TITUTION NOlinillSNI^NVINOSHlIWS^a I MVa a n 'Ll BRAR S ES^ SMITHSONIAN^ INSTITUTION' NO 1 m V g S g jf|* a ° “ * " ^ ^ W‘ ‘ *' H „ A/ PI 2 XJWlVS?^ PI ^ PI UVH a II “LIBRARIES SMITHSONIAN^NSTITUTION^NOliniSISNrNVINOSHlMS SiSaVaail LI “"* Z. '* CO 2 Co Z z OT . w . _ .^. _ _ .. .. > s TITUTION NOlinillSNI NVINOSHlIIAIS^Sa I d VH 8 ll^LI B RAR I ES^SMITHSONIAN INSTITUTION NO to 5 \ co _ . — m . 5 to K «WE/ g S Q 2 =a 2 j 2 -J awnan libraries Smithsonian institution NoiiniiiSNi nvinoshiiws saiavaan n z C z r* 2 ~ m XgvDC^ ^ m w n^S2X m X ^ Nguiisgx m to ± CO £* ^ CO \ 5 to TITUTION NOlinillSNI NVINOSH1UMS S3IHVHSn LIBRARIES SMITHSONIAN INSTITUTION NC ... CO Z __ CO Z ■ tO 2 ... CO #•• S •* /^§&v * S f _ ^ 2 2 > I HVH a |]Z LI B RAR I ES^SMITHSONIAN^ INSTITUTION ''‘NOlinillSNI NVINOSBilWS^Sa I HVU 0 11 LI CO 5 CO — . CO uj ^ tu xoiiv^s. ^ ?^S\\ w ^ „ , . . .3 1 5 | ^0^ “ 5 ^ o TITUTION^ NQfiniliSNI^MVINOSHltSAIS SBlUVUan LIBRARIES SMITHSONIAN^INSTITUTION NC Z r~ > z i- z z O ^ o /^Tin^x “ O UJ1 sj > 33 — \-V7 ___ m 'C' ^ XjvAsi^X rn X^ost^' invaan libraries SMiTHS0NiAN-tNSTiTUTi0NC/>N0iiniiiSNi~NviN0SHiiSNS saiavaan LI ~ Z » CO ~ •A ^ .s _) X to o ± wwio/ ci z ,N , _, _ __ > ■' 5 Xcvosv^ > 2 TITUTION NOlinillSNI NVINOSHlIWS^ S 3 I B V H 9 ll^L I B R A R I ES^SMITHSONIAN INSTITUTION NC CO Z V to — CO ™ CO tr S 5 W “ — o =■ -J _J Z J Z -S iava a n libraries Smithsonian institution NoiiniiiSNi nvinoshhws saiavaan^u r- z r- z JNV!N0SHJlllAIS2S3 ! ava a ll^LI B RAR I ES^ SMITHSONIAN INSTITUTION ^ NOliniliSNI ""WlNOSHlIWS r v Z r= z z «“ \„ PO 2 ™ y^°yo>K 2 *Mr,. rr, 2 s xmi 511 ' z X^vas^' rn J5 c ' m x^vosv<^y w/ xjvasviv m £/)'■'— CO “ CO _ CO SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIIIAIS SBiavaail LIBRARIES SMITHSONIAN £ ^ z \ Of z W z > ^ ^ ^ | | I '■sr > z OT ■■••*' Z OT 2 w -V 2 _NVIN0SH1HMS S3iavdan LIBRARIES SMITHSONIAN INSTITUTION NO!iniliSN!_NV!NQSHl!WS w — 5 w “ co w S ^so^x W fj S, - \v*ALm»z ~i W _ i 7 5 ^ ~ v"Xdl^/ V~ WlNOSHlIWS S3 I avaail LIBRAR I ES^SMITHSONIAN”* INSTITUTION ^ NOliniliSNI "NVINOSHIIIAIS : v Z r- Z r] z i“ v O “ /tJSX O O x > 'v^vaI!^ s '\^v > 2 >• 2 '" \^S NVIN0SH1IWSWS3 I BVH a n^LI B RAR I ES^SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIIIAIS O __ O Nx£fosw£/ _ 2 =J Z SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIIIAIS S3 I avaail LIBRARIES SMITHSONIAf 2 3= 4*' .. O - ?= 5 «A P Palaeontology VOLUME 31 • PART 1 FEBRUARY 1988 O.E 101 Nrt' Published by The Palaeontological Association • London Price £23 00 THE PALAEONTOLOGICAL ASSOCIATION The Association was founded in 1957 to promote research in palaeontology and its allied sciences. COUNCIL 1987-1988 President : Dr. L. R. M. Cocks, Department of Palaeontology, British Museum (Natural History), Cromwell Road, London SW7 5BD Vice-Presidents: Dr. D. E. G. Briggs, Department of Geology, University of Bristol, Bristol BS8 1RJ Dr. L. B. Halstead, Department of Geology, University of Reading, Reading RG6 2AB 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. W. Owen, Department of Geology, The University, Dundee DD1 4HN Secretary: Dr. P. W. Skelton, Department of Earth Sciences, The Open University, Milton Keynes MK7 6AA Circular Reporter: Dr. D. J. Siveter, Department of Geology, University of Hull, Hull HU6 7RX Marketing Manager: Dr. V. P. Wright, Department of Geology, University of Bristol, Bristol BS8 1RJ Public Relations Officer: Dr. M. J. Benton, Department of Geology, The Queen’s University of Belfast, Belfast BT5 6FB Editors Dr. M. J. Benton, Department of Geology, The Queen’s University of Belfast, Belfast BT5 6FB Dr. P. R. Crowther, City of Bristol Museum and Art Gallery, Bristol BS8 1 RL Dr. D. Edwards, Department of Plant Science, University College, Cardiff CF1 1XL Dr. T. J. Palmer, Department of Geology, University College of Wales, Aberystwyth SY23 2AX Dr. C. R. C. Paul, Department of Geology, University of Liverpool, Liverpool L69 3BX Dr. P. A. Selden, Department of Extra-Mural Studies, University of Manchester, Manchester M13 9PL Other Members Dr. H. A. Armstrong, Newcastle upon Tyne Dr. G. B. Curry, Glasgow Dr. M. E. Collinson, London Professor B. M. Funnell, Norwich Dr. J. A. Crame, Cambridge Dr. P. D. Taylor, London 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. J. Cuffey, Department of Geology. Pennsylvania State University, Pennsylvania 16802 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 1987 are: Institutional membership £50 00 (U.S. $79) Ordinary membership £2100 (U.S. $38) Student membership . . £1 1 -50 (U.S. $20) Retired membership £10-50 (U.S. $19) There is no admission fee. Correspondence concerned with Institutional Membership should be addressed to Dr. A. W. Owen, Department of Geology, The University, Dundee DD1 4H V 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, Aston Triangle, 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 1988 will receive Palaeontology, Volume 30, Parts 1-4. Back numbers still in print may be ordered from Basil Blackwell, Journals Department, 108 Cowley Road, Oxford OX4 1JF, England. Cover: The brachiopod Meristina obtusa (J. de C. Sowerby, 1823), a life position assemblage from the Much Wenlock Limestone Formation, Abberley Hills, Hereford (Specimen no. BB 12345, x 1). Photograph by Harry Taylor of the British Museum (Natural History) Photographic Studio. PRESERVATION OF FISH IN THE CRETACEOUS SANTANA FORMATION OF BRAZIL by DAVID M. MARTI LL Abstract. Early diagenesis of calcareous concretions in the Santana Formation (Cretaceous) of north-east Brazil has allowed some fishes killed in mass mortality events to be preserved three dimensionally. Fluctuating salinities may have been responsible for the mass deaths of the dominantly marine fish fauna. Early phosphatization, brought about by bacterial activity, has allowed a variety of soft tissues to be preserved within the body cavity of a variety of fish taxa. Both high and low pH micro-environments existing within the body cavity of the fishes allowed the precipitation of microcrystalline francolite within the partly decomposing soft tissues, while early non-ferroan calcite was produced within the coelomic cavity. The Santana Formation (?Upper Aptian) of the Chapada do Araripe, north-east Brazil, is one of the most important Cretaceous lagerstatten (sensu Seilacher et al. 1985), and has long been famous for the beautiful preservation of its fossil fish fauna. An examination of the large collections of Santana Formation fossil fishes at the Field Museum of Natural History, Chicago, the American Museum of Natural History, New York, and the British Museum (Natural History), London, suggests that soft-tissue fossilization in this formation is relatively common. In some fish specimens nearly the entire musculature of the trunk may be preserved, and in most of the three-dimensional specimens there is some phosphatization of soft tissues. Also, most of the specimens contained in calcareous concretions are only slightly compacted, and the three-dimensional shape of the specimen can frequently be determined. Previous investigations of the fossil fishes have also revealed the presence of phosphatized soft tissues in associated ostracods (Bate 1972) and copepods(Cressey and Patterson 1973), while Campos et al. (1984) announced the discovery of pterosaurian wing membranes preserved in concretions from this formation. The following work is based on observations made on over 300 fossil-bearing concretions held in the Field Museum of Natural History (FMNH), Chicago, the American Museum of Natural History (AMNH), New York, the British Museum (Natural History)(BMNH), London, and the University of Leicester (LEIUG). Specimens used for destructive analysis were obtained from ‘Rock Art’, 4-6 Gypsy Lane, Leicester. LOCALITY The fossil-bearing concretions occur at a number of localities at the foot of the Chapada do Araripe, at the boundary between Ceara, Pernambuco, and Piaui provinces in north-east brazil (see text-fig. 1). The Chapada lies between 7° and 7° 45' S. and between 39° and 41° W., forming an east-west trending plateau approximately 150 km long, with a general elevation of between 600 and 900 m above sea-level. The plateau consists largely of flat lying Cretaceous sediments lying unconformably on Devonian and Pre-Cambrian basement. Exposure is poor and restricted to a few stream sections and gypsum mines. Most of the fossil fish concretions are collected by local farmers or fossil dealers and sold to tourists at markets in Sao Paulo, or exported to foreign fossil dealers. There are restrictions controlling the export of large quantities of fossils from Brazil. [Palaeontology, Vol. 31, Part 1, 1988, pp. 1-18, pis. 1-4.| © The Palaeontological Association 2 PALAEONTOLOGY, VOLUME 31 text-fig. 1 . Locality map of the Chapada do Araripe, Brazil. Outcrop of the Santana Formation is indicated by light stipple. STRATIGRAPHY The fish-bearing concretions are from the Santana Formation of the Araripe Series. The main outcrop lies at the foot of the Chapada do Araripe, but a few small outliers occur to the south and west of the chapada. The stratigraphy of the Santana Formation has been examined by Beurlen (1962, 1963, 1971), Braun (1966), and Silva Santos and Valenca (1967) and is generally considered to comprise three members (text-fig. 2). The lowest (Crato) member consists of a series of alternating shales and limestones (mainly biomicrites). At the top of the Crato Member a unit of finely bedded micrites (plattenkalk facies) yield abundant insects and the small fish Distilbe elongatus Silva Santos. This is overlain by a middle (Ipubi) member which is dominated by evaporites, including gypsum and anhydrite. The upper (Romualdo) member contains a variety of clays, limestones, and sandstone. Towards the base of the Romualdo Member there is a unit of bituminous shales and limestones (biomicrites) in which the fossil-bearing concretions occur. The Brazilian Cretaceous is divided into a number of isolated basins which cannot be easily cross correlated. Due to the lack of diagnostic fossils in these basins the precise age of the Santana Formation is uncertain. Diagnostic marine invertebrates are almost unknown in the Santana Basin, and the vertebrates can only be used stratigraphically in a broad sense. Thus the age is considered to be Aptian or Lower Albian. Brito (1984) has proposed the erection of the Vinctifer Biozone to allow correlations to be made with other Brazilian Cretaceous basins. Unfortunately, the fish Vinctifer (= Aspidorhynchus) is a long ranging genus, and it is clear that its distribution is palaeoecologically controlled. The use of Vinctifer as a zone fossil must be treated with caution. Sedimentation in the Brazilian Cretaceous basins was controlled by the separation of the South American and African plates, and the formation of the South Atlantic Ocean. Four distinct sedimentological episodes can be recognized (Brito and Campos 1982, 1983). The lower part of the Santana Formation lies within the Alagoan (non-marine) Stage. The passage from non-marine to marine sediments is thought to approximate to the Middle Aptian/Upper Aptian boundary (Brito 1984). SANTANA FORMATION EXU TEXT-FIG. MARTILL: PRESERVATION OF FOSSIL FISH 3 50 m Shales Biosparite Plattenkalk Sands Evaporites Concretions Fish Simplified stratigraphic section through the Santana Formation of the Chapada do Araripe. Sedimentological data from Mabesoone and Tinoco (1973). 4 PALAEONTOLOGY, VOLUME 31 PALAEONTOLOGY The Santana Formation is well known for the abundance and exceptional preservation of its contained fish fauna, the first records of which go back to the last century (Agassiz 1841, 1844; Gardner 1841; Woodward 1887) and early part of this century (Jordan and Branner 1908; Woodward 1908). The fish have received continued attention (Jordan 1923; D’Erasmo 1938; Schaeffer 1947; Silva Santos 1945, 1947, 1950, 1960, 1968, 1970a, b , 1974; Silva Santos and Valenca 1968; Campos and Wenz 1982) and are actively being studied as more taxa are being discovered. In recent years the Santana concretions have gained further importance for yielding amongst the worlds best-preserved pterosaur material. The pterosaurs were first described by Price (1971), and have been further investigated by Buisonje (1980), Campos (1983), Leonardi and Borgomanero (1983), Wellnhofer (1977, 1985), and Wellnhofer et al. (1983). Campos et al. (1984) reported a pterosaur in which wing membranes with supporting fibres are preserved. Other tetrapods reported from the Santana Formation concretions include crocodilians (Beurlen and Buffetaut 1981; Price 1959), turtles (Beurlen and Barreto 1968; Price 1973), and a possible dinosaur (Campos, D. de A. 1985). The invertebrate fauna is less diverse and there appears to be little information on the stratigraphic distribution of the various taxa. Smooth shelled ostracods are abundant in some of the fish-bearing concretions and occasionally copepods can be found. I have failed to observe any calcareous shelled invertebrates other than ostracods in the concretions. The invertebrate fauna of the formation as a whole has been reviewed by Lima (1979) and Mabesoone and Tinoco (1973). They list ostracods, decapod crustaceans (Beurlen 1963), and conchostracans, while Cressey and Patterson (1973) described the earliest known parasitic copepods. Gastropods, bivalves, and echinoids (Beurlen 1966) also occur. Recently, insects have been recorded from the top of the Crato Member (Maisey, pers. comm.) where rarely scorpions also occur (Campos, D. de R. B. 1 985). The flora has been described by Lima (1978) and fossil resin is reported by Castro et al. (1970). PALAEOECOLOGY The palaeoecology of the various members of the Santana Formation has been discussed by Beurlen (1971), and by Mabesoone and Tinoco (1973). Most authors conclude that the various units of the Santana Formation show a trend from lacustrine (Crato Member), through restricted marine/ evaporitic basin (Ipubi Member) to marginal marine environments (top of Romualdo Member). There is, however, little agreement on the environment of the concretion bearing horizon. Silva Santos and Valenca (1968) considered this level to be brackish on the basis of the contained fish fauna. Schaeffer (1947) thought the fish represented fully marine forms, while Beurlen (1971) suggested deposition under hypersaline conditions. The salinity of the concretion bearing horizon is difficult to establish, and it is possible that surface water salinity differed significantly from that of the bottom waters. The fish fauna from the concretions contains forms that are well known from fully marine environments, including Lepidotes (Semionotidae), Aspidorhynchus [= Vinctifer ] (Aspidorhynchiformes), Microdon (Pycnodonti- formes), Cladocyclus (Ichthyodectiformes), and Rhinobatos (Rhinobatidae). Cressey and Patterson (1973) have demonstrated the presence of marine parasites on the gills of Cladocyclus sp. but considered also that the fish may have entered fresh water from the sea. The absence of a marine nektonic invertebrate fauna has led workers to consider that the concretion horizon must represent either fresh, brackish, or hypersaline conditions. Hypersaline conditions must be ruled out as the normal environment for the surface waters of the