compression revealed by degaging and transferring clearly shows that there was a narrow band of compression extending into the matrix around the oval. This compression can be traced upwards into what must have been a persistent leaf and downwards into a distinctive heel. The ovals cannot therefore be true abscission scars. Instead, they must be artificial scars produced by the splitting of the carbon compression during the fracturing of the rock. Such breakage is similar to that described in Angaropldoios Meyen (1976) and in Tomiodendron Radczenko by Meyen (1976) and Thomas and Purdy (1982). It also parallels the artificial loss of leaves shown by Chaloner (1967) and Mensah and Chaloner (1971). With this new concept of leaf cushion in mind, we can interpret other features shown on the stems. The central ridges shown on the ovals are most probably the very basal expressions of laminae midribs (keels) rather than any features formed by abscission. Some stems have ovals which seem to extend downwards into tails (4034/3-2). This would originally have suggested different-shaped scars, but now they can be interpreted as being the result of heel impressions extending the more normal oval leaf cushion impressions. In this case the heels were closely pressed to the stem and were not separated by rock matrix (text-fig. Id). This formation of an apparently different character by sedimentary effects is paralleled by similar effects on the ligule pit. The pits can sometimes become filled with sediment producing a pit cast. Meyen (1972, 1976) has clearly shown this to happen in Tomiodendron Radczenko, Ursodendron Radczenko, and Angarodendron Zalessky (see also Thomas and Purdy 1982). If a ligule pit cast is formed in the impression, the corresponding stem compression will obviously have no ligule pit cuticle (text-fig. lc). Cuticle preparation from the upper angle of such ovals will never reveal the plant to be ligulate. If the ligule pit does not become filled with sediment, it will separate with the main part of the stem compression during the fracturing of the rock (text-fig. Id). Maceration will then reveal a cutinized ligule pit as described by Chaloner (1967) and Thomas (1968). Such effects of sedimentary infillings on the resultant fossils are all summarized in text-fig. 1 . Both collections showed sedimentary infillings of the ligule pits, so no cuticular linings could be prepared as was achieved with Eskdalia minuta. THOMAS AND MEYEN: REAPPRAISAL OF ESKDALIA 709 text-fig. 1. Diagram to illustrate the interrelationships of sedimentation and rock fracture upon the Eskdalia type of leaf cushion, a, e, and G represent plant tissues seen in section before fossilization (leaf = LF, ligule pit = LP, heel = FI, and lateral Wings = W). The others represent compressed plant material (dark stippling) embedded in rock matrix (light strippling). a, b, and c represent successive stages of plant compression and rock fracture as seen in longitudinal section. In c, the ligule pit cast is formed (LPC) and the leaf and heel are lost. D represents an alternative situation to c after rock fracture. Here there is no ligule pit cast although the pit is retained as a compression (LP') and the heel is pressed against the stem. Either may occur independently so there are four possibilities of preservation, e and f represent similar stages to a and c except that a ligule pit compression or cast is formed further up the leaf lamina and is never seen. The heel may also be visible as in d. g and h are similar stages seen in transverse section. Here the lateral wings may be lost or shown as side extensions to the cushion. 710 PALAEONTOLOGY, VOLUME 27 One other sedimentary effect is also of interest to us here. Both types of stem can be seen in transverse section either on the cut side of the core or on the fracture surface of the rock. In both cases there is a compressed central core of coaly plant material inside the matrix of the stem. This is almost certainly the remains of the protostele even though no tracheids can be seen or prepared by maceration. Stem cuticle was prepared from both collections by maceration with Schulze’s solution. Both showed a simple epidermal arrangement with polygonal equal-sided to longitudinally elongated cells. The anticlinal walls are straight and the periclinal walls are flat, smooth, and without the cuticular lumps described from E. minuta by Thomas (1968). No cuticle could be prepared from the ovals, although cell outlines were clearly visible and similarly none was obtained from the leaves uncovered from the matrix. DISCUSSION The available information therefore allows us to rediagnose the genus Eskdalia and to name the two new species E. kidstonii and E. siberica. The morphological and cuticular characters of the two new species together with those of E. minuta Kidston, are summarized in Table 1 . There are differences in leaf cushion and ligule pit sizes, although these two are not directly related. E. kidstonii shows much more regularity in cushion size and cushion spacing than the other species, whereas E. siberica shows the greatest variation. The latter species also differs in having longitudinal furrows running between the leaf cushion on some of the stems. Stem epidermal details are not sufficiently distinctive to separate the species, although the ranges of cell sizes are not identical. The epidermal lumps described from E. minuta are not, however, found in the other two species. E. kidstonii alone shows occasional dichotomies, although, as mentioned earlier, this feature is not related to the comparative lengths of the specimens. The other species include large stems but show no evidence of branching. Leaf, heel, and stele sizes are unfortunately not known for all the species. COMPARISON Eskdalia in its new concept is a sparsely branched ligulate lycopod shoot. It has persistent leaves borne on leaf cushions which also possess distinct heels at their lower ends. There are many other similar genera and doubtfully placed specimens which now need to be reconsidered in the light of this new interpretation of Eskdalia. Some are clearly better understood than others and some may never table 1. Comparison of Eskdalia kidstonii sp. nov. and E. siberica sp. nov. with the type species of the genus, E. minuta E. minuta E. kidstonii E. siberica Size of leaf cushions (mm) 4-5-5-5 x 3-4 2x1-2 1-2x0-61-3 Ligule pits (/un) flask shaped cylindrical globular 400 x 1 50 200 x 100 300 x 150 Heels (mm) ? 0-7 7 Leaves (mm) ? 4 4 Stem epidermal cells (/xtu) 130 x 25-35 60-125 x25 90-150x25 with lumps no lumps no lumps Observed stem sizes (mm) 1 10x4-25 75x4 130x 13 Branching none occasional dichotomies none Stele/axis ratio 7 1:3 1:4 table 2. Comparison of Eskdalia and other genera of ligulate lycophytes (ZL(>\) sSuiuu3f U0Jpu3p0J3A3UI/a/ ( (0861) UMOJQ pul? 3u3jOUI3q pil3lU3 l|nt!U3-g sisdoipodojA'rj (L96\) sriuoqx pusuia nyi uojpuapo/f) (0961) mjuszopny ucupuaposjfl (ZL6\) puaiiia o>(U3ZDpR^j uo.ipudponuoj' jS It 3 C txO O a w> ^ TJ g -O + ii + I + ~. + + I I I I I I + I I I I + + + I 1 1 33 ~ £ •5^3? € B ^ -js c: -o I T3 a o 53 TJ >■ 3 ■3 g c- C (W.6I ) ASiUBuy a. uojpuapopida/opnjsj I + + I I I V — O' (S 1 6 1 ) ^SS3|RZ uojpuapouoj I + I I I (6L6I) Pur[Asa\ pur |3sni:j>i ouvjpSisopida 7 + 111 + I -x (Z961) -<33131 uintvJo/jad uoupuapopida 7 I + I I 7? E (£061) uojsprjj »5TO (9Z.6I) us-fafM PU3UI3 A3(ss3]rz uojpuapojvSuy >? o cd 3 -O I + + I + + + I • S2 O — I O :d -) < 3: U > (often equally) (often equally) Angarodeitdron Zalessky emend. Meyen (1976) o 2. Eskdalia — J Kidston (1903) Lepidodendron perforatum Lacey (1962) o o. Lepidosigillaria " ° Krauscl and Weyland (1949) Porodendron Zalessky (1915) 3 2 $ 2. Pseudolepidodendron ~" y •. 4,.> *"V“ * } 9m t P 1 ^ , -** srf Jj * il ! H* «£ li V 'fc'V . BENTON, rhynchosaur dentitions 746 PALAEONTOLOGY, VOLUME 27 since these teeth do not occlude in the adult. Relative growth causes the maxilla to curve upwards away from the dentary, and progressively more posterior portions of the tooth rows come into occlusion. It should be noted that the degree of tooth wear is not related to skull size, but probably depends on the individual’s diet as in Sphenodon (Robinson 1976). Tooth form In general, the teeth of Hyperodapedon are deeply fixed in the bone of the jaw and they have open roots with pulp cavities of varying size. The bulk of the tooth is composed of dentine, and enamel may be seen capping the upper portion in some cases. The maxillary teeth have short roots and shallow conical pulp cavities. They may be nearly circular in cross-section with a central root canal. In microscopic sections they show occasional longitudinal fluting on the surface, and radial dentinal tubules (PI. 67, fig. 4). These features have been described in detail in H. huxleyi by Chatterjee (1974, pp. 228-229). More information is available on the dentary teeth of Hyperodapedon. A series of vertical transverse sections (text-fig. 3) shows the form and emplacement of the teeth and variation in their pattern along the jaw. Anterior portions (text-fig. 3 a) lack teeth, probably as a result of wear and migration of the teeth occlusally, or the teeth are small and the roots closed (text-fig. 3 b\ PI. 66, fig. 4). Further back, a series of buccal teeth is seen in every section. This is not a succession of teeth or a Zahnreihe. The buccal teeth slope forwards, laterally, and upwards, so that most of them are cut obliquely in vertical sections. Each tooth is waisted at about the mid point which is shown by unworn crowns (text-fig. 3; PI. 66, fig. 2). Wear on the inner surface produces a medial concavity and the outer surface retains its original convex profile (PI. 66, fig. 3). At the base, most buccal teeth display an axial pulp cavity that may be relatively large or small in an apparently random way and it is not dependent on the level of the section (text-fig. 3c-c). Vertical longitudinal sections through the dentary show the curved shape of the buccal teeth (text-fig. 4 b-d), and horizontal sections (text-fig. 5 a-c) show that they are nearly circular in section, rather than compressed, as suggested by Chatterjee (1974, p. 230) for H. huxleyi , except when they are worn to a knife-like edge on the top of the jaw (text-fig. 5c). The lingual teeth are shorter and shaped like thick-walled thimbles. Because they are directed medio-dorsally, the standard sectioning planes do not cut them axially, but these teeth were clearly cylindrical with a deep and narrow conical pulp cavity (text-figs. 3c, <7, 4 a, b, 5b). Bone and tooth histology The serial peels and microscopic sections from the lower jaw of Hyperodapedon reveal a great deal of detail concerning the histology of the bone and teeth. The bone may be divided into three types: 1 . Laminar fibrolamellar with parallel longitudinal primary osteons: in regions with teeth (PI. 66, figs. 2-4; PI. 67, figs. 1,2, 4). The osteons are occluded to a greater extent towards the edge of the jaws text-fig. 3 (opposite). Transverse vertical sections of the upper part of the mandible of Hyperodapedon gordoni (NUGD B) showing buccal and lingual teeth. The locations of the sections are indicated on an outline mandible. a , anterior mandible; teeth are absent, b , anterior mandible, further back; heavily worn teeth with closed roots, and a large vessel canal, c, mid-jaw; a heavily worn lingual tooth and a series of buccal teeth sectioned at different positions along their length. The upper one shows a characteristic wear shape; the lower ones show open roots near the apex. These are not successional teeth, but sections at an angle through teeth that slope up and forwards, and all are functional. The area shown in PI. 66, fig. 3 is outlined, d , further back; a rare abnormality where a small additional lingual tooth occurs, and erosion of a buccal tooth by the growth of a lingual tooth is also seen. e, posterior part of dentary, behind the lingual tooth row; a series of sections of buccal teeth with broad root canals and the characteristic ‘waisted’ appearance of the barely erupted dorsal tooth. Abbreviations: bt, buccal teeth; It, lingual teeth; me, Meckel’s canal; mnf, mandibular foramen; sp, splenial. BENTON: RHYNCHOSAUR DENTITIONS 747 text-fig. 4. Longitudinal vertical sections of the upper part of the mandible of Hyperodapedon gordoni (NUGD B), showing the buccal and lingual teeth. The locations of the sections are indicated on an outline mandible, a, medially located section; lingual teeth only are seen, b, slightly more laterally; the lingual teeth are much reduced and the roots of the buccal teeth appear, indicating that they slope up, forwards, and laterally. The area shown in PI. 67, fig. 3 is outlined, c, more laterally; the full shape of longitudinal sections through the closely packed buccal teeth is clear. The area shown in PI. 67, fig. 4 is outlined, d , towards the lateral margin of the jaw; only partial oblique sections through buccal teeth appear, and show their very close packing. Abbreviations: bt, buccal teeth; It, lingual teeth. BENTON: RHYNCHOSAUR DENTITIONS 749 and on either side of the teeth where the bone itself may be subject to wear. This may be functionally equivalent in part to the highly calcified layer of bone observed on the sides of the jaw of Sphenodon by Harrison ( 1901, pp. 200-201). Further forwards in the jaw, where the teeth are older and they have closed root canals, the bone becomes almost wholly compact in the vicinity of the teeth (text-fig. 3a, b\ PI. 66, fig. 4). 2. Recticular (cancellous) fibrolamellar: in the centre of bones and away from the teeth (text-figs. 3, 4; PI. 66, figs. 2, 4). 3. Bone of attachment: around the sides of the teeth, the bone is remodelled and takes on a regular appearance with tightly packed small osteons (PI. 66, figs. 2, 3; PI. 67, fig. 3). These may have a compressed appearance when they occur between two closely spaced teeth (PI. 67, figs. 4, 6). At the base of the growing teeth the bone is highly vascular (e.g. text-fig. 6a c; PI. 66, fig. 2; PI. 67, figs. 2, 3, 4), and this is particularly marked in posterior parts of the jaw. In the front of the jaw, a track of extensively remodelled bone may be seen at the base of certain teeth (PI. 66, fig. 4) — this probably marks the passage of the tooth through the bone during jaw growth when the tooth was maintaining its position in occlusion. The remodelled bone may also indicate resorption of a tooth that is no longer in occlusion. In micrographs of the Hyperodapedon jaw (e.g. PI. 67, fig. 6), osteocyte lacunae are clearly visible in the bone matrix. In some cases there appear to be transverse fibres that run from one tooth to the next (text-fig. 6c). These may be traces of collagen fibre directions, or they may be preservational artefacts. Enamel is not always present on the dentary teeth, but it may be represented by some white, radially prismatic, deposits (text-fig. 6 d). Enamel is probably present initially on the tooth crowns and is later worn off, rather than being totally absent as suggested by Sill (19716) for Scaphonyx. In text-fig. 5. Horizontal sections of the upper part of the mandible of Hyperodapedon gordoni (NUGD B) showing the buccal and lingual teeth. The locations of the sections are indicated on an outline mandible. a, ventrally placed section; only the roots of some buccal teeth are seen, some with open root canals, b, higher up; both lingual and buccal teeth appear, and the latter can be seen to be more closely packed than the former, c, section near the occlusal margin of the dentary; the worn, closely packed buccal teeth occupy nearly the whole width of the jaw. Abbreviations: bt, buccal teeth; It, lingual teeth; rc, root canal. 750 PALAEONTOLOGY, VOLUME 27 text-fig. 6. Sections of dentary teeth of Hyperodapedon gordoni (NUGD B). Scale bars all measure 0-5 mm. a-c, transverse sections of buccal teeth near the root apices, showing open root canals and some erosion of tooth material by neighbouring teeth; d , transverse vertical section through the upper part of a buccal tooth, showing a thin enamel cap with perpendicular prismatic fabric; e , detail of the compact bone between two buccal teeth in horizontal section, showing the diffuse tooth margin and ‘fibres’ in the bone perpendicular to the tooth surface; /, detail of the bone between two buccal teeth in vertical section, showing the closely packed primary osteons and the dentinal tubules. Abbreviations: de, dentine: e, enamel; rc, root canal. EXPLANATION OF PLATE 67 Histology of the bone and teeth of the dentary of Hyperodapedon gordoni. Sections from NUGD B. Fig. 1. Horizontal section taken above Meckel’s canal, showing laminar fibrolamellar bone with largely occluded canals, x 30. Fig. 2. Horizontal section taken at the root apex of a buccal tooth, showing compact laminar fibrolamellar bone, and some bone of attachment, x 30. Fig. 3. Longitudinal vertical section taken towards the medial side of the dentary, showing part of a lingual tooth with open root canal and loose bone at the base (top), and the lower part of a buccal tooth; the shiny mineral that infills the cavities is goethite; this section is located in text-fig. 56, x 13. Fig. 4. Longitudinal vertical section through the bases of four buccal teeth showing the laminar fibrolamellar bone, and the diffuse tooth/bone of attachment margin; this section is located in text-fig. 4c, x 13. Fig. 5. Transverse vertical section through the roots of some buccal teeth and a lingual tooth, showing erosion of the former by the latter; the erosion occurs along a typical arcuate front, x 12. Fig. 6. Horizontal section through two buccal dentary teeth, showing the margins of two teeth with clear radial dentinal tubules, and the close-packed bone of attachment between the teeth, x 80. PLATE 67 BENTON, rhynchosaur dentitions 6 752 PALAEONTOLOGY, VOLUME 27 Hyperodapedon , the bulk of the tooth is composed of orange-yellow orthodentine which clearly shows radial dentinal tubules in microscopic section (PI. 67, fig. 6), and these are picked out by the iron oxide infill of the pulp cavity on etched polished surfaces. The tubules run from the pulp cavity to a diffuse area at the junction of tooth and bone with no apparent cement layer. In horizontal section, many buccal dentary teeth show possible growth rings in the dentine (PI. 67, fig. 2). These mark the boundary between primary dentine, laid down initially around the circumference of the tooth, and secondary dentine. Secondary dentine seems to be laid down in most teeth of Hyperodapedon , occluding the pulp cavity, but there is no regular pattern. Adjacent teeth may have root canals occluded to completely different extents, and the canal may become completely closed in the middle portion of the mandible (text-figs. 3c-e, 5). The root canal generally appears to remain open in all but the most anterior teeth (text-fig. 3b). The apical foramen is also usually open which indicates continued deposition of secondary dentine. The apical foramen may be axial, but it is frequently lateral, in which case the root tip is crescentic in section (text-fig. 6 a-c; PI. 66, fig. 2). These lower parts of the teeth also show resorption effects. The tips of the roots are often randomly arranged (text-fig. 6 a-c) and the growth of one clearly causes resorption of another along an arcuate front. This also occurs higher up where lingual teeth are growing close to buccal teeth, and one tooth achieves its normal form to the detriment of the other (text-fig. 3d; PI. 67, fig. 5). THE DENTITION OF STENAULORHYNCHUS Arrangement of teeth in the maxilla Stenaulorhynchus has two grooves on the maxillary tooth-plate. In juveniles, these lie between three distinct rows of teeth which are raised on sharp ridges (text-fig. 76, c). However, in adults (text-fig. 7e, g ), the grooves become apparently less regular as the teeth and bone are worn down. The medial groove runs the length of the maxilla, but the lateral one is restricted to the posterior portion. The grooves become shallower and rounded, and the teeth are not wholly restricted to the ridges. These facts suggest that the grooves in Stenaulorhynchus are initiated in the bone between the juvenile tooth rows, but that their subsequent appearance depends on wear to a far greater extent than in Hyperodapedon and other late Triassic rhynchosaurs, where the groove is a regular inherent part of the maxillary tooth-plate (Chatterjee 1974; Benton 19836). The teeth are arranged in longitudinal rows. In juvenile specimens (text-fig. 76, c ) there are three rows on the occlusal surface of the maxilla, the two outer ones running the length of the jaw, and the middle one only occupying the posterior third of the length. On the medial side of the maxilla (text-fig. la) a series of longitudinal rows of teeth may be seen running diagonally down and forwards to the crest of the jaw. Similar features may be seen in adult specimens (text-fig. Id-g) where additional longitudinal rows are added to the occlusal surface from the medial diagonal rows. The pattern of tooth rows is generally regular, but a scattering of odd teeth that cannot easily be assigned to rows may occur in the medial groove (text-fig. Ig). The tooth pattern of the maxilla of Stenaulorhynchus has been interpreted as consisting of distinguishable occlusal and lingual teeth (Huene 1938; Chatterjee 1974, 1980), but this is not an appropriate description. The lingual' diagonal rows all run to the jaw margin and continue without break on to the medial portion of the occlusal surface. It is not possible to distinguish between the ‘occlusal’ and the ‘lingual’ series of teeth in any specimen, and in terms of tooth growth the distinction is meaningless since ‘lingual’ teeth become ‘occlusal’ as the jaw is remodelled and worn (see below). In order to check if any differences existed between longitudinal tooth rows, a series of longitudinal vertical sections through a complete maxillary tooth-plate was made. The three-dimensional reconstruction (text-fig. 8) shows the pattern of teeth in several longitudinal rows through the maxilla. Unfortunately, the most posterior portion of the tooth-plate was missing, and the youngest teeth cannot be shown. The tooth arrangement is clearly rather irregular within individual longitudinal rows. Even the lateral row is not quite as simple as it appears in occlusal view (text-fig. Ig). In the example sectioned, BENTON: RHYNCHOSAUR DENTITIONS 753 two series of teeth of rather different shape are involved, and they are cut at an angle by the plane of section (text-fig. 8). The anterior set consists of nine short teeth (a-i) with closed roots and enamel caps. None of these anterior teeth is particularly worn— they barely reach the occlusal margin in the side of the lateral groove— and the posterior one (/') does not erupt at all. The posterior set consists of seventeen long-rooted teeth (1-17), all of which have erupted. The anterior nine teeth (I 9) text-fig. 7. Tooth-bearing bones of Stenaulorhynchus stockleyi : maxillae ( a-g ) and dentaries (/, /) (cf. text-fig. 14). a , 6, juvenile left maxilla (BM(NH) R9279) in medial and occlusal views, c, juvenile left maxilla (CUMZ T993) in occlusal view, d, e , right maxilla (BM(NH)R9281 ) in medial and occlusal views. /, g , right maxilla (CUMZ T1 138) in medial and occlusal views. /;, right lower jaw (CUMZ T1 1 12), lacking the splenial, in medial view (cf. text-fig. 9c). i,j, left dentary (BM(NH) R9273) in medial and occlusal views. All x 1, except /;, x 0-71. 