concretion horizon because of the diversity of the nektonic fish fauna, but an influx of hypersaline water may be considered as a possible mechanism for producing a mass mortality of nektonic organisms. Mabesoone and Tinoco ( 1973) concluded that the concretions formed around fishes washed on to a palaeoshoreline by accretion of sediment on to an adipocere coated carcase. They argued that rolling around of the fish carcases caused breakage of the distal portions of the fish fins, hence the common MARTILL: PRESERVATION OF FOSSIL FISH 5 text-fig. 3. Common morphologies of fish bearing concretions from the Santana Formation of north-east Brazil, a, formed around parts of skeleton only, based on FMNH PF 5493 and PF 8333. b, formed on body and skull only, most appendages remain outside concretion, based on FMNH PF 9778, PF 9669, and PF 9779. c, all of specimen enclosed within concretion, concretion shape reflects shape of fish, based on FMNH PF 10372. d, elliptical concretion envelopes entire fish, based on FMNH PF 9616, PF 9631, and PF 9632. e, concretion develops around several fish, based on FMNH PF 9625. preservation of fish bodies only. A detailed investigation of the numerous concretions in the collections of FMNH, BMNH, and AMNH now show that this model cannot be accepted. The concretions The Santana concretions are subspherical bodies composed of laminated carbonate rich sediment, usually enclosing a vertebrate fossil (text-fig. 3). Unweathered concretions usually have a blue/grey core with an outer buff coloured zone. Weathered concretions are buff coloured throughout. The laminae appear as fine alternating light and dark bands varying in thickness from 100 mm to less than 0-25 mm. The sediment type varies between a poorly laminated ostracod biomicrite, to a micro-laminated calcareous clay, both with organic rich laminae. Laminae persist horizontally through the concretion, with slight attenuation due to compaction towards the edges. There is no evidence of accretionary accumulation of sediment comprising the concretion. In all of the concretions I have examined there has been a complete absence of benthic invertebrates except for ostracods, which are often present in vast numbers and probably of a single species. The ostracods are mostly articulated and filled with coarse sparry non-ferroan calcite. Cements vary from sparry calcite in the ostracod rich bio-micrite concretions, to micro-spar in the calcareous clay concretions. All of the concretion cements are early diagenetic non-ferroan calcite. Fracture within the concretions and void spaces within enclosed fossils are frequently filled with coarsely crystalline brown non-ferroan calcite. Larger voids within fossils occasionally show a late generation of ferroan calcite. Incompletely filled voids are lined with white rhombs of calcite approximately 100 mm across (text-fig. 4). 6 PALAEONTOLOGY, VOLUME 31 Skeletal elements Sediment invaginations Gut infills Early brown calcite Fringing calcite Late sparry calcite Voids text-fig. 4. Cross-section through three-dimensional specimen of Rhacolepis sp., FMNH PI 2174, showing geopetal collapse of axial skeleton, upward development of gas cavities, and several generations of calcite cements. Pyrite is abundant in unweathered concretions as patchily distributed crystal aggregates scattered throughout the concretion and lining cavities within trabecular bone. In some cases pyrite is found lining thin septarian cracks, and clearly pre-dates the non-ferroan calcite fills. Phosphate occurs in some concretions as a replacement for soft tissues and as a surface coating of bones (the bones themselves are phosphatic, but this is original bio-genic phosphate and remains relatively unaltered) (PI. 3, figs. 1-4). Phosphate also occurs in coprolites. The phosphate, a cryptocrystalline francolite, pre-dates all other diagenetic mineral phases. It is found replacing muscle fibres, some of which remain in myomeres (text-fig. 5). Some myomers clearly began to decompose before phosphatization occurred and appear as a mass of disordered muscle fibres (see PI. 1, figs. 2 and 3). A vertical thin section of Notelops brama (AMNH 1 1753, PI. 4, fig. 5) shows the upper surface of the specimen is devoid of scales. Body segments of muscle fibres (myomeres) are preserved in phosphate, and are in direct contact with the overlying sediment. Thus, phosphatization must have occurred prior to burial, as no sediment has invaded the exposed body cavity or tissues. This is strong evidence that phosphatization occurred extremely early. Accessory minerals within the concretions include sphalerite, which is usually associated with bone, baryte, and malachite, the latter probably being derived from a sulphide precursor. Sections through concretions containing fishes with ruptured body walls, reveal gas escape structures through the concretion. These emanate from the body cavity of the fish. In some the body wall of the fish has parted from the enclosing concretion as the gas pressure within the fish was released (PI. 4, fig. la, b ). There is some plastic deformation of the sediment above the gas cavity due to this pressure release. Rarely, small quantities of sediment have invaded the body cavity of some EXPLANATION OF PLATE 1 Figs. 1-4. Carbonate concretions with fish from the Santana Formation (?Aptian), Brazil. 1, Brannerion vestitum Jordan and Branner, FMNH PF 9607, specimen showing preservation of entire trunk musculature in intact myomeres, xf. 2, Notelops brama (Agassiz), AMNH 11753, muscle fibres preserved, but intact myomeres restricted to central core of trunk, x^. 3, N. brama (Agassiz), FMNH PF 9626, all muscle fibres preserved, but no myomeres remain intact, xf. 4, Rhacolepis sp., FMNH PF 10771, multi-specimen concretion with orientated, articulated fish, x f. PLATE I MARTILL, fish preservation PALAEONTOLOGY, VOLUME 31 text-fig. 5. Diagrammatic representation of Brannerion vestitum Jordan and Branner, FMNH PF 9607 (see also PI. 1, fig. 1), showing musculature preserved in myomeres. Note the outline of the fish is an artefact of the preparator as indicated by the posterior continuation of the axial skeleton beyond the outline of the ‘caudal fin’. specimens. More commonly, the surrounding sediment has remained intact with little or no invasion of the body cavity, indicating some coherence of the sediment at a very early stage. Slight compaction of the sediment within the concretion around enclosed fossils is indicated by the draping over of laminae. The formation of concretions began during the decomposition of the vertebrate carcases. Gas pressure within the carcases allowed small (up to 40 cm) fishes with intact body walls to remain three-dimensional until the surrounding sediment had become partly lithified. The sediment probably behaved like a stiff gel. Eventually gas pressure within the body cavity increased sufficiently to explode from the body cavity. The sediment was soft enough to allow upward escape of the gas, but was sufficiently firm to resist collapsing into the cavity of the fish. The diagenetic history of concretion formation is summarized in text-fig. 6. In many fish specimens, scales from the upper surface of the fish have become detached and lie a few millimetres from the carcase. This indicates that the fish must have begun to decompose while lying exposed on the sea floor. Under normal conditions, a carcase exposed for only a few days would be scavenged or dispersed by current activity. Low oxygen levels may have reduced scavenging to a minimum, but the abundance of articulated ostracods indicates well-oxygenated bottom water was present. Current activity must have been almost zero as even the smallest of fish scales and fin rays EXPLANATION OF PLATE 2 Figs. I 6. Muscle preservation in Brazilian teleost fishes. 1, Brannerion vestitum Jordan and Branner, FMNH PF 9607, intact muscle myomeres with well-preserved muscle fibre, x 7. 2, SEM of muscle fibres from Rhacolepis sp. prepared in dilute acetic acid, BMNH P62145, x 80. 3, surface detail of fig. 2 showing granular texture produced by masses of phosphatic spherulites, x 615. 4, rib of Rhacolepis sp. with surface coating of phosphatic spherulites, x 780. 5, mass of phosphatic spherulites replacing muscle fibre of Rhacolepis sp., x 12 000. 6, detail of phosphatic spherulites showing individual lath-like crystallites, x 64 000. PLATE 2 MARTILL, fish preservation 10 PALAEONTOLOGY, VOLUME 31 text-fig. 6. Geochemical model for the mass mortality of the Santana biota and the preservation of soft tissues by phosphate authigenesis under oxic conditions. MART1LL: PRESERVATION OF FOSSIL FISH II remain articulated, and detached scales only lie a few millimetres from the main portion of the carcase. The few separated scales were presumably detached due to the early escape of decomposition gasses. Cypridid ostracods have been used by Bate ( 1 972) to show that the fauna is non-marine, but marine cypridid ostracods are known (e.g. Suzinia, Moore 1961). Bate (1972) also considered that the ostracods must have been feeding on the fish carcases, but thin sections through ostracod bearing fish containing concretions show that ostracods were present in the sediment before and after the fish carcases arrived on the sea floor. No ostracods occur within the body cavity of the fish. The general lack of other benthic organisms in the concretions may be due to diagenetic removal, but most likely the restricted benthos was the result of slightly hypersaline bottom water. EODIAGENETIC ENVIRONMENT Prior to the event which killed large numbers of fishes in the Santana basin, bottom waters were probably well oxygenated and hypersaline. Organic rich laminae within the concretions indicate the presence of a widespread prokaryotic mat covering the sea floor. Presumably such an extensive mat could survive due to a lack of grazing invertebrates inhibited by the hypersaline water. A similar situation has been postulated for the Solnhofen Limestone (Jurassic) of Germany (Keupp 1977). The mat may also have had a role in restricting colonization of the sea floor by hypersaline tolerant infauna and epifauna, but provided a suitable substrate for salt tolerant ostracods. After the mass mortality of fishes living at all levels in the water column, many thousands of fishes descended to the sea floor. It is also likely that many floated for a prolonged period and drifted away from the area of the event to be washed on to nearby shores, as happens from time to time on the west coast of Africa (Brongersma-Sanders 1949). Many of the fish carcases that descended to the sea floor may have been overgrown by the prokaryotic mat, or became entangled in it by their fins. This would prevent the carcase from rising to the surface as decomposition gases slowly built up inside the body cavity. The accumulation of vast numbers of fishes on the sea floor undergoing decomposition may have rapidly depleted the level of dissolved oxygen in the water column. The bottom water, especially adjacent to the carcases may have become anoxic. This could have the effect of killing the benthic ostracods, and of inhibiting continued growth of the prokaryote mat. The presence of a mat trapping a carcase may have allowed the build up of a variety of decomposition products. If C02 was allowed to build up initially, the pH of the environment immediately surrounding the fish would decrease slightly. This would have two major effects, first to inhibit carbonate precipitation, and secondly to enhance the precipitation of francolite. Low pH may also have been maintained by the oxidation of sulphides migrating upwards from the sulphate reduction zone (Coleman 1985). Phosphate is relatively rare in sea water, but it builds up in sediment pore waters due to microbial breakdown of organic material (Berner 1980). Concentrations of phosphate in the sediment increase with depth, and a concentration gradient exists between the deeper buried sediment and the sediment/water interface. Thus, dissolved phosphate diffused into the carcases both from the sea water and from pore water. This and additional phosphate, liberated by bacteria feeding on proteins, RNA (Lucas and Prevot 1984), and other phosphorous rich bio-molecules within the carcase was unable to remain in solution, and was rapidly precipitated as francolite micro-spheres (text-fig. 7). Presumably larger authigenic crystals did not grow due to the presence of inhibitory magnesium (McConnell 1973). This initial low pH micro-environment persisted only briefly, and was brought to an end by the production of vast amounts of ammonia which rapidly increased the pH of surrounding waters. In this way phosphatization of the soft tissues took place within a few seconds of the bacteria metabolizing the tissue (see text-fig. 5), and probably only a few hours after the dead fish arrived on the sea floor. Phosphatization of soft tissues did not take place in all of the fish specimens, but in most specimens the same type of phosphate occurs on bone surfaces. Petrographic study of the bone suggests that no 12 PALAEONTOLOGY, VOLUME 31 text-fig. 7. Diagrammatic representation of muscle fibre bacterial breakdown, with subsequent replacement of fibres by francolite micro-spheres, scale bar 20 /mi. alteration of the bone has taken place, and therefore bone phosphate has not been a source of phosphate for the preservation of the soft tissues. In some of the fish specimens (PI. 1, figs. 2-4) almost the entire musculature of the trunk has been preserved. It is difficult to see how so much phosphate can be rapidly dumped unless vast quantities of phosphate-enriched pore water are flushed through the system. It is equally difficult to explain why some specimens have no soft tissues preserved at all, although this latter observation may be due to derivation of concretions from different localities. During the initial phase of decomposition with accompanied phosphatization, sedimentation continued. Further decomposition beneath a few millimetres of sediment by anaerobic bacteria (sulphate reducers), continued to liberate ammonia which maintained high pH in pore water and created an environment for the early precipitation of carbonate to begin concretion formation. This prevented compaction of the smaller fish under the increasing weight of overburden. EXPLANATION OF PLATE 3 Figs. 1-4. Soft tissues from Brazilian Cretaceous teleost fishes. 1, Rhacolepis sp., LEIUG 94515. a , secondary lamellae of gills attached to gill ray, x 50; h , oblique view of secondary lamellae, x 100; c, detail of surface of secondary lamellae displaying rope-like texture, x 930. 2, Rhacolepis sp. Phosphatized stomach wall, removed from BMNH P62101. SEM stub now LEIUG 94516. a , portion of stomach wall, x 30; h, detail of reticulated surface, x 200. 3, possible nerve fibre entering section of fibrous muscle, x 300, LEIUG 94517. 