754 PALAEONTOLOGY, VOLUME 27 text-fig. 8. Graphic reconstruction of the dentition of the right maxilla of Stenaulorhynchus (BM(NH) R 10008). Drawn from serial sections taken at 0-5 mm spacing, and traced on to glass plates at a magnification of 2 x . The four block diagrams (a-d) are drawn as if the bone is transparent, and individual teeth are numbered or lettered in sequence from oldest to youngest in each row (see the text). The lines of section that separate the blocks are indicated on the occlusal view of the tooth-plate. The medialmost rows of teeth are missing owing to damage of the specimen. Abbreviations: ANT, anterior; LAT, lateral; MED, medial. BENTON: RHYNCHOSAUR DENTITIONS 755 are heavily worn and rather mixed up with the anterior set just described. Of the posterior eight teeth, six (10, 13-17) still have their points and enamel caps, but the other two (II, 12) are worn. Two teeth (11, 17) have widely open roots — which indicates active deposition of dentine — and it could be that the posterior tooth series (1-17) splits into two longitudinal series (1 11, 12-17) where teeth 1 1 and 1 7 were initiated latest of all. In this case, it would be difficult to assign tooth 10 to either series. The middle longitudinal series of fifteen teeth is easier to interpret. The teeth occur in one sequence, with the anterior group heavily worn, and the posterior group just coming into occlusion. The teeth in the latter group (11-15) have open roots. There is a considerable space between the middle row and the rows of the lingual (medial) side of the tooth-plate. The anterior teeth of the lingual rows are most heavily worn, and the posterior ones have open roots and unworn crowns. The twenty-one teeth are divided tentatively into three series (1-8, 9-15, 16-21), where the youngest teeth of each group are 6, 1 5, and 21 . However, the sectioned maxilla does not show a well-developed battery of ‘lingual’ teeth, as in other jaws of similar size (e.g. text-fig. 7/, g), possibly as a result of damage, and this makes a full reconstruction difficult. In summary, the teeth in the maxilla of Stenaulorhynchus are arranged in several longitudinal rows. There are three tooth-bearing areas in the tooth-plate: lateral, median, and medial. In the medial area, several diagonal rows of teeth pass into occlusion in sequence (text-fig. 86), and there is no clear distinction between lingual' and ‘occlusal’ teeth. Arrangement of teeth in the dentary The lower jaw of Stenaulorhynchus (text-fig. Ih) is longer and lower than that of Hyperodapedon (c.f. text-fig. lc, d) and the tooth-bearing portion is concentrated further forward. Well-preserved dentaries (text-fig. 7 i, j) show that there is a raised longitudinal row of buccal teeth (lateral) and several diagonal rows of lingual teeth (medial) that run up into occlusion on top of the jaw. The lowest lingual teeth are small and occasional (?) replacement pits are seen (text-fig. li). The teeth are separated by a broad, shallow groove which changes in shape along its length. At the very back of the tooth row the posterior ten or twelve buccal teeth, which are not in occlusion, are raised on a high ridge and separated from the coronoid by a clear groove (text-fig. 9c). A three-dimensional reconstruction of the Stenaulorhynchus dentary (text-fig. 9a) shows how the buccal teeth are deep-rooted and slope up and forwards, as in Hyperodapedon. The lingual teeth also slope up and forwards, but they also slope medially when low on the inside of the jaw. The most lateral tooth of the lingual series at any point is generally the largest and nearly all have open roots. The lingual teeth remain distinct from the buccal teeth, but there is little difference in shape or size between the two kinds, as seen in Hyperodapedon. Jaw occlusion and tooth wear Tooth and bone wear is clearly shown on the occlusal surfaces of the maxilla and dentary of Stenaulorhynchus . The relative curvature of the two tooth-bearing elements is not as great as in Hyperodapedon , and larger areas are in contact. Juvenile specimens show little wear (text-fig. 7c/-c), but adult maxillae and dentaries (e.g. text-fig. 7 g, j) are worn smooth except at the very back. The small enamel cap of each tooth is stained dark brown or black in some specimens, and this highlights the degree of wear (text-fig. 7/). As in Hyperodapedon , teeth generally bite against bone, although some medial teeth occlude (text-fig. 96). Buccal teeth may be heavily worn on the lateral side of the dentary as well as on the occlusal surface (text-fig. 96, c ). Tooth and bone seem to wear at the same rate, and distinct pits or striations are not seen. Stenaulorhynchus probably had a precision-shear bite, as in Hyperodapedon. The quadrate- articular joint appears to be rather tight (Huene 1938), and the strong symphysis would prevent rotation of the lower jaws— a necessary feature if they are to move back and forwards in such a broad skull. These points, and the apparent precise fit of dentary and maxilla, would prevent any marked fore and aft sawing of the jaws. text-fig. 9. The right dentary of Stenaulorhynchus (BM(NH) R 10007). a, graphic reconstruction of the dentition, drawn from serial sections taken at 0-5 mm spacing, and traced on to glass plates at a magnification of 2 x . The single row of buccal teeth is numbered in sequence from the front backwards. The lingual teeth occur in several longitudinal rows, but these are rather confused by bone remodelling and relative tooth movement. b , transverse vertical section through the dentary and maxilla with the jaws slightly apart, to show the nature of the occlusion. A tooth may bite against another tooth, or against bone, c, medial view of a right lower jaw (CUMZT1 1 1 2, Songea district; cf. text-fig. Ih) which lacks thesplenial (cf. Hyperodapedon gordoni, text-fig. 1 d). The boxed region of the dentary is equivalent to that shown in the graphic reconstruction. Four cross-sections of the lower jaw at different positions along its length (a, b, c, d) are also given. Abbreviations: an, angular; ANT, anterior; art, articular; bt, buccal teeth; c, coronoid; d, dentary; dig, dental lamina groove; LAT, lateral; It, lingual teeth; me, Meckel's canal; MED, medial; mx, maxilla; pra, prearticular; sa, surangular; sp, splenial. BENTON: RHYNCHOSAUR DENTITIONS 757 Tooth form and histology The teeth of Stenaulorhynchus are generally long-rooted and deeply fused in the bone of the jaw. The shapes are less regular than in Hyperodapedon , and the root canal is open in less teeth. The bulk of the tooth is composed of dentine, and unerupted or unworn teeth may display a small cap of enamel. Tooth shape and size vary slightly across the maxilla. In juveniles, most teeth are unworn and their round conical pointed shape may be seen (text-fig. la-c). It is not possible to distinguish ‘lingual’ from ‘buccal’ teeth as in Hyperodapedon. In older specimens, most of the occlusal teeth are worn flush with the surrounding bone (text-figs. 7 d-g, 10 a, b ). Teeth of the lateral row are nearly always larger than the others (2-3 mm in diameter, compared with 1 -2 mm or less). They are rather compressed, or oval, in shape, with the long axis directed transversely— this may be a result of their tight packing with only a thin wall of intervening bone. Other maxillary teeth are more circular in cross-section. The teeth on the lingual surface are small and unworn. They are pointed and conical and still retain their enamel caps (text-figs, la, d,f, 1 1 b). Teeth are generally absent from the most anterior portion of the maxilla — they have presumably been completely worn away. Tooth size increases backwards from the short heavily worn teeth at the front to the middle of the area currently in occlusion, where the teeth have the longest roots (text-figs. Ig, 10). They then diminish to the back of the jaw, where the most posterior, youngest, teeth have not yet come into occlusion. Some anterior teeth in the lateral series are short and rounded (text-figs. lOu, 12c). Other teeth are long and compressed, or resorbed in an irregular way under the influence of neighbouring teeth. A selection of tooth shapes may be seen in text-figs. 10 and \2a-d. Newly formed teeth have open apical foramina and wide root canals surrounded by a thin tube of dentine (text-fig. 1 2a), while older teeth show signs of irregular resorption and irregular closure of the root canal (text-fig. 12 b). The root canal becomes occluded (text-fig. 1 2c), and finally, heavily worn teeth may be resorbed at the base by the surrounding bone (text-fig. 12 d). The teeth are circular to oval (long axis transverse to axis of jaw) in cross-section. A transverse section of a Stenaulorhynchus maxilla (text-fig. 116) shows how successive longitudinal rows of ‘lingual’ teeth grow down into the teeth below. Each tooth bears an enamel cap that extends further medially where it erupts on the inside of the jaw, and the teeth in occlusion are worn flush with the surface of the bone. Teeth are present in middle and posterior portions of the occlusal edge of the dentary. Teeth are absent from the very front of the jaw (text-fig. 1 lc), but they may be seen in the buccal row a short distance back (text-fig. Hr/). Here, the lower portion of the next tooth has an open root canal. Further back, the buccal teeth are deeply rooted, and they slope up and forwards (text-fig. 1 lc,/). The lingual teeth slope up, forwards, and medially. They often have open root canals, and the enamel caps may be seen when the teeth are not in occlusion (text-fig. 1 1/). The teeth are cylindrical, but the shape may be disturbed by irregular resorption as a result of close-packed neighbouring teeth. Tooth shape is just as variable in the maxilla. Newly formed teeth have large root canals (text-fig. 12c), which become partly closed off (text-fig. 126) and occluded, often in an irregular way (text-fig. 12g). The relationships between neighbouring lingual teeth may be complex (text-fig. 126). Resorption of teeth at the base is not seen in the dentary as much as in the maxilla. Bone and tooth histology As with Hyperodapedon , the serial peels and microscopic sections of Stenaulorhynchus jaws have shown a great deal of histological detail. The bone of the jaws includes four types: 1. Avascular lamellar-zonal periosteal bone in the anterior occlusal edge of the maxilla (text-fig. 10a) and around blood vessel canals (PI. 68, fig. 1 ) where the ‘track’ of the vessel as the bone grows may be seen. 2. Laminar fibrolamellar bone with parallel longitudinal primary osteons in regions with teeth (text-figs. 10, 11; PI. 68, fig. 5). text-fig. 10. Longitudinal vertical sections of the maxilla of Stenaulorhynchus (BM(NH) R10008) showing teeth. The locations of the sections are indicated on an outline maxilla, a, lateral tooth row; several largely unworn teeth are shown, one with a wide open root canal. This shows teeth numbered as a-d, 10, 13-17 (text-fig. 8). The boxed area is shown at higher magnification in PI. 68, fig. 2. b , middle tooth row; the youngest (left-hand) tooth has an open root, while the oldest (right-hand) teeth show resorption of the base. This shows teeth numbered as 2 15 (text-fig. 7). c, medial tooth row(s); anterior teeth are heavily worn and resorbed. This shows teeth numbered as 5-21 (text-fig. 9). text-fig. 1 1. Transverse vertical sections of the right maxilla (a, b ) and right dentary (c-f) of Stenaulorhynchus. The locations of the sections are indicated on an outline maxilla and lower jaw. a , b, cross-sections of the maxilla, traced from polished end sections ( BM(N H ) R9276, R9277), and showing the arrangement of teeth in the lateral, middle, and medial rows. Lingual teeth of young (high) Zahnreihen can be seen to have caused erosion of older teeth in b. The dentary: c, anterior portion where the bone is compact and the teeth have been worn away and resorbed; d , further back, worn buccal teeth may be seen on the lateral side; e, in the occlusal area, large buccal and lingual teeth are present, with open root canals; /, at the back, some teeth are in occlusion, and others with unworn enamel caps have just been implanted. The boxed area is shown at higher magnification in PI. 68, fig. 6. Abbreviations: LAT, lateral; me, Meckel’s canal; MED, medial. 760 PALAEONTOLOGY, VOLUME 27 3. Reticular (cancellous) fibrolamellar bone in the centre of bones and away from the teeth (text-figs. 10, 1 1). 4. Bone of attachment: secondary reticular fibrolamellar bone surrounding teeth and cross- cutting laminar fibrolamellar bone. The sheath of reticular bone surrounds each tooth and accompanies it through the jaw (text-figs. 10, 1 1; PI. 68, figs. 3, 4, 6). Between close-packed teeth, the osteons of the bone of attachment may be distorted and flattened (PI. 68, figs. 4, 6). The bone is especially cancellous at the base of teeth, and spongy reticular bone may fill the wide open root canals of newly formed teeth (PI. 68, fig. 2). Micrographs of the bone of Stenaulorhynchus maxillae and dentaries show osteocyte lacunae, and the centripetal arrangement of finely lamellated bone in primary osteons around vascular canals is clear (e.g. PI. 68, figs. 5, 7). Enamel is present in a cap on most unerupted or unworn maxillary and dentary teeth (PI. 68, figs. 3, 6), but the crystal structure is not seen. The enamel cap is very small and covers only the tip of the tooth and the exposed sides of those that erupt first on the lingual side of the maxilla or dentary. The enamel is soon worn away when teeth come into occlusion (text-fig. Ij). The dentine clearly shows radial dentinal tubules (PI. 68, figs. 7, 8), as in Hyperodapedon. Growth lines (contour lines of Owen) in the dentine are very clear (text-fig. 12; PI. 68, figs. 3, 4, 6-8) and these are marked by bends in the dentinal tubules. They clearly show the sequential centripetal filling of the pulp cavity, with periods of slow and fast deposition of secondary dentine. Adjacent teeth show matching sequences of dark and light, and broad and narrow bands (PI. 68, fig. 4), and these catalogue periods of growth (food availability/seasonality?). The effects of resorption of tooth material may be seen in several ways. Adjacent teeth may cause extensive resorption along an arcuate front where they contact their neighbours and give rise to irregular constrictions and bends (e.g. text-figs. 10, 1 1), and, in some cases, teeth are excluded from the jaw margin by others (text-fig. 106). The resorption cuts through the incremental lines in the EXPLANATION OF PLATE 68 The dentition and bone of the maxilla (figs. 1-4) and dentary (figs. 5-8) of Stenaulorhynchus. Sections from BM(NH) R 10008 (figs. 1-4) and BM(NH) R 10007 (figs. 5-8). Fig. 1. Longitudinal vertical section through the anterior portion of the maxilla, showing the laminar fibrolamellar bone with the canals running up and backwards, and two small blood-vessels and their ‘tracks’ of remodelled bone also running up and backwards, x 4-5. Fig. 2. Longitudinal vertical section through the root apex of a posterior tooth, showing the widely open root canal filled with spongy bone and the bone of attachment, set in the normal laminar fibrolamellar bone of the jaw; this section is located in text-fig. 13 a, x 17. Fig. 3. Longitudinal vertical section through some unerupted anterior teeth of the lateral series, showing the bone of attachment, the circumferential growth lines in the dentine and erosion of an older (lower) tooth in the bottom right-hand corner, by a younger (higher) tooth, x 12. Fig. 4. Longitudinal vertical section through some anterior heavily worn teeth (occlusal margin at foot of picture), showing dentine growth lines, fully closed root canals, irregular resorption of the base of the teeth, and associated irregular bone, x 12. Fig. 5. Transverse vertical section, showing typical primary osteons of the laminar fibrolamellar bone, x 30. Fig. 6. Transverse vertical section through two lingual teeth, showing the unworn enamel caps, growth rings in the dentine, and erosion of the side of the older (left-hand) tooth by the younger (right-hand) tooth; this section is localized in text-fig. 11/, x 12. Fig. 7. Transverse vertical section through part of a lingual tooth showing the bone of attachment, the root canal, secondary dentine with growth rings, and radial dentinal tubules, x 25. Fig. 8. A similar transverse vertical section through parts of two lingual teeth with occluded root canals, showing the relationship between the growth lines in the secondary dentine and the radial dentinal tubules, x 45. PLATE 68 BENTON, rhynchosaur dentitions text-fig. 12. Sketches of individual teeth of Stenaulorhynchus : a-d , from the maxilla with the crown facing downwards; e-h, from the dentary, with the crown facing upwards, a, newly implanted tooth with open root canal, some secondary dentine deposition and an unworn enamel cap; b , slightly older tooth, partly worn, with more deposition of secondary dentine, and irregular closure of the root canal owing to extensive erosion by neighbouring teeth; c, two small unerupted teeth with complete enamel caps, and 'interference' in which the younger (upper) one causes erosion of the older (lower) one; d , an old heavily worn tooth in which the root canal has been completely occluded, and resorption of the base has begun; e, a newly implanted lingual dentary tooth in which little secondary dentine has been deposited;/, an older lingual tooth in which the crown is worn, but the root canal is still open; g, a large lingual tooth with unworn enamel cap and extensive occlusion of the root canal by secondary dentine; /?, graphic reconstruction of two lingual teeth which have been implanted so close together that each has interfered with the normal development of the other. Abbreviations: d, dentine; e, enamel; rc, root canal. BENTON: RHYNCHOSAUR DENTITIONS 763 dentine, and may open up the root canal (text-fig. 126). In other cases, the crown of a growing tooth may pass through the root of a tooth in occlusion and cause loss of dentine in the latter (text-fig. 12c; PI. 68, fig. 3). This pattern of extensive resorption is more common in Stenaulorhynchus than in Hyperodapedon , as is resorption of the base of old worn teeth by the surrounding bone (text-figs. 106, 1 2c/). This resorption occurs especially in the maxilla where the bone of attachment can be seen to invade the dentine, often along particular growth lines, leaving the root of the tooth ragged and incomplete (PI. 68, fig. 4). TOOTH IMPLANTATION IN RHYNCHOSAURS Rhynchosaurs have deeply rooted teeth fused to bone of attachment. They show a combination of features of both the thecodont and acrodont systems, and Chatterjee ( 1974) has termed this mode of attachment ‘ankylothecodont’. A characteristic feature is the secondary bone of attachment which has also been identified in Scaphonyx fischeri (Sill 19716, pi. 4c) and H. huxleyi (Chatterjee 1974, p. 230). The latter author described this bone of attachment as ‘spongy in appearance resembling a foam of very small bubbles’, and it is clearly demarcated from the surrounding bone. H. gordoni does not show the ‘bony layered structure ... at the base of some teeth, invading the pulp cavity’ observed by Chatterjee (1974) in H. huxleyi , but reticulate bone has been noted above in the pulp chamber of Stenaulorhynchus . Most early reptiles (e.g. pelycosaurs, captorhinomorphs, early diapsids) had subthecodont ( = protothecodont) tooth implantation (Edmund 1969). In Captorhinus the subthecodont teeth have relatively shallow roots and they are ankylosed into a socket by bone of attachment with ‘no space . . . for a periodontal ligament or other soft tissues between the socket and the base of the tooth’ ( Bolt and DeMar 1975). Most early diapsids have also been stated to have subthecodont teeth, and that is probably the primitive character for the group (Benton 19836), although there is much confusion about the terminology here, and detailed histological information is needed. Some diapsids evolved thecodont teeth in the Triassic (thecodontians, dinosaurs, crocodiles), while others evolved acrodont teeth (sphenodontids) or ‘subpleurodont’ teeth (early squamates). The rhynchosaurs evolved a fourth system— ankylothecodont teeth (deeply rooted teeth surrounded by bone of attachment which may also invade the pulp chamber; no ‘socket’ with soft tissues around the teeth; no typical reptilian tooth replacement). The acrodont agamid lizard Uromastix may also show a bony core in the pulp chamber of posterior teeth, but its function is uncertain (Throckmorton 1 979). An analogous condition also occurs in adult Sphenodon. Secondary bone grows round the bases of the teeth and encloses many of them in shallow alveoli. The teeth are still firmly fused with the base and sides of the alveoli and are thus not thecodont. Howes and Swinnerton (1901 ) and Harrison (1901) describe this condition in Sphenodon as ‘hyperacrodont’. TOOTH ADDITION IN RHYNCHOSAURS The multiple row dentition of the rhynchosaur maxilla and dentary has been interpreted as the retention in the adult of an embryonic type of dentition, with the diagonal tooth rows equivalent to the Zahnreihen of Woerdemann (1921) (Edmund 1960, p. 59; 1969, p. 153; Chatterjee 1974, p. 234). The captorhinomorphs Captorhinus , Labidosaurikos , and Moradisaurus also retain apparently similar diagonal rows of functional teeth in the adult, and a brief consideration of these early reptiles may throw light on the development of the rhynchosaur dentition. Multiple tooth rows in captorhinomorphs Early captorhinomorphs, such as Eocaptorhinus laticeps from the lower Permian of Texas and Oklahoma, had a single-row dentition of subthecodont teeth in premaxilla- maxilla and dentary (text-fig. 13 a). There is evidence of tooth replacement from below (replacement gaps, replacement scars), and the tooth rows have been divided tentatively into Zahnreihen (Heaton 1979). 764 PALAEONTOLOGY, VOLUME 27 The slightly later C. aguti (lower Permian, Texas, New Mexico, Oklahoma) normally has three or four clear subparallel postero-medially directed rows (total range, one to eight rows) of subthecodont teeth on premaxilla-maxilla and dentary (text-fig. 13fi, c). Bolt and DeMar (1975) and Ricqles and Bolt ( 1 983) have demonstrated that all teeth were continuously replaced and that the number of teeth and of rows did not depend on the size of the animal. Each diagonal row is interpreted as a Zahnreihe, and tooth replacement proceeded in such a way that matching diagonal rows of lower and upper teeth were maintained for an efficient shearing jaw action. Teeth were replaced from a lingually situated dental lamina at the posterior end of each Zahnreihe, and teeth were lost anteriorly/labially. There was a limit to this addition, and at times whole Zahnreihen would be lost anteriorly. ANT text-fig. 13. Occlusal views of the jaws of captorhinomorphs, showing the teeth in outline: a , right dentary of Eocaptorhinus ; b, right dentary of Captorhinus ; c, right maxilla of Captorhinus ; d, right maxilla of Labidosaurikos\ e, right maxilla of Moradisaurus. Zahnreihen are indicated with thin lines; two alternative interpretations are given for Labidosaitrikos, in terms of Zahnreihen ( GG QJ Oh £ c w) •G S-H (75 £ e £ £ * o 2 c o >>T3 o .2 £ G r? o >5 D- .& x S2 d) •S > g ^ o ^ £ 2 2 CD 33 GG CD -(— 1 G U-4 o O CD s dJ £, ’o bX) CD Cg 3 C/5 *-• — - C/5 G II Cm O > C/5 G O G CD CD £ T3 C (D — 33 G C/5 G CD G CD O £ CD C/5 C . r. <33 15 c/5 G a O C/5 gxj ’ c ’ C 15 o CD II ‘5b JO O o g: CG O 2 UJ J CQ < H GG CD Q* G o3 o3 G bij G~ § Z o a ^ C 5b £ c ^ c CD td) b -Jx E X - V a.l»I .x £ * 60 E “cZ c “oo .2 2 § *So” — S E o a3 aj Gh c C/D co C - CD : g > o : g 3 03 ! . -G ( /ML)! r T3 _J > G r : £ cs j 0.2 o ia ,ZS UE o. ca ■3! X g“ E « ' x*'1 c y- ’—l o3 <4- c G G Gy • Gy x O > U £ c 03 (D §g GG 03 0> o £ a _ ^ •H 03 fj.H • — £ £ G £ £ C/5 G 33 GO c/5 03 2(J C3 03 03 £ 03 o3 S udSS E E 3 e £ ’5b T3 E 2 r- vo ov GG A\ .g gg A\ A o3 33 oa o- P 8 ■§ § ^ Oh S §- r- vo G o bD G GG O G O ? GO O GG A a VJ table 1. Morphological characteristics and measurements of all species of the genus Pulvinites; junior synonyms in parentheses. Under lateral profile, O = oval; T = trigonal; st, cone, and conv = straight, concave, and convex posterior dorsal margin respectively. Species; synonyms in inverted commas Lateral profile Inflation weak, moderate, or strong Max. length (1) mm Max. height (h) mm ]>, =, or < h Length of ligament mm Number of ligamer pits Length 1 Age Geographic occurrence ALMER: Length of ligament No. of pits adansonii T, st-conc weak 53 46 1 Ssh 6-9 7-8 0-9 1-2 44-6-3 Maastrichtian Fresville, Manche, 73 m Normandy ‘ auriculus ’ T cone weak 51 47 1 h 6 Cenomanian Hajula, Lebanon 00 ‘ anlarclica ' O, T cone weak 52 49 1 > h 10-11 9-11 10- 11 4-3-52 U. Campanian or Seymour Is., O Maastrichtian Antarctica z argenteus T conv, O, weak- 53 52 1 >, = 9 15 10-15 0-9 10 2-9-3-9 Maastrichtian Texas, Mississippi, o st or cone mod. or < h Tenessee, Alabama ■n rupellensis O-oval mod. 30 50 1 < h 5-11 5-12 0-8- 1-5 2-1-54 Kimmerideian nr. La Rochelle, strong Charente Maritime c ‘ triangularis ' oval w. mod. — — 1 exempla T cone weak 131 111 1 > h 36 20 1-8 3-7 Recent Gabo Is. m Victoria, Aus. PALAEONTOLOGY, VOLUME 27 Genus Pulvinites Blainville, 1824 Type Species. Pulvinites adansonii Blainville, 1824, by monotypy. Synonomy. Hypotrema d’Orbigny, 1853; Foramelina Hedley, 1914 Original diagnosis. ‘Coquille mince, ovale, equivalve, subequilaterale, a sommets bien marques et a peine inclines en avant; charniere composee par huit ou dix dents un peu divergents du sommet et separees par autant de fossettes pour les ligamens; impressions musculaires inconnues’ (Blainville 1824, p. 316). Amended diagnosis. Pulvinitidae of medium size; not auriculate. RV flat or slightly concave, with scar of anterior byssal (pedal) retractor muscle displaced posteriorly by migration of foramen to occupy position just posterior to foramen’s dorsal margin; scar of posterior retractor lies just ventral to this, off the foramen’s ventro-posterior margin. LV weakly to strongly convex, with anterior byssal (pedal) retractor scar lying just beneath centre of ligamental area, just posterio-dorsally of the apex of the posterior retractor scar (text-fig. 1). Discussion. All previous accounts have credited Defrance with authorship of both the genus and the type species. However, Defrance’s (1826a, b) entry in the Dictionnaire des Sciences Naturelle under the entry ‘Pulvinite’ was preceded by that of Blainville (1824) under ‘Mollusca’, and it is as a sub- category of this entry that the above diagnosis, albeit using the name ‘ Pulvinites Adansonii Defr.’, was first given. Blainville’s entry refers to the volume of plates of the same work which bears a publication date of 1826. However, even if this date is wrong and the relevant plate (pi. 88, fig. 3) had in fact already been published by the time that Blainville’s entry was published, it can still not be regarded as a valid description since the plate caption uses only the vernacular name ’Pulvinite d’Adanson. {Defy . There is no doubt, therefore, that Blainville, 1824, is the true author of both genus and type species. The type material has been destroyed and is represented only by rather poor photographs (Freneix 1956). Some of these, however, show muscle impressions and were presumably collected after the original diagnosis, in which it is stated that the scars were unknown, was written. The de Gerville collection topotypes in the British Museum (Natural History) clearly show muscle impressions and details of the foramen and suture. Blainville and Defrance seem not to have been aware of these features. The material from the upper Jurassic (‘Corallien’, now known to be of Kimmeridgian age — see Arkell 1956) of La Rochelle, for which d’Orbigny (1853) created the genus Hypotrema, was discovered in 1826 (d'Orbigny 1853) and seems first to have been referred to in the Palaeontologie Franqaise (d’Orbigny 1847) under the name P. oblonga ( lapsus calami). In the Prodrome (d’Orbigny 1850), however, this species is not mentioned and the La Rochelle material is referred to Pulvinites rupellensis. Presumably the names Pulvinites oblonga and Pulvinites rupellensis applied to the same material. D’Orbigny subsequently, however ( 1 853), noting that the foramen had not been mentioned in the original description of P. adansonii , decided that Pulvinites was no more than an ordinary Perna ( = Isognomon) and that the genus was therefore invalid. It was for this reason that Hypotrema was erected, to encompass Isognomon- like forms which, like the La Rochelle material, truly had an enclosed foramen in the right valve. Since d'Orbigny’s reasoning was erroneous, and since examination of de Gerville’s collection or of Sowerby’s (1833) plate would have proved the validity of Pulvinites Blainville, Hypotrema d'Orbigny, 1853, is an unequivocal synonym. This has been suspected by other workers (Bronn 1852; Deshayes 1864; Stoliczka 1871; Fischer 1886; Stephenson 1941; Vokes 1941; Freneix 1956). Cox (1969) re-defined Hypotrema and distinguished it from Pulvinites on three points: (i) Hypotrema is taller and narrower; (ii) in Hypotrema the byssal retractor and adductor scars touch in the left valve, and (iii) the ligamental area is arched in Hypotrema but straight in Pulvinites. Examination of a representative suite of material shows that valve profile is variable in both forms, and that the taller, narrower condition seen in some Hypotrema is not fully diagnostic and only of specific importance. Similarly, the ligamental area is variably straight or PALMER: REVISION OF PULVINITIDAE 819 arched in both forms. This leaves the relative position of the muscle scars in the left valve: in P. mackerrowi sp. nov., they touch in some specimens and not in others from the same population. None of these features is therefore likely to be of generic significance. Foramelina Hedley, 1914 was erected for the newly discovered F. exempla. Hedley was aware of the similarity of his shell to both Hypotrema d’Orbigny, and to P. argentus Conrad, 1858. However, Hypotrema was rejected as a possible genus because of d’Orbigny’s (erroneous) opinion that the foramen served for passage of a muscle, not a byssus as in Hedley’s shell in which an uncalcified byssus was preserved. P. argenteus Conrad was stated to be assigned to that genus only with doubt (Hedley 1914, p. 71), and this gave him justification for establishment of his new genus. In fact, F. exempla is very similar to both P. adansonii and P. argenteus , differing principally in its larger size (length c.130 mm, cf. not more than 60 mm in the other two Pulvinites', see Table 1), and the corresponding larger number and wider spacing of the ligament pits (both of which increase during ontogeny). Apart from size, the details of musculature, profile, foramen, and suture are virtually identical in both forms and there is no doubt that they are congeneric species. Zinsmeister’s (1978) reservations on inclusion of Foramelina in the Pulvinitidae, based on his perceived differences between the ‘sub-umbonal orifice and groove’ described in Foramelina , and the narrow suture of Pulvinites , are unwarranted. The two conditions are identical, and wholly different from that shown by the species which Zinsmeister calls Melina percrassa Tate (and which he considers a possible relative of Foramelina ), in which the byssal slit, though deep, is both open to the anterior and located in the left valve. All the species considered in this study are represented by at least some specimens whose muscle scars are well-enough preserved to indicate the points of insertion of the byssal (pedal) retraction muscles, as well as the adductor muscle, in both valves (text-fig. 1). These insertion points are in equivalent positions in each species, and their identity is confirmed by examination of the soft parts of P. exempla. Pulvinites adansonii Blainville, 1824 Plate 72, fig. 1 1824 Pulvinites Adansonii Blainville, p. 316. 1826a Pulvinites Adansoni Defrance, p. 107. 18266 Pulvinite d’Adanson Defrance, pi. 88, fig. 3. 1833 Pulvinites Adansonii Defrance; Sowerby, pi. 104. 1941 Pulvinites auriculas Vokes, p. 8, figs. 14, 15. 1956 Pulvinites adansoni Defrance (partim.); Freneix, fig. a, 1 -6. 1978 Pulvinites antarctica Zinsmeister, p. 567, pi. 1 , figs. 1 -4. Types. Syntypes: Defrance Coll., University of Caen, destroyed in 1944, figured by Freneix (1956), figs. 14. Topotypes: Gerville Coll, figured by Freneix ( 1956), figs. 5, 6 (incorporated with Defrance Coll, and destroyed with them); Gerville Coll., British Museum (Natural History), BML 63618, 63619, 65685; British Museum (Natural History), BMLL 40007. Original diagnosis. As for genus (Blainville 1824). Emended diagnosis and description. Orbicular, or trigonal with a straight, slightly convex or slightly concave posterio-dorsal shell margin. In adult, LV weakly inflated with length around 53 mm; length usually slightly greater than height; c.7-1 1 ligament pits on ligamental area of length c.6-1 1 mm; ratio of valve length to length of ligament greater than 4 (Table 1). Scars of adductor and byssal retractor muscles do not touch in LV; division of adductor scar into quick and catch portions is sometimes discernable. Discussion. Sowerby’s (1833) figure of topotypes (BML 63618, 61619) remains the best published illustration of this species, those of Freneix (1956) being of poor quality. Vokes’ (1941) P. auriculus was distinguished from P. adansonii on the basis of its being 'more regularly rounded anteriorly and ventrally, and in being somewhat more produced posteriorly’. Presumably he only had Defrance’s 820 PALAEONTOLOGY, VOLUME 27 original figure to go on which is somewhat pointed anteriorly and not particularly produced posteriorly. In fact, Sowerby’s (1833) figure shows the typical outline of P. adansonii, which is identical to that shown in Voices’ figures of P. auriculus. Zinsmeister (1978) did not consider the characteristics of P. adansonii when he erected his species P. antarctica. The holotype of his species has a slightly more orbicular outline than typical adansonii (not diagnostic), but the paratype is well within the limits of variation of that species. Pulvinites argenteus Conrad, 1858 Plate 72, figs. 3, 4 1858 Pulvinites argentea Conrad, p. 330, pi. 34, fig. 5. 1956 Pulvinites adansoni Defiance (partim.); Freneix. Types. Conrad’s types from Owl Creek, 4 km north-east of Ripley, Tippah Co., Mississippi are stated by Stephenson (1941 ) to be lost. He further states that specimens USNM 20667 and USNM 73635 in the National Museum in Washington (Smithsonian Institution) are good topotypes. A topotype (USNM 32741) is also illustrated (PI. 72, figs. 3, 4). Original diagnosis. ’Transversely subovate, compressed; perforated valve slightly concave; foramen oval, from its upper margin a channel extends to the apex; substance of shell silvery, thin; cardinal plate broad, with about thirteen radii or teeth; the lower valve presents within the appearance of two muscular impressions, the one above the other, the former striated’ (Conrad 1858, p. 330). Amended diagnosis. Orbicular, or trigonal with a straight, convex, or slightly concave posterio-dorsal shell margin. In adult, LV weakly or moderately inflated with a length of around 53 mm; length equal to, slightly greater than, or slightly less than height; 10 1 5 ligament pits on a ligamental area of length c.9-15 mm; ratio of valve length to length of ligament less than 4 (Table 1). Scars of adductor and byssal retractor muscles do not touch in LV. Discussion. Freneix (1956) wrongly regards this species as identical to P. adansonii. Although similar, the hinge plate is consistently broader for the equivalent valve length (Table 1) as was stated by Conrad in the original description. Conrad’s original figure showed a particularly orbicular specimen; the more typical outline is the trigonal one figured by Wade (1926) and reproduced in Cox (1969). Pulvinites rupellensis d'Orbigny, 1850 Plate 72, fig. 2 1847 Pulvinitis oblonga d’Orbigny; p. 522 ( nomen nudum). 1850 Pulvinites rupellensis d'Orbigny; p. 24. EXPLANATION OF PLATE 72 Pulvinites spp.: s = suture; f = byssal foramen; a = adductor muscle; apr = anterior pedal retractor; ppr = posterior pedal retractor. Fig. 1. P. adansonii , latex impression of mould of interior of right valve of topotype; Gerville Coll., British Museum (Natural History) BML 63618. x 1-3. Fig. 2. P. rupellensis , internal mould of left valve of topotype d’Orbigny Coll. Museum National d’Histoire Naturelle, I.P.M. —ORB 4314A. x 2- 1 . Figs. 3, 4. P. argenteus , internal views of right and left valves of hypotype; Ripley Formation, Coon Creek, Tennessee, USNM 32741. x 1-3. Fig. 5. P. mackerrowi , internal view of right valve from which the inner shell layer has been lost; holotype, x 2-7. Wood Eaton quarry, nr. Oxford; Ardley Member of White Limestone Formation. Oxford University Museum J40183. Fig. 6. P. exempla, internal view of right valve of paratype. Australian Museum, Sydney, C37004. x 0-5. PLATE 72 PALMER, Pulvinites 822 PALAEONTOLOGY, VOLUME 27 1853 Hypotrema rupellensis d’Orbigny; p. 437, pi. 10, figs. 1-5. 1853 Hypotrema triangularis d’Orbigny; p. 437, pi. 10, figs. 6-12. 1929 Hypotrema Rupellensis d’Orbigny; Cottreau, pi. 20, figs. 17-19. 1931 Hypotrema Rupellensis d'Orbigny; Cottreau, p. 1. Types. Twenty-two specimens from d’Orbigny’s collection are preserved in the Museum National d’Histoire Naturelle in Paris. These are now all labelled H. rupellensis, and come from La Rochelle (5 specimens, collective No. 4314) and Estre (17 specimens, collective No. 4314a). Estre is today a suburb of La Rochelle, called Aytre. The Paris specimens include those figured by d’Orbigny (1853) as H. rupellensis, and one of the figured syntypes of H. triangularis (labelled as H. rupellensis). The other figured syntypes of H. triangularis appear to be missing. Clearly, the material that d’Orbigny considered to constitute the second species has now been combined with that of the first. At the time of erection of the nomen nudum, Pulvinitis oblonga, d’Orbigny (1847) stated that he had only eight specimens. Whether this number constitutes the type suite of Pulvinites rupellensis d'Orbigny, 1850, and when the other specimens now in the d’Orbigny collection were added, are unknown. Other material from the La Rochelle region includes specimens nos. 255-258 in the collections of the Museum d’Histoire Naturelle at La Rochelle, and Oxford University Museum nos. JZ 1791-JZ 1801 from the coast at Marsilly just north of La Rochelle (author’s collection). All specimens are of Kimmeridgian age. Original diagnosis and description. ‘Espece remarkable, ovale ou oblongue, fixee sur le polypiers’ (d’Orbigny 1850, p. 24). Amended diagnosis and description. Orbicular, sub-trigonal, or, more commonly, dorso-ventrally elongated oval, with moderately or strongly inflated LV. Height in adult up to c.50 mm and usually much higher than long; 5-12 ligament pits on a ligamental area up to c. 11 mm (Table 1). Adductor muscle scar in LV more dorsal than in Cretaceous species so that its dorsal edge is in contact with the ventral side of the main byssal retractor scar. Widely spread threadlike radial riblets on exterior of RV. Discussion. The main component of growth in this species is in the ventral direction, rather than anterio-posteriorly. The precise rate at which size increase in the anterior-posterior direction occurs is variable, so that some specimens are more or less symmetrically oval, whereas others have a more triangular profile like that of an egg. These latter specimens, with the shorter ligamental area (and thus less ligament pits), were separated into the species triangularis by d’Orbigny ( 1853) but in fact the range of variation is continuous. The most inflated specimens show the plane of the ligamental area rotated through 90° so that, as in Anotnia with a similar lifestyle, the opening thrust of the ligament must have been dorsally directed. Pulvinites mackerrowi sp. nov. Plate 72 fig. 5; text-fig. 2 1969 Nucula sp.; McKerrow, Johnson and Jakobson, Table 8. 1969 Exogyra sp.; McKerrow, Johnson and Jakobson, Table 8. 1969 Epithyris oxonica Arkell (partim.); McKerrow, Johnson and Jakobson, Tables 8, 9. Types. (Oxford University Museum Collections); holotype: J43401; paratypes: J43402-J43486, J28030, J28035-J28038, J28041. Deriviation of name. After Dr. W. S. McKerrow, who first collected large populations of the species. Diagnosis. Orbicular to (rarely) trigonally suboval, opisthogyrate with strongly inflated LV and well- developed umbo; approximately equidimensional, with length and height up to c.30 mm (Table 1). Outer shell surface smooth, or with xenomorphic ornament. Scars of adductor and byssal retractor muscles in LV may or may not touch; see Table 1 for measurements. Discussion. This species is found in the Bathonian White Limestone Lormation of Oxfordshire, where it occurs commonly in beds with a high proportion of corals and other epifauna (Palmer 1979). It is often found in life position on coral fronds or brachiopods; the xenomorphic ornament is one reason why it has been overlooked in the past. The combination of strong inflation, relatively small size, and the orbicular outline distinguish it from other species. PALMER: REVISION OF PU L VI N ITI D A E 823 text-fig. 2. Pulvinites mackerrowi sp. nov., left valves of paratypes, both coated with ammonium chloride, a, sediment mould of internal, showing ligament pits and muscle scars, J43486, x 3 0. b, external view, J43403, x 3 0. Pulvinites exempla (Hedley, 1914) Plate 72, fig. 6 1914 Foramelina exempla Hedley, p. 71; pi. 11, fig. 6; pi. 12, figs. 7, 8. Types. Australian Museum, Sydney, No. C37003 (holotype); C37004 (paratype). Original diagnosis and description. ‘Shell equivalve, equilateral, large, flat, discoidal, margin irregular in outline. Valves thick, composed of brittle, imbricating lamellae which both include and are overlaid by a thin membranous epidermis, where eroded of a silvery sheen, interior a dark bronze. Perforation in the right valve about 10 mm in diameter, median and subumbonal, internally with a raised margin, externally excavated as an oblique furrow ascending to the vertex. From the margin of the perforation a suture leads to the anterior extremity of the hinge plate. Byssus a dense bundle of threads about half an inch long. Hinge line about 43 mm long. Area much eroded above and traversed by about twenty-two narrow ligamental grooves which slightly radiate from above. Externally the hinge gapes when the valves are closed as in Melina [ = Isognomon Lightfoot], Adductor small, subcircular about its own breadth below the perforation. Byssal retractor rather larger than the adductor and immediately above it in the left valve. Pallial margin entire, about half an inch within the ventral margin. Height, 120 mm; length, 130 mm’ (Hedley 1914, p. 71). Amended diagnosis and description. Distinguished from other species of Pulvinites by its large size when adult, and by features related to size such as the greater number of ligament pits and their greater size and spacing (Table 1 ). Discussion. P. exempla is identical to what would be expected in a scaled-up P. adansonii. Hedley (1914) was aware of the similarities of his species to Hypotrema d'Orbigny and to P. argenteus Conrad, but stated that reference of the latter species to the genus Pulvinites was doubtful. Now that the Cretaceous members of the genus are well known, it is clear that the inclusion of P. exempla in a genus of its own is quite unjustified. Acknowledgements. My grateful thanks go to W. .1. Kennedy, N. J. Morris, and A. Wyatt for discussions. 824 PALAEONTOLOGY, VOLUME 27 REFERENCES arkell, w. J. 1956. Jurassic Geology of the World , 806 pp. Oliver and Boyd, Edinburgh and London. blainville, H. M. D. de. 1824. Mollusca. Dictioiwaire des Science Naturelles, 32, 316. bronn, H. G. 1 852. Lethaea Geognostica , V, Vierte Periode , Kreide-Gebirge, 3rd edn., 412 pp. Schweizerbart’sche Verlagshandlung, Stuttgart. conrad, t a. 1858. Observations on a group of Cretaceous fossil shells found in Tippah County, Miss., with a description of 56 new species. J. Acad. Nat. Sci. Philadelphia , series 2d, 3, 323-336. — 1860. Descriptions of new species of Cretaceous and Eocene fossils of Mississippi, and Alabama. Ibid. 4, 275-298. — 1867. Descriptions of new genera and species of fossil shells. Am. J. Conchology , 3, 8-12. cottreau, J. 1929. Types du Prodrome de Paleontologie Stratigraphique Universelle d’Alcide d’Orbigny. Annals de Paleontologie, 18, pi. 20. 1931. Types du Prodome de Paleontologie Stratigraphique Universelle d’Alcide d’Orbigny. Ibid. 20, 1 -40. cox, L. r. 1969. Family Pulvinitidae. In moore, r. c. (ed.). Treatise on Invertebrate Paleontology, Part N, Mollusca 6, Bivalvia. Geological Society of America and University of Kansas Press, Lawrence, Kansas, N326. defrance, m. j. l. 1826a. Pulvinite. Dictionnaire des Sciences Naturelles, 44, 107. — 1826 b. Dictionnaire des Sciences Naturelles, volume des planches, Zoo/ogie, Conchy liologie et Malacologie, pi. 88. deshayes, g.-p. 1864. Description des animaux sans vertebres decouverts dans le Bassin de Paris, 2nd edn., vol. 2, 968 pp. Bailliere et Fils, Paris. d'orbigny, a. c. v. d. 1847. Paleontologie francaise. Description de Mollusques et Rayonnes fossiles. Terrains Cretaces; tome troisieme; Lamellibr cinches, 807 pp. V. Masson, Paris. — 1850. Prodrome de Paleontologie Stratigraphique Universelle; quatorzieme etage, Coral lien, 427 pp. V. Masson, Paris. — 1853. Note sur le nouveau genre Hypotrema. J. Conchyliologie, 4, 432-438. fischer, p. h. 1886. Manuel de Conchyliologie, fasicule 10, 1369 pp. F. Savy, Paris. freneix, s. 1956. Paleontologie Universalis, N.S. No. 277 Pulvinites adansoni, 4 pp. Centre d’Etudes et du Documentation paleontologiques. Museum National d’Histoire Naturelle, Paris. hedley, c. 1914. Zoological results of the fishing experiments carried on by F.I.S. ‘Endeavour’ 1909-10, under H. C. Dannevig, Commonwealth Director of Fisheries: Mollusca. Publication of the Commonwealth of Australia Department of Trade and Customs; Fisheries, 2, 63-74. mckerrow, w. s., Johnson, r. t. and jakobson, m. e. 1969. Palaeoecological studies in the Great Oolite at Kirtlmgton, Oxfordshire. Palaeontology, 12, 56-83. palmer, r. j. 1 979. The Hampen Marly and White Limestone Formations: Florida-type carbonate lagoons in the Jurassic of central England. Ibid. 22, 189-228. sowerby, G. b. 1833. The Genera of Recent and Fossil Shells, part 40, 14 pp. London. Stephenson, l. w. 1914. Cretaceous deposits of the Eastern Gulf region. U.S.G.S., Professional Papers, 81, 1 -78. — 1926. The Mesozoic rocks. Geol. Surv. Alabama, Special Reports, 14, 231-250. — 1941. The larger invertebrate fossils of the Navarro Group of Texas. Bull. Texas Univ. 4101, 1-641. stoliczka, F. 1871. Cretaceous fauna of Southern India , Vol. 3, Pe/ecypoda, 409 pp. Palaeontologica Indica (Memoirs of the Geological Survey of India), London. vokes, h. e. 1941. Contributions of the palaeontology of the Lebanon Mountains, Republic of Lebanon. Am. Mus. Novitates, 1145, 1-13. wade, B. 1926. The fauna of the Ripley Formation on Coon Creek, Tennessee. U.S.G.S. Professional Papers, 137, 1-192. zinsmeister, w. j. 1978. Three new species of Pulvinites (Mollusca: Bivalvia) from Seymour Island (Antarctic Peninsula) and Southern California. J. Paleont. 52, 565-569. T. J. PALMER Department of Geology University College of Wales Typescript received 9 October 1983 Aberystwyth Revised typescript received 24 April 1984 Dyfed SY23 3DB STEM MORPHOLOGY OF THE RECENT CRINOID CHLADOC RINUS ( NEOCRINUS ) DECORUS by S. K. DONOVAN Abstract. The modern crinoid Chladocrinus ( Neocrinus ) decorus (Wyville Thomson) has a column which is divided into a short proximal growing region, and a longer distal portion in which the arrangement of columnals remains constant. Lumen shape is pentagonal just beneath the cup but is circular for most of the length of the stem. Articular facet morphology shows considerable variation proximally but is constant in the dististele except between nodals and infranodals, where articulation is synostosial rather than symplexial. Cirri are quite different from the column, being elliptical with elliptical lumina and synarthrial articulation between cirral ossicles. Lumen shape is more constant than columnal outline or facet morphology, so it is concluded that the Russian system of naming morphogenera, which emphasizes lumen shape, is preferable to that of Moore (1939a), which stresses columnal outline. I t is possible to describe crinoid columnals and to group those of a similar morphology together, but the reconstruction of complete crinoid stems from dissociated ossicles is generally impracticable. New vertebrate taxa can often be described from minimal evidence (e.g. Rieppel 1982) because the number of bones in the skeleton and their arrangement remains reasonably constant throughout the group, despite the great variation in vertebrate morphology. Crinoid columns, however, differ from the vertebrate skeleton by showing great variation in the number, shape, and arrangement of units (i.e. columnals, which are broadly analagous to bones), even within members of the same genus (compare, for example, the columnals of Colpodecrinus quadrifidus Sprinkle and Kolata, 1982 and C. forbesi Donovan, 1983a). Dissociated columnals are not usually sufficiently distinct for them to be classified with confidence, and can usually only be named by inclusion within artificial morphogenera (Moore 1939a; Yeltysheva 1955, 1956). Few detailed studies have been made of the variation shown within a crinoid column, so there is only minimal reference material available. Jeffords and Miller (1968) examined four columnal taxa for which numerous good specimens were available. Growth stages were determined by examination of internodal insertion, the relationship between columnal diameter and the number of culmina (radiating ridges on the articulation surface), and the nature of longitudinal sections. Each of the columnal types showed a distinct sequence of development. Comparison of columnal diameter with the number of culmina was shown to be a useful method of ontogenetic analysis, although good preservation is necessary for this to be applied to fossil material. Roux graphically determined the relationship between columnal height and diameter in the Bourgueticrinina (1977a) and the Millericrinidae (1978). This is reasonable if good material is available, but Ordovician columnals, for example, are often preserved as external moulds. Unless a counterpart is available, it is not usually possible to determine the height of columnals preserved in this manner. A number of parameters were used in bivariate analyses of Devonian columnal morphospecies by Le Menn (1981); of these, graphs of articular facet diameter against columnal diameter, lumen diameter against articular facet diameter, and columnal height against columnal diameter are of the greatest general utility. Plots of columnal diameter against lumen diameter and against columnal height have proved most useful in the analysis of Ordovician columnals (Donovan 1 9836, pp. 68-74). The problem remains, however, that little detailed information is available concerning the morphology of columns in individual crinoids. In this paper the stem of the recent isocrinid Chladocrinus ( Neocrinus ) decorus (Wyville Thomson, 1864) (Breimer 1978, p. T9, footnote; Rasmussen 1978, p. T857) is considered. Reichensperger (1905) described the anatomy of this species (Palaeontology, Vol. 27, Part 4, 1984, pp. 825-841, pis. 73-76.