4, ?phosphatized bacteria on surface of fish bone, x 59 600. All photographs are of acetic acid prepared specimens. PLATE 3 MARTILL, fish preservation 14 PALAEONTOLOGY, VOLUME 31 Subsequent burial of the fish carcase passed it from an oxidizing or suboxic zone to the sulphate reducing zone. Here precipitation of iron monosulphides with later conversion to pyrite assisted in cementing the surrounding sediment, and produced authigenic pyrite on the surfaces of bones. At the same time precipitation of non-ferroan carbonates centripetal to the carcase formed pre-compaction concretions with non-ferroan sparry calcite void fills. With increased depth of burial the concretion enclosed carcase passed into the methanogenic zone. Continued calcite precipitation in voids within the body cavity produced fringes of ferroan calcite, but no ferroan calcite was precipitated on the fringes of the concretions, possibly because the pH was too low beyond the concretion limits. MASS MORTALITY The great abundance of fish at the concretion horizon(s) of the Santana Formation is suggestive of one or more mass mortality events. The number of concretion horizons within the Romualdo Member is not recorded, but it is likely that there are several concretion bearing layers over the area of the outcrop. Most concretions contain only a single fish skeleton, but a few concretions contain several fish (PI. 1, fig. 4), often of the same taxon, and of similar size. This is suggestive of the death of an entire shoal rather than the accumulation of individual fish killed in separate events through time. The cause of the mass mortality events cannot easily be established, but it seems likely that a catastrophic change of chemistry or temperature could have occurred in the surface water. In marginal marine environments changes in salinity may occur relatively fast. An upward migration of the halocline to the surface could result in the sudden death of organisms in the nekton. Sudden changes of temperature in surface waters periodically kill vast numbers of fishes in the Gulf of Mexico (Gunter 1947), as can blooms of surface living micro-organisms (Brongersma-Sanders 1949), either by clogging the gills and causing asphyxiation or by producing toxins in the water column. DISCUSSION The importance of prokaryotic scums on sediment surfaces is not yet fully appreciated, but it is becoming clear that thin microbial films of cyanobacteria play an important physical and biochemical role in sediment diagenesis and fossilization processes. In the finer grained Santana Formation concretions, fine carbonaceous laminae lie between equally thin microcrystalline calcareous laminae. Sheets of this carbonaceous material can easily be extracted from the concretion by dissolution in dilute acetic acid. The thin sheets are dark brown, amorphous flakes which resemble modern dead bacterial scums found on dried lake beds. Some poorly preserved filaments can be seen after treatment of the mat with concentrated nitric acid. Such microbial films play two important roles. First, they prevent colonization of the sea floor by epibenthos and endobenthos (Keupp 1977) and as a result protect the sediment from bioturbation, and secondly, such mats provide a plentiful source of organic matter for feeding autotrophic bacteria EXPLANATION OF PLATE 4 Figs. 1-5. Thin sections of Santana Formation concretions, la, section through three-dimensional specimen of Rhacolepis sp. showing geopetal collapse of axial skeleton (black arrow) and gas escape structure. The thin laminae above the fish have collapsed as gas escaped, FMNH PF 10765, x 1-8. 16, detail of gas escape structure in same specimen showing fringing calcites, x 4-5. 2, section through laminated concretion showing numerous articulated cypridid ostracodes, FMNH PF 11846a, x 10. 3, section through vertebral centra of a flattened Enneles sp. showing that despite flattening the bone has remained uncrushed, FMNH PF 11846a, x 10. 4, stained section through cavity in trunk of Rhacolepis sp. showing several generations of carbonate cement and euhedral calcite crystals lining the cavity, FMNH PF 10765, x 5. 5, phosphatic fringes on surface of bone in Notelops brama , AMNH 11753, x 10. PLATE 4 MARTILL, fish preservation r'S*- 16 PALAEONTOLOGY, VOLUME 31 within the sediment. Consequently, they may fuel complex diagenetic reactions. In the Santana Formation such reactions have enabled delicate soft tissues to be preserved in a three-dimensional state. Seilacher et al. (1985) point out that today cyanobacterial films are mainly restricted to hypersaline environments. The low diversity of the Santana benthos (mainly smooth shelled ostracods) may be due to increased salinity; thus a partly stratified water column with hypersaline bottom water may have existed within the basin, which could be used to explain the rarity of largely benthonic fish such as Rhinobatos and the mollusc eating pycnodonts. This palaeoecological model should be used to assist in the discovery of other sites of exceptional fossil preservation. Acknowledgements. Special thanks to Dr C. Patterson for bringing the Santana concretions to my attention as a source of fossilized soft tissue. Many thanks also to Dr Lance Grande (FMNH) and Dr J. Maisey ( AMNH) for the loan of specimens in their care. I also take this opportunity to thank my wife Jill for patiently typing the manuscript. George McTurk and Rod Branson kindly operated the SEM and discovered numerous interesting objects when they modified the stage to take an entire concretion. The main part of this work was carried out in the Geology Departments of the Field Museum of Natural History, Chicago, during tenure of a Harkness Fellowship, and at the University of Leicester during tenure of a University of Leicester Research Scholarship. REFERENCES agassiz, L. 1841. On the fossil found by Mr Gardner in the province of Ceara in the north of Brazil. Edin. New Philos. J.. 30, 82-84. - 1844. Sur quelques poissons fossiles du Bresil. C.R. hebd. Seanc. Acad. Sci., Paris , 18, 1007-1015. bate, r. h. 1972. Phosphatised ostracods with appendages from the Lower Cretaceous of Brazil. Palaeontoloqv, 15 (3), 379A-393. benmore, r. a., coleman, M. L., and mcarthur, J. M. 1983. Origin of sedimentary francolite from its sulphur and carbon isotope composition. Nature, Land. 302, 516-518. berner, r. a. 1968. Calcium carbonate concretions formed by the decomposition of organic matter. Science, N.Y. 159, 195-197. 1980. Early diagenesis, a theoretical approach, 241 pp. Princeton University Press. beurlen, k. 1962. A geologia da Chapada do Araripe. An. Acad, brasil, Cienc. 34 (3), 365-370. 1963. Geologia e stratigraphia da Chapada do Araripe. XVII Cong. Bras. Geol. Publ. SUDENE, 1-47. 1966. Novos equinoides no Cretaceo do Nordeste do Brazil. An. Acad, brasil, Cienc. 38 (3-4), 455-464. 1971. As condicoes ecologicas e faciologicas da formacao Santana na Chapada do Araripe (Nordeste do Brasil). Ibid. 43, 411-415. and barreto, a. 1968. Noticia sobre uma tartaruga fossil da regiao do Araripe. SUDENE-Dep. Rec. Nat. Bol. Est. 4, 27-37. braun, o. p. G. 1966. Estratigrafia dos sedimentos da parte interior da regiao Nordeste do Brasil (Bacias de Tucano-Jatoba, Mirandiba e Araripe). D.N.P.M., Div. Geol. Miner. Bol. 236, 1-15. brito, i. m. 1983. The Brazilian Cretaceous. Zitteliana, 10, 277-283. Miinchen. 1984. The upper Lower Cretaceous in Brazil, its divisions and boundaries. An. Acad, brasil, Cienc. 56 (3), 287-293. brongersma-sanders, m. 1949. On the occurrence of fish remains in fossil and recent marine deposits. Bijdr. Dierk. 28, 65-76. buffetaut, e. 1981. Die biogeographische Geschiste der krokodilier, mit Beschreibung einer neven Art, Araripesuchus wegeneri. Geol. Rdsch. 70 (2), 611-624. Stuttgart. — and taquet, p. 1979. An early Cretaceous terrestrial crocodilian and the opening of the South Atlantic. Nature, Lond. 280, 486-487. buisonje, p. h. de. 1980. Santandactylus brasiliensis nov. gen., nov. sp., a long necked, large pterosaurian from the Aptian of Brazil. Proc. K. ned. Akad. Wet. B 83 (2), 145-172. Amsterdam. burnett, w. c. 1977. Geochemistry and origin of phosphorite from off Peru and Chile. Bull. geol. Soc. Am. 88, 813-823. campos, d. de a. 1983. Un novo pterosauro do Chapada do Araripe. An. Acad, brasil, Cienc. 55 (1), 141-142. 1985. Ocorrencia de un novo Arcossauro na Chapada do Araripe. Ibid. 57 (1), 140-141. and wenz, s. 1 982. Premiere decouverte de Coelacanthes dans le Cretace inferieur de la Chapada do Araripe (Bresil). C.R. hebd. Seanc. Acad. Sci., Paris, 11, 294. MARTILL: PRESERVATION OF FOSSIL FISH 17 ligabue, G. and taquet, p. 1984. Wing membrane and wing supporting fibres of a flying reptile from the Lower Cretaceous of the Chapada do Araripe (Aptian, Ceara State, Brazil). Ill Symp. Mesoz. Terrestr. Ecosystems , Short papers: 37-39. Tubingen. Campos, de r. b. 1985. Primeiro registro fossil de scorpionoidea na Chapada do Araripe (Cretaceo Inferior), Brasil. An. Acad brasil, Cienc. 57 (1), 136-137. castro, c., menor, e. a. and campantra, v. a. 1970. Descoberta de resinas fosseis na Chapada do Araripe Municipio de Porteiras, Ceara. Univ. Fed. Pernambuco , Dep. Petrol. Ser. C. Not. Prev. 1 (1), 1-11. COLEMAN, m. L. 1985. Geochemistry of diagenetic non-silicate minerals: kinetic considerations. Phil. Trans. R. Soc. A.315, 39-56. cressey, r. and patterson, c. 1973. Fossil parasitic copepods from a Lower Cretaceous fish. Science , N.Y. 180, 1283-1285. d’erasmo, g. 1938. Ittioliti cratacei del Brasile. Atlas R. Accad. Sci. Fis. Mat ., 3a, Ser. 1 (111), 1-44. Gardner, G. 1841. Geological notes made during a journey from the coast into the interior of the province of Ceara, in the north of Brazil, embracing an account of a deposit of fossil fishes. Edin. New. Philos. J. 30, 75-82. gunter, G. 1947. Catastrophism in the sea and its paleontological significance, with special reference to the Gulf of Mexico. Am. J. Sci. 245 (11), 669-676. Jordan, D. s. 1923. Peixes cretaceos do Ceara e Piauhy. Serv. Geol. Min. Bras. Monogr. 3, 4-97. and branner, j. c. 1908. The Cretaceous fishes of Ceara, Brazil. Smithson, misc. Colins , 25 (1), 1-29. keupp, h. 1977. Ultrafazies und Genese der Solnhofener Plattenkalk (Oberer Malm, Sudliche (Frankenenalb). Abh. naturhist. Ges. Niirnberg. 37, 1-128. leonardi, G. and borgomanero, G. 1983. Cearadactylus altrox nov. gen., sp.: nova pterosauria (pterodactyloidea) da Chapada do Araripe, Ceara, Brasil. VIIII Cong. Brasil , Paleonl. Resumos dos communicacoes, 1 17. Rio de Janeiro. lima, M. R. 1978. Palinolologia da Formacao Santana (Cretaceo do Nordeste Brasil ). Univ. Sao Paulo-Tese de Doctoramento (unpublished), 1-335, pis. 1-27. 1979. Paleontologia da Formacao Santana (Cretaceo do Nordeste do Brasil): Estagio actual de conheci- mentos. An. Acad, brasil, Cienc. 51 (3), 545-556. LUCAS, J. and prevot, l. 1984. Synthese de l’apatite par voie bacterienne a partir de nratiere organique phosphatee at de divers carbonates de calcium dans de eaux douce at marine naturelles. Chemical Geology, 42, 101-118. mabesoone, j. m. and tinoco, i. m. 1973. Palaeoecology of the Aptian Santana Formation (North eastern Brazil). Palaeogeogr. Palaeoclimatol. Palaeoecol. 14, 97-118. mcconnell, D. 1973. Apatite, its crystal chemistry, mineralogy, utilization, and geologic and biologic occurrences, 1 1 1 pp. Springer Verlag, New York and Wien. moore, R. c. 1961. Treatise on Invertebrate Paleontology, Part Q, Arthropoda 3, 442 pp. Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas. price, L. i. 1959. Sobre um crocodilideo Notossuquio do Cretacico brasileiro. D.N.P.M., Div. Geol. Miner. Bol. 188, 1-155. 1971. A precenca de pterosauria no Cretaceo Inferior de Chapada do Araripe, Brazil. An. Acad, brasil, Cienc. 43, suppl., 451 461. 1973. Quelques Amphychelidia no Cretaceo Inferior do Nordeste do Brazil. Rev. Bras. Geoc. 3 (2), 84-95. Schaeffer, b. 1947. Cretaceous and Tertiary actinopterygian fishes from Brazil. Bull. Am. Mus. nat. Hist. 89, 1 40. seilacher. A., reif, w. e. and westphal, f. 1985. Sedimentological, ecological and temporal patterns of fossil Lagerstatten. Phil. Trans. R. Soc. Fond. B.311, 5-23. silva, m. d. da. 1970. Notas preliminares sobre o genero Cypridae em Exu. Univ. Fed. Pernambuco ser. Cient. Paleont. 5, 1-13. 1978. Ostracodes da Formacao Santana (Cretaceo Inferior)- Grupo Araripe- Nordeste do Brasil. I. Nova especies du Genero Bisulcocypris. An. XXX Cong. Bras. Geol. 2, 1014-1022. silva santos, r. da. 1945. Revalidacao de Aspidorhynchus comptoni Agassiz do Cretaceo do Ceara, Brasil. D.N.P.M. (Div. Geol. Miner.). Not. Prelim. Est. 29, 1-10. — 1947. Uma redescricao de Distilbe elongatus com algumas consideracos sobre o genero Distilbe. Ibid. 42, 1-7. 1950. Anaedopogon, Chiromystis e Ennelichthys como sinonimos de Cladocyclus, da familia Chirocentridae. An. Acad, brasil, Cienc. 22 (1), 123-134. 18 PALAEONTOLOGY, VOLUME 31 silva santos, r. da. 1960. Leptolepis diasii , nova peixe fossil da Serra do Araripe, Brasil. Univ. Fed. Pernambuco , Inst. Pesq , Agron. Arg. 5, 21-31. — 1968. A paleoictiofauna da Forraacao Santana-Eusalachii. An. Acad, brasil, Cienc. 40 (4), 491-498. — 1970a. A paleoictiofauna da Formacao Santana. Ffolostei: Familia Girodontidae. Ibid. 42 (3), 445-452. — 1970b. Novo genero e nova especie de Elopideo da Bacia sedimentar do Araripe. Ibid. 42 (4), 853 -R. — 1974. Paleoictifaunula da Formacao Codo (reultados da expedicao paleontologica e stratigrafica da Academia Brasieira de Ciencias a bacia do Parnaiba). Ibid. 46 (3-4), 710. — and valenca, j. o. 1968. A Formacao Santana e sua paleoictiofauna. An. Acad, brasil , Cienc. 40, 339-360. wellnhofer, p. 1977. Araripedactylus dehmi nov. gen., nov. sp., ein neuer flugsaurier aus der Unterkreide von Brasilien. Mitt. Bayer. Staatsslg. Palaont. hist. Geol. 17, 157-167. Miinchen. 