| 826 PALAEONTOLOGY, VOLUME 27 and determined the organization of soft tissues in the axial canal (see Jefferies 1 968, p. 259, fig. 5, after Reichensperger). Roux (19776), in his review of isocrinid stalk joints, described the articular facets of C. decorus. In the present study the whole stem of C. decorus is described; particularly the changes that occur between different parts of the column, so that comparison can be made when similar variations are detected in fossil columns. C. decorus is an articulate, and therefore in a different subclass to all Palaeozoic crinoids, but it is probable that the morphological variations which it shows are determined by stem function and that similar variations which occur in other pelmatozoans, regardless of age or systematic position, are also functional in origin. A recent crinoid, rather than a fossil species, has been chosen for this study of gross stem morphology because of the ease with which the column can be disarticulated and sectioned, examined functionally (e.g. comparing the flexibility of the proxistele and dististele), and prepared for SEM study of the microstructure. Stem terminology follows Moore, Jeffords, and Miller (1968), Roux (19776), Ubaghs (1978), and Webster (1974). Features of the articular facet of C. decorus are illustrated in text-fig. 3. Other terminology applied to the crinoid stem is explained in text-figs. 1, 5, and 6. Terminology of stereom microstructure follows Smith (1980). The stem is divided into two distinct skeletal systems: the column (composed of individual calcite plates called columnals) and associated cirri (jointed appendages, in this example adapted for attachment, composed of cirral ossicles). MATERIALS AND METHODS This study is based on two specimens exchanged with the Smithsonian Institution, Washington DC, USA. The exchange was initiated by Dr. David L. Pawson of the Department of Invertebrate Zoology at my request. Both specimens formed part of a group registered with the United States National Museum as USNM 12356 Isocrinus decorus and are referred to in the text as USNM 12356/1 and 12356/2. Labels attached to the specimens state ‘ Pentacrinus decorus Wyv. Th. Off Havana, Cuba; Sta. 2,319-2,350. u.s. fish commission. Steamer Albatross, 1885’. Both were damaged but the most complete specimen (USNM 12356/1) had an entire stem, broken into a number of pieces, which was disarticulated to obtain most of the ossicles shown in Plates 73-75. Ossicles from USNM 12356/2 are illustrated in Plates 73 and 76, and the 'whole’ specimen in text-fig. 2. Both specimens are to be returned to the Smithsonian Institution. Ossicles were disarticulated using Milton 2 sterilizing fluid diluted with an equal volume of water. The process was observed using a 'Wild’ binocular microscope so that ossicles could be removed from the solution as they dissociated. This enabled ossicles to be selected for examination before complete disarticulation into a mass of plates. Columnals treated in this manner had all soft tissues dissolved. The ossicles were then washed and dried with a gentle heat source. Plates for examination under SEM were mounted on stubs using 'Durofix' glue or doubled-sided 'Sellotape.' Coating of stubs with 60% gold-palladium was carried out in the Department of Botany, University of Liverpool. COLUMN AND COLUMNALS The stem of C. decorus can be divided into two distinct regions, a proxistele (i.e. the stem proximal to the cup), and a relatively longer dististele (at least five times as long, although this is probably variable due to autotomy; Emson and Wilkie 1980) (text-fig. 1a). Stems such as this example, which can be divided into two or more morphologically distinct regions, are called xenomorphic. Using Webster’s notation (1974; N = nodal, 1 = priminternodal, etc.; see text-fig. 5), and dividing the column into noditaxes (text-fig. Ib), the columnals which can be seen in the proximal region of USNM 12356/2, without sectioning, are: (cup)-2 1 2N -2 1 2N (first cirrinodal)-2 1 2N -2 1 23N-323 1 323N-323 1 434243N- 3424341434243N (25-0 mm below base of cup)-3424341434243N-3424341434243N (54-5 mm). The proximal stem of USNM 12356/1 was (before disarticulation): (cup)-?N-N-lN (first cirrinodal)-lN-lN-212N-212N-3231323N- 3231323N (11-5 mm)-3424341434243N (19-0 mm)-3424341434243N (25-5 mm). cup ® text-fig. 1 (left), a-c, hypothetical stem showing three morphologically distinct regions (only two are seen in Chladocrinus (Neocrinus) decorus ), the proxistele, mesistele, and dististele. A, schematic diagram of the complete stem; cirri (attachment structures) only occur in the dististele of this example, unlike the stem of C. (N.) decorus. b, part of the mesistele (middle stem). C. (TV.) decorus has a column which can be divided into two distinct regions only, and is not regarded as having a ‘middle stem’ unlike many other crinoids. The largest columnals are nudinodals (i.e. nodals that do not bear cirri). Priminternodals are larger than secundinternodals. A taxis is a sequence of ossicles. The internoditaxis includes all ossicles (internodals) between two sequential nodals. The noditaxis comprises a nodal with associated internoditaxis. c, cirrinodal of the dististele; cirri articulate on small facets, called cirrus scars, on the nodal latera. text-fig. 2 (right). Chladocrinus (Neocrinus) decorus (Wyville Thomson). USNM 12356/2. The most distal part of the column, many of the cirri, and some of the arms are missing. Scale bar represents 2cm. 828 PALAEONTOLOGY, VOLUME 27 LUMEN text-fig. 3. Chladocrinus (Neocrinus) decorus. USNM 12356/1. Proximal articular facet of a nodal (traced from an enlargement of PI. 74, fig. 2). The Webster chart for USNM 12356/2 represents the entire proxistele, whereas the chart for 12356/1 is slightly less. The variation which can occur in the arrangement of columnals in the proxistele is apparent when the two charts are compared. Breimer (1978, p. T23, fig. 10) illustrates the most proximal part of the column of another specimen of C. decorus. The most proximal columnals apparent are very thin. Between these and the base of the cup occur a number of cryptic ossicles (for example, three between the cup and ?N of USNM 12356/1 ) which are very small, pentastellate in outline, a single stereom layer thick, and extremely delicate; these are presumably incipient nodals. Similarly, internodals of the same appearance are formed in the proxistele between more mature columnals, and are also hidden from view until they are as wide as the adjacent columnals. These cryptic columnals are accommodated during their early growth stages in concave regions of the proxistele, either at the base of the cup (PI. 73, figs. 2, 8) or in the petaloid zone of concave columnals (PI. 74, fig. 7; the petaloid zone is the principle area of ligamentation of EXPLANATION OF PLATE 73 Figs. 1-8. Chladocrinus (Neocrinus) decorus (Wyville Thomson). USNM 12356/1 (figs. 1, 2, 6-8) and 12356/2 (figs. 3-5). Scanning electron micrographs of the stem and cup. 1 , noditaxis N3424341434243; the upper nodal is the more proximal, x 12. 2, 7, 8, the base of the cup: 2, general view of the whole cup, x 6; 7, lumen at the base of the cup, x 80; 8, a single basal plate, x 25. 3, 5, cryptosymplexies of distal nodal facets, x 56 and x 120 respectively. 4, articular facet showing ligament fibres concentrated in petaloid zones, x 12. 6, poorly preserved proximal internodal (cf. text-fig. 4c), x 60. PLATE 73 • • *-•••« U’( DONOVAN, Chladocrinus ( Neocrinus ) 830 PALAEONTOLOGY, VOLUME 27 text-fig. 4. Chladocrinus (Neocrinus) decorus. a-c, USNM 12356/1. Transverse (a) and longitudinal (b) sections through nodals at the junction between the axial canal and cirral canals, i.e. canaliculi. a, three canaliculi cut by section, all five being completed as dashed lines. Arrows point down dip. Orientation of diagram looking distally, i.e. away from the cup. Tubuli shown as black dots and axial canal of the columnal shaded black. Irregular oval areas correspond to rectilinear stereom. b, longitudinal section x-y from a, showing the orientations of the axial canal and canaliculi (cup towards the top of the page), c, reconstruction of the facet of a proximal internodal (cf. PI. 73, fig. 6); some stereom pores outlined (lower left), d-g, sequence of formation of new cirrals (numbers refer to order of formation, N = nodal), d, most proximal cirrus, formed of primary cirrals 1 and 2. e, cirral ossicle 3 intercalated between 1 and 2 on a slightly more distal nodal, f, cirral ossicle 4 appears between 1 and 3. G. cirral ossicle 5 is intercalated between I and 4. h, USNM 12356/2. Junction between the dististele (the first columnal of which is the lowest quartinternodal) and the proxistele; fossulae (between columnals) and canaliculi shaded black, a-c and h are camera lucida drawings. The longest scale bar refers to a and b. DONOVAN: RECENT CRINOID STEM 831 a columnal facet). A damaged internodal in its early growth stages has been examined (PI. 73, fig. 6) and a reconstruction of the columnal made (text-fig. 4c). A slightly more advanced columnal is also shown (PI. 74, fig. 6). The crenularium (i.e. the alternating arrangement of ridges and grooves— culmina and crenellae, respectively— which interlock to provide a rigid suture between adjacent columnals) has yet to develop. The proxistele is much more flexible than the dististele. This is because it is composed of thin columnals of variable height and diameter which are able to adjust to changes of orientation of adjacent columnals more easily than the thicker columnals of the dististele. The number of internodals increases away from the cup until a maximum number of columnals in a noditaxis is attained i.e. fourteen columnals, 342434 1434243N (PI. 73, fig. 1). No further internodals are added after this configuration has been attained, and all noditaxes of the dististele have this arrangement of columnals. A theoretically perfect noditaxis would have two further quart- internodals, i.e. 434243414342434N. This is not the case, however, and the reason why this particular internoditaxis configuration is attained remains unknown. The internoditaxis (text-fig. 1b) pattern is symmetrical about the priminternodal, implying that formation of new internodals has occurred evenly throughout the noditaxis. The sequence of columnal insertion is similar (but not identical) to that of the Muschelkalk (middle Triassic) species Encrinus liliiformis (text-fig. 5; Linck 1954; Raup and Stanley 1978, p. 48 fig. 3.2) and numerous other examples from the fossil record. The first cirrinodals occur close to the bottom of the cup, although the nodals closest to the cup are nudinodals; therefore, cirrinodals develop from nudinodals. Columnal calcite is resorbed to form the axial canals (canaliculi) of the cirri which are then formed externally. The cirri are initially extremely short and non-functional as holdfast structures. The first cirri of USNM 12356/1 were composed of two ossicles only. The terminal cirral ossicle is conical and retains the same position throughout growth of the cirrus, becoming claw-like when mature (PI. 76, fig. 5). The most proximal cirral ossicle © © text-fig. 5. Idealized growth sequence of a heteromorphic stem, based on Raup and Stanley (1978, fig. 3.2) and Linck (1954, fig. 8). A nodal, N (b), is formed beneath the cup. A second nodal is intercalated (c). Prim- (d), secund- (e), and tertinternodals (f) are intercalated between these two nodals (internodal orders labelled 1, 2, and 3, respectively). Further calcite secretion results in all columnals attaining a similar diameter (g). The parts of the stem where stages b-g occur are shown in a. 832 PALAEONTOLOGY, VOLUME 27 has a domed proximal facet (PI. 76, fig. 4) which articulates on the saucer-like cirrus-scar (PI. 74, fig. 8). The cirri most proximal to the cup do not seem to have this arrangement, but such domed proximal cirral ossicles have developed by the third cirrinodal beneath the cup in USNM 12356/2. If new cirral ossicles are added at the cirrus scar, the domed facet would have to become modified to a planar configuration as each new proximal plate is intercalated. This is unlikely, and it is more probable that new plates are formed between the first two cirral ossicles. These plates are thus of primary importance in cirrus formation, always remaining the most proximal and most distal ossicles, and are here called primary cirrals. The process of cirral ossicle formation is shown diagrammatically in text-fig. 4d-g. The columnals of the proxistele are initially extremely thin, irregular pentastellate in outline, and lack a crenularium (PI. 73, fig. 6; text-fig 4c). Columnals grow by increasing in thickness and becoming more regular in outline (PI. 74, fig. 6). Such columnals are still thin and lack crenulae but show definite tubuli (broad stereom canals arranged around petaloid zones; text-fig. 3). Columnals become apparent externally only when a crenularium is developed, i.e. when the columnals are seen to articulate with adjacent ossicles. This is associated with a further thickening of the columnal and an alteration in outline from pentastellate to petaloid pentagonal (PI. 74, fig. 1). The ossicle now shows all features of the mature, symplexial articular facet (text-fig. 6b) of all columnals of the dististele (some distal articula have a secondary synostosial articulation, described below) (PI. 74, fig. 2; text-fig. 3). This final stage of development is attained by further increasing the height and diameter of columnals, increasing the number and amplitude of the crenulae, and reducing the facet concavity of the early formed columnals. The change from proxistele to dististele occurs after the last noditaxis in which adjacent columnals share five intercolumnal fossulae (Franzen-Bengston 1983). The fossulae of C. decorus are radially orientated, tubular passageways which lie at 72° to each other and are seen as grooves on the articular facet (PI. 74, fig. 1 ), i.e. the grooves of adjacent articula combine to produce a tube. The positions of fossulae in the dististele are represented by the axial grooves. The last proximal columnal of USNM 12356/2 is the most distal nodal of the sequence described above (text-fig. 4h). It is also the first nodal to show a good synostosial articulation (i.e. facets are smooth; text-fig. 6a) with the next distal internodal. © SYNARTHRY PROXIMAL GLYPTOCYSTITID text-fig. 6. Articulations between columnals (ligaments shown as vertical rulings in a-c). a. a synostosis; adjacent facets are planar, b, a symplexy; two interlocking crenularia (text-fig. 3) composed of ridges (culmina) and grooves (crenellae). c, a synarthry; opposed faces rock on adjacent ridges. D, proximal glyptocystitid; a ‘see-saw’, in which a planar facet rocks on a fulcral ridge, (a-c, after Lewis 1980; D, after Paul 1968). EXPLANATION OF PLATE 74 Figs. 1 10. Chladocrinus ( Neocrinus ) decorus (Wyville Thomson). USNM 12356/1. Scanning electron micrographs of columnals. 1. 6, 7, columnals from the proxistele: 1, articular facet, x 12; 6, articular facet of a cryptic internodal, x 12; 7, oblique view of a nodal facet showing the dish made by the interrays (note broad meshwork of lateral stereom which extends into the fossulae), x 25. 2, proximal facet of a nodal of the dististele, x 12. 3, 4, synostoses of the dististele: 3, stereom growth into the lumen of a nodal, x 200; 4, nodal facet, x 12. 5, 8, cirrus scars: 5, the left-hand scar encroaches on to the adjacent infranodal, the other being limited to the nodal, x 1 5; 8, cirrus scar, showing the elliptical lumen and the synarthrial ridges, x 50. 9, 10, stereom of the symplexial facet: 9, petaloid zone, x 50; 10, interpetaloid zone (note tubuli), x 50. PLATE 74 DONOVAN, Chladocrinus ( Neocrinus ) 834 PALAEONTOLOGY, VOLUME 27 The dististele is composed of noditaxes of a constant configuration i.e. 342434 1434243N (PI. 73, fig. 1). All internodals are pentagonal in outline, with well-rounded angles and sides infolded to give a flower-like appearance to the facet. Latera of internodals are gently convex. A reducing sequence of columnal heights from priminternodals to quartinternodals distinguishes the various internodal orders. The nodals are taller than the internodals, with latera that are more convex. Nodals are thus the widest columnals of the dististele. Nodals are further distinguished by having five distally orientated cirrus scars in radial positions on the lower half of the latus, i.e. they are cirrinodals (text-fig. lc). Almost all articula of the dististele resemble those of PI. 74, fig. 2 and text-fig. 3. The only exceptions are the distal facets of nodals and the proximal facets of adjacent internodals, which have a synostosial articulation. The terminology used in text-fig. 3 follows Roux (19776, p. 46) and Moore et al. (1968), except that the term ‘large meshes’ is replaced by tubuli (singular, tubulus; see explanation above). The petaloid zones are interradial. The configuration of the facets of my specimens does not agree completely with Roux’s diagram of the same area (19776, fig. 19); they have rectilinear stereom (Smith 1980) with diamond-shaped meshes in the five areola pits (PI. 75, fig. 7). This is indicative of strong ligamentation in these areas. In transverse section the areas of rectilinear stereom of the columnal appear oval to pear-shaped in outline (text-fig. 4a). The areolar pits also have pear-like outlines. Nine to twelve crenulae border each petaloid zone. The tubuli form an irregular border around the areola petals but do not reach the margins of the columnal. The crenularium can be open or closed and an interpetaloid axial groove separates each pair of crenularia (PI. 74, figs. 9, 10). PI. 74, fig. 2 and text-fig. 3 show only areolae with open crenularia, but this is not always the case (PI. 74, fig. 9; Roux 19776, fig. 5c). Roux did not mention the possibility of open crenularia when discussing this species. The distal articulum of each nodal, and the proximal facet of the sequential, distal tertinternodal (called the infranodal by Breimer 1978, p. T24), have synostosial articulation facets (PI. 73, figs. 1, 3, 5; PI. 74, figs. 4, 5). These are secondarily modified from symplexial articulations (compare text-fig. 6a and b) by infilling with stereom (Roux 1974, p. 6) and have been called cryptosymplexies by some authors (e.g. Moore et al. 1968). Relict crenulae are apparent at the edge of some synostosial articula (PI. 73, figs. 3, 5). It is almost impossible to break (mechanically) a column along such a stem joint; the stem will normally break through a symplexial suture, or partly through both suture and plate. Synostosial joints therefore have extremely strong ligamentation. Such stem joints, however, are thought to be adapted for autotomy, i.e. self-mutilation as a defence mechanism (Emson and Wilkie 1980). Lumina (singular, lumen: the intersection between the axial canal of the column and an articular facet) show little variation in shape or diameter along the whole column. The lumen at the base of the cup is pentagonal with rounded angles (PI. 73, figs. 2, 7). This pentagonal outline is retained by columnals in the most proximal part of the stem (PI. 73, fig. 6; text-fig. 4c) but most plates have circular lumina. For example, in USNM 12356/1 the most distal columnal with a pentagonal lumen is the second priminternodal beneath the cup. The transition from pentagonal to circular is gradual, not sudden. EXPLANATION OF PLATE 75 Figs. 1-9. Chladocrinus ( Neocrinus ) decorus (Wyville Thomson). USNM 12356/1. Scanning electron micrographs of stereom microstructure. 1, 2, 4, 8, longitudinal section through a pluricolumnal of three internodals from the dististele: 1, rectilinear (centre) and labyrinthic stereom, the latter concentrated in the interpetaloid zones, x 30; 2, rectilinear stereom adjacent to the axial canal (right), x 100; 4, sequence of columnals (taxis) in section, x 15; 8, section through a petaloid zone, showing the interlocking crenulae, x 60. 3, rectilinear stereom of a crenula, x 60. 5, interpetaloid zone of a synostosis, showing tubuli, x 55. 6, stereom infilling axial canal at distal end of the most distal columnal, x 100. 7, rectilinear stereom of a petaloid zone, x 400. 9, stereom at the edge of a columnal, x 550. PLATE 75 DONOVAN, Chladocrinus ( Neocrinus ) 836 PALAEONTOLOGY, VOLUME 27 The shape of the lumen is modified in two areas of the dististele. The lumina of synostoses are partially infilled by long, spike-like, labyrinthic stereom fingers which grow into the axial canal (PI. 74, fig. 3; Roux 19776, fig. 8). This must limit the amount of soft tissue in the axial canal adjacent to synostoses (cf. PI. 74, fig. 3 and PI. 75, fig. 2). Such an ingrowth of stereom suggests that the axial canal can be infilled rapidly by calcite growth if autotomy occurs at a synostosial stem joint. The axial canal of the synostosial facet of the most distal columnal is completely filled by labyrinthic stereom (PI. 75, fig. 6) and no suggestion of the outline of the lumen is preserved. In all other respects this columnal resembles the distal facet of any nodal of the dististele (e.g. PI. 74, fig. 4). The axial canal is also modified in the regions within nodals in which the lateral extensions into the cirri arise (canaliculi; text-figs. 1, 4). In these regions the axial canal is pentagonal in outline, becoming pentastellate at the junction with the cirral canals. Cirri are attached to the column at the five cirrus scars which are present on each nodal (PI. 73, fig. 1; PI. 74, figs. 5, 8). These occur on the distal part of the latus and are orientated so that cirri always point away from the cup (PI. 74, fig. 5; text-fig. 2; Breimer 1978, p. T24, fig. 11). Cirrus scars may be limited to the nodal or they may overlap on to the adjacent infranodal (PI. 73, fig. 1; PI. 74, fig. 5). They are elliptical in outline with a central, elliptical axial canal which is Hanked by a pair of synarthrial articular ridges (PI. 74, fig. 8). The ectoderm covering the column is probably continuous over the cirri. The cirri of the proxistele have been discussed above. Cirri and cirral ossicles of the dististele are illustrated in Plate 76. An entire cirrus is approximately 17 mm long and is composed of twenty-six cirrals (PI. 76, fig. 1 ). The most proximal and distal cirral ossicles (PI. 76, figs. 4, 5, respectively) are the two primary cirrals, which were defined above. The proximal primary cirral has a domed proximal facet which fits the cirrus scar. The lumen is approximately elliptical (PI. 76, fig. 3) but the axial canal of the cirral ossicle has two central ridges on the long sides which gives the appearance of a figure ‘8’ to the aperture. The lumen is flanked on the articulum by two notches which articulate against the ridges of the cirrus scar. This differs from columnals with synarthrial ridges (text-fig. 6c), such as platycrinitids (Moore 19396) and Ristnacrinus (Chauvel and Le Menn 1972), in which the articulation is formed by two opposed ridges. It is, perhaps, more analogous with the ‘see-saw’ articulation of the glyptocystitid cystoids (text-fig. 6d; Paul 1968). The arrangement of ridge and groove is presumably more resistant (than two ridges) to a force acting to twist the cirral ossicle against the cirrus scar, e.g. eddy currents. The distal facet is approximately planar, with a slightly raised margin of stereom, so that the next distal cirral ossicle fits into it like a cup on to a saucer (PI. 76, fig. 6). Similar articulation is shown by all cirral ossicles, except the distal primary cirral. The proximal cirral ossicles are low but cirral height increases distally until they become slightly higher than wide (PI. 76, fig. 1). The most distal cirral ossicles taper and are reduced in height, terminating in the claw-like, distal primary cirral (PI. 76, figs. 1, 2, 5), i.e. cirral height is a maximum in the centre of the cirrus. The lumen becomes excentric in the middle cirrus (PI. 76, fig. 7). The articulation surfaces are dorsal to the axial canal in this area, so that the cirrus has a preferred distal flexure. This effect is further aided by the articular facets being angled to the latera (PI. 76, fig. 1 ). This excentricity and the nature of the stereom around the axial canal suggest unequal development of EXPLANATION OF PLATE 76 Figs. I -8. Chladocrinus ( Neocrinus ) decorus (Wyville Thomson). USNM 12356/2. Scanning electron micrographs of cirri. I , complete, mature cirrus (proximal end at the bottom of the photograph), x 8. 2, 5, 8, distal, claw-like primary cirral: 2, articular facet, x 120; 5, latus, x 55; 8, stereom microstructure of the latus, x 200. 3, 4, proximal facet of the most proximal cirral: 3, lumen, x 400; 4, oblique view of curved articular facet, slightly damaged to expose the internal stereom, x 55. 6, articulation between sequential cirrals, x 55. 