1985. Neue pterosaurier aus der Santana Formation (Apt.) der Chapada do Araripe, Brasilien. Palaeonto- graphica Abt. A. 187 (4-6), 105-182. Stuttgart. — buffetaut, E. and GIGASE, p. 1983. A pterosaurian notarium from the Lower Cretaceous of Brazil. Palaont. Z. 57 (1-2), 147- 157. Stuttgart. Typescript received 29 May 1986 Revised typescript received 2 April 1987 DAVID M. MARTILL Department of Earth Sciences Open University Milton Keynes MK7 6AA MIDDLE CRETACEOUS WOOD FROM THE NANUSHUK GROUP, CENTRAL NORTH SLOPE, ALASKA by judith t. parrish and Robert a. spicer Abstract. Analysis of growth rings in Albian and Cenomanian (late Cretaceous) coniferous wood from the North Slope of Alaska (74-85° N. palaeolalitude) has shown that tree growth was rapid and steady during the growing season, resulting in wide growth rings and few false rings, that narrow late wood, as little as one cell wide, indicates that tree growth ceased abruptly at the end of the growing season owing to rapid onset of winter darkness, and that inter-annual growth was variable. Water being in abundant supply, this variability was likely to have been caused by fluctuations in soil drainage or in summer mean temperatures. In mid-Cretaceous times, the North Slope of Alaska was situated 75-85° N. (Smith and Briden 1977; Ziegler et al. 1983) and was probably the nearest land to the North Pole. The mid-Cretaceous South Pole was in Antarctica, but no part of the continent that was within 5° of the pole is currently accessible. Mid-Cretaceous global climate generally is regarded to have been warm relative to the present, particularly at high latitudes (Savin 1977; Barron 1983). Therefore, the North Slope is a critical region for Cretaceous climatic studies. The Albian-Cenomanian Nanushuk Group of the North Slope contains a rich leaf megafossil flora (Smiley 1966, 1967, 1969a, 6; Scott and Smiley 1979; Spicer 1983) and abundant fossil wood. In 1985 we collected 45 samples of this wood from several localities along the Colville River (text-fig. 1; Spicer and Parrish 1986). They are coniferous and comprise three taxa, Xenoxylon latiporosum (Cramer) Gothan and two previously undescribed taxa. Thirteen were preserved well enough to yield some anatomical information, and seven yielded sequences of seven or more measurable growth rings. Certain characteristics of the growth rings provide information about the palaeoclimate. Although this information is not conclusive by itself, analysis of the growth rings supports our previous conclusions about the polar climate during the mid-Cretaceous. PREVIOUS WORK Fossil woods. Studies on the palaeoclimatic implications of growth rings are few for wood older than Pleistocene, although many pre-Pleistocene fossil woods, including those from northern Alaska (Arnold 1952), have been described and named (Jefferson 1982; Creber and Chaloner 1984, 1985). Palaeoclimate-orientated studies include Francis (1984) on Purbeck wood, Isle of Portland, England, Jefferson (1982) on early Cretaceous wood, Alexander Island, off the Antarctic Peninsula, Francis (1986) on Cretaceous and early Tertiary wood of the Antarctic Peninsula, and Creber and Chaloner (1984, 1985) who surveyed Cretaceous and Tertiary wood worldwide. Polar palaeoclimatology. Until recently, only qualitative information existed on Cretaceous polar temperatures. Based on the leaf megafossil assemblages and noting particularly the presence ofcycads, Smiley (1969a) implied a warm-temperate climate for the North Slope but did not define 'warm-temperate’. A similar conclusion was reached by May and Shane (1985), using palynofloral assemblages, and Roehler and Strieker (1984), based on the presence of dinosaurs and wide growth rings in fossil woods. Following Savin (1977), Barron (1983) estimated mean annual polar temperatures of 0-15 °C, based on extrapolations from seawater temperatures at lower palaeolatitudes as determined from isotopic analyses. This estimate encompasses the range of climates from subpolar to subtropical (cf. Wolfe 1979), but was the most reasonable quantitative estimate at the time. (Palaeontology, Vol. 31, Part 1, 1988, pp. 19-34, pis. 5-6.| © The Palaeontological Association 20 PALAEONTOLOGY. VOLUME 31 Subsequently, we added to and reassessed the leaf megafossil data of the Cretaceous North Slope (Spicer and Parrish 1986). Using leaf-margin analysis of the angiosperms, we concluded that the mean annual temperature in the Cenomanian was 10 + 3°C, or cool temperate in Wolfe’s (1979) scheme. GEOLOGY AND DEPOSITIONAL ENVIRONMENTS OF THE WOOD The fossil plants are found in nonmarine rocks of the Nanushuk Group (text-fig. 2), although associated marine rocks contain rare leaf remains and abundant finely disseminated carbonaceous debris that was derived from terrestrial vegetation. The Nanushuk Group consists of a fluvial-deltaic assemblage (Ahlbrandt et al. 1979; Huffman et al. 1985) and comprises several depositional facies along the Colville River. These and the environments of the wood and associated fossil plant material are described in Table 1. DESCRIPTION OF THE WOOD 1. Xenoxylon latiporosum (Cramer) Gothan (PI. 5, figs. 1-5) Transverse section. Secondary wood consists of tracheids and parenchymatous rays only. Cross-sectional area of early-wood tracheid lumina typically is 2100 pm2. Transition from early wood to late wood is abrupt; late-wood zone is narrow. Tracheids tend to be rectangular in cross-section with early-wood walls 5 pm thick. Rays are uniseriate, 10 12 pm wide. Resin ducts are absent. Radial longitudinal section. Bordered pits on the radial sides of the tracheids are vertically flattened, uniseriate, and contiguous, measuring 15 /im vertically and 20-23 pm horizontally. Pit apertures are round or slightly elliptical and typically 8 pm in diameter (or 4 x 8 pm where elliptical). Pits are not uniformly distributed along the tracheid wall but are in groups of variable number separated by more or less smooth tracheid walls. Transverse septae within the tracheids are common and spaced at intervals typically ranging from 60 pm to 90 pm. The rays PARRISH AND SPICER: ALASKAN CRETACEOUS WOOD 21 text-fig. 2. The Nanushuk Group, North Slope, Alaska (modified from Huffman 1985). Position of Albian-Cenomanian boundary based on Foraminifera (Sliter 1979) and palaeobotanical evidence (Spicer and Parrish 1986). The Grandstand and Tuktu Formations contain Albian molluscs and the Ninuluk Formation contains the Cenomanian bivalve, Inoceramus dunveganensis (Detterman et al. 1 963; Chapman et cd. 1 964; Brosge and Whittington 1966). table 1. Depositional environments, lithologies, and palaeontology of the Nanushuk Group. Depositional environments Lithology Palaeontology Fluvial channels: distributaries Very fine to fine, broad, lenticular sand- Wood; G, platanoid A, E main channels stone occasionally cross-bedded, ripple- marked Narrow, lenticular very fine sandstone to Wood, commonly flow orientated Overbank deposits pebble conglomerate Very poorly bedded, mottled brown, grey, Abundant root casts, rhizome systems of Swamps yellow mudstone; thin coal lenses; dark red, fossiliferous ironstone concretions, nodular or tabular, probably represent- ing wetter areas Coal, few cm to > 3 m thick Equisetites , diverse leaf assemblages in- cluding Co, G, Cy, F, A; wood, includ- ing stumps in growth position; clams, turtle Wood; impressions of Podozamites leaves Ponds Poorly consolidated to indurated grey clay- in paper coal; seat earths contain Equi- setites rhizomes Abundant leaves, including entire abscissed stones, weathering light yellow Co shoots, cones, twigs. A, G Leaves: A, angiosperms; G, ginkgophytes; Co, conifers; Cy, cycadophytes; F, ferns; E, Equisetites. 22 PALAEONTOLOGY, VOLUME 31 appear to be entirely parenchymatous. Cross-field pits are large and fenestrate with more or less square to rhombic outlines. Usually only one pit per cross field exists but rarely two are present. Tangential longitudinal section. Rays are uniseriate, typically 12-15 cells high (but extremely variable from 1 to 30 cells high), each cell measuring 20-21 gm in height. Pits on the tangential side of the tracheids are rare and somewhat smaller than on the radial side. Comments. Wood of this kind is widespread at high northern latitudes in the Mesozoic. Although the wood possesses pineoid cross-field pitting, it cannot be assigned to any living family of conifers (Arnold 1952). 2. Taxon A (PI. 6, figs. 1-3) Transverse section. Secondary wood consists of tracheids and ray parenchyma. Transition from early wood to late wood is abrupt with minimal late wood. Early-wood cell lumina typically are 50 gm across, usually more or less square in section. Cell walls typically are 8- 10 jum thick. Resin ducts apparently are absent. Radial longitudinal section. Tracheid pitting is mixed uniseriate and biseriate. Uniseriate pits typically measure 23 /nn in height and 26-28 gm in width (occasionally up to 38 x 23 / 0 This theorem shows that curvature k and torsion x respectively indicate the revolution rates of the unit tangent vector and the tangent plane with the change of arc length X. On the other hand, if the differentiable functions k(X) (> 0) and t(2) are given in the interval I, there is a definite regular space curve, in which the curvature and torsion are given by k and r with arc length X, respectively. If we regard the curves, which fit each other by revolution and parallel dislocation, as identical, there is only one space curve fitting the given condition. Generally speaking, there are many formulae for expressing a space curve, e.g. z = /(x, y); x = / (f), y = g(t\ z = h(t), and so on. But these equations show different forms in accordance with the setting of a coordinate system. On the contrary, k(X) and x{X) with a parameter of arc length have geometrical significance, and define a unique space curve independent of any coordinate axis. Therefore, k = k(X) and t = r(2) can be regarded as equations of a space curve. They may be called ‘natural equations of a space curve’. Standardization of moving frame The tube model for real coiling has an increasing whorl radius r throughout growth. To describe the pattern of heteromorph coiling geometry more precisely, it is necessary to consider not only the locus of the tube centre but also the tube radius. It is, therefore, impossible to use the natural equations of a space curve directly for a growing tube. In order to establish a method of differential geometric analysis for a coiling shell, it is necessary to devise some modifications of these parameters X, k, and x. In a Frenet frame, each coordinate axis is defined as a unit vector t, n, or b. This frame is not suitable for a coiling tube because it is independent of the tube size. In this case, the unit length of a moving frame at an arbitrary stage should be defined as a length proportional to the tube radius. Therefore, I adopt three dimensional vectors rt, rn and rb as a standardized moving frame instead of Frenet’s. These may be adequate standards to estimate the mode of coiling corresponding to shell size and growth stage. Use of these vectors enables arc length, curvature, and torsion to be standardized as set out below. Growth stage s In the description of any growth pattern, time might be considered the most appropriate parameter. Among fossil organisms, however, time scale in the growing process cannot be detected. In the natural equations of a space curve, curvature k and torsion x are expressed as the functions of arc length X. But arc length is not always a suitable parameter for expressing the growth of a coiling shell because it is independent of size. The description of growth should reflect an organism’s size at any time. The concept of relative growth developed by Huxley (1932) is based upon this principle. If the concept of relative growth is applied to the tube model, scale can be defined differentially. For the analysis of tube coiling, I introduce a parameter s that indicates the growth stage of the coiling tube instead of X. In the time interval from t to t + dt, the increase of the growth stage ds(t) is related to the increase of arc length dX(t) and tube radius r(t) as follows: d Jt s(f) = 1 d r(t) dt X(t) Radius enlarging ratio E The radius enlarging ratio is based upon the radii at two growth stages. If a slight advance of growth stage from s to s + e produces a change in tube radius from r to r + Ar, then the radius enlarging ratio E is given by: r + Ar d OKAMOTO: GEOMETRY OF HETEROMORPH AMMONOIDS 41 In this definition, £ is a function of s, and prescribes the size of r at the next stage. On the other hand, the value of r is influenced by change in s, by which the radius enlarging ratio E is defined. Note that E and s are determined recurrently. This definition is generally more suitable for describing the mode of growth of a coiling shell. Standardized curvature and torsion The helicoid, gastropod-like model tube in text-fig. 6a appears to have a constant mode of coiling throughout growth, but the curvature k and torsion i calculated along the tube centre are not constant. Although the figure shows a proportional spiral curve, k and t must decrease with growth. The three tubes shown in text-fig. 6b D are instructive: b and c show the same curvature of their centre line but have different radii, while B and d are similar in shape but different in size. Thus d must have a different value of curvature k from b and c. In the tube model, however, it would be more convenient for b and d to be the same in their mode of coiling, and different from c. The curvature and torsion of a tube can be standardized so that the differential parameters are constant in such proportional growth as shown in text-fig. 6a, and so that text-fig. 6b and d have the same values. Therefore, I introduce new parameters standardized curvature C and standardized torsion T instead of k and x. The curvature and torsion of the tube model can be defined as the revolution rate of the standardized moving frame. Finally, the parameters C and T are given by: C = rx, T = rx. By using these parameters, all geometrically similar figures can be expressed as the same coiling pattern. Description of a growing tube In the new tube model, I have now defined three parameters: E, radius enlarging ratio; C, standardized curvature; and T, standardized torsion. These describe the differential characters of a coiling tube in general, and are given by the parameter s, indicating growth stage, as follows: E = E(s), C = C(s), T = T(s ) These three equations describe only the mode of coiling, not its size. For the description of shell size, a constant r0 indicating initial tube radius must be introduced. If the three equations and one constant are given, a unique tube conforming with the conditions is obtained. Finally, the combination of E(s), C(s), T(s), and r0 may be regarded as a natural equation of the tube model. To the extent that a tube grows proportionally, the parameters E, C, and T are constants. Moving frame analysis for the growing tube model One of my main purposes here is to establish a method of analysis and description of the regular but free coiling of heteromorph ammonoids. Text-fig. 7 makes the geometric meaning of the parameters of the growing tube model more explicit. Given a circular generating curve with centre Qs and radius rs text-fig. 6. Four hypothetical tubes, a, a proportional spiral, like a gastropod, b and c have the same curvature of the tube centre line, but a different tube radius, b and d are similar in shape but different in size. 42 PALAEONTOLOGY, VOLUME 31 Differential parameters text-fig. 7. Growing tube model show- ing the three differential parameters E , C, and T. In this theory the three parameters are calculated as a limit value of e = 0. at growth stage s, at the next growth stage s + e the centre of the circular generating curve shifts to Qs+e along a line normal to this circle’s plane. This direction indicates the tangential vector of the standardized moving frame. On the generating curve, there is a special point MGPS which signifies the point of maximum growth at this growth stage. The normal vector of the standardized moving frame indicates the point MGPS from Qs. Then we can define the standardized moving frame (rs t, rs n, rs b) forming a right-hand system at an arbitrary growth stage. Consequently, the generating curve can be explained as follows: 1, during growth of the tube from s to s + £, the centre of the generating curve Qs moves srs in length (6 radians) around the point Os, in the tangent plane; 2, in the normal plane, the maximum growth point MGPS revolves cp radians around Qs; 3, the radius of the generating curve increases from rs to rs + c. In this growth model for a generating curve, the parameters £, C, and T are given by: In £ = (In rs+£ — In rs)/e C = 0/e T =