7, facet of a cirral from the central part of the cirrus (note the excentric axial canal), x 55. PLATE 76 SSI: ■MM mmm mm® DONOVAN, Chladocrinus ( Neocrinus ) 838 PALAEONTOLOGY, VOLUME 27 a contractile apparatus and a preferred flexure away from the cup, as in comatulids (Holland and Grimmer 1981). The distal primary cirral has an articular facet which is angled in the area of synarthrial articulation (PI. 76, figs. 2, 5). The lumen is elliptical, excentric, and situated on the axis of articulation. The five axial canals of the cirri are connected to the main axial canal of the column by curved canaliculi in the nodals (text-fig. 4a, b). The canaliculi are curved in the same direction as the preferred curvature of the cirri. It is possible that a swollen termination of nervous tissue exists in the zone where the cirral axial canals diverge. In the event of autotomy and loss of the stem distal to any nodal of the distal stem, the rapid growth of stereom is assumed to infill the axial canal up to the base of this termination. STEREOM MICROSTRUCTURE Stereom microstructure was examined on cleaned ossicles (PI. 75; the columnal illustrated in PI. 73, fig. 4 shows the concentration of ligament fibres in the petaloid zones). The stereom of the petaloid zones (PI. 74, figs. 9, 10; PI. 75, figs. 3, 7, 8), cirrus scars (PI. 74, fig. 8) and cirral facets (PI. 76, figs. 2, 4, 7) is of the type called a-stereom by Roux (1970, 1975, 19776). Smith (1980), in his revision of stereom microstructure, recognized two types of a-stereom: rectilinear and galleried. The stereom of the petaloid zones (PI. 75, fig. 7) is rectilinear, although that of the culmina (PI. 75, fig. 3) is apparently galleried. Latera (PI. 75, fig. 9; PI. 76, fig. 8) appear to be composed of the simple perforate stereom of Smith. The stereom of the interpetaloid zones (PI. 74, fig. 10) is labyrinthic ( = /3-stereom of Roux). Longitudinal sections of pluricolumnals show that rectilinear stereom is concentrated in the region of the axial canal and of the petals (PI. 75, figs. 1 , 2, 4), while other regions are composed of labyrinthic stereom. Sectioning has also revealed the close contact between articulating crenulae (PI. 75, fig. 8). The stereom of the synostosial articula (PI. 74, fig. 4; PI. 75, fig. 5) is a labyrinthic layer (synostosial stereom of Roux 19776, p. 47) which has overgrown a normal symplexy. Tubuli are present on these articula, in similar positions to those of the symplexial articula. It is probable that these canals penetrate the entire column. BIVARIATE ANALYSIS OF COLUMNALS Only three features discussed in previous columnal studies are considered in this section: columnal diameter (KD), columnal height (KH), and lumen diameter (LD). Features such as the length/ breadth ratio of the petaloid zone show little variation, and the number of culmina varies in an irregular manner even between petals of a single articular facet (text-fig. 3; Roux 19776, p. 62, fig. 19). In an attempt to generate an artificial ‘palaeontological’ sample, the KD/LD and KD/KH graphs (text-fig. 7a and b, respectively) are based on a random sample of columnals from the proxistele of IJSNM 12356/1 and the dististeles of both specimens. These graphs are somewhat artificial because the forty-one plotted points are based on information derived from only two individuals, whereas a similar collection of fossil columnals could come from forty-one individuals. The KD/LD plot (text-fig. 7a) shows that lumen diameter remains almost constant as columnal diameter increases, although there is a slight decrease in lumen diameter with columnal growth. This differs from many examples from the fossil record, in which lumen diameter increases with increased columnal diameter. The KD/KH plot (text-fig. 7b) gives a good linear grouping of columnals apart from one narrow, low columnal, an internodal from the proxistele. Such comparatively thin columnals are less likely to be preserved than the more robust, and plentiful, ossicles of the dististele. Lines of best fit have been calculated using the Bartlett method (Fryer 1966). It is not intended to discuss here the functional morphology of the isocrinid column, which is summarized in Rasmussen (1977). Similarly, ecology of modern stalked crinoids is discussed by Macurda and Meyer (1974, 1983). Soft tissues of these specimens has deteriorated over the past hundred years and has not been studied in detail. The anatomy of modern crinoids has been reviewed by Breimer (1978). DONOVAN: RECENT CRINOID STEM 839 A B text-fig. 7. KD/LD (a) and KD/KH (b) graphs for Chladocrinus (Neocrinus) decorus. KD, columnal diameter; LD, lumen diameter; KH, columnal height. CONCLUSIONS From the above survey of the column morphology of C. decorus , the following comments are made which may prove to be of general significance in all crinoids. 1. The diameter and shape in cross section of the axial canal is almost constant along the whole length of the stem. The only columnals which do not have a circular lumen are those of the most proximal part of the column and the distal facets of cirrinodals. This arrangement agrees with the intuitive conclusion that adjacent columnals must have lumina of similar morphology. It would not be expected that, say, a columnal with a pentastellate lumen would articulate with a columnal in which the lumen is circular. The shape of the axial canal, however, is not necessarily the same as that of the column. 2. Columnals change shape during growth, although their arrangement in the stem (of this example, at least) is regular. Adjacent articula always have similar morphologies, e.g. symplexial and synostosial articula are never in direct contact with each other. This is true in both columnals and cirrals. Cirral facets are not necessarily the same as those of columnals from the same stem. 3. Noditaxes become regular and fixed in their arrangement of columnals, particularly away from the main growing area. Only nodals bear cirri. Only nodals and infranodals have different articula on the same columnal, i.e. both have a symplexy on one articulum and a synostosis on the other. 840 PALAEONTOLOGY, VOLUME 27 4. Cirrus morphology is usually quite different from that of the associated column. Cirrus growth is different from that of crinoid arms (in which new ossicles are added at their tips) and columns, with new cirral ossicles being intercalated just distal to the most proximal primary cirrus. 5. Some columnals from the growing region (proxistele) are too small to be seen externally and can only be detected when the column is disarticulated or sectioned. Information about them will be lost in fossil pluricolumnals which are preserved as external moulds. 6. Some fine details which can be seen in C. decorus may be too delicate to be fossilized, e.g. the stereom infills of some axial canals (PI. 74, fig. 3). However, good stereom is known to be preserved in some of the earliest echinoderm fossils (Donovan and Paul 1982). The observation that axial canal outline is less variable than the columnal shape is important when assessing the relative merits of the two existing systems for naming dissociated crinoid columnals. That of Moore (1939a; Wright 1983) is based on both columnal and lumen outline, the name which is generated describing these features, e.g. a columnal with a pentagonal outline and a circular lumen would be placed in the morphogenus Pentagonocyclopa. In this system the first part of the generic name describes the columnal outline, which is presumably regarded as being more important than the lumen outline. The method of columnal naming used by Russian authors (Yeltysheva 1955, 1956) is similar, but the lumen outline is regarded as being the more important of the two parameters, e.g. the above columnal would be placed in the morphogenus Cyclopentagoncilis. The stem morphology of C. decorus suggests that the latter system is to be preferred. The lumen outline and facet articulation are more important than columnal shape in columnal classification, and for reconstruction of stems from dissociated ossicles. Acknowledgements. I thank Dr. David L. Pawson, Smithsonian Institution, for the loan of the described specimens; Dr. Christopher R. C. Paul, Department of Geology, University of Liverpool, for critically reading any early draft of this paper; and Mr. Cornelis J. Veltkamp, Department of Botany, University of Liverpool, for taking the scanning electron micrographs. This work was carried out during the tenure of Natural Environment Research Council research studentship GT4/80/GS/55 in the Department of Geology, University of Liverpool. REFERENCES breimer, A. 1978. General morphology. Recent crinoids. In moore, r. c. and teichert, c. (eds.). Treatise on Invertebrate Paleontology. Part T. Echinodermata 2 ( 1 ). Geological Society of America and University of Kansas Press, New York and Lawrence, T9-T58. chauvel, J. and le menn, j. 1972. Echinodermes de l'Ordovicien Superieur de Coat-Carrec, Algol (Finistere). Bull. Soc. geol. miner. Bretagne, ser. C, 4, 39-61. donovan, s. k. 1983«. Tetrameric crinoid columnals from the Ordovician of Wales. Palaeontology , 26, 845-849. 19836. Evolution and biostratigraphy of pelmatozoan columnals from the Cambrian and Ordovician of Britain. Ph.D. thesis (unpubl.), University of Liverpool. — and PAUL, c. R. c. 1982. Lower Cambrian echinoderm plates from Comley, Shropshire. Geol. Mag. 119, 611-614. emson, R. G. and wilkie, i. c. 1980. Fission and autotomy in echinoderms. A. Rev. Oceanogr. Mar. Bio. 18, 155-250. franzen-bengston, c. 1983. Radial perforations in crinoid stems from the Silurian of Gotland. Lethaia , 16, 291-302. fryer, H. c. 1966. Concepts and methods of experimental statistics. Allyn and Bacon, Rockleigh, N.J. Holland, n. d. and grimmer, j. c. 1981 . Fine structure of the cirri and a possible mechanism for their motility in stalkless crinoids (Echinodermata). Cell Tissue Res. 214, 207-217. jefferies, R. p. s. 1968. The subphylum Calcichordata (Jefferies 1967), primitive fossil chordates with echinoderm affinities. Bull. Br. Mus. nat. Hist. (Geol.), 16, 241-339. jeffords, R. m. and miller, T. h. 1968. Ontogenetic development in late Pennsylvanian crinoid columnals and pluricolumnals. Paleont. Contr. Univ. Kans., Echinodermata Art. 10, 1-14. lemenn, j. 1981. LesCrinoides. In morzadec, p., Paris, f. and racheboeuf, p. (eds.). Le Devonien inferieurde la tranchee de la Lezais. Mem. Soc. geol. miner. Bretagne, 24, 261-273. lewis, R. D. 1980. Taphonomy. In broadhead, t. w. and waters, j. a. (eds.). Echinoderms: notes for a short course. University of Tennessee, Atlanta, 27-39. DONOVAN: RECENT CRINOID STEM 841 LINCK, o. 1954. Die Muschelkalk-Seelilie Encrinus liliiformis. Naturw. Mschr. dt. Naturk. ‘ Aus der Heimat' , 62, 225-235. macurda, d. b., Jr. and meyer, d. l. 1974. The feeding posture of modern stalked crinoids (Echinodermata). Nature , Load. 247, 394-396. - - 1983. Sea lilies and feather stars. Am. Scient. 71, 354-365. moore, r. c. 1939a (for 1938). The use of fragmentary crinoidal remains in stratigraphic paleontology. J. scient. Labs. Denison Univ. 33, 165-250. — 1939 b. Platycrinitid columnals in Lower Permian Limestone of Western Texas. /. Paleont. 13, 228-229. — Jeffords, R. m. and miller, t. H. 1968. Morphological features of crinoid columns. Paleont. Contr. Univ. Kans., Echinodermata Art. 8, 1-30. Paul, c. R. c. 1968. Macrocystella Callaway, the earliest glyptocystitid cystoid. Palaeontology, 11, 580-600. rasmussen, h. w. 1977. Lunction and attachment of the stem in Isocrinidae and Pentacrinitidae: review and interpretation. Lethaia , 10, 51-57. — 1978. Articulata. In moore, r. c. and teichert, c. (eds.). Treatise on Invertebrate Paleontology. Part T. Echinodermata 2 (J). Geological Society of America and University of Kansas Press, New York and Lawrence, T813-T928. raup, d. m. and Stanley, s. m. 1978. Principles of Paleontology (2nd. edition). Lreeman, San Lrancisco. reichensperger, a. 1905. Zur Anatomie von Pentacrinus decorus Wy. Th. Z. wiss. Zool. 80, 22-55. rieppel, o. 1982. A new genus of shark from the Middle Triassic of Monte San Giorgio, Switzerland. Palaeontology, 25, 399-412. roux, m. 1970. Introduction a l’etude des microstructures des tiges de crinoi'des. Geobios, 3, 79-98. 1974, Les principaux modes d’articulation des ossicules du squelette des Crinoi'des pedoncules actuels. Observations microstructurales et consequences pour l’interpretation des fossiles. C. r. hebd. Seanc. Acad. Sci., Paris, 278, 2015-2018. -1975. Microstructural analysis of the crinoid stem. Paleont. Contr. Univ. Kans., Pap. 75, 1-7. — 1977a. Les Bourgueticrinina du Golfe de Gascogne. Bull. Mus. natn. Hist. nat. Paris, ser. 3, 426 (Zool. 296), 25-83. — \911b. The stalk-joints of recent Isocrinidae (Crinoidea). Bull. Br. Mus. nat. Hist. (Zool.), 32, 45-64. — 1978. Ontogenese, variabilite et evolution morphofonctionnelle du pedoncle et du calice chez les Millercrinida (Echinodermes, Crinoi'des). Geobios, 11, 213-241. smith, a. b. 1980. Stereom microstructure of the echinoid test. Spec. Pap. Palaeont. 25, 1-81. sprinkle, j. and kolata, d. r. 1982. ‘Rhomb-bearing’ camerate. In sprinkle, j. (ed.). Echinoderm faunas from the Bromide Lormation (Middle Ordovician) of Oklahoma. Paleont. Contr. Univ. Kans., Monogr. 1, 206-21 1 . Thomson, c. w. 1864. Sea lilies. Intellect. Obs. 6, 1-11. ubaghs, G. 1978. Skeletal morphology of fossil crinoids. In moore, r. c. and teichert, c. (eds.). Treatise on Invertebrate Paleontology. Part T. Echinodermata 2 (7). Geological Society of America and University of Kansas Press, New York and Lawrence, T58-T216. webster, G. d. 1974. Crinoid pluricolumnal noditaxis patterns. J. Paleont. 48, 1283-1288. wright, d. k. 1983. Crinoid ossicles in upper Ordovician benthic marine assemblages from Snowdonia, North Wales. Palaeontology , 26, 585-603. yeltysheva, r. s. 1955. Klass Crinoidea — morskie lilii — stebli morskikh lilii. In Nikiforova, o. i. (ed.). Field Atlas of Ordovician and Silurian faunas of the Siberian Platform , 40-47. VSEGEI, Leningrad. — 1956. Stebli morskikh lilii i ikh klassifikatsiya. Vest. Leningr. gos. Univ. (Geol. Geog.), 12, 40-46. Typescript received 21 November 1983 Revised typescript received 1 Lebruary 1984 S. K. DONOVAN Jane Herdman Laboratories of Geology University of Liverpool P.O. Box 147 Liverpool L69 3BX Present address Department of Geology, Trinity College, Dublin 2, Eire ARTH RO PLEURA TRAILS FROM THE WESTPHALIAN OF EASTERN CANADA by DEREK E. G. BRIGGS, A. GUY PLINT and RON K. PICKERILL Abstract. The trace fossil Diplichnites cuithensis Briggs, Rolfe and Brannan, 1979 is described from the Tynemouth Creek Formation of southern New Brunswick, and is interpreted as a trail of the giant Carboniferous myriapod Arthropleura. The arthropod was weaving through a forest of calamites which formed a single species stand on a sheetflood deposit in an alluvial fan environment. Comparison with other Arthropleura trails suggests that, in this case, drier conditions prevailed, supporting the interpretation of an essentially terrestrial habit for the arthropod. This trail is the first of Arthropleura to be described in detail from North America, and provides evidence that the myriapod’s mode of turning was similar to that of modern myriapods, and unlike trilobites. The ichnogenus Diplichnites should not be applied to trilobite trails. Trace fossils can rarely be ascribed with confidence to a particular organism, but when this is possible they provide otherwise unavailable evidence of its mode of life and habitat. They can prove particularly important in interpreting the degree of terrestrialization achieved by arthropods in transitional environments, such as eurypterids (Briggs and Rolfe 1983) and the giant Carboniferous myriapod Arthropleura , because, unlike body fossils, traces are not transported (Rolfe 1980, p. 131 ). The first trails attributed to Arthropleura were reported and figured by Ferguson (1966, 1975) from the celebrated Joggins section (Westphalian B) in Nova Scotia (text-fig. 1 ). An analysis of an older Namurian example from Arran, Scotland, by Briggs et at. (1979) revealed details of the morphology and locomotory capability of the arthropod. Smaller arthropleurid trails are known from the Stephanian of Montceau-les-Mines, France (Langiaux and Sotty 1977c/; Rolfe et al. 1982). The new example described here, however, is the first to preserve evidence of the arthropod making a pronounced change in direction, its path being constrained by calamite Trees’ growing in sheetflood deposits near the margin of an alluvial fan. The resultant trail provides information on the mode of cornering employed by Arthropleura , and its habitat. It also extends the geographical range of the trace fossil to New Brunswick, where unequivocal body fossils of Arthropleura have yet to be reported (Briggs et al. 1979, p. 287). The trace fossil locality was discovered by A.G.P. in 1981; he and R.K.P. made a latex cast of the best-preserved part of the trail in August 1 982. Unfortunately, most of the bedding plane was buried by a landslip the following winter. The trail was mapped (text-fig. 3b) using both a mosaic of enlarged photographs (cf. text-fig. 3a), and the latex cast. The position of the calamite stems was also recorded in the field. The latex cast (GSC 76665) is housed in the Geological Survey of Canada, Ottawa; sets of photographs of the trace fossil are held by the Department of Geology, University of New Brunswick, and the Hunterian Museum, University of Glasgow. GEOLOGICAL SETTING The trace fossil occurs in the Tynemouth Creek Formation ( Flint and Poll 1982) which outcrops in the area around Tynemouth Creek, on the south coast of New Brunswick (text-fig 1 ). The locality is about 200 m south-west of Gardner Creek Bridge. The trail horizon lies 28-5 m above the base of the section exposed on the shore, immediately to the west of Gardner Creek (text-fig. 1 ; cf. Plint and Poll 1 982, fig. 2). A less well-preserved example, about I m long, occurs at approximately the same horizon about 100 m along strike to the north-east. The succession shown in text-fig. 2a is approximately | Palaeontology, Vol. 27, Part 4, 1984, pp. 843-855.] 844 PALAEONTOLOGY, VOLUME 27 equivalent (along strike) to the lowest 30 m shown by Plint and Poll (1982, fig. 3). Spore analysis (Barss in Plint and Poll 1982, p. 106) indicates a Westphalian A or B age for the trace fossil; poor preservation of the spores makes a more precise age determination impossible. Two similar poorly preserved trails occur in a cliff exposure 300 m east of Tynemouth Creek (text-fig. 1). Both lie at the top of channel sandbodies beneath overbank sediments, 75 m and 90 m respectively above the base of Section 2 of Plint and Poll (1982, fig. 1 8). The trail horizons at Gardner Creek and Tynemouth Creek cannot be accurately correlated, but it seems likely that they are approximately contemporaneous. text-fig. 1 . Location map. The Tynemouth Creek Formation consists predominantly of red siltstones, red and grey sandstones, and coarse conglomerates, and shows an overall upward-coarsening. Rare freshwater limestones are locally present. The sequence containing the trace fossil consists of fine, red, slightly silty, tabular sandstones, interbedded with red and green siltstones (text-fig. 2a). The sandstones are dominantly massive, but include plane and cross-laminated units, and vary in thickness from thin laminae within siltstones to units up to 2 m thick. The thicker units are usually composed of several decimetre-thick sandstone beds, separated by silt laminae. The top 10-20 cm of the thicker sandstone units are usually mottled pale green and bioturbated. Both siltstone and, in particular, sandstone beds contain numerous upright and obviously in situ calamite stems, up to 10 cm in diameter. The best-preserved myriapod tracks at Gardner Creek occur within a 1-5 m sandstone which rests on 40 cm of siltstone. Bed 1 (text-fig. 2b) comprises fine, grey sandstone, containing large calamite ‘tree stumps’ and Stigmaria (otherwise rare in the Tynemouth Creek Formation) which radiate to a distance of about 5 m. It was not possible to determine whether the Stigmaria trees project above this bed, due to insufficient exposure. Bed 1 is overlain by 12 cm of siltstone (Bed 2), and 40 cm of fine sandstone (Bed 3) with numerous in situ calamites rooted in Bed 2. The top of Bed 3 grades up into a few millimetres of siltstone, and it is on this surface that the trace fossil is preserved. The trail surface undulates slightly but shows no evidence of sedimentary structures (e.g. ripple marks) or other trace BRIGGS. PUNT AND PICKERILL: CARBONIFEROUS MYRIAPOD TRAILS 845 Sond LEGEND catamite Burrows Stigmaria Current ripples A A Crossbedding Plane lamination Mottled & rooted palaeosol text-fig. 2. Stratigraphic sections: a, lowest 30 m immediately west of Gardner Creek, showing position of trail horizon; b, detail of beds including trail horizon. fossils. Bed 4 comprises 50 cm of fine sandstone, at the top of which is a green-mottled palaeosol. The calamites in Bed 3 can be traced up through the trail horizon into Bed 4. The sandstones are interpreted as the deposits of major sheet floods; palaeocurrents indicate a flow toward the north and north-west. The intervening siltstones were probably deposited during less vigorous floods. Thin sandstone laminae within the siltstones suggest a pulsatory flow. Green- mottled and bioturbated horizons at the top of sandstone units are interpreted as palaeosols which probably formed during prolonged breaks in sedimentation. The upper trail east of Tynemouth Creek occurs on the top of a highly bioturbated sandstone palaeosol which contains Stigmaria and locally displays low-amplitude, straight-crested ?wave ripples, suggesting periodic shallow immersion. The environment, therefore, appears to have been stable for relatively long periods, favouring colonization by plants, except for short intervals of very rapid sedimentation during major floods. The depositional area was of low relief and gradient, and lay towards the margin of a major alluvial fan that was prograding toward the north-west (Plint and Poll 1982). The large numbers of calamites, and the apparent absence of other plants (except for rare 846 PALAEONTOLOGY, VOLUME 27 030' 1 22 50 cm text-fig. 3. Diplichnites cuithensis Briggs, Rolfe and Brannan 1979, Tynemouth Creek Formation (Westphalian A or B), 200 m south-west of Gardner Creek Bridge, near Tynemouth Creek, New Brunswick, a, photomosaic of the trail as it was in 1982 (20 cm scale bar), b, plan of the trail to show the position of the preserved tracks and calamites (represented by subcircular outlines); box marks position of text-fig 4. BRIGGS, PLINT AND PICKERILL: CARBONIFEROUS MYRIAPOD TRAILS 847 Stigmaria ), suggest that catamites alone were well suited to colonizing this type of sedimentary environment, and formed essentially single species forests. Pfefferkorn and Zodrow (1982) recorded similar standing forests of calamites in the Pennsylvanian of Nova Scotia; they concluded that calamites and lycopods grew where sedimentation rates were high, in areas ‘that were generally not occupied by other plant groups’. DESCRIPTION OF THE TRACE FOSSIL The most extensively exposed trail follows a sinuous course, over 5-5 m long, between calamite stems (text-fig. 3). The trail varies between 29 5 and 36-5 cm in total width (the trails at Tynemouth Creek are 30 and 27 cm wide), and is preserved in a layer of siltstone about 5 mm thick grading into the underlying sandstone (text-fig. 2b). The siltstone parts readily along planes parallel to the bedding, and the effects of differential erosion were quite evident even after one winter (1981 1982). Thus the majority, if not all, of the imprints are preserved as undertracks (Goldring and Seilacher 1971), and in places erosion has removed short sections of the trail completely. The detailed morphology of individual imprints is not preserved (contrast Briggs et al. 1979, pi. 30, figs. 5, 6), and the tracks vary in size and shape. This may result from: (1) the superimposition or coalescing of two or more footfalls; (2) the water content of the sediment; (3) slight erosion during deposition of the overlying fine sand. The maximum width (normal to the axis) of both the entire trail (36 5 cm) and of the right and left rows of tracks (c. 1 1 cm) is reached roughly at the points of maximum curvature of the trail. Here the lateral spread of right or left tracks is over twice that in the straight sections of the trail (although the number of footfalls remains the same). Thus the proportion of the total width of the trail occupied by imprints increases from less than 50 to about 60 %. The curve at the bottom of text-fig. 3 a, b displays a linear density of about twenty imprints in 20 cm, i.e. about 1 per cm (text-fig. 4). There appears to be a lower density in the straight sections of the trail, but this is due to the superimposition of a greater proportion of footfalls. The pronounced elongation of some of the tracks transverse to the trail is also due to the coalescing of adjacent footfalls. A deep depression hidden in shadow (text-fig. 3a, bottom right) marks the site of a calamite which appears to lie only just on the edge of the trail. There is some equivocal evidence that its close proximity to the course taken by the arthropod may have prompted slight ‘side stepping' by the limbs on the side in question. The surface has been eroded in the vicinity of the depression, however, and the critical imprints either lost or impaired as a consequence. There is no reliable evidence for the direction of progress of the arthropod (cf. Briggs et til. 1979, p. 278). text-fig. 4. Well-preserved portion of the trail (down-dip; position marked on text-fig. 3b) showing increase in the lateral spread of tracks in curve (20 cm scale bar). 848 PALAEONTOLOGY, VOLUME 27 INTERPRETATION AND DISCUSSION The trailmaker . The size of the trace fossil, the large number of regularly spaced tracks, and the sedimentary environment all indicate that the trail was made by Arthropleura (see Briggs et al. 1979, p. 278 for a fuller discussion). Environmental setting. Following the deposition of Bed 1 (text-fig. 2b), probably by sheetflooding, a Stigniaria ‘forest’ was established. Low-energy floods deposited the siltstone of Bed 2 in which stands of calamites later rooted. A subsequent sheetflood deposited Bed 3, burying the lower parts of the calamites. A thin layer of silt accumulated on the top of Bed 3, probably during the waning phase of the flood. Arthropleura then walked through the area, following subaerial emergence, perhaps searching for food among the detritus carried in by the floodwaters. The cohesive nature of the mud was probably essential for preservation of the tracks which would have been easily eroded by the next flood had they been made in sand alone. Calamites in Bed 3 extend through the trail horizon into Bed 4. There is evidence of syndepositional scour of the sediment of the trail horizon around some of the stems. There is no sign of upward disruption of the bedding, such as would have occurred if the calamites had grown up through the trail horizon after it had been deposited. It is highly unlikely that the bed could have preserved the trail while remaining sufficiently wet to allow a large calamite to grow through it without causing disruption. It is also improbable that the calamites grew in positions flanking the trail by chance. The ‘trees’ were therefore standing when the trail was made. The sinuous course represents Arthropleura picking its way through this ‘forest’ of calamite stems. Number of limbs and size. Due to the small number of complete specimens known, the ontogeny of Arthropleura is poorly understood. The apparent variation in the number of somites in near complete specimens (Rolfe 1969, p. 608) may indicate that development was partially anamorphic. If somites were indeed added during growth, estimating the dimensions of the trail-maker (apart from width) from the trace-fossil is not straightforward, particularly in the absence of evidence for the number of appendage-bearing somites. A reconstruction by Rolfe and Ingham (1967, p. 121, fig. 2) was based on the largest, most complete specimen known (Rolfe 1969, p. 607). Their reconstruction shows an individual 85 cm long with twenty-eight limb-bearing somites, which would produce a trail about 24-5 cm wide, assuming that Rolfe and Ingham have reconstructed the attitude of the appendages correctly (note that the magnification of x 0-2 given for the same figure reproduced in Rolfe 1969, p. 609, fig. 387, is extrapolated for an animal 1 -8 m long). The Arthropleura trail from the Namurian of Arran (Briggs et at. 1979) was made by an individual with only twenty-three limb-bearing somites (assuming that all the limbs were used in walking) but is none the less much wider (36 cm) than predicted by Rolfe and Ingham’s (1967) reconstruction. Reducing the length of the reconstruction by five somites (assuming anamorphic development) indicates that the Arran individual was about 105 cm long. The near-complete juvenile figured by Rolfe (1969, p. 608, fig. 386), however, has at least twenty-three postcephalic (and presumably limb-bearing) somites although it is only 65 mm long. The smallest Arthropleura known (Rolfe et al. 1982, p. 426) is 29 mm long and appears to have twenty to twenty-two somites (Secretan 1980, p. 32). The data available, although unsatisfactory, therefore suggest some variability in rates of development in different examples (or species) of Arthropleura. The number of body segments commonly varies in living adult myriapods with more than twenty (Lawrence 1952). Preliminary observations by John Almond (pers. comm.) suggest that two rather than one pair of limbs correspond to each of the more posterior tergites (at least) of small Arthropleura from Montceau-les-Mines (see Secretan 1980). If true, this does not necessarily imply a return to Waterlot’s (1934) interpretation of Arthropleura limbs as biramous (see discussion in Rolfe and Ingham 1967, p. 1 1 8); it may, however, indicate that each tergite of Arthropleura corresponds to some sort of diplosegment. If Almond’s observation can be confirmed and shown to apply to large arthropleurids, the basis for the reconstruction in text-fig. 5 ( Rolfe 1 969, p. 609) will require revision. In addition, a reconsideration of the estimate of the number of tergites in the individual which made BRIGGS, PLINT AND PICKERILL: CARBONIFEROUS MYRIAPOD TRAILS 849 text-fig. 5. Reconstruction of Arthropleura making the trail. The detailed morphology of the head is unknown. The arthropod is depicted walking around the corner at the bottom of text-fig. 3a, b in a northerly direction (toward the left of the page). There is no evidence, however, to indicate that this was the more likely direction of progress. The position of the calamites is somewhat schematic; that in the left foreground has been displaced to one side to avoid concealing part of the arthropod. (For discussion see text; drawing by Annemarie Burzynski.) the Arran trail (Briggs et al. 1979) will be necessary. Twenty-three pairs of walking limbs would then imply about half that number of tergites— an unlikely total for such a large individual. Determination of the number of walking appendages from the trace-fossil depends on identifying two successive imprints of the same limb (i.e. evidence of a stride) and counting the number of footfalls between them ( Briggs et al. 1 979, p. 282). Unfortunately, the preservation of the present trail is inadequate to provide the necessary evidence. For the purpose of reconstruction (text-fig. 5) the number of walking limbs (twenty-eight pairs) and relative proportions of the Rolfe and Ingham 850 PALAEONTOLOGY, VOLUME 27 (1967) reconstruction are assumed; the arthropod is unlikely to have exceeded 102 cm in length (based on the 29-5 cm width of the straight sections of the trail). Gait. Evidence for the gait employed by the arthropod is largely circumstantial, but provides a basis for the reconstruction (text-fig. 5). The trace-fossil provides no unequivocal evidence that the limbs on opposite sides of the body moved in phase, but an out-of-phase mode is highly improbable (Rolfe and Ingham 1967, fig. 2; Briggs et al. 1979). Rolfe and Ingham (1967) reconstructed a slow gait of 30 : 7-0 (ratio of duration of forward to backstroke; Manton 1977), i.e. with 70% of limbs in contact with the ground, suitable for pushing through the vegetation and plant debris on the coal-forest floor. Analysis of the Arran trace fossil (Briggs et al. 1979) revealed a gait of 5-5 : 4-5 (45% of limbs on the ground), when the arthropod was apparently unimpeded by vegetation. An intermediate pattern of 4 0 : 6 0 is adopted here (text-fig. 5) as a likely gait for the arthropod on open ground walking between calamite stems. A phase difference between limbs of 01 gives the most even spacing (cf. Briggs et al. 1979, pp. 283-284), hence Arthropleura is reconstructed (text-fig. 5) with ten limbs in a metachronal wave. Mode of cornering. This trace fossil provides direct evidence for the cornering capability of Arthropleura. The configuration of imprints suggests that it changed direction in a fashion similar to living myriapods. The series of papers on arthropod locomotion by Manton (1977 and references therein) does not include a detailed discussion of cornering in the Myriapoda, but her observations on turning in the onychophoran Peripatus (1950, p. 561) explain how this is achieved. When the arthropod changes direction the body follows a turn of the head — ‘the legs of both sides are displaced ' V, ■ Gy. v- Gjgjlll ' - • . .. ...... . - *-> ~ ' ' -‘.''A' ” • . % . • . ,■ "• .v *, . * . - - i-- ’Vi 5 1 ' * . . % , B 8® - v . tK ■ •• . •••*»<-. . .. T*&'- ■ ^v« ->'C- - sT^ - ■ • >y>- V* ' '•**- •, X- , • 'a text-fig. 6. Trails of a recent millipede Scaphiostreptus seychellarum (130 mm long) made in wet mud, x 0-7. a, curved, to show the increase in the lateral spread of tracks, b, straight. (Research and photographs by E. F. Walker.) BRIGGS, PLINT AND PICKERILL: CARBONIFEROUS MYRIAPOD TRAILS 851 laterally in the direction of the turn, and the angle of swing of the legs on the outer side is increased without alteration of the pattern of the gait. If the turn is acute . . . the posterior part of the body does not follow the path of the anterior end but becomes progressively displaced towards the side.’ Although Manton’s studies of the locomotion of myriapods were based in part on records made by the arthropods walking on smoked paper, very little work has been done on the traces produced by living arthropods walking on soft substrates. Some preliminary work by Rolfe (1980, p. 135, fig. 5) and Elaine Walker (Manchester University) has emphasized what a range of trails an individual arthropod can produce. Walker has provided photographs (text-fig. 6) of trails produced by the millipede Scaphiostreptus seychellarum which, although made by an individual a mere 130 mm long with about one hundred pairs of limbs, provides a basis for comparison with Arthropleura. Walker describes (pers. comm. ) how the less dense part of the curved trail (text-fig. 6a) is made by the anterior limbs of the millipede as it probes forward, while the more pronounced lineation is the product of the overlapping imprints of the posterior limbs. The apparent 'doubling’ of the right and left rows of prints on the corners of the Arthropleura trail (text-fig. 3a, b, left curve) indicates that, in this case also, the posterior part of the body did not precisely follow the anterior as the arthropod probed ahead to find a course between the calamite stems. Briggs et al. (1979, p. 287) noted that the Joggins Arthropleura trails (Ferguson 1966, 1975) showed a wider spacing of imprints across the width of the right and left rows than the Arran example, and they considered that this suggested 'a greater variation in appendage length and flexibility in the smaller Joggins arthropleurids’. It is perhaps more likely that this wider spacing is the result of the arthropod 'probing’ forward with the anterior appendages, although the figured examples from Joggins (Ferguson 1966, fig. 2; 1975, fig. 4) do not show as pronounced a change of direction as the example described here. Lateral flexure. The deduced length of the trailmaker indicates that significant lateral flexure of the body must have taken place. Stormer (1976, p. Ill, fig. 43) figured a posterior axial doublure on the tergites of Arthropleura extending forward about 25% of the length of each somite (a length equivalent to the overlap between tergites reported and reconstructed by Rolfe 1969, p. 608, fig. 387). Fateral flexure of the trunk would have been limited by the length of this doublure, and by the pronounced anterior keel on the para tergal folds (cf. Richardson 1959, fig. 43). Text-fig. 7 shows that sufficient curvature could be achieved within these constraints to allow Arthropleura to produce the trace fossil. text-fig. 7. Reconstruction of Arthropleura armata displaying the lateral flexure required to produce the trail (after Rolfe and Ingham 1967, fig. 2), x 014. Both the telson and the detailed morphology of the head are unknown. 852 PALAEONTOLOGY, VOLUME 27 DISTRIBUTION OF ARTHROPLEURA TRAILS Arthropleura trails have been reported from both North America and Europe, and range in age from Namurian (Pendleian) to Stephanian B (Table 1). In addition to those figured in the literature (Table 1 ), large, poorly preserved trails from the Westphalian D north of Florence, Cape Breton Island, Nova Scotia, have been attributed to Arthropleura (Baird in Carroll et al. 1972, p. 54). M. Gibling (pers. comm.) also reported an example from the upper Westphalian-Stephanian Morien Group in the Sydney Basin, Cape Breton Island. table 1 . Arthropleura trails figured in the literature Locality Age Width of trail Source Arran, Scotland Namurian (Pendleian) 36 cm Briggs et ah 1979 Gardner Creek, New Brunswick Westphalian A or B 29-5-36-5 cm This paper Joggins, Nova Scotia Westphalian B Up to 26 cm Ferguson 1966, 1975 Montceau-les-Mines, France Stephanian B Up to 10-8 cm Langiaux and Sotty 1977a Rolfe et al. 1982 The sedimentary environment of the Montceau-les-Mines trails has yet to be described (Langiaux and Sotty 1977/5; Rolfe et al. 1 982), but they appear to occur in fluvial flood-plain overbank deposits (J. E. Pollard, pers. comm.). The environment of the New Brunswick locality described here and those at Joggins and in Scotland are similar, but the New Brunswick occurrence differs in detail. The specimens at Joggins occur in a sheet sandstone that thickens laterally into a channel-filling sandstone (Bed 39/S2 of Duff and Walton 1973). The sheet sandstone possibly represents a crevasse splay that was subaerial at the time the tracks were made. The overall sedimentary environment was interpreted by Duff and Walton (1973) as an upper delta plain, characterized by laterally migrating fluvial channels with intervening low-lying floodbasins, lakes, and coal-swamps. In Arran, the well- preserved trail also occurs in a proximal deltaic environment, near the top of a fluvial channel-fill, in rippled, flaser-bedded, and rooted sandstones that were probably deposited in shallow water, close to the channel margin. The relatively good preservation of the tracks suggests that they were made subaerially, after the water level in the channel had dropped (Briggs et al. 1979). The Tynemouth Creek Formation (Plint and Poll 1982) contrasts with the previously described depositional settings in that it apparently represents a much drier alluvial fan environment characterized by periodic sheetfloods across an otherwise quiescent area of relatively slow deposition. Desiccation, bioturbation, and weathering considerably modified the sediments under these conditions of intermittent deposition. Although smaller Arthropleura may have sought the relatively humid environment provided by hollow trunks (Rolfe 1980, p. 149), this refuge was presumably not as readily available to larger individuals such as the trail-maker in this case. The occurrence of Arthropleura in this environment thus provides additional evidence for an essentially terrestrial rather than amphibious or aquatic habitat (Rolfe 1969; Briggs et al. 1979). A specimen of an Arthropleura limb with Monoletes pollen grains attached has recently been reported from the middle Pennsylvanian Mazon Creek biota (Richardson 1980). This suggests that the arthropod may have pollinated medullosan seed ferns while brushing flood-plain scrub (Scott and Taylor 1983; Taylor and Scott 1983), thus supporting a terrestrial habitat. TAXONOMY Briggs et al. (1979) referred the Arthropleura trail from Arran to Diplichnites Dawson, 1873, pointing out that this genus was originally described from a similar non-marine environment in the Westphalian at Joggins. BRIGGS. PLINT AND PICKERILL: CARBONIFEROUS MYRIAPOD TRAILS 853 The holotype of the type species, D. aenigma Dawson, 1873, has not been located, and the details of the specimen are not clear on the original woodcut (Dawson 1873, fig. 3). Briggs et al. (1979) established a new species, D. cuithensis for the Arran example, in recognition of the morphology of the individual tracks, and the size attained by the trace. Although in agreement with the need to maintain a morphological rather than biological basis for trace fossil taxonomy, they (1979, pp. 288-289) argued against the current tendency to extend the concept of Diplichnites (Seilacher 1955) to include what are obviously marine trails and probably the work of trilobites. In doing so they pointed out that such a restriction would not necessitate the erection of new taxa for these marine trace fossils, as a number have long been available in the literature (see Osgood 1970; Anderson 1975). The ichnogenus Diplichnites has been applied to non-marine traces made by animals other than arthropleurids and myriapods (Tevesz and McCall 1983). Savage (1971) described traces from later Carboniferous or early Permian periglacial lake sediments in Natal which he assigned to Diplichnites and interpreted as trails of syncarid or peracarid crustaceans. Bromley and Asgaard (1979, p. 64) referred traces from Triassic freshwater sediments in East Greenland to D. triassicus which they also considered to be the work of crustaceans (branchiopods). Detailed study of well-preserved examples of such traces should reveal the number of walking limbs employed by the animal. This would provide a means of distinguishing crustacean walking trails from those of the more numerous-limbed myriapods. Dawson (1862), in his first report of the trails which he subsequently named Diplichnites , observed that ‘their direction curves abruptly’; the original concept for the ichnogenus therefore included curved trails. The straight portions of the trail described here widen gradually into the curved portions which are characterized by a greater width of the rows of tracks (text-figs. 3, 4). In part of the trail (text-fig. 3a, b, left curve) the lateral spread of imprints in the opposing rows of tracks appears to divide for a short distance where the posterior end of the arthropod has not precisely followed the anterior. Examples of arthropod trace-fossils are known where different sections are referable to different ichnotaxa; Crimes (1970, pi. 12, figs, a , b), for example, figured specimens of Rusophycus continuous with Cruziana. These traces normally occur separately, however, are clearly distinct morphologically, and represent different behaviour patterns. In the present example it would seem unnecessary and counterproductive to assign the curved portions of the trail to a new ichnogenus, separate from the straight portions. It would be impossible to decide exactly where one taxon ends and the next begins! Thus the diagnosis of Diplichnites is emended below, as Dawson (1862, 1873) presumably intended, to include the curved parts of trails. The trail described here is referred to D. cuithensis Briggs, Rolfe and Brannan, 1979. This more complete diagnosis of Diplichnites reinforces the observation of Briggs et al. (1979) that the ichnogenus should not be applied to traces attributed to trilobites. The opposing rows of imprints in trilobite traces differ in showing no obvious tendency to expand in width on corners (see Osgood 1970, for example). The articulation of the trilobite thorax does not permit significant lateral flexure. Thus unlike myriapods, including Arthropleura (as evidenced by this trail), the anterior of trilobites could not ‘probe’ forward and follow a slightly different line to the posterior. SYSTEMATIC PALAEONTOLOGY Ichnogenus diplichnites Dawson, 1873 (emended) Type ichnospecies. D. aenigma Dawson, 1873, by original monotypy. Emended diagnosis. Morphologically simple trail, up to 37 cm wide, consisting of two parallel rows of tracks (each up to 1 1 cm wide); width of opposed rows increasing on curves corresponding to greater lateral separation of individual tracks; each row may divide into two on acute curves; individual tracks elongate roughly normal to trail axis, spaced closely and regularly at as few as one per cm in large examples. Diplichnites cuithensis Briggs, Rolfe and Brannan, 1979 Text-fig. 3 Type locality. Salt Pans harbour quarry, Laggan, Arran, Scotland. Additional localities. Gardner Creek, Tynemouth Creek, southern New Brunswick, Canada. Horizon. Carboniferous. Namurian, Pendleian Stage (Ej) to Westphalian A or B. 854 PALAEONTOLOGY, VOLUME 27 Acknowledgements. We are grateful to the following for advice and information: J. Almond, P. R. Crane, J. E. Dalingwater, R. M. Feldmann, J. T. Hannibal, H. W. Pfefferkorn, J. E. Pollard, W. D. E Rolfe, and E. Walker. Text-fig. 5 was drawn by Annemarie Burzynski; E. Walker provided the photographs for text-fig. 6. D.E.G.B.’s work was completed at the Field Museum of Natural History, Chicago, under the auspices of the Visiting Scientist Programme; A.G.P. undertook this research while in receipt of a U.N.B. Postdoctoral Fellowship; R.K.P. gratefully acknowledges the support of N.S.E.R.C. Canada Grant A3857. REFERENCES Anderson, a. m. 1975. The ‘trilobite’ trackways in the Table Mountain Group (Ordovician) of South Africa. Palaeont. afr. 18, 35-45. briggs, d. e. G. and rolfe, w. d. i. 1983. A giant arthropod trackway from the lower Mississippian of Pennsylvania. J. Paleont. 57, 377-390. — and brannan, j. 1979. A giant myriapod trail from the Namurian of Arran, Scotland. Palaeontology, 22,273-291. bromley, r. and asgaard, u. 1979. Triassic freshwater ichnocoenoses from Carlsberg Fjord, East Greenland. Palaeogeogr. Palaeoclimat. Palaeoecol. 28, 39-80. CARROLL, R. L., belt, E. s., dineley, D. L., baird, D. and mcgregor, D. c. 1972. Vertebrate paleontology of Eastern Canada , 1 13 pp. Guidebk, Excursion A59, 24th Internat. geol. Congr., Montreal. crimes, t. p. 1970. Trilobite tracks and other trace fossils from the Upper Cambrian of North Wales. Geol. J. 7, 47-68. dawson, j. w. 1862. Notice of the discovery of additional remains of land animals in the Coal-Measures of the South Joggins, Nova Scotia. Q. Jl geol. Soc. Load. 18, 5-7. - 1873. Impressions and footprints of aquatic animals and imitative markings, on Carboniferous rocks. Am. J. Sci. Ser. 3, 5, 16-24. duff, p. mcl. d. and walton, e. k. 1973. Carboniferous sediments at Joggins, Nova Scotia. Septieme Congr. internat. Strut, geol. Carb., Krefeld 1971, C. R. 2, 365-379. ferguson, l. 1966. The recovery of some large track-bearing slabs from Joggins, Nova Scotia. Marit. Sediments, 2,128-130. 1975. The Joggins section. Ibid. 1 1, 69 -76. ( = 1976. In Ancient sediments of Nova Scotia. Fid Trip Guidbk Eastern Sect. Soc. econ. Paleontologists Mineralogists, 111-118.) goldring, r. and seilacher, a. 1971. Limulid undertracks and their sedimentological implications. Neues Jb. Geol. Paldont. Abb. 137, 422-442. langiaux, j. and sotty, d. \911a. Ichnologie 4: pistes et empreintes dans le Stephanien de Blanzy-Montceau. La Physiophile, 86, 74-91. - \911b. Elements pour une etude ecologique d’un paysage de I'epoque houillere. Ibid. 87, 35-60. Lawrence, r. f. 1952. Variation in the leg-numbers of a South African millipede, Gymnostreptus pyrrocephalus C. Koch. Ann. Mag. nat. Hist., Ser. 12, 5, 1044-1051. manton, s. m. 1950. The evolution of arthropodan locomotory mechanisms. Part 1. The locomotion of Peripatas. J . Linn. Soc. (Zook), 41, 529-570. — 1977. The Arthropoda: habits, functional morphology and evolution, 527 pp. Clarendon Press, Oxford. osgood, r. G. 1970. Trace fossils of the Cincinnati area. Palaeontogr. am. 6 (41), 277-444. pfefferkorn, h. w. and zodrow, e. l. 1982. A comparison of standing forests from the Pennsylvanian of Nova Scotia with modern tropical forests. Bot. Soc. Am., Abstracts, Misc. Publ. 162, 62-63. plint, a. G. and poll, h. w. van de, 1982. Alluvial fan and piedmont sedimentation in the Tynemouth Creek Formation (Lower Pennsylvanian) of southern New Brunswick. Marit. Sediments All. Geol. 18, 104-128. richardson, e. s. 1959. Pennsylvanian invertebrates of the Mazon Creek area, Illinois: Trilobitomorpha, Arthropleurida, II. Fieldiana, Geol. 12, 79-82. 1980. Life at Mazon Creek. In langenheim, r. j. jr and mann, c. j. (eds.). Middle and late Pennsylvanian strata on the margin of the Illinois Basin. 10th Ann. Fid Conf. Great Lakes Sect. Soc. econ. Paleontologists Mineralogists, 217-224. Univ. Illinois, Urbana. rolfe, w. d. i. 1969. Arthropleurida. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part R, Arthropoda, 4 (2), 607-620. Geol. Soc. Am. and Univ. Kansas Press, Boulder, Colorado, and Lawrence, Kansas. 1980. Early invertebrate terrestrial faunas. In panchen, a. l. (ed.). The terrestrial environment and the origin of land vertebrates. 1 17-157. Academic Press, London and New York. BRIGGS, PLINT AND PICKERILL: CARBONIFEROUS MYRIAPOD TRAILS 855 — and ingham j. k. 1967. Limb structure, affinity and diet of the Carboniferous ‘centipede’ Arthropleura. Scott. J. Geo/. 3, 118-124. — schram, f. r., pacaud, G., sotty, d. and secretan, s. 1982. A remarkable Stephanian biota from Montceau-les-Mines, France. J. Paleont. 56, 426-428. savage, n. m. 1971. A varvite ichnocoenosis from the Dwyka Series of Natal. Lethaia , 4, 217-233. scott, a. c. and taylor, t. n. 1983. Plant/animal interactions during the Upper Carboniferous. Bot. Rev. 49, 259-307. secretan, s. 1980. Les arthropodes du Stephanien de Montceau-les-Mines. Bull. Soc. Hist. nat. Autun , 97, 23-35. seilacher, a. 1955. Spuren und Lebensweise der Trilobiten. In schindewolf, h. and seilacher, a. (eds. ). Beitrage zur Kenntnis des Kambriums in der Salt Range (Pakistan). Abh. math.