1, this point would lie inside the generating curve, a state never found in real coiled shells. T represents the revolution rate of MGP (maximum growth point) in the generating curve, to which there is no theoretical limit. Whether coiling is dextral or sinistral is determined by the sign of T, and if T = 0 the tube is planispiral. The moving frame in this growing tube model means a coordinate system that is always situated at the last generating curve, i.e. at the aperture of a coiling shell. In other words, this frame travels along the centre line of the coiling tube throughout its growth. Therefore, moving frame analysis is a method, using three parameters £, C, and T, that describes how the frame behaves in space. By applying this method to actual coiling shells, it is possible to analyse and express not only complex shell coiling but also any ontogenetic changes in mode of coiling. One of the striking merits of moving frame analysis in the growing tube model is its ability to determine uniquely the changing pattern of £(s), C(s), and T(s), corresponding to each growth stage. Traditional methods, using a fixed coordinate OKAMOTO: GEOMETRY OF HETEROMORPH AMMONOIDS 43 text-fig. 8. Three-dimensional block diagram showing the spectrum of hypothetical shells, when the three differential parameters E, C, and T are changed. system, require extremely variable formulae according to slight differences of the fitting model. Furthermore, even for one and the same tube model, many equations are possible, dependent on different definitions of the axis or coordinate system. In the growing tube model, by contrast, any gently curved tube can be visually expressed by a diagram showing the change of the three parameters during growth. The more accurate the computer graphics representation becomes, the more closely a graph of the three parameters must approach a single pattern. 44 PALAEONTOLOGY, VOLUME 31 text-fig. 9. Spectrum of computer-produced hypothetical shell forms with various values of C and T, and a constant value of E. This corresponds to a horizontal section through the block diagram in text-fig. 8, near the base. APPLICATION TO HETEROMORPH AMMONOIDS Many well-preserved heteromorph ammonoids from the Upper Cretaceous of Hokkaido were described by Yabe (1904), Matsumoto (1967, 1977), Matsumoto and Kanie (1967), and others. Some species, especially those belonging to the Nostoceratidae and Diplomoceratidae, show peculiar three-dimensional coiling patterns; Tanabe et al. (1981) and Okamoto (1984) studied their coiling OKAMOTO: GEOMETRY OF HETEROMORPH AMMONOIDS 45 geometry. Here I analyse by moving frame analysis the growth patterns of several characteristic species of Nostoceratidae and Diplomoceratidae. The repositories of specimens are as follows: UMUT, University Museum, University of Tokyo; GK, Department of Geology, Kyushu University; WEA, Institute of Earth Science, Waseda University; and KPMG, Kanagawa Prefectural Museum. Several methods can be used to estimate values for the parameters £, C, and T from actual specimens. One is to calculate them directly from specimen measurements. This method is effective for the estimation of standardized curvature C or radius enlarging ratio £ in certain growth stages, but these values are difficult to measure continuously throughout growth: moreover, it is almost impossible to estimate the standardized torsion T by this method because the maximum growth point MGP is not evident on the shell surface. Alternatively, we can employ the tube model to model a specimen using a fixed coordinate system to obtain the equations of the centre line of the tube and its radius. The coefficients in the equations can then be determined from measurements of actual specimens. When the equations of the tube model and their coefficients are known, it is possible to calculate the three differential parameters £, C, T and corresponding growth stage s. Thirdly, there is trial and error, using computer graphics after making rough estimates of £, C, and T; by comparing the result with actual specimens, the pattern of these differential parameters can be precisely determined. In practice I employed all three methods until I obtained a satisfactory approximation to the actual fossils. Some of the results are shown in text-figs. 10 and 1 1, together with graphs of £, C, and T. The following specimens were used for the comparisons: Eubostrychoceras japonicum (Yabe, 1904). KPMG 6373, text-fig. 11b; PI. 7, fig. 9. £. muramotoi Matsumoto, 1967. WEA 003T-1, text-fig. 10c; PI. 7, figs. 3 and 4. Nipponites mirabilis Yabe, 1904. UMUT MM 17738, text-fig. 11d; PI. 7, fig. 10. Muramotoceras yezoense Matsumoto, 1977. WEA 001Y, text-fig. 10b; PI. 7, fig. 2. Hyphantoceras orientate (Yabe, 1904). WEA 002K, UMUT MM 17741, text-fig. 11a; PI. 7, figs. 7 and 8. Ainoceras kamuy Matsumoto and Kanie, 1967. GK H5575, text-fig. 10d; PI. 7, fig. 5. Scalarites scalaris (Yabe, 1904). UMUT MM17739, 17740, text-fig. 10a; PI. 7, fig. 1. Polyptychoceras sp. KPMG 6374, text-fig. 11c; PI. 7, fig. 6. The computer-produced figures represent well the fundamental coiling properties of real specimens (text-figs. 10 and 1 1). DISCUSSION Ontogenetic change The shell growth of heteromorph ammonoids often consists of a few stable stages divided by abrupt changes of coiling pattern. For example, M. yezoense and £. muramotoi show a transitional interval between two stable stages (early orthoconic stage and helicoid stage; text-fig. 10b, c). In each stable stage, the three differential parameters maintain nearly constant values, but in the transitional interval standardized curvature C and torsion T change abruptly. A. kamuy shows essentially the same coiling pattern (text-fig. 10d) in its early-middle growth, but then goes through a short transitional interval and finally forms a retroversal hook, which also has comparatively stable differential parameters. In N. mirabilis (text-fig. 1 Id), two stages are clearly discriminated. Early on, the whorl forms a loose open helix with constant differential parameters; later, however, the whorl meanders intensely around the earlier helicoid. C and T oscillate regularly during this meandering, but the stage should still be regarded as stable because the variation is regular. Changes of coiling pattern during ontogeny can be understood clearly using moving frame analysis. Rapid change of C indicates rapid change of the whorl’s direction of growth. Provided the ventral margin of the organism roughly coincides with the whorl maximum growth point (MGP), a rapid change of T suggests a rapid twist of the ventral side of the living chamber. In a transitional interval 46 PALAEONTOLOGY, VOLUME 31 B text-fig. 10. Diagrams showing the results of a moving frame analysis of some actual specimens. Right-hand figures are computer-produced profiles corresponding to the diagrams, a, Scalarites scalaris, UMUT MM 17739, 17740. B, Muramotoceras yezoense, WEA 001 Y. c, Eubostrychoceras muramotoi , WEA 003T-1. D, Ainoceras kamuy, GK H5575. between stable growth stages, C and T often change simultaneously. Such abrupt changes of coiling mode between stable stages suggest changes in mode of life. Interspecific comparison In early growth, many heteromorph species possess a nearly orthoconic shell, with very small values of C and T. After the first transitional interval, shell forms become variously diversified. Similar OKAMOTO: GEOMETRY OF HETEROMORPH AMMONOIDS 47 Hyphantoceras orientale .5 -.5 Eubostrychoceras japonicum .•..■-J ■ . ~*’-s 20 40 60 80 : ■ early stage middle st I.Or .5 -.5 • 77 Polyptychoceras . ■- i f c — s ; ” 20 40 ' SO ' 80 T E early stage ^ middle stage ^ ^ late stage 1.0 .5 • Nipp oni tes mira bit is -.5 text-fig. 1 1. Diagrams showing the results of a moving frame analysis of some actual specimens. Right-hand figures are computer-produced profiles corresponding to the diagrams, a, Hyphantoceras orientale , WEA 002K, UMUT MM 17741. b, Eubostrychoceras japonicum, KPMG 6373. c, Polyptychoceras sp., KPMG 6374. d, Nipponites mirabilis, UMUT MM 17738. coiling patterns may occur in different lineages. For example, a similar change of coiling pattern is found in M. yezoense, E. muramotoi, and A. kamuy. These ammonoids have orthoconic shafts in early growth; after the quick turn up, a helical whorl forms and coils around the earlier orthocone. N. mirabilis and Madatjasearites ryu have a similar meandering coiling pattern in their middle growth stage (Matsumoto and Muramoto 1967), yet the former belongs to Nostoceratinae and the latter to 48 PALAEONTOLOGY, VOLUME 31 Hyphantoceratinae (Matsumoto 1967); this morphological convergence suggests a similar mode of life. On the other hand, some heteromorph ammonoids show quite different coiling patterns in spite of a close phylogenetic relationship. For instance, N. mirabilis and E. japonicum share similar surface ornamentation and loose helical coiling in early growth. Matsumoto (1977) suggested that Nipponites was derived from Eubostrychoceras. During their middle growth stages, however, the two species are quite different in coiling pattern, and no transitional form has been found. Also, in the differential parameters, a transition between the two species is difficult to envisage; some saltation of coiling geometry must have occurred, if this phylogenetic relationship is true. Elypothetical shell coiling Analysis using these differential parameters appears to elucidate the mechanism of shell coiling. Nostoceratid and diplomoceratid ammonoids show considerable intraspecific variation and coiling diversity. If the three parameters were freely variable, a tremendous range of shell form would result. An analysis of the coiling of actual specimens, however, shows that changes in these parameters are closely related to one another and, consequently, produce well-regulated shell forms. For example, I have successfully reconstructed the trombone-like profile of Polyptychoceras sp. mainly using a trial and error method of computer graphics (text-fig. 11c), and values 10 °/0 larger or smaller than the best fit values of C and T produce biologically impossible shell shapes (text-fig. 12). This result strongly suggests that some regulatory mechanism affects the pattern of shell growth, so as to produce ‘well-proportioned’ coiling, which occupies only a very narrow band within the imaginable spectrum; this probably has high adaptability to the environment. CONCLUSIONS The shells of many invertebrates are formed by accretionary growth. Therefore, if growth at the aperture is exactly described, shell form can be determined absolutely. The growing tube model was derived from such a recognition, and is applicable to any pattern of shell coiling as a first approximation. One of the most practical merits of the model is that perfect similitude is kept at any growth stage because growth patterns are described relative to tube radius. This model is probably the most appropriate one available for the recognition of the actual growing process of tubular shells. Moving frame analysis enables the highly allometric and complicated coiling patterns of tubular EXPLANATION OF PLATE 7 Fig. 1. Scalarites scalaris (Yabe). UMUT MM 17739, Middle Yezo Group, Turonian; Tappu area, central Hokkaido. Lateral view, x 1-5. Fig. 2. Muramotoceras yezoense Matsumoto. WEA 001 Y, Middle Yezo Group, Turonian; Oyubari area, central Hokkaido. Upper view, x 1. Figs. 3 and 4. Eubostrychoceras muramotoi Matsumoto. WEA 003T-1, Upper Yezo Group, Coniacian; Tappu area, central Hokkaido. 3, lateral and 4, apical views, x 1-5. Fig. 5. Ainoceras kamuy Matsumoto and Kanie. GK H5575, Upper Yezo Group, Campanian; Saku area, north Hokkaido. Lateral view of retroversal hook, x 1. Fig. 6. Polyptychoceras sp. KPMG 6374, Upper Yezo Group, Santanian-Campanian; Saku area, north Hokkaido. Lateral view, x 0-67. Figs. 7 and 8. Hyphantoceras orientate (Yabe). Upper Yezo Group, Santonian; Kotambetsu area, central Hokkaido. 7, UMUT MM17742,lateral view of middle helicoid stage, x 1. 8, WEA 002K, lateral view of early stage, x 1-5. Fig. 9. Eubostrychoceras japonicum (Y abe). KPMG 6373, Middle Yezo Group, Turonian; Kiritachi area, central Hokkaido. Ventral view of early stage, x 1. Fig. 10. Nipponites mirabilis Yabe. UMUT MM 17738, Middle Yezo Group, Turonian; Oyubari area, central Hokkaido. Lateral view, x 1. PLATE 7 OKAMOTO, heteromorph ammonoids 50 PALAEONTOLOGY, VOLUME 31 text-fig. 12. Two hypothetical but unlikely shell forms of Polyptychoceras sp. a, differential parameters C and T 10% larger than actual values of text-fig. 11c. b, 10% smaller. shells to be described, analysed, and compared. If ontogenetic similarity indicates close phylogenetic relationship, moving frame analysis may be useful for the classification of heteromorph ammonoids and the reconstruction of evolutionary lineages. The growing tube model has other possible applications. If the growth of a whorl is successfully reproduced, the volume or capacity of the shell, its surface area, centre of gravity (or buoyancy), and many other physical quantities can be easily computed by integrating the differential parameters. Trueman (1941) made some qualitative inferences about the living position of some heteromorph ammonoids by considering the relationship of centre of gravity to centre of buoyancy. Klinger (1981) discussed qualitatively the mode of life of some heteromorph ammonoids from the standpoint of possible buoyancy control. By combining his concept of life orientation with the growing tube model, any changes in life position during ontogeny can be predicted for various heteromorph ammonoids (Okamoto, in press). Acknowledgements. I thank Itaru Hayami (University of Tokyo) for his critical reading of the manuscript, and Kiyotaka Chinzei (Kyoto University), David M. Raup (University of Chicago), Kazushige Tanabe (University of Tokyo), and Richard D. Norris (Harvard University) for their valuable suggestions. I am also indebted to Tatsuro Matsumoto (Kyushu University), Hiromichi Hirano (Waseda University), and Yoshiaki Matsushima (Kanagawa Prefectural Museum) for the loan of specimens. I am grateful to Tatsuo Oji (University of Tokyo), Haruyoshi Maeda (Kochi University), and Takehiro Koyaguchi (Ehime University) for constructive discussions during my laboratory work. REFERENCES fukutomi, t. 1953. A general equation indicating the regular forms of Mollusca shells, and its application to geology, especially in paleontology (I). Hokkaido Univ. Geophys. Bull. 3, 63-82. [In Japanese.] huxley, J. 1932. Problems of relative growth, 276 pp. Methuen, London. klinger, h. c. 1981. Speculations on buoyancy control and ecology in some heteromorph ammonites, 337-355. In house, m. r. and senior, j. r. (eds.). The Ammonoidea. Spec. Vol. Syst. Ass. 18, 593 pp. matsumoto, T. 1967. Evolution of the Nostoceratidae (Cretaceous heteromorph ammonoids). Mem. Fac. Sci. Kyushu Univ. Ser. D, Geol. 18, 331-347, pis. 18 and 19. 1977. Some heteromorph ammonites from the Cretaceous of Hokkaido. Ibid. 23, 303-366, pis. 43-61. and kanie, y. 1967. Ainoceras, a new heteromorph ammonoid genus from the Upper Cretaceous of Hokkaido. Ibid. 18, 349-359, pis. 20 and 21. OKAMOTO: GEOMETRY OF HETEROMORPH AM MONOIDS 51 — and muramoto, T. 1967. Two interesting heteromorph ammonoids from Hokkaido. Ibid. 361-366, pis. 22-24. moseley, h. 1838. On the geometrical forms of turbinated and discoid shells. Phil. Trans. R. Soc. 1838, 351-370. okamoto, T. 1984. Theoretical morphology of Nipponites (a heteromorph ammonoid). Kaseki (Fossils). Palaeont. Soc. Japan , 36, 37-51, pi. I. [In Japanese.] — In press. Changes in life orientation during the ontogeny of some heteromorph ammonoids. Palaeontology , 31 (2). raup, D. M. 1966. Geometric analysis of shell coiling: general problems. J. Paleont. 40, 1178-1190. 1967. Geometric analysis of shell coiling: coiling in ammonoids. Ibid. 41, 43-65. tanabe, k., obata, i. and futakami, m. 1981. Early shell morphology in some Upper Cretaceous heteromorph ammonites. Trans. Proc. palaeont. Soc. Japan, ns 124, 215-234, pis. 35-38. Thompson, d’a. w. 1942. On growth and form , 1116 pp. Cambridge University Press, Cambridge. trueman, a. e. 1941. The ammonite body chamber, with special reference to the buoyancy and mode of life of the living ammonite. Q. Jl geol. Soc. Lond. 96, 339-383. yabe, h. 1904. Cretaceous Cephalopoda from the Hokkaido. Part II. J. Coll. Sci. imp. Univ. Tokyo, 20, 1-45, pis. 1-6. Typescript received 10 September 1986 Revised typescript received 29 June 1987 TAKASHI OKAMOTO Geological Institute University of Tokyo Hongo 7-3-1, Bunkyo-ku Tokyo 1 13, Japan 52 PALAEONTOLOGY, VOLUME 31 APPENDIX The SNAKY program was written in N-88 BASIC for a 16-bit personal computer NEC PC-9801 series, interfaced with PC-8853n CRT and PC-PR101F (for hard copy production). The abridged version for Palaeontology does not provide some supplementary functions (e.g. elimination of back lines and calculation of physical quantities), and requires the input of sequential data for parameters E , C, T, and s before running. 1 ‘ ************************************************************* 2 ' + label : SNAKY. abg * 3 '+ programmed by T. Okamoto 1985/2/12 * 4 '* abridged version for 'Palaeontology' 1987/5/24 * 5 '+**********+*+**********+************************************ 6 CONSOLE 0,25,0,0: SCREEN 3: CLS 3 1000 •************+******+**+******+**+* Data Input ******* 1000 1010 INPUT "Label of data " ; LBL$ 1020 OPEN "2 : "+LBL$ FOR INPUT AS #1 1030 INPUT #1 ,COMENT$,NUM 1040 DIM S(NUM) ,E(NUM) ,C(NUM) ,T(NUM) 1050 FOR 1=1 TO NUM 1060 INPUT #1 ,DAM,S( I ) ,E( I ) ,C( 1 ) ,T( I ) 1070 NEXT I: CLOSE #1 1080 X0=300 : YO=2 00 : XS = 5: YS = 5: PI=3. 14159 1090 INPUT "VIEW ANGLE [p . q , r ] " ; PI , Q1 , R1 2000 •*******************+**++*********+ First Setting **** 2000 2010 IF P1=0 AND Q1 =0 THEN COC=l : SIC=0: GOTO 2040 2020 COC=Ql /SQR(P1 ' 2 + Q1 '2 ) 2030 SIC=P1/SQR(P1'2+Q1'2) 2040 COD=SQR ( PI ' 2 + Q1 ‘ 2 ) /SQR ( PI '2+Q1 ~2+R 1 '2 ) 2050 SID=R1/SQR(P1*2+Q1'2+R1'2) 2060 • [ coordinates ] 2070 P0=40: Q0=0 : R0=0: GOSUB +ANGLE ' X-axis 2080 LINE (XO-XS*S,YO+YS*U)- (XO+XS+S ,YO-YS*U) , 1 2090 P0 = 0 : Q0= 40 : R0 = 0: GOSUB *ANGLE ' Y-axis 2100 LINE ( XO-XS+S ,YO+YS*U)-( XO+XS+S, YO-YS+U) , 1 2110 P0=0 : Q0=0 : R0=40: GOSUB *ANGLE ’ Z-axis 2120 LINE (XO-XS+S, YO+ YS*U )-( XO+XS+S , YO-YS+U) , 1 2130 ' [ starting condition ] 2140 X=0 : Y=0 : Z=0 2150 P=0 : Q=0 : R= 1 2160 RA = 2 2170 MGX = RA : MGY = 0 : MGZ = 0 3010 FOR 1=1 TO NUM- 1 3020 IF IOl THEN GOSUB +MOVEMENT 3030 ' [ view 3040 P0=P : Q0 = Q : R0=R 3050 GOSUB +ANGLE 3060 PP=S : QQ=T : RR=U 3070 P0=X : Q0=Y: R0=Z 3080 GOSUB +ANGLE 3090 XX=S : YY=T : ZZ=U 3100 P0=MGX : P0=MGY: R0=MGZ 3110 GOSUB +ANGLE 3120 MGX1=S : MGY 1 =T : MGZ1=U 3130 GOSUB *GRAPH I CS 3140 NEXT I : END 4000 +MOVEMENT 4000 4010 EPS I LON = S ( I + 1 ) -S ( I ) 4020 RA=RA*E( I ) 'EPSILON 4030 CUR = C ( I > *EPS I LON 4040 TOR = T ( I ) *EPS 1 LON 4050 ' t next condi tion ] 4060 GOSUB +ROTAT I ONI ' maximum growth point 4070 FX4=RA: FY4=0 : FZ4=0 4080 GOSUB +ROTAT I ON2 4090 FZ5=FZ5+RA*EPSI LON 4100 GOSUB +ROTATION3 4110 GOSUB +ROTAT I ON4 4120 MGX=FX8: MGY=FY8 : MGZ=FZ8 4130 FX4=0 : FY4=0: FZ4=0 ' centre of tube 4140 GOSUB * ROT AT I ON2 4150 FZ5=FZ5+RA*EPS I LON 4160 GOSUB +ROTAT I ON3 4170 GOSUB +ROTATION4 4180 X=FX8 : Y=FY8 : Z=FZ8 4190 FX4=0 : FY4=0: FZ4=1 ' growth direction 4200 GOSUB +ROTAT I ON2 4210 GOSUB +ROTAT I ON3 4220 P=FX7 : Q=FY7 : R=FZ7 4230 RETURN growth direction radius maximum growth point Growing Process ** 3000 angle ] growth direction centre of tube maximum growth point 4240 4250 4260 4270 4280 4290 4300 4310 4320 4330 4340 4350 4360 4370 4380 4390 4400 4410 4420 4430 4440 4450 4460 4470 4480 4490 4510 4520 4530 4540 4550 4560 4570 4580 4590 4610 4620 4630 4640 4650 4660 4670 4680 4690 4710 4720 4730 4740 4750 5000 5010 5020 5030 5040 6000 6020 6030 6040 6050 6060 6070 6080 6090 6100 6110 6120 6130 6140 6160 6160 [ revolutionary subroutines ] ♦ROTATION1 ' *** R1 *** ' F: (1) FX1=MGX-X FY1 =MGY- Y FZ1 =MGZ-Z ’ F: (2) IF P' 2 + Q'2 = 0 THEN C01=l ELSE COl =P/ SQR ( P'2 + Q' 2 ) IF P'2+Q'2=0 THEN SI1 = 0 ELSE SI 1 =Q/SQR( P'2+Q'2 ) C02=R/SQR(P'2+Q'2+R'2 ) SI 2=SQR( P'2+Q'2 ) /SQR (P'2 +Q'2 +R' 2 ) FX2 = FX1 *C01 +C02+FY 1 * S I 1 +C02-FZ1+S I 2 FY2=-FX1*SI 1+FY1+C01 FZ2=FX1 *C01 +SI2+FY1+SI 1 *S I 2 + FZ1 *C02 1 F: (3) C03=C0S (TOR) SI 3 = SIN(TOR) FX3 = FX2+C03-FY2*S I 3 FY3=FX2*SI 3+FY2+C03 FZ3=FZ2 ’ F: (4 ) IF FX3'2+FY3'2=0 THEN C04=l ELSE C04=FX3/SQR( FX3'2+FY3'2 > IF FX3'2+FY3'2=0 THEN SI4=0 ELSE SI 4=FY3/SQR ( FX3 '2 +FY3 '2 ) FX4=FX3*C04+FY3*S I 4 FY4=-FX3*SI 4+FY3+C04 FZ4=FZ3 : RETURN ♦ ROT AT I ON2 * *** R2 *** ’ F: (5) GR1=RA*(EPSI LON + CUR) GR2=RA*( EPSILON-CUR) C05=2*RA/SQR( (GR1-GR2 )'2+(2*RA>'2) S I 5= (GR1 -GR2 ) /SQR( (GR1 -GR2 ) ' 2 + ( 2 *RA ) '2 ) FX5=FX4*C05-FZ4*SI 5 FY5=FY4 FZ5=FX4*SI 5+FZ4+C05 : RETURN +ROTAT I ON3 1 *** R3 *** ' F:(6) rev. of F( 4) • • • • FX6 = FX5+C04-FY5*S I 4 FY6=FX5*SI 4+FY5+C04 FZ6 = FZ5 FX7 = FX6*C01 +C02-FY6 + S1 1 + FZ6+C01 *S I 2 FY7=FX6*SI 1+C02+FY6*C01+FZ6*SI 1 *S 1 2 FZ7=-FX6*SI 2+FZ6+C02 : RETURN ♦ROTATION4 Of F( 2 ) • *** R 4 *** of F( !)•••• FX8 = FX7 + X FY8=FY7+Y FZ8=FZ7+Z : RETURN ♦ ANGLE ' 5000 S=PO*COC-QO*S I C T=P0*S I C*COD+QO*COC*COD+RO*S I D U=-P0*SIC*SI D-QO+COC+S I D+RO+COD RETURN ♦ GRAPHICS ' 6000 IF PP' 2 + QQ' 2=0 THEN COA=l ELSE COA=PP/SQR(PP'2+QQ'2 ) IF PP' 2 + QQ' 2 = 0 THEN SIA = 0 ELSE SI A = QQ/SQR(PP'2 + QQ'2 ) COB=RR/SQR(PP'2+QQ'2+RR'2 ) S I B=SQR ( PP'2+QQ' 2 ) /SQR ( PP'2+QQ' 2+RR'2 ) FOR J=0 TO 360 STEP 9 X2=X1 : Y2=Y1 : Z2=Z1 XO=RA*COS (J+PI/180) Y0=RA*SIN( J+PI/180) Z0 = 0 XI =XO*COA*COB-YO*S I A+ZO+COA+S I B Y 1 =X0*S I A+COB+ YO+COA+ZO+S I A*S I B Z1 =-X0*S I B+ZO+COB IF J = 0 THEN 6160 LINE (X0+XS*(X1+XX) , Y0-YS*(Z1+ZZ) )- ( XO+XS+ ( X2 + XX ) , YO-YS*(Z2+ZZ) ) .6 NEXT J : RETURN A REVIEW OF THE LATE ORDOVICIAN FOLIOMENA BRACHIOPOD FAUNA WITH NEW DATA FROM CHINA, WALES, AND POLAND by L. R. M. cocks and rong iia-yu Abstract. The late Ordovician Foliomena fauna is now known from five palaeocontinents. South China, Avalonia, Baltica, north-west Gondwana, and Laurentia, but has not yet been recorded from the others. New records of the fauna are presented from the Tangtou Formation, South Jiangsu Province, China, and from the Staurocephalus clavifrons Beds in Poland, and illustrated for the first time from the Crugan Mudstones of North Wales. The occurrences and ages of all known Foliomena faunas are documented and reviewed, and it is concluded that the most typical Foliomena fauna, including true F. folium, occurred from the latest Caradoc (Onnian) beds to the early upper Ashgill (mid-Rawtheyan). Its ecology is discussed, and evaluated as deep-water, but not necessarily ocean-facing, and marginal to continents. Comparable faunas, one including the closely related ribbed Proboscisambon , are identified as occurring in Bohemia, Canada (Perce, Quebec), and Wales (Garth). The taxonomies of Foliomena and the Eoplectodonta genus-group are briefly discussed, with Kozlowskites recognized as a sub-genus of Eoplectodonta and Foliomena confirmed as a strophomenacean. Knowledge of late Ordovician brachiopods has increased enormously during the past quarter century, in particular those faunas which occur peripherally to the main continental cratons or in deeper-water aulacogens within them. One such fauna, a representative of which was monographed for the first time as recently as 1973 (Sheehan 1973), is the Foliomena fauna, named after a smooth strophomenacean, which occurs in a distinctive assemblage with other brachiopods listed below. Because trilobites, ostracods, and other animals are sometimes dominantly associated with this fauna, it is not termed here a 'community', since it often forms only a small proportion of the total fossils found, and it is also uncertain how many of the other fauna were benthic and associated with the brachiopods: some of the trilobites, for example, were almost certainly mesopelagic. The brachiopods of the Foliomena fauna are all of very small size, are thin-shelled, and never occur very abundantly, which is the chief reason for their lack of systematic attention; and moreover, since the marginal sites in which the fauna often occurs have usually been structurally deformed, the fauna has seldom been systematically collected. However, the fauna, and a related one containing Proboscisambon instead of Foliomena , have been recognized from various places, and a review is now timely. DISTRIBUTION OF FOLIOMENA FAUNAS (TABLE 1) 1. South China (text-fig. 1) (a) North Guizhou. Mu En-Zhi, Zhu Zhao-1 ing, and Rong Jia-yu collected a Foliomena fauna in 1972 from the Linhsiang Formation at Ganxi, Yanhe County, which overlies the Pagoda Limestone with a Sinoceras chinense fauna and underlies the graptolitic Wufeng Formation, and consists of green-yellow calcareous mudstones with a thickness of 3-2 m. Brachiopods are uncommon (Rong 1984) and there were fewer than five specimens each of F. /o/mm (Barrande), Kassinella incerta(X u and Rong), Christiania nilssoni Sheehan, Dedzetina sp., and Aegiromena sp. Trilobites dominate the fauna and include N ankinolitlms. Trinodus, Hammatocnemis, Corrugatagnostus, Shumardia, Ampyx, Calymenesum, and others. The age is pr e-szechuanensis Zone and thus probably early Ashgill. A small collection including F. folium, K. incerta, and a lingulid (BC 7320-2) has also been made by R. P. Tripp from the Chiencaokou Formation at its type locality at Jiancaohe, Donggongsi, Zunyi County. (Palaeontology, Vol. 31, Part 1, 1988, pp. 53-67, pis. 8—9. | © The Palaeontological Association 54 PALAEONTOLOGY, VOLUME 31 table 1. The constitution of the various collections of the true, restricted Foliomena fauna from localities detailed in the text. The actual numbers of specimens collected are shown where known. 3 o D O N 3 C/3 W) c _ N 3 bf) 3 '3 O a pj G .5 G < 43 N z z c/5 GO cd £ O z o o g cd > j-t a cd GO TD c cd T3 cd 'O G jd :0 bij :cd > “O G ^cd O cu -C o OQ lingulide indet. Philhedra ? sp. X 1 X X Acrothelel sp. 2 craniacean X eocramatiid X Orbiculoidea sp. dolerorthid 1 1 exopunctate orthid 1 Glyptorthis sp. 1 7 4 Dedzetina sp. X 1 3 X X X X X 1 11 2 X Heterorthiiial sp. Karlicium karlicum 4 X Aegiromena sp. X Kassinella incerta Eoplectodonta ( Kozlowskites ) X X nuntius /ragnari 4 X 21 3 Leptestiina prantli 2 40 X X X 23 X 5 X Leptestiina sp. Durranellal sp. 2 1 1 1 leptellinid indet. Anoptambonitesl sp. Leangella sp. Anisopleurella sp. 1 1 1 1 1 3 Sericoidea plectambonitacean indet. 1 X 1 8 1 Foliomena folium X X 4 5 x 27 X X X X 1 250 x 3 X Christiania nilssoni X 5 5 x 37 1 X X 1 36 x 3 Eostropheodonta sp. 1 1 Leptaena sp. 1 1 Holtedahlinal sp. 1 Eopholidostrophia sp. Parastrophinella'I sp. 1 1 Cyclospiral scanica Zygospiral sp. rhynchonellid indet. 1 1 3 1 ? ? X X 1 190 4 (b) South Jiangsu. The Tangtou Formation in the Nanjing area has yielded the following trilobites of the N. nankinensis Zone (Lu and Zhou 1981): Trinodus , Corrugatagnostus , Shumardia , Telephina , Nileus , Cyclopyge, Bumastus , Magydenia, Phillipsinella , Nankinolithus , Dionide , Lonchodomas , Encrinurella , Hammotocnemis , Atractopyge , and Diacanthaspis. From this formation at Tangtou, Jianging County, about 20 km east of Nanjing, R. P. Tripp has collected the following brachiopods, which we identify as Christiania nilssoni Sheehan (PI. 9, fig. 9), Eoplectodonta ( Kozlowskites ) nuntius (Barrande) (PI. 9, figs. 2, 3, 5, 6), F. folium (Barrande) (Plate 9, fig. 8), Leptestiina prantli COCKS AND RONG: ORDOVICIAN FOLIOMENA FAUNA 55 text-fig. 1. Locality map of the Foliomena fauna in South China, including the boundaries of the provinces and also the boundary of the South China Plate (after Rong and Chen 1987). Havlicek, Leptestiina sp., Acrothelel sp., Dedzetinal sp.. Cyclospiral scanica Sheehan, Philhedral sp., Parastrophinellal sp., Durranellal sp., indeterminate rhychonellid, and a further indeterminate articulate. The age of this fauna is also early Ashgill, since it lies below szechuanensis Zone graptolites and above the Sinoceras chinense cephalopod zone. (c) South Anhui. A Foliomena fauna was collected in 1974 from the Huangnehkang Formation at Jiugongmiao, Ningguo County, which overlies the Yenwashan Formation with a S. chinense faunas and underlies the graptolitic Wufeng Formation, and consists mainly of yellow mudstones bearing trilobites of the N ankinolithus Fauna (Rong 1984). Brachiopods are rare but include F. folium (Barrande) (PI. 9, fig. 7), Christiania nilssoni Sheehan, Cyclospiral cf. scanica Sheehan, Leptestiina sp., indeterminate leptellinid, and an indeterminate plectambonitacean. The age is the same as north-east Guizhou. (d) West Zhejiang. The Foliomena fauna occurs in two formations in a hill close to the northern border of Jiangshan County, western Zhejiang (text-fig. 1), which was collected by Han Nai-ren. A few specimens from the Huangnehkang Formation of Pusgillian age included F. folium (Barrande) (PI. 9, fig. 14) and Christiania nilssoni Sheehan (Rong 1984) (PI. 9, fig. 10). In addition, a single specimen of F. folium was collected from the overlying Changwu Formation, together with five specimens of Tcherskidium sp. (PI. 9, figs. 17-20), four of Kassinella anisa Percival (PI. 9, figs. 11-13, 15), and two each of Cyclospiral cf. scanica Sheehan and Leptestiina sp. from an horizon overlying szechuanensis Zone graptolites. This fauna is discussed further in the section on ecology below. 56 PALAEONTOLOGY, VOLUME 31 2. Wales ami Belgium ( Avalonia ) (a) North-west Wales. In the Llyn Peninsula of western North Wales (Gwynedd) the Crugan Mudstone Formation outcrops in two geographically separated areas (Price 1981), and the Foliomena fauna has been collected from the following localities within the formation, chiefly by S. F. Morris, B. Roberts, and D. Price: 1, a road-side quarry in Crugan Lane, 1 km north-east of Llanbedrog, Grid Ref. SH 33323241; 2, a temporary exposure in Dynana farmyard, 1 km north-north-east of Llanystumdwy, SH 48 1 396; 3, an exposure at Berllan Cottage, 1 km north of Llanbedrog, SH 32483365; and 4, section along Afon Penfhos, north-east of Llanbedrog, SH 33503245. The fauna from Locality 1 has been listed by Cocks (in Price 1981, p. 203) as numerous L. prantli (Havlicek); Christiania nilssoni Sheehan (PI. 8, figs. 2-4); F. folium (Barrande) (PI. 8, fig. 1); and Dedzetina sp.; and one specimen each of Anoptambonitesl sp., Eostropheodonta sp., Leangella sp., Zygospiral sp., and Orbiculoidea sp.; Durranelhf! sp. (PI. 8, fig. 5) is also now recorded. Collections from the other localities have also all yielded Leptestiina prantli , F. folium, and Dedzetina sp., with, in addition, C. nilssoni and Leptaena sp. from Locality 4; Cyclospira sp., Eopholidostrophia sp., Leptaena sp., Anisopleurella sp., Glyptorthis sp., Christiania nilssoni , and Holtedahlinal sp. from Locality 3; and Lingula sp., another indeterminate lingulide, Cyclospira sp., and Eoplectodonta sp. from Locality 2. A small locality in the hillside south-east of Locality 2 yielded the first listed record of the Foliomena fauna as such (Temple in Roberts 1967, p. 378). The age of the Crugan Mudstone has been determined by Price (1981, p. 206) as fairly high in the Rawtheyan Stage, correlating with Zones 5 to 6 of Ingham (1966). EXPLANATION OF PLATE 8 Figs. 1-11. Foliomena fauna from the Crugan Mudstone Formation, North Wales. 