-naturw. Kl. Akacl. I Viss. Main: , 10, 86-143. st0rmer, l. 1976. Arthropods from the Lower Devonian (Lower Emsian) of Aiken an der Mosel, Germany. Part 5: Myriapoda and additional forms, with general remarks on fauna and problems regarding invasion of land by arthropods. Senckenberg leth. 57, 87-183. taylor, T. N. and scott, a. c. 1983. Interactions of plants and animals during the Carboniferous. Bioscience , 33, tevesz, m. j. s. and 'mccall, p. l. 1983. Geological significance of aquatic nonmarine trace fossils. In tevesz, m. j. s. and mccall, p. L. (eds.). Animal-sediment relations , 257-285. Plenum Press. waterlot, G. 1934. Etude de la faune continentale du terrain houiller sarro-lorrain. Etud. Gites miner. Fr. Bassin houiller de la Sarre et de la Lorraine, 2, Faune fossile, 320 pp. 488-493. DEREK E. G. BRIGGS Department of Geology Goldsmiths’ College University of London Creek Road London SE8 3BU A. GUY PLINT and RON K. PICKERILL Department of Geology University of New Brunswick Typescript received 10 January 1984 Revised typescript received 20 Febuary 1984 P.O. Box 4400 Fredericton, New Brunswick Canada E3B 5A3 NEW EVIDENCE OF A SPIRIFERIDE ANCESTOR FOR THE THECIDEIDINA ( BRACH IOPODA) by P. G. BAKER Abstract. Investigation of the microstructure of the ventral interarea of a juvenile denticulate spiriferacean assignable to Unispirifer reveals rod-like structures which, apart from a difference in size, are structurally almost identical with the tubercle cores of a recently discovered Aalenian thecideidine species Mimikonstantia sculpla Baker and Elston, 1984. The coincidence of cyrtomatodont teeth, shell resorption, and secondary fibrous shell, together with rod-like granular calcite structures ensheathed in secondary fibres, links the thecideidines with denticulate spiriferaceans. Comparison of the thecideidine shell microstructure with that of a stropheodontid strophomenide Amphistrophia has failed to reveal comparable microstructural elements. The new evidence indicates that the spiriferacean denticle is a structural homologue of the thecideidine tubercle and, from a systematic point of view, removes any remaining objection to the formal assignment of the Thecideidina as a suborder of the Spiriferida. The morphological similarity between thecideidines, suessiacean spiriferides, and certain davidsoniacean and productidine strophomenides is now regarded as an expression of homoeomorphy. In recent years attention has centred on the question of whether the thecideidine brachiopods share affinity with the Strophomenida or the Spiriferida. The main arguments in favour of derivation from strophomenide stock were advanced by Rudwick (1968) and Baker (1970) who supported the idea of descent from the Davidsoniacea. Pajaud (1970) and Grant (1972) similarly argued for a stropho- menide ancestor but were in favour of derivation from productidine stock. The only strong dissent was voiced by Williams (1968, 1973) who, on the basis of shell microstructure, suggested that the thecideidines were derived from spiriferide stock and, more specifically, from the Suessiacea. The importance of neoteny in thecideidine evolution has been repeatedly stressed (Elliott 1953; Pajaud 1970; Williams 1973). If neoteny has exerted the profound influence which most workers believe to be true, the early thecideidines are likely to bear a much closer resemblance to juveniles of ancestral forms than to their adult counterparts. Unfortunately, early juveniles are not only less common than adults of the species but also more difficult to identify. After reviewing the morphological and microstructural evidence, Williams (1973, p. 466) concluded that certain persistent characters in the various thecideidine lineages were of fundamental phylogenetic significance. He identified shell microstructure as probably the most important character. The question posed, therefore, is whether the shell microstructure of early middle Jurassic thecideidines represents an ontogenetic ancestral character which through neoteny became ‘frozen’ into an adult shell fabric. Work in progress on the preservation of ontogenetic relics in the shell fabrics of adult articulate brachiopods pinpoints the umbonal region of the brachial valve and, where no shell resorption has occurred, the pedicle valve also, as an area of great importance in the determination of phylogeny. However, because of the effects of neoteny, the solution to the problem of thecideidine affinity may never be reached through study of adult shell fabrics of even immediate potential ancestors. The aim of the current investigation has been, therefore, to establish whether the shell fabrics of juvenile representatives of Palaeozoic spiriferide and strophomenide genera provide unequivocal evidence of spiriferide or strophomenide affinity. Unfortunately, within the Thecideidina the shell micro- structure exhibits such a bewildering variety of detail that it becomes difficult to isolate those characters which are of major significance. However, in spite of the drastic changes which affected the shell microstructure of the later representatives of the group, studies have shown that tubercles and secondary fibrous shell are characteristic and persistent features of the thecideidine shell (Baker 1970; Baker and Laurie 1978; Smirnova 1979; Williams 1973). When traced back to early representatives of the group, the tubercles are found to originate as cored structures in forms with a normal (sensu | Palaeontology, Vol. 27, Part 4, 1984, pp. 857-866, pis. 77-78.| 858 PALAEONTOLOGY, VOLUME 27 Williams 1968) secondary fibrous layer. Since tubercle cores seem to be a fundamental feature of the thecideidine shell structure, persisting throughout the history of the group, it seems reasonable to assume that similar structures would be a character of the ancestral stock. The shell microstructure of juveniles of certain spiriferide and strophomenide genera was investigated with this in mind. Registration of material. The material investigated in this study is to be housed in the British Museum (Natural History) as BB81 1 15-81 119. The BM(NH) specimens referred to in the discussion are relocated as numbers BB84702 (complete shell ex tubed specimens B32375D) and BB84703 (pedicle valve ex tubed specimens B32376B). Preparation of material. Sufficiently large specimens were cut at the required orientation using a Logitech ‘Trimsaw’. The cut face was then finished, using F800 C6 black silicon carbide abrasive powder, followed by etching for ten seconds in 5% hydrochloric acid. Small specimens were mounted in cold-setting resin before being sub jected to the above preparation technique. All material selected for stereoscan electron microscopy was gold-coated before photography. SHELL MICROSTRUCTURE Spiriferide shell microstructure. The shell microstructure of spiriferide brachiopods has been the subject of a detailed study by MacKinnon (1974) and it is not necessary to add to his account of the general shell fabric encountered. He appears, however, to have overlooked aspects of the shell microstructure of denticulate spiriferaceans such as Unispirifer. This is unfortunate as investigation of the umbonal region of juveniles of lower Carboniferous specimens has provided the first clear evidence of structures in spiriferides which, apart from differences in size and orientation relative to the external surface of the shell, are otherwise identical with the tubercles found in the pedicle valve of certain basal middle Jurassic thecideidines. The general shell microstructure of Unispirifer is identical with that of other spiriferacean genera studied by MacKinnon (1974). Study of the surface of the ventral interarea of well-preserved juveniles, however, reveals the presence of a parallel series of granular calcite-filled grooves aligned perpendicular to the hinge line (PI. 77, fig. 1). Unfortunately, the umbonal regions of all the available specimens have suffered some abrasion or exfoliation so that the primary shell is nowhere complete. The best-preserved material, however, clearly shows that on the ventral interarea the striae, orientated at right angles to the hinge axis, are really in-sunk areas of primary shell which become more pronounced as the primary layer is lost (PI. 77, fig. 2). Sections parallel with the hinge axis and perpendicular to the surface of the interarea show that the grooved areas are underlain by trough-like invaginations around which the secondary shell mosaic is deflected (PI. 77, figs. 3, 4), indicating that the intervening ridges are an artefact produced by removal of shell material from the grooves. EXPLANATION OF PLATE 77 Figs. 1 -8. Unispirifer sp., juvenile specimens, north Derbyshire (precise horizon and locality unknown), Visean limestones, lower Carboniferous. BB81 1 17 (figs. 1 -5), BB81 1 18 (fig. 8), and BB81 1 19 (figs. 6, 7). 1, oblique view of ventral interarea, hinge-line upper right, showing the ridges and grooves formed by removal of the majority of the primary shell layer, x45. 2, surface view showing the granular calcite-filled troughs which deflect the secondary shell fibres, x 80. 3, transverse section, parallel with hinge axis, through ventral interarea to show detail of the way in which the granular calcite filling a trough deflects the secondary shell fibres, x 400. 4, exfoliated region, same orientation as fig. 3, showing detail of a trough from which the granular calcite has been removed, x 250. 5, oblique view, ground surface top right, to show the rod-like (accentuated by etching) nature of the granular calcite body occupying the trough, x 250. 6, section parallel with surface of ventral interarea showing the granular calcite, together with traces of a gross pseudo-fibrous mosaic, of a denticle and its associated core in longitudinal section, x 400. 7, detail of the granular calcite of the denticle core in fig. 6, location as indicated, x 3000. 8, surface view of ventral interarea showing detail of primary layer (left) overlying a denticle core, and the fibrous secondary shell (right) adjacent to it, x 800. Stereoscan photomicrographs. PLATE 77 BAKER, Unispirifer 860 PALAEONTOLOGY, VOLUME 27 Specimens in which the structures are partially exfoliated (PI. 77, fig. 5), and sections parallel with the surface of the interarea (PI. 77, fig. 6), show that the troughs are occupied by rod-like bodies of granular calcite approximately 80 ^.m in diameter, whose outer surface remains in contact with the primary layer along their length (text-fig. 1b-d). Longitudinal sections through the rod-like bodies (PI. 77, fig. 6) show them to be continuous with the denticles developed along the hinge margin. Although the shell material of the denticle core is clearly granular calcite (PI. 77, figs. 7, 8), a gross fibrous mosaic appears to be discernible (PI. 77, fig. 6) and is discussed later. The in-sunk condition of the denticle cores of Unispirifer is apparently different from that of the upper Devonian Tenticospirifer whose denticles were depicted by Williams and Rowell (1965, p. H94, fig. 100c, e) as forming denticular ridges on the surface of the interarea. Thecideidine shell microstructure. The presence of cored tubercles in both valves of the shell of M o or ellina granulosa (Moore) has been demonstrated by Baker (1970). The tubercles of the brachial valve were shown to have granular cores aligned almost perpendicular to the primary shell layer, whilst those of the pedicle valve had cores of conically arranged fibres aligned almost parallel with the inner surface of the primary layer (Baker 1970, p. 91, text-fig. 6). Williams (1973), in his detailed and comprehensive study of Recent and the majority of fossil thecideidine taxa, demonstrated that the evolution of the group was characterized by the neotenous suppression of the fibrous secondary layer; he identified the main onset of the sporadic secretion of secondary shell as a late Jurassic or early Cretaceous event. The recent discovery of a new genus (Baker and Elston 1984) clearly shows that in one stock at least the suppression of secondary shell was well advanced by basal middle Jurassic times. Mimikonstantia sculpta possesses cored tubercles and those of the brachial valve have granular calcite cores (Baker and Elston 1984, pi. 71, fig. 2) almost identical with those found in Moorellinci granulosa. The tubercles of the pedicle valve of Mimikonstantia sculpta , however, have attenuated granular calcite cores (Baker and Elston 1984, pi. 71, figs. 6, 7) approximately 40 ^ m in diameter, in sharp contrast with the fibrous cores of Moorellinci granulosa although their orientation approximates even more closely to an alignment parallel with the primary layer (text-fig. 1f-h). A significant difference between the shell fabric of M. granulosa and Mimikonstantia sculpta is that in the latter species the secondary fibrous shell layer is greatly reduced in thickness and is underlain by granular calcite. In this respect the shell of M. sculpta very closely resembles that of Cretaceous genera such as Thecidiopsis. Strophomenide shell microstructure. Laminar shell (Williams 1968) is, with few exceptions (Williams 1970), a fundamental character of the strophomenide shell although, as Williams concluded (1970, p. 339), the strophomenides must have evolved from fibrous-shelled ancestors. It was important therefore, because of the possibility of a paedomorphic origin of the thecideidine shell fabric, to examine the shell microstructure of certain juvenile strophomenides to try to ascertain whether any of the characters could be correlated with those observed in the thecideidine shell. The discovery of the granular calcite core in the denticles of the spiriferacean ventral interarea necessitated an investigation of the interareas of stropheodontid strophomenaceans, in order to establish whether the microstructure of the stropheodontid denticle exhibited characters correlateable with the thecideidine tubercle. Horizontal and transverse sections through the ventral interarea of an upper Silurian juvenile Amphistrophia show that the denticles are cored structures (PI. 78, figs. 1, 2). Like pseudopunctae, however, the denticles have a core of crystalline calcite enveloped in typical laminar shell in which the laminae are deflected distally (PI. 78, figs. 3-5). The denticles, therefore, appear to originate in the same way as pseudopunctae and, in view of the distribution of the occurrence of pseudopunctae throughout the Strophomenida, may be regarded as modified pseudopunctae. DISCUSSION If the thecideidines arose neotenously (Elliott 1953; Pajaud 1970; Williams 1973) or paedomorphically (Williams and Rowell 1965), the validity of conclusions drawn from comparison of morphological BAKER: THECIDEIDINE BRACHIOPOD ANCESTRY 861 text-fig. 1. Exploded block reconstructions (not to scale), a-d, Unispirifer , small section of the ventral interarea (location as indicated) of a juvenile to show the relationship between the primary shell layer and the denticle cores. E-H, Mimikonstantia , small section of the free ventral wall (location as indicated) of the pedicle valve to show the relationship between two comparable tubercle cores and the primary shell layer, d., denticle; d.c., denticle core; g.l., growth line; p.!., primary shell layer; st., striation; t., tubercle; t.c., tubercle core. 862 PALAEONTOLOGY, VOLUME 27 characters may be considerably weakened. There is also an explanation other than genetic relationship for the morphological similarity between thecideidines, davidsoniacean stropho- menides, and suessiacean spiriferides. Work by Cooper and Grant (1974) on the beautifully preserved reef-associated Permian faunas of West Texas showed that many of the Permian forms which most closely resemble thecideidines were an abundant element of patch reef faunas. Other studies (Baker 1981, 1983) established that middle Jurassic thecideidines were characteristically associated with patch reefs or coralliferous debris adjacent to patch reefs. Among the brachiopods there is a clear correlation between typically conical pedicle valves, loss of pedicle, complete delthyrial covers, weakly convex brachial valves, and the association with reefs. In short, the characters may be regarded as a response to environmental pressure, a view which is confirmed by the characteristic morphology of richthofenid brachiopods and rudistid bivalves. Williams (1973) drew attention to the danger of ascribing genetic significance to characters of convergent origin. It now seems probable that the morphological similarity between davidsoniaceans, suessiaceans, and thecideidines is environmentally induced; in which case, the davidsoniacean and suessiacean genera previously regarded as genetic relatives of the thecideidines are nothing more than heterochronous homoeomorphs. Therefore, in attempting to trace affinity, attention must be focused on characters which are likely to be less susceptible to environmental pressure. Although morphological comparison is suspect, there are certain characters which appear to be so fundamental to thecideidines and their ancestral stock that none of the caenogenetic changes contributing to the emergence of the thecideidines (Williams 1973, p. 469) was sufficiently profound to be able to mask them. Jaanusson (1971) noted that brachiopod teeth were either deltidiodont or cyrtomatodont, and that with only two exceptions forms with deltidiodont teeth did not acquire the ability to use resorption for the construction of their shells. This frequently overlooked observation has profound implications for any proposed strophomenide (deltidiodont) line of descent because the ability to resorb shell material is of such crucial importance for the construction of thecideidine (cyrtomatodont) shells that it is almost certain to be a capability shared by their immediate ancestors. Similarly, tubercles are such a persistent feature of the thecideidine shell that they should, together with fibrous secondary shell, be traceable back to the ancestral stock. Baker (1970) concluded that a better knowledge of the Triassic genus Thecospira was critical to an understanding of thecideidine systematics. Three significant contributions (Dagis 1973; Williams 1973; MacKinnon 1974) were soon forthcoming and it is perhaps unfortunate that they served only to polarize still further the views already held. Dagis (1973) supported the views of Rudwick (1968, EXPLANATION OF PLATE 78 Figs. 1-5. Amphistrophia funiculata (M'Coy), juvenile specimens. Wren's Nest, Dudley, Much Wenlock Limestone Formation, upper Silurian. BBS 1115 (figs. 1, 4, 5), BB81116 (figs. 2, 3). 1, transverse section through hinge area showing crystalline calcite denticle core in longitudinal section, flanked by laminar shell, x 400. 2, horizontal section through lunge area showing crystalline calcite denticle core in transverse section, flanked by laminar shell, x 1000. 3, horizontal section through lateral region showing crystalline calcite taleola in transverse section, flanked by laminar shell, x 1000. 4, transverse section through the shell showing laminar shell deflected by a taleola, x400. 5, detail of laminar shell in transverse section, x 1000. Figs. 6-8. Aft'. Moorellina. BB84702, Dundry Hill, Bristol, Inferior Oolite (precise horizon and locality unknown). 6, brachial view, showing striations on ventral interarea, x 15. 7, angled view (tilt angle 25°) of ventral interarea with ridge and groove structure perpendicular to hinge axis, x 35. 8, part of ventral interarea showing detail of ridge and groove structure, and granular nature of the shell, x 400. Figs. 9 12. Cf. Moorellina. BB84703, Dundry Hill, Bristol, Inferior Oolite (precise horizon and locality unknown). 9, angled view (backward rotation 50°) showing row of small tubercles along hinge-line, x 12. 10, part of hinge-line showing tubercles in more detail, x 100. 11, broken hinge tubercle showing its apparently granular structure, x 1000. 12, surface view of part of hinge tooth showing clearly defined fibrous structure, for comparison with hinge tubercle in fig. 1 1, x 1500. Stereoscan photomicrographs. PLATE 78 BAKER, Amphistrophia and Moorellina 864 PALAEONTOLOGY, VOLUME 27 1970) and Baker (1970) that Thecospira was of strophomenide affinity. Williams (1973), however, reiterated his earlier (1968) view that Thecospira was of spiriferide affinity. Williams’s view was supported by MacKinnon (1974) in his comprehensive survey of spiriferide shell structure which included a detailed study of the microstructure of Thecospira from the Triassic St. Cassian Beds of northern Italy. Dagis (1973) studied a wider range of material than was available to MacKinnon and established beyond doubt that the shell structure of thecospirids is almost identical with that of middle Jurassic thecideaceans, such as Moorellina and Mimikonstantia , even to the extent of the dissimilarity of the structure of the brachial and pedicle valves. Thecospirids, therefore, both morphologically and in the organization of their secondary shell fabric, are much closer to thecideidines than they are to any undoubted spiriferides and there can be little doubt that the relationship is a genetic one. Dagis (1973, p. 367) concluded that among the thecospirids Hungaritheca is in all probability ancestral to the Thecideidina. I agree that Hungaritheca , if not actually ancestral to the thecideidines, is certainly very close to the thecideidine line of descent. The plexus of descent of the thecideidines is thus inextricably linked with the derivation of the thecospirids also. Prior to Dagis’s (1973) evidence, I had previously expressed the opinion that thecideidine tubercles might be the functionally modified homologue of the strophomenide pseudopunctae (Baker 1970, p. 97). This opinion must now be revised as granular calcite appears to be the primitive tubercle core material in thecideidines. The evidence is also weak from a paedomorphic point of view as the strophomenide taleola, even at a very early age, is characteristi- cally flanked by laminar shell (PI. 78, fig. 5). It now seems, therefore, that the strophomenide pseudopunctae have no counterpart among the Thecideidina. Consequently, the contention that the thecideidines might be descended from strophomenide ancestors is not supported by the present study. In view of the demonstrable link between thecideidines and thecospirids, an important consideration arising from the new evidence is whether a similar link exists between the thecideidines and the spiriferaceans. In both Unispirifer and Mimikonstantia the rod-like granular calcite bodies possessed a distal accretion zone that remained slightly in advance of the secretion of secondary fibres which were subsequently deflected. Their growth pattern caused them to emerge along the hinge line as a row of tiny denticles in Unispirifer (text-fig. 1a) and along the inner margin of the valve edge as a row of small tubercles in the pedicle valve of Mimikonstantia (text-fig. 1e). In Unispirifer the denticle cores maintain a connection with the primary layer throughout their length (text-fig. 1b-d). Such cores would most easily have been formed by an invagination of the primary layer which, in the case of Mimikonstantia , may be regarded as having become ‘pinched off’ from the primary layer during their development (text-fig. If, g) to become totally ensheathed in secondary fibres. In Unispirifer the orientation of the denticles must mean that they were generated consecutively (text-fig. 2a, b) as the hinge line increased in length. Once initiated, the structures apparently continued to develop during the life of the animal. In Mimikonstantia on the other hand the tubercles show a systematic intercalary generation pattern (text-fig. 2d, e) which may be reconciled with the need to ensure the effective maintenance of the continuity of arrangement and location of the tubercles along the inner margin of the pedicle valve as growth proceeded. The difference in the location of the structures in the two genera is not a serious problem. Both sets of structures are sequential, with early and later representatives, and the arrangement of the tubercles bears the same relationship with the ventral umbo (text-fig. 1 ) irrespective of whether the structures are located in the ventral interarea or in the wall of the ventral valve. All that is required is a slight change in the growth pattern of the shell. Thecideidines have a relatively much shorter hinge line than spiriferaceans; therefore, if the trend towards the shortening of the hinge line was independent of the tubercle generation pattern, a slight change in the growth pattern of the shell could have resulted in the development of the structures along the lateral (text-fig. 2c) and, ultimately, the anterior margin of the valve. The same development pattern may have been triggered by a slight change in functional requirements (e.g. reef-association), in which case the change to an intercalary development pattern and the elimination of denticles and the appearance of tubercles in