1, F. folium (Barrande 1879), internal mould of pedicle valve, BB 33232, x 6, from roadside quarry in Crugan Lane 1 km north-east of Llanbedrog, Gwynedd, Grid Ref. SH 33343241, collected by S. F. Morris. 2-4, Christiania nilssoni Sheehan 1973, internal and external moulds of a brachial valve, fig. 2 taken obliquely to show details of cardinalia, BB 33251, x 3, same locality as fig. I. 5, Durranellal sp., internal mould of brachial valve, BB 33242, x 6, same locality. 7 and 8, F. folium (Barrande 1879). 7, internal mould of pedicle valve, BB 95938, x4, quarry 150 m west of Crugan Farm, 1 km north-west of Llanbedrog, Gwynedd, Grid Ref. SH 33323241, collected D. Price; 8, latex cast of external mould of pedicle valve showing the unique specimen with a single median costella, BB 33196, x 8, exposure at Berllan Cottage, north of Llanbedrog, Gwynedd, Grid Ref. SH 32483365. 9, Dedzetina sp., internal mould of brachial valve, BB 33167, x 6, temporary exposure in Dynana farmyard, 1 km north-north-east of Llanystumdwy, Gwynedd, Grid Ref. SH 481396, collected by S. F. Morris. 6, 10, 11, Leptestiina prantli (Havlicek 1952). 6, internal mould of pedicle valve, BB 33224, x 8 from Afon Penfhos, 1-3 km east of Llanbedrog Church, Gwynedd, Grid Ref. SH 33503245; 10 and 11, two views of internal mould of pedicle valve, BB 33248, x 4, same locality as fig. 1. Figs. Hand 15. Foliomena fauna from the Slade and Redhill Mudstone Formation, South Wales. 14, Aegironetes sp., internal mould of pedicle valve, BB 26136, x 12, from quarry 50 m south of Rudbaxton Church, Dyfed, Grid Ref. SM 961206. 15, C. nilssoni Sheehan 1973, internal mould of brachial valve, BB 26687, x 6, from Prendergast Place Quarry, Haverfordwest, Dyfed, Grid Ref. SM 956166. Fig. 16. Proboscisambon quaesitus (Barrande 1879), internal mould of a pedicle valve, MM 039, x 6, from Cryptolithus kosoviensis horizon, Kraluv Dvur Formation, foot of Kosov hill, near Kraluv Dvur, Czechoslovakia, collected by M. Mergl. Figs. 12, 13, 17-24. Foliomena fauna from the Staurocephalus clavifrons Bed, Wolka, near Nowa Stupia, Holy Cross Mountains, Poland, collected by W. T. Dean. 12 and 13, Glyptorthis sp., internal and external moulds of pedicle valve, BC 7165, x 8. 17, 18, 22, exopunctate orthid gen. et sp. indet.; 17 and 18, internal mould and latex cast of the external mould of a pedicle valve, BC 7157, x 4; 22, enlargement of the external mould of the same valve, x 20, to show the exopunctate ornament. 19, Cyclospira sp., internal mould of pedicle valve, BC 7169, x 8. 20 and 21, Foliomena folium (Barrande 1879), external and internal moulds of a brachial valve, BC 7177, x4. 23, Glyptorthis sp., internal mould of a brachial valve, BC 7163, x 8. 24, Leptestiina prantli (Havlicek 1952), internal mould of a pedicle valve, BC 7171, x 4. PLATE 8 COCKS and RONG, late Ordovician brachiopods 58 PALAEONTOLOGY, VOLUME 31 ( b ) South-central Wales. In the Garth district, Powys, Williams and Wright (1981) have recorded a Foliomena fauna from unnamed siltstones of Rawtheyan age (Zones 6 to 7 of Ingham) based on trilobites. From Cwm Clyd Quarry (Grid Ref. SN 946509), they have collected a sparse but diverse fauna which they identified as Leptestiina sp., Cyclospira sp„ plectambonitacean, dalmanellid, Dedzetina sp., Leangella sp., Dalmanella cf. sculpta (Cooper, 1956), Eoplectodonta sp., ? Leptestiina, leptestiinid nov., Kozlowskites sp., F. cf. joliensis Sheehan & Lesperance, Eospirigerina sp., Trimurellina sp., Onniella sp., Cremnorthis sp. nov., Orthambonitesl sp., Plaesiomys cf. porcata (M’Coy), Aegiromena sp. nov., Chonetoidea sp., Leangella cf. scissa (Davidson), leptestiinid, sowerbyellid?, Gunnarellal , Leptaena rugosa Dalman, Christiania sp., strophomenacean, spire-bearer indet, and articulate indet., a total list of twenty-nine species out of only eighty-two specimens. However, as mentioned in the section on ecology below, this fauna does not fall within the range of a typical Foliomena fauna, even though F. cf. joliensis is recorded from it. (c) Belgium. Passing references to a Foliomena fauna in Belgium have been made by Sheehan (e.g. 1979, p. 69), but no details, locality, or age have been published. 3. Scotland , Ireland , and Canada ( southern margin of Laurentia) (u) Western Scotland. Harper (1979, 1984) has recorded a Foliomena fauna from the red mudstone member of the Myoch Formation, in the upper part of the Whitehouse Group of Girvan, Strathclyde, from five separate localities all within 100 m of each other on the Whitehouse foreshore, south-west of Girvan (Grid Ref. NX 175954 to 176955). Harper (1979, p. 440) provisionally recorded Lingula , Dedzetina, Sericoidea, Foliomena, Christiania, Cyclospira, a craniacean, and a new genus of eocramatiid. The associated trilobites and graptolites indicate an age which lies within the latest Onnian Stage of the Caradoc (Harper 1984, p. 3). Harper (1984) has redescribed Lingula sp. and craniid gen. et. sp. indet. 1, but the remainder of the fauna will appear in later parts of his monograph. It is of importance as being the oldest known Foliomena fauna, and indeed the only one of probable Caradoc age. Harper (1979, p. 441) also recorded a small fauna from the overlying Shalloch Formation of Cautleyan age in green shales associated with Dicellograptus anceps itself and consisting only of Dedzetina sp. and Cyclospira sp.: he regarded this fauna as a restricted Foliomena fauna, and we agree. (b) Ireland. Harper (1980) recorded and illustrated Dedzetina sp., Sericoidea sp., Foliomena sp., Christiania sp., Cyclospira sp., and a leptellinid gen. et sp. indet. from the Ballyvorgal Group at Slieve Bernagh, County Clare, but he informs us (pers. comm. 1986) that only one specimen of each taxon was found. Unfortunately there are no other stratigraphically significant fossils known from these beds. (c) Canada. Sheehan and Lesperance (1978) described a Foliomena fauna from the base of the sea cliffs at Mount Joli, Perce, Quebec Province, which at that time they attributed to the White Head Formation, but which Skidmore and Lesperance (1981, p. 40) later excluded from the White Head Formation and simply placed within the Matapedia Group. They list the brachiopods Lingulacid gen. et sp. indet., orthacean fam. gen. et sp. indet., dalmanellacid gen. et sp. indet., plectambonitacean fam. gen. et sp. indet., IDiambonia septata (Cooper 1930), F. joliensis Sheehan and Lesperance, and C.? minuscula Cooper 1930, as well as 205 specimens of trilobite attributable to seven species, and ‘common’ crinoid columnals. The trilobites indicate an age of somewhere within the Ashgill. 4. Scandinavia and Poland ( Baltica ) (a) South Sweden. The first proper description of any Foliomena fauna as such was by Sheehan (1973) from the Jerrestad Mudstone in the Koangen borehole, near Fagelsang, east of Lund, Scania, where elements of the fauna were found in different proportions through 13 m of core. The fauna is dominated by F. folium (Barrande) and C.? scanica Sheehan, and also contained Christiania nilssoni Sheehan, Leptestiina prantli Havlicek; Eoplectodonta ( Kozlowskites ) ragnari Sheehan, Dedzetina sp., Sericoidea sp., Glyptorthis sp., Heterorthinal sp., Anoptambonites sp., and indeterminate dolerorthid, and Aegiromeninae. Associated trilobites and graptolites give the age as within the complanatus COCKS AND RONG: ORDOVICIAN FOLIO MENA FAUNA 59 graptolite zone and the Tretaspis granulata trilobite zone, which is Pusgillian. Nilsson (1979) recorded the fauna from the same formation in a borehole 850 m north of Sodra Sandby Church, western Scania, and which consisted of F. folium (Barrande), DP. cf. honorata (Barrande), and ‘some indeterminable specimens and a couple of orthids and strophomenids’ (Nilsson 1 979, p. 1 1 ). However, these occurred in the Stawocephalus clavifrons Zone, which is of Rawtheyan age, and hence younger than the fauna described by Sheehan (1973). (, b ) Vastergotland, Sweden. Jaanusson {in Sheehan 1979, p. 69) records Foliomena, Leptestiina, and Christiania from the Jonstorp Mudstones in Vastergotland, and we here confirm the identification of F. folium (Barrande) on two specimens (Stockholm Riksmuseet Br 10402 and Br 10625) from the Red Jonstorp Mudstone at Mosseberg, of probable Cautleyan age. Jaanusson (1982, pp. 173-174) also recorded a fauna from the middle beds of the Ulunda Formation in the Phillipsinella parabola trilobite zone, which is also probably of Cautleyan age, and includes F. folium (Barrande), Rugosowerbyella rosettana (Henningsmoen), and C. nilssoni Sheehan. Finally, there is a single specimen, Br. 134001, from a limestone in the base of the Dalmanitina Beds, and therefore perhaps of early Hirnantian age, from Kullsberg, Vastergotland, which has four or five primary costellae and definitely no parvicostellae, and which we can thus identify only as FoliomenaP sp. (c) Poland. A hitherto unrecorded occurrence of the Foliomena fauna is from the S. clavifrons Bed at Wolka, near Nowa Stupia, in the Holy Cross Mountains, Poland, where a small number of blocks originally collected by W. T. Dean in 1961 has yielded to us the following brachiopods: L. prantli Havlicek (PI. 8, fig. 24), Cyclospiral ragnari Sheehan (PI. 8, fig. 19), Glyptorthis sp. (PI. 8, figs. 12, 13, 23), Christiania nilssoni Sheehan, F. folium (Barrande) (PI. 8, figs. 20 and 21), Foliomena 1 sp., exopunctate orthid (PI. 8, figs. 17, 18, 22), Eoplectodonta sp., and Dedzetina sp. The S. clavifrons Beds are Rawtheyan in age (Kielan 1960). 5. Czechoslovakia (north-west Gondwana) Havlicek & Vanek (1966) described a Foliomena fauna from an old (now lost) locality within Prague from the upper part of the Kraluv Dvur Formation, 15-20 m under the base of the Kosov Formation, in compact grey-green micaceous shales of the T. seticornis horizon, of anceps Zone age. They recorded F. folium (Barrande), L. prantli Havlicek, D. macrostomoides Havlicek, and Karlicium karlicum Havlicek (see Havlicek 1982, p. 126). A further related fauna from the uppermost Kraluv Dvur Formation and containing Proboscisambon quae situs (Barrande) is discussed below in the section on ecology. AGE OF THE FOLIOMENA FAUNA In South China the Linhsiang, Huangnehkang, and Tangtou Formations, all bearing the true Foliomena fauna, all lie above the Pagoda or Yenwashan Limestone bearing the Sinoceras chinense nautiloid fauna and below the Wufeng Shale, which carries a Dicellograptus szechuanensis graptolite zone fauna at its base. These formations were correlated with the Shikoan Stage (Sh 1) or late Caradoc, but have been reassigned to the early Ashgill (Pusgillian or early Cautleyan) because they immediately underlie the szechuanensis Zone, which is usually correlated with the lower to middle anceps Zone (Rong 1984). Thus the Foliomena fauna is only known from the early Ashgill of China, apart from the specimen of F. folium itself from the Changwu Formation of West Zhejiang, which is of Middle Ashgill age. In Avalonia, the only locality with an accurately dateable true Foliomena fauna is within the Crugan Mudstone of North Wales, which Price (1981) had dated as lower Rawtheyan on the basis of the associated trilobite fauna; the other Welsh fauna, that from Garth (Williams and Wright 1981), although accurately dated as mid-Rawtheyan, is one of the closely comparable faunas containing FP d.joliensis rather than F. folium. The Belgian occurrence has not yet been dated. From Baltica, which was at least in close faunal contact with Avalonia at the time (Cocks and Fortey 1982), and may even have collided with it, the Foliomena fauna is known from both the Pusgillian/Cautleyan and also the Rawtheyan in Scania, south Sweden, and from the Middle Ashgill in Vastergdtland, central 60 PALAEONTOLOGY, VOLUME 31 Sweden, the Rawtheyan horizon in Scania is from the Staurocephalus clavifrons trilobite zone, in which the Foliomena fauna also occurs in the new fauna from the Holy Cross Mountains, Poland, near the southern margin of the palaeocontinent. From the north-western margin of Gondwana, the Foliomena fauna is known only from the upper part of the Kraluv Dvur Formation of Bohemia, which lies within the anceps Zone, and is probably of Rawtheyan age (Havlicek 1982). From Laurentia, the best record comes from the Myoch Formation of Girvan, which correlates with the upper part of the linearis Zone and probably with the late Onnian stage of the Caradoc (Harper 1 984). The correlation is also based on the occurrence of T. ceriodes from the basal beds of the Upper Whitehouse Group (Ingham 1978), suggesting that the early and middle linearis Zone is no younger than Onnian in age (Williams and Bruton 1983). A restricted Foliomena fauna also occurs in the overlying Shalloch Formation of anceps Zone age, which may be early Cautleyan (Harper 1979). Along strike, at Perce, Canada, the previously reported Foliomena fauna of Sheehan and Lesperance (1978) cannot be dated exactly within the Ashgill. To sum up, the age of the Foliomena fauna mainly falls within the Pusgillian, Cautleyan, and early to middle Rawtheyan, with a single fauna from Girvan being perhaps of late Caradoc (Onnian) age. The comparable Proboscisambon fauna is only known undoubtedly from Rawtheyan rocks, and all the occurrences of both faunas appear to be stratigraphically older than the Hirnantia faunas of the latest Ordovician (text-fig. 2). ECOLOGY OF THE FOLIOMENA FAUNA Without exception, all the known occurrences of the Foliomena fauna are in fine-grained rocks, usually fine mudstones. In addition, all of the brachiopods are small, with very few specimens over 10 mm in width. They are also invariably very sparsely distributed in low density through the sedi- ment, and this latter fact has also meant that the sample sizes of the collections available both to us and to other workers have very often been pitifully small. This collecting failure exacerbates the problem of comparing the relative diversities of the various collections of brachiopods. Brachiopods EXPLANATION OF PLATE 9 Figs. 1 and 4. Foliomena fauna from Chiencaokou Formation, Donggongsi, Zunyi City, West Guizhou Province, China, collected by R. P. Tripp. 1, Foliomena folium (Barrande 1879), internal mould of pedicle valve, BC 7426, x 4; 4, Kassinella incerta (Xu, Rong and Liu 1974), internal mould of pedicle valve, BC 7424, x 4. Figs. 2, 3, 5, 6, 8, 9. Foliomena fauna from Tangtou Formation, Lunshan trench, 20 km north-east of Nanjing, Jiangsu Province, China, collected by R. P. Tripp. 2, 3, 5, 6, Eoplectodonta ( Kozlowskites ) nuntius (Barrande 1879). 