thecideidines may be explained as an evolutionary development. THECIDEIDINE BRACHIOPOD ANCESTRY 865 BAKER: text-fig. 2. Generalized diagrams to compare consecutive development patterns, a, b. Unispirifer, denticles in the ventral interarea as the hinge-line increases in length, d, e, Mimikonstantia, essentially intercalary pattern of the tubercles round the anterior border of the pedicle valve as the shell increases in size, c, cf. specimen BB84703 (PI. 78, figs. 9, 10) hypothetical transitional stage with denticles along the hinge-line and tubercles on the lateral margins of the valve. The above argument would obviously be strengthened if any early thecideidines showing some of the juvenile Unispirifer characters could be located. A search of the British Museum (Natural History) collections has revealed two specimens from the Inferior Oolite (horizon uncertain) of Dundry Hill near Bristol. BB84702 is a complete shell, 3-3 mm in width and assignable to aff. Moorellina , which has a ventral interarea (PI. 78, figs. 6-8) showing clearly developed striations perpendicular to the hinge axis. BB84703 is a pedicle valve, 4 mm in width and assignable to cf. Moorellina , with a row of tubercles (denticles?) along the hinge line (PI. 78, figs. 9, 10) in a position similar to that occupied by the denticles of Unispirifer. A broken tubercle (PI. 78, fig. 1 1) appears to be composed of granular material, whereas the hinge tooth (PI. 78, fig. 12) clearly displays a fibrous structure. This evidence, considered in conjunction with the microstructure described, convinces me that the spiriferacean denticle is homologous with the thecideidine tubercle. With regard to the thecospirids, Dagis (1973) described a pseudo-fibrous texture for the tubercle cores of the pedicle valve of Thecospira communis in which the individual fibres were composed of acicular grains of calcite. Williams (1973) described a similar arrangement in the secondary fibres of the teeth of Thecidellina barretti (Davidson). It is interesting that the granular denticle cores of the Unispirifer studied here have the appearance of a gross pseudo-fibrous fabric (PI. 77, fig. 6) which may be correlated with the thecospiriid tubercle cores described by Dagis. Nalivkin (1976, p. 70) noted the presence of low narrow ridges and striae on the interareas of a number of spiriferides and concluded that during life the structures were associated with a covering of byssal attachment filaments. None of Nalivkin's material is available for study but the evidence from the British lower Carboniferous material, which enables the external ridges and grooves to be correlated with underlying structural elements of the interarea, is incompatible with his (1976) interpretation. CONCLUSIONS Characteristic though all the features evaluated by Williams (1973) are, the evidence indicates that cyrtomatodont teeth, secondary fibrous shell, tubercles, and the ability to resorb large tracts of shell material must be regarded as criteria of paramount importance in the indication of thecideidine ancestry. The morphological similarity previously cited as evidence of strophomenide or spiriferide affinity may simply reflect homoeomorphy. Shell microstructure, however, demonstrates unequivo- cally that the thecideidines are genetically related to the thecospirids and it is here recommended that the thecospirids should be assigned, as a taxon of superfamily rank, to the Thecideidina. The establishment of a genetic link between the shell microstructure of early middle Jurassic thecideidine and lower Carboniferous spiriferacean brachiopods leads to the conclusion that, however unlikely it may seem on morphological grounds, spiriferacean forms probably include the ancestral stock from 866 PALAEONTOLOGY, VOLUME 27 which the thecideidines and the thecospirids were derived. In this respect it is noted that the upper Devonian Tenticospirifer with its hemipyramidal pedicle valve and relatively flattened brachial valve is beginning already to approximate to thecideidine external morphology. Acknowledgements. I acknowledge the help received from Mr. E. F. Owen, British Museum (Natural History) in locating and facilitating access to the museum collections. I thank Sir Alwyn Williams, University of Glasgow, for the stimulation and encouragement of my ideas, and Dr. P. H. Bridges, Derbyshire College of Higher Education, for helpful comments on the manuscript. REFERENCES baker, p. G. 1970. The growth and shell microstructure of the thecideacean brachiopod Moorellina granulosa (Moore) from the Middle Jurassic of England. Palaeontology , 13, 76-99. 1981. Interpretation of the Oolite Marl (Upper Aalenian, Lower Inferior Oolite) of the Cotswolds, England. Proc. Geol. Ass. 92, 169-187. 1983. The diminutive thecideidine brachiopod Enallothecidea pygmaea (Moore) from the Middle Jurassic of England. Palaeontology , 26, 663-669. and elston, d. g. 1984. A new polyseptate thecideacean brachiopod from the Middle Jurassic of the Cotswolds, England. Ibid. 27, 777-791. and laurie, k. h. 1978. Revision of Aptian thecideidine brachiopods of the Faringdon Sponge Gravels. Ibid. 21, 555-570. cooper, G. a. and grant, r. e. 1974. Permian brachiopods of West Texas, II. Smithson. Contr. Paleobiol. 15, 233-793. dagis, a. s. 1973. Ultrastructure of thecospirid shells and their position in brachiopod systematics. Paleont. J. 6, 359-369. elliott, G. f. 1953. The classification of the thecidean brachiopods. Ann. Mag. nat. Hist., Ser. 12, 6, 693-701. grant, r. e. 1972. The lophophore and feeding mechanism of the Productidina (Brachiopoda). J. Paleont. 46, 213-249. jaanusson, v. 1971. Evolution of the brachiopod hinge. Smithson. Contr. Paleobiol. 3, 33-46. muckinnon, D. I. 1974. The shell structure of spiriferide Brachiopoda. Bull. Br. Mus. nat. Hist. (Geol.), 25, 189-261. nalivkin, d. v. 1976. The spiriferid interarea. Paleont. J. 10, 67-71. pajaud, d. 1970. Monographic des Thecidees (Brachiopodes). Mem. Soc. geol. Fr. 112, 1-349. rudwick, m. j. s. 1968. The feeding mechanisms and affinities of the Triassic brachiopods Thecospira Zugmayer and Bactrynium Emmrich. Palaeontology, 11, 329-360. 1970. Living and fossil brachiopods , 6-199. Hutchinson, Univ. Libr., London. Smirnova, T. N. 1979. Shell microstructure in Early Cretaceous thecidean brachiopods. Paleont. J. 13, 339-344. williams, a. 1968. Evolution of the shell structure of articulate brachiopods. Spec. Pap. Palaeont. 2, 1-55. — 1970. Origin of laminar-shelled articulate brachiopods. Lethaia, 3, 329-342. — 1973. The secretion and structural evolution of the shell of thecideidine brachiopods. Phil. Trans. R. Soc., Ser. B, 264, 439-478. — and rowell, a. j. 1965. Morphology. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part H. Brachiopoda. Pp. H57-H155. Geological Society of America and University of Kansas Press, New York and Lawrence. p. g. baker Department of Geology Derbyshire College of Higher Education Typescript received 1 1 January 1984 ' Kedleston Road Revised typescript received 13 March 1984 Derby DE3 1GB PYROTHERIU M , A LARGE ENIGMATIC UNGULATE (MAMMALIA, INCERTAE SED IS) FROM THE DESEADAN (O LIGOCENE) OF SALLA, BOLIVIA by BRUCE J. MACFADDEN and CARL D. FRAILEY Abstract. A well-preserved sample of Pyrotherium is described from at least two stratigraphic horizons in the Salla Beds of Bolivia. This sample is essentially indistinguishable from, and therefore conspecific with, the species P. romeri from Argentina. This represents the first description of this large herbivorous mammal outside of the classic 'Pyrotherium Beds’ of Argentina. The presence of this biochronologically diagnostic taxon further supports previous assignments of a Deseadan (Oligocene) age for the Salla Beds of Bolivia. T he Age of Mammals in South America is characterized by a highly unique and endemic fauna. The pyrotheres, or ’fire-beasts' (in reference to their occurrence in volcanic ash deposits), which are known from the late Palaeocene to the Oligocene in South America, are among the most enigmatic of eutherian mammals. Although of uncertain affinities, the Oligocene terminal member of this group is so distinctive and commonly encountered that it gave rise to the term 'Pyrotherium Beds’ to characterize the Deseadan land mammal age as it was originally defined from Argentina (e.g. Ameghino 1895). The original description of pyrotheres was based on the Deseadan genus Pyrotherium which has a diagnostic suite of characters including prominent upper and lower tusks, bilophodont cheek teeth, and large, graviportal limbs. Subsequently, more primitive forms have been described from earlier Tertiary localities in Argentina (see Simpson 1968), Colombia (Hoffstetter 1970), and Venezuela (Patterson 1977). The present paper describes new material from the Deseadan (Oligocene) Salla Beds of Bolivia. This paper is principally based on specimens collected during the 1 960s by L. Branisa (then of LaPaz, Bolivia) for Princeton University. Other, more fragmentary, specimens were collected by the authors and associates during our 1981 and 1983 field seasons. The latter specimens do not contribute to an increased understanding of the morphology of Salla Pyrotherium. However, they are associated with precise stratigraphic data and they extend the palaeobiogeographic range of this distinctive taxon outside the classic Deseadan localities of Argentina. The ‘Estratos de Salla’ have attracted much attention in the literature because of a very diverse and abundant Deseadan fauna (text-fig. 1 ). A general introduction to the geology of the Salla-Luribay basin has been presented by Hoffstetter (1968, 1976), Hoffstetter et at. (1971), and Villarroel and Marshall (1982). Various parts of the mammalian fauna have been presented including rodents (Hoffstetter 1976; Patterson and Wood 1982), marsupials (Villarroel and Marshall 1982), the earliest South American primate Branisella (Hoffstetter 1969; Wolff, in press, 6), Argyrolagidae (Wolff, in press, a), and ungulates (Cifelli and Soria 1983a, b). The remainder of the rich Salla fauna presently is undescribed. In early descriptions of Pyrotherium from the Deseado of Argentina, Ameghino (1895, 1902) believed these mammals to be similar and closely related to primitive proboscideans like Palaeomastodon from the late Oligocene Fayum of Egypt and Deinotherium from the early Miocene of the Old World. He believed that pyrotheres were the stem group from which the Old World proboscideans were ultimately descended. Loomis (1914) also considered Pyrotherium to be most | Palaeontology, Vol. 27, Part 4, 1984, pp. 867-874.| 868 PALAEONTOLOGY, VOLUME 27 closely related to proboscideans. Gaudry ( 1909) concluded that Pyrotherium was unlike any known large mammal and did not fit into any then existing family. Patterson (1977), based on several basicranial characters, concluded that pyrotheres are notoungulates, although the characters used for this assessment have been criticized elsewhere (Simpson 1978, 1980; McKenna 1980). Other work (e.g. on tarsal bones) suggests similarities between pyrotheres and the Embrithopoda or Proboscidea (Cifelli 1983). All of the above mentioned hypotheses of relationships between pyrotheres and other eutherians require rather interesting palaeobiogeographical speculation. In the absence of any M ACFADDEN AND FRAILEY: BOLIVIAN PYROTHERIUM 869 unambiguous diagnostic characters, the pyrotheres are considered here as a separate order of eutherian mammals incertae sedis. Although the sample of Pyrotherium from Salla does not elucidate the phylogenetic affinities of this group, it is nevertheless important to place on record a description of this material from this relatively new and significant locality. The following abbreviations are used in the text: LACM, Natural History Museum of Los Angeles County, Vertebrate Paleontology Collection; PU, Princeton University, Vertebrate Paleontology Collection; R, right side; L, left side; x, mean; s, standard deviation; V, coefficient of variation; OR, observed range; mm, millimeters. For relevant cheek teeth, the following measurements were taken: (1) greatest anteroposterior length including cingulum; (2) greatest transverse width across anterior loph and lophid; and (3) greatest transverse width across posterior loph and lophid. Because the dental homologies of advanced pyrotheres are uncertain, topographic names for dental characters are used in this paper (e.g. anterior loph rather than protoloph). SYSTEMATIC PALAEONTOLOGY Class mammalia Linnaeus, 1758 Order pyrotheria Ameghino, 1895 Family pyrotheriidae Ameghino, 1 895 Pyrotherium romeri Ameghino, 1889 Text-figs. 2-3 Referred material. Uppers: PU 20693, maxillary fragment with RP2 M3, LP2, P2-Mfi PU 21919, LP3-M1; PU 21917; RM’-M3; PU 22146, anterior part LM3; PU 22143, LM3 fragment; PU 22143, posterior loph of M2 or M3; PU 22144, posterior loph LM3; PU 22145, LM2; PU 22142, 21917, upper cheek teeth; PU 20683, premaxilla with parts of four incisors. Lowers: PU 20679, mandible with fragment LI, R and LP3-M3; PU 20695, mandible with alveoli and RM3; PU 21918, L ramus with M2, M3; PU 21989, R ramus with alveoli P4, Ml5 fragmentary M2, M3; PU 21921, fragmentary ramus with partial ?M3; PU 22141, RP3 fragment; PU 22147, LM3 fragment; PU 20694, ramus with LP3-M3; PU 20684, mandible with R and LI, RP3-M3, LM , -M3 (also see uppers above); PU 20692, mandible with fragment R and LI, RP3-M3, LP3, M3-M3; LACM 1 17571, LP3-M3; LACM 1 17572, RP3, RM3; LACM 117573, LP3. PU 22096, PU 22097, PU 22148; cheek tooth fragments. Also numerous uncatalogued specimens mostly consisting of tooth and postcranial fragments. There also are several uncatalogued fragments in the University of Florida — Servicio Geologia de Bolivia collection. Stratigraphic distribution and age. Based on our field collecting and from specimens with relevant locality data in the PU Branisa collection, Pyrotherium occurs from at least two zones within the Salla Beds: ( 1 ) the lower part of the section above the Luribay Conglomerates (Branisa’s Quebrada Chala Jahuira, Anchallani, V- 1 2) and from the same general area, but c. 100 m below the base of the principal guide level, or ‘Nivel Guia’ (also see Villarroel and Marshall 1982); (2) the middle part of the section, which based on our field-work, includes the most fossiliferous concentrations (V-3, Tapial Pampa; also see Villarroel and Marshall 1982). Some of the better- preserved specimens in the Branisa collection come from Pasto Huarante. Unfortunately, the exact stratigraphic position of this locality is unknown (Branisa, pers. comm. 1983). Description. As also noted for Pyrotherium from Argentina, the premaxillary region has four tusks; two on each half side and one behind the other. They emerge from the alveoli almost horizontally and curve downward to end almost vertically. The enamel is relatively thin, e.g. in contrast to proboscideans. The upper dentition consists of six cheek teeth, which are probably P2 M3. Relative to other Oligocene mammals from South America, Pyrotherium is very large as is reflected in its dental measurements (see Table 1). The P2 is triangular in shape and consists of an anterior cone and two posterior cones connected by a loph (text-fig. 2). The P3 and P4 are molariform, i.e. they consist of well-developed anterior and posterior lophs as do M1 to M3. There is a well- developed ‘heel’ on M3. On P3 M3 the lophs nearly comprise the total occlusal area of the tooth. In the premolars, these lophs sometimes merge at the external portion of the tooth almost forming an ectoloph. The enamel is breeched during early wear exposing the dentine (this character persists throughout later wear stages). In the upper cheek teeth, the plane of shear on the lophs is anteriad. The cingulum varies from well developed to 870 PALAEONTOLOGY, VOLUME 27 A B 0 2 4 6 8 10 cm text-fig. 2. Pyrotherium romeri from the Salla Beds, a, PU 20693, lateral view of R maxillary fragment with P3-M3. b, PU 20693, occlusal view P2 M3. c, PU 20694, occlusal view of R P3-M3. d, PU 20694, lateral view of mandible with anterior tusks and R P3-M3. absent; characteristically it is strong on the anterior and internal parts of the teeth and poorly developed, rudimentary, or absent on the external and posterior portions of the teeth. In both the upper and lower dentitions the enamel is frequently crenulated, particularly on the cingulum. The mandible is relatively robust (text-fig. 2). The bony portion of the symphysis extends posteriorly to a position below P4. The ascending portion of the ramus is dorsally abbreviated. Anteroventrally, the mandible is enlarged at the symphysis to accommodate the large pair of incisor tusks. The tusks are relatively robust and elongated with enamel only on the ventral side; dorsally dentine is exposed. The anteromost tip of the incisor MACFADDEN AND FRA I LEY: BOLIVIAN PYROTHERIUM 871 tusks is characteristically flattened by wear. The lower dentition consists of five teeth, presumably P3-M3. P3 consists of an anteromost conid followed posteriorly by two lophids. P4-M3 are generally similar in the presence of well-developed anterior and posterior lophids. As is also seen in the upper dentition, the enamel is breeched during early wear exposing the dentine. The principal shear on the lophids is posteriad. The cingulum varies from well developed and continuous, particularly on the posterior portion of the teeth, to rudimentary or absent. In M3 and, to a lesser extent in M2, there is a well-developed heel which seems to consist of an inflated cingulum. DISCUSSION The genus Pyrotherium was originally proposed by Ameghino (1889) to include the species P. romeri based on material from Patagonia, southern Argentina. As was characteristic of his taxonomic philosophy, Ameghino (1895, et seq.) later named at least four other species of Pyrotherium of which P. sorondoi has been most commonly cited in the literature. Although all of these 'species' were of roughly similar size, Ameghino and some later workers (e.g. Loomis 1914) mistakenly believed that P. romeri could be distinguished from the other species by the presence of P1. In a recent summary of pyrotheres, Patterson (1977) concluded that: (1) the 'P1' of P. romeri is actually a deciduous tooth; and (2) all the Deseadan species of Pyrotherium are synonymous. Therefore, P. romeri stands as the senior species and it is used in the present report. So far as can be determined, in all characters of the dentition, the Salla sample of Pyrotherium is indistinguishable from that of Argentina. Therefore, the Bolivian material is confidently referred to P. romeri. In the Salla sample of P. romeri there is considerable variation in size as evidenced by V s greater than 10 for certain characters (Table 1; text-fig. 3). With possible exception of one seemingly aberrant specimen (PU 21918, text-fig. 3), the Salla P. romeri corroborates Patterson’s (1977) idea of a single morph. Comparisons of this sample with the measurements of three specimens of P. romeri from Argentina (Loomis 1914, p. 181) suggest that the Bolivian sample might be slightly smaller than X h- Q LLI C/D DC LLI > CO z < DC h- 654- 60 — ^ 55 — 50 — 45 — 40 — 35 SALLA PYROTHERIUM LOWER M3 • PU 21918 35 40 45 50 55 60 65 ANTEROPOSTERIOR LENGTH (MM) text-fig. 3. Bivariate plot of anteroposterior length versus transverse width of anterior loph for M3 of Salla Pyrotherium romeri in the PU and LACM collections. 872 PALAEONTOLOGY, VOLUME 27 E ClT (L) H E u. P c/f S- jd C/3 cd u O e s 'U S-H Oh C/3 c V-i 1) _o £ o jD JL> s_. o o o a X ’TD c Cd X a P jD CL J-H dJ o x ■ £ C/3 2 c cd *> 5-1 o V CQ C E cd o -*-» <— dj ? cd CQ u. X < X H cd soj G v> I O VI O L- (N ON NO (C) -H oo rt — « N/O t"- CO I 't rf (N (N (N r" r-~ vo I oo »n no ' o 9 on NO — cO ■cf NO ^ •'T op ^ O co I — — CJ NO — O O I O oo no co n VN T OO I On NO NO Vi U- 6 P' cb (N CO CO CO _h I no co oo 1 r- -T (N CO VI (N O N 6 OO co CO co co ro I - -H U- OO in I NO — NO OO U ro VI Tt NO NO OO On I NO - V V ro V (V (N -- 6 a. b. smith. 81 pp., 20 text-figs., 23 plates. Price £15 (U.S. $26.50). 26. (for 1981): The Fine Structure of Graptolite Periderm, by p. r. crowther. 119 pp., 37 text-figs., 20 plates. Price £25 (U.S. $44). 27. (for 1981): Late Devonian Acritarchs from the Carnarvon Basin, Western Australia, by G. playford and r. s. dring. 78 pp., 10 text-figs., 19 plates. Price £15 (U.S. $26.50). 28. (for 1982): The Mammal Fauna of the Early Middle Pleistocene cavern infill site of Westbury-sub-Mendip, Somerset, by m. j. bishop. 108 pp., 47 text-figs., 6 plates. Price £25 (U.S. $44). 29. (for 1982): Fossil Cichlid Fish of Africa, by], a. h. van couvering. 103 pp., 35 text-figs., \0plates. Price £30(U.S. $52.50). 30. (for 1983): Trilobites and other early Arthropods. Edited by d. e. g. briggs and p. d. lane. 276 pp., 64 text-figs., 38 plates. Price £40 (U.S. $70). 31. (for 1984): French Coniacian Ammonites by N. j. Kennedy. 160 pp., 42 text figs., 33 plates. Price £0000. 32. (for 1984): Autecology of Silurian Organisms. Edited by m. g. bassett and j. d. lawson. 295 pp., 75 text-figs., 13 plates. Price £0000. Field Guides to Fossils 1. (1983): Fossil Plants of the London Clay, by m. e. collinson. 121 pp., 242 text-figs. Price £7-95 (U.S. $14). Other Publications 1982. Atlas of the Burgess Shale. Edited by s. c. morris. 31 pp., 24 plates. Price £20 (U.S. $44). © The Palaeontological Association , 1984 Palaeontology VOLUME 27 • PART 4 Review of the distribution of the commoner animals in Lower Silurian marine benthic communities l. r. m. cocks and w. s. mckerrow 663 Growth analysis of Silurian orthoconic nautiloids R. A. HEWITT 671 Mode of life and autecology of Silurian-Devonian Grammysiidae (Bivalvia) L. F. marsh 679 Janeia silurica, a link between nuculoids and solemyoids (Bivalvia) LOUIS LILJEDAHL 693 Autecology and distribution of the Silurian brachiopod Dubaria BRIAN JONES and JOHN M. HURST 699 A reappraisal of the Lower Carboniferous lepidophyte Eskdalia Kidston B. A. THOMAS and S. V. MEYEN 707 A new conifer genus from the Lower Cretaceous Glen Rose Formation, Texas JOAN WATSON and HELEN L. FISHER 719 Crane Gfus fossils from the Maltese Pleistocene E. MARJORIE NORTHCOTE 729 Tooth form, growth, and function in Triassic rhynchosaurs (Reptilia, Diapsida) MICHAEL J. BENTON ' 737 A new polyseptate thecideacean brachiopod from the Middle Jurassic of the Cotswolds, England P. G. BAKER and D. G. ELSTON • 111 Ichnological nomenclature of clavate borings SIMON R. A. KELLY and RICHARD G. BROMLEY 793 An immature specimen of the crocodilian Bernissartia from the Lower Cretaceous of Galve (Province of Teruel, Spain) A. D. BUSCALIONI, E. BUFFETAUT, and J. L. SANZ 809 Revision of the bivalve family Pulvinitidae Stephenson, 1941 T. J, PALMER 815 Stem morphology of the Recent crinoid Chladocrinus ( Neocrinus ) decorus S. K. DONOVAN 825 Arthropleura trails from the Westphalian of eastern Canada DEREK E. G. BRIGGS, A. GUY PLINT, and RON K. PICKERILL 843 New evidence of a spiriferide ancestor for the Thecideidina (Brachiopoda) p. G. baker 857 Pyrotherium , a large enigmatic ungulate (Mammalia, Incertae Sedis) from the Deseadan (Oligocene) of Salla, Bolivia BRUCE J. MACFADDEN and CARL D. FRAILEY 867 A new actinolepid arthrodire from the Lower Devonian of Arctic Canada D. L. DINELEY and LIU YU H A I 875 It Printed in Great Britain at the University Press, Oxford by David Stanford, Printer to the University CO NSTITUTION NOIinillSNI_NVINOSHllSNS S3iavaai1 LIBRARIES SMITHSONIAN INSTITUTION NC m \ c° ^ CO ” CO oz < _ NviVAS^X o ■ o Xfrosv£Z - o -J 2 _1 z J 2 _8 S3 I 8 VH 3 11 LIBRARIES SMITHSONIAN INSTITUTION NOIlfllllSNI NVIN0SH1IINS S3!HVHail LI C 2 ^ 2 f“ CO m XL* p.c^ m z; xgmisgx- m CO — CO “ co\± CO INSTITUTION NOlinillSNI NVIN0SH1IWS S3iaVaai1 LIBRARIES SMITHSONIAN INSTITUTION Nl 2 00 2 .... CO 2 ,v < ✓<\v'sovyx s . s / co co S3 co co S3 i ava a > ^ 5 5 > 2 CO z co * Z CO k Z II LIBRARIES SMITHSONIAN INSTITUTION NOIlfUllSNI NVINOSHIIINS S3iaVaai1 LI co Z CO — co 1X1 w - /^S>v W ^ w ' — X\ - ,y •« /■&grr&\ — /o'^ro^x mv - a: < ;c^*6^L°y — \f^KS^sy m ~ ” ^ m o INSTITUTION NOlinillSNI ~NVIN0SH1I IAIS SBiavaail L I B R A R I ES“ SMITHSONIAN INSTITUTION Nl rn v v*st>’ ^sJvasv^x m co = co S: co S3 lava an libraries Smithsonian institution Nonniiism nvinoshiiins ssiavaan U 5 ^ ^ ^ — . z \ co z z > ' W > " 2 co 2 ^ '* 2 CO 2 co INSTITUTION NOlinillSNI NVI NOS Nil INS SHIHVHail LIBRARIES SMITHSONIAN INSTITUTION N< co = ••» CO — CO “ CO O X^os^J^ Z' O — Z j z -» saiavaan libraries Smithsonian institution NOlinillSNI nvinoshiiins saiavaan l rn t&ocjty ^ '^4$^ m w ^ co m INSTITUTION °° NOlinillSNrNVINOSHlIWS S3 I HVH 8 I T™L I B RAR 1 ES~)SMITHS0NIAN_INST1TUTI0N Nl •NX-. w 2 ^ w Z ,w. W Z ,v. iS.vC E 1 . V- . aWx E ^ 1 LIBRARIES SMITHSONIAN co ES SMITHSONIAN INSTITUTION NOIiflillSNI NVINOSHiliNS z o S3 1 avaa n BRARIES SMITHSONI/ z «£- NVIN0SH1IINS HVaail LIBRARIES SMITHSONIAN 1 ES SMITHSONIAN INSTITUTION NOIifUilSNI NVINOSHilWS"#S3 CO UJ avea n LIBRARIES SMITHSONS CO sni_nvinoshjliiais S3 1 a va an libraries Smithsonian INSTITUTION NOIifUilSNI CO NVINOSH1II' I ES SMITHSONIAN z INSTITUTION g/yos*£ NOliniliSNI NVINOSH1I1NS S3 1 avaa n" LIBRARIES SMITHSON!/