2 and 6, internal and external moulds of a pedicle valve, BC 7126, x4; 3, internal mould of brachial valve, BC 7125, x 4; 5, internal mould of small pedicle valve, BC 7148, x 4; 8, F. folium (Barrande 1879), internal mould of pedicle valve, BC 7144, x 8; 9, Christiania nilssoni Sheehan 1973, internal mould of pedicle valve, BC 7133, x 6. Fig. 7. F. folium (Barrande 1879), from Huangnekang Formation, Jiugongmiao, Ningguo County, South Anhui Province, China, internal mould of pedicle valve, NIGP 101831, x 6. Figs. 10 and 14. Foliomena fauna from Huangnekang Formation, hill to west of Jiangshan County, south-west Zhejiang Province, China. 10, C. nilssoni Sheehan 1973, internal mould of pedicle valve, NIGP 101829, x 8. 14, F. folium (Barrande 1879), internal mould of pedicle valve, NIGP 101830, x4. Figs. 11 13, 15-20. Fauna allied to Foliomena fauna from Changwu Formation, hill to west of Jiangshan County, south-west Zhejiang Province, China. 11-13, 15, Kassinella anisa Percival 1979; 11 and 12, internal moulds of pedicle valves, NIGP 101833 and NIGP 101834, x 4; 13 and 15, internal moulds of brachial valves, NIGP 101835 and NIGP 101836, x 4. 16, F. folium (Barrande 1879), external mould of conjoined valves, NIGP 101832, x 6. 17-20, Tscherskidium sp. 17 and 20, internal mould of brachial valve viewed from posteriorly and above, NIGP 101827, x 2; 18 and 19, internal mould of a pedicle valve viewed from above and posteriorly, NIGP 101828, x 2. PLATE 9 COCKS and RONG, late Ordovician brachiopods 62 PALAEONTOLOGY, VOLUME 31 text-fig. 2. Correlation of the true Foliomena fauna (shown by triangles), the Hirnantia fauna (shown by stars), and some faunas related to the Foliomena fauna, with the European and Chinese graptolitic zonations (Williams 1986; Mu et al. 1979). The numbers indicate: (1) A shallower brachiopod faunule with F. folium, Kassinella anisa, and Tcherskidium sp„ (2) Foliomena sp. associated with a high-diversity brachiopod fauna; (3) A faunule allied to the Foliomena fauna but with no F. folium, (4) Foliomena ? sp.; (5) Proboscisambon with no Foliomena, Quebec’s data with no precise age, and the Belgian fauna (unpublished) are not included. very often form only a small proportion of the total fauna, which is commonly dominated by trilobites and sometimes ostracods. All previous workers, e.g. Sheehan (1973), Harper (1980) have considered the fauna to have been deposited under deep-water, and we agree with this analysis, but for additional reasons. First, the fauna is known from five different palaeocontinents in late Ordovician time (text-fig. 3), and moreover continents whose palaeolatitudes differed very considerably (Cocks and Fortey 1988). It has already been stated (e.g. Cocks and Fortey 1982) that the deep-water faunas are the ones most likely to be of a more even temperature, and hence less latitudinally dependent, and so the disparate occurrences of the Foliomena fauna in both space and time might be expected from their deeper-water mode of life. This is in contrast with the adjacent shallower-water communities, the occurrence of which reflect continental distributions and palaeolatitudes (Cocks and Fortey 1982, fig. 1). The total biota in which the Foliomena fauna is found is dominated by trilobites, very often including the cyclopygids, and this trilobite biofacies is also consistently indicative of a deeper-water habitat (Fortey 1985), which strengthens our conclusions drawn from the brachiopods. However, we agree with Sheehan (1977) that the brachiopod fauna was indigenous to the substrate in which it is found, for the following reasons: (a) a high proportion of shells are found still articulated, which argues against transportation; (b) the Foliomena fauna is found consistently together, whereas if it lived epiplanktonically at or near the surface its remains ought to be found scattered more widely and also over a greater range of lithofacies, including shallower ones; and (c) a uniformly smaller and thinner-shelled fauna is more likely to be found in deep rather than shallow water (Fiirsich and Hurst 1974). This is not to say that the Foliomena fauna depths were very great, and probably not off the continental shelves, as can be seen by its distribution in, for example, south China and Sweden (Vastergotland): these areas, although undoubtedly relatively deep, only represented depressions in the continental landmasses. The other faunas closely related to the Foliomena fauna are also relevant here. The first is the so-called Proboscisambon Community of Havlicek (1982), which is known only from the uppermost COCKS AND RONG: ORDOVICIAN FOLIOMEN A FAUNA 63 text-fig. 3. The distribution of the Foliomena fauna during Ashgill times. The palaeogeographic map, which omits Siberia, Kazakhstan, and North China, is taken from Cocks and Fortey (1988). Kraluv Dvur Formation at Kosov Hill, near Kraluv Dvur, Zadni Treban, Liten, and Jezerka; all in the Prague area, Czechoslovakia. Like the typical Foliomena fauna (which occurs lower down in the same formation: see above), it occurs in calcareous silty shales and contains low-density, high-diversity brachiopods with greater numbers of trilobites and ostracods. The brachiopods include abundant (more than 100 specimens) P. quaesitus (Barrande), Kozlowskites ragnari Sheehan, Ravozetina honorata (Barrande), R. opima Havlicek & Mergl, relatively common (50-100 specimens) Aegironetes tristis (Barrande) and Epitomyonia dorcicava Havlicek & Mergl, fewer than 50 specimens of Durranella moneta (Barrande), and rarer specimens of ‘Salopian', Jezercia ostaria Havlicek & Mergl, Boticium , Cliftonia, Leptaena, Cryptothyrella, Ornothyrella , and others. True Foliomena, Christiania, Leptestiina, and Cyclospira are all absent from that list, but nevertheless it seems probable that the Proboscisambon fauna does represent a comparable ecological position to the 64 PALAEONTOLOGY, VOLUME 31 Foliomena fauna, particularly in view of the ‘ Foliomena faunas of Perce, Canada (Sheehan and Lesperance 1978) and Garth, Wales (Williams and Wright 1981), where the place of the smooth F. folium itself is taken by the ribbed and closely related Foliomena! sp. (although final attribution must await elucidation of the internal features of the Canadian and Welsh forms). The other fauna is the one listed above from the Changwu Formation of West Zhejiang, China, in which a single (but undoubted) specimen of Foliomena and rare Cyclospiral and Leptestiina occur with Tscherskidium and Kassinella. This is one of the few places in eastern China where early pentameraceans (Tscherskidium) are known from mid-Ashgill rocks; the other is from the Sanjushan Formation of Zhuzhai, Yushan County, north-eastern Jiangxi Province (Rong and Han 1986), where argillaceous limestones bearing Tscherskidium are interbedded with coral horizons including abundant Acjetolites , Kolymopora, Fletcheriella , Plasmoporella, Heliolites, and Stelliporella, which have also been found in the north-eastern USSR (Yang 1984). Tscherskidium itself is characteristic of the Rawtheyan fauna of the north-east USSR (Koren et al. 1983) and it has also been doubtfully recorded from the western slope of the northern Urals and southern Kazakhstan (Rukavishnikova and Sapelnikov 1973; Sapelnikov and Rukavishnikova 1975). Another distinctive brachiopod occurring in the Chungwu Formation with Foliomena is Kassinella anisa Percival, a plectambonita- cean hitherto recorded only from New South Wales, Australia (Percival 1979), where it apparently occupied a relatively shallow-water environment. We interpret the Changwu Formation fauna as having lived in medium shelf depths, with the odd record of Foliomena as probably representing the extreme shallow end of its depth range. SYSTEMATIC NOTES Superfamily plectambonitacea Jones, 1928 Family sowerbyellidae Opik, 1930 EOPLECTODONTA Kozlowski, 1929 Eoplectodonta was originally erected as a subgenus of Plectodonta by Kozlowski (1929), but has been treated as a separate genus for many years (e.g. Williams et al. 1965): the differences between Eoplectodonta and Plectodonta were tabulated by Cocks (1970, p. 167). Unknown to Kozlowski, Eoplectodonta had a long history prior to Silurian times, and is known from rocks at least as old as Llandeilo in age. Various genera have been erected which are very closely related to Eoplectodonta, including Eochonetes , Kozlowskites , and Thaerodonta from the late Ordovician. All have denticulate hinge lines, but Eochonetes is distinct from the others in possessing, in addition, a perforate hinge line. Thaerodonta has been revised by Rdomusoks (1981), and is not considered further here: of the group, only Eoplectodonta s.s. and Kozlowskites are known from the Foliomena fauna. Havlicek (1952) erected Kozlowskites as a separate genus, upon which he elaborated further in 1967, but Williams (in Williams et al. 1965, p. H381) regarded it as only a subgenus of Eoplectodonta, an opinion followed by Sheehan (1973). Havlicek (1967) listed the chief difference as the absence of the pedicle valve median septum in Kozlowskites (although the ‘septum’ in Eoplectodonta is more in the nature of an elongated myophragm between the pedicle valve muscle scars). We follow Williams and Sheehan in considering Kozlowskites to be a subgenus of Eoplectodonta. Although the two stocks are very closely related, we regard them as more than specifically distinct because (i) since the pedicle opening of Kozlowskites is larger, the muscle scars diverge directly from either side of it with a slight space between them, unlike Eoplectodonta where they are closer together and separated only by the variably developed myophragm, and (ii) the brachial valve muscle area is relatively shorter (always less than half the valve length) and more likely to be enclosed posteriorly by the bounding ridge in Kozlowskites than in Eoplectodonta where the scars are always over half the length and almost always open posteriorly. Although shells in the Foliomena fauna are usually rather small, when they are well enough preserved to be identified accurately, they are usually found to be Eoplectodonta ( Kozlowskites ) rather than E. ( Eoplectodonta ), although contemporary assemblages from shallower palaeodepths (e.g. Cocks 1982) usually contain only E. (Eoplectodonta). COCKS AND RONG: ORDOVICIAN FOLIOMENA FAUNA 65 Superfamily strophomenacea King, 1846 Family foliomenidae Williams, 1965 foliomena Havlicek, 1952 Since Foliomena was proposed (Havlicek 1952), with its monotypic species from the Ashgill Kraluv Dvur Formation of Bohemia, only one other species has been assigned to the genus, F. joliensis from the Ashgill of Perce, Canada (Lesperance and Sheehan 1978). Another genus, Proboscisambon , was established by Havlicek and Mergl (1982), with type species P. quaesita (Barrande), a species which Havlicek had previously (1967, p. 36) assigned to the plectambonitacean Anoptambonites. Havlicek and Mergl listed F. joliensis as a synonym of P. quaesita and placed Proboscisambon with the Foliomenidae, which Williams (1965) had erected as a monogeneric family within the Strophomenacea. Various questions have arisen in the course of our work. (a) Fs Foliomena a plectambonitacean or a strophomenacean? Havlicek (1952) placed his new genus within the Strophomenacea, and this was followed by Williams (1965) when he elevated the taxon to familial rank. However, Bergstrom ( 1968, p. 480) wrote that 'it seems probable that this family ought to be placed in the vicinity of Leptestiidae Opik within Plectambonitacea rather than within Strophomenacea’, and Sheehan (1973, p. 65) followed Bergstrom’s assignment. However, the basic difference between the two superfamilies is the bilobed cardinal process of the Strophomenacea and the trilobed cardinal process of the Plectambonitacea. There is no doubt that the cardinal process of Foliomena is bilobed (Havlicek 1952, p. 31; 1967, pi. 9, figs. 2 and 13), and this is confirmed by the examination of our own specimens described in this paper. Thus, the Foliomenidae must be placed within the Strophomenacea. (b) What are the differences between Foliomena and Proboscisambon? The two genera are very similar in shell size, outline, and convexity, in their prominent pedicle sheath, in lacking dental plates, and in the size and proportions of the faintly impressed muscle fields in both valves. However, Proboscisambon differs from Foliomena in the following ways: (i) it has radial ornamentation of very fine parvicostellae and stronger axial costellae (3-7 in number), in contrast to the smooth shell of Foliomena ; (ii) it lacks a cardinal process (Havlicek 1982, p. 45), in contrast to the bilobed cardinal process of Foliomena', ( iii) it lacks any septa in the brachial valve, in contrast to Foliomena, which has a prominent pair of slightly divergent septa; and (iv) it has prominent dispersed papillae on the inner surface of the shell, in contrast to Foliomena which has very fine and dense papillae almost all over the shell interior. From this, Foliomena can be rediagnosed as follows: shell small to medium size, gently concavo-convex, nearly flat; pedicle sheath prominent; ornament only of fine growth lines and occasional rugae; no dental plates; pedicle valve muscle scars faintly impressed and divided by a small myophragm; small bilobed cardinal process; socket plates nearly parallel with hinge line; pair of lateral septa; fine papillae irregularly distributed over most of the internal shell surface. ( c ) Should ‘F.’ joliensis be placed within Proboscisambon? There is no doubt that the specimens of F. joliensis illustrated by Sheehan and Lesperance ( 1 978, pi. 1 , figs. 7 13) have costellae. However, the problem is that in neither case are internal features available, and thus it is not known whether the other very distinctive features of Proboscisambon (the lack of cardinal process and of brachial valve septa) are developed in the Canadian shells. It is still possible that F. joliensis has all the features of true Foliomena apart from the development of costellae and parvicostellae. Thus, we have identified the Canadian specimens as FP joliensis until full details of their internal morphology become available. The Garth specimen illustrated as F. cf. joliensis by Williams and Wright (1981, fig. 2f) is quite similar to the single specimen, Br 134001, from the base of the Dalmanitina Beds at Vastergotland: they both have three to five primary costellae and no fine parvicostellae, although their internal details are unknown. They differ from the true FP joliensis of Quebec in their lack of fine parvicostellae. In addition, a single one of the many specimens of F. folium from the Crugan Mudstones of North Wales (BB 33196; PI. 8, fig. 8) has a single marked central costella without any sign of fine parviscostellae. It seems to us that this specimen may have been a precursor of the Garth and Vastergdtland specimens, which we have provisionally identified here as Foliomenal sp. 66 PALAEONTOLOGY, VOLUME 31 Acknowledgements. We are most grateful to Dr D. A. T. Harper for his comments on the first draft of this manuscript, to Professor V. Jaanusson, Drs V. Havlicek and M. Mergl for discussion and providing specimens from Sweden and Czechoslovakia respectively, and to the Royal Society for an exchange fellowship for one of us (R. J.-y.) REFERENCES bergstrom, J. 1968. Review of V. Havlicek, Brachiopoda of the Suborder Strophomenidina in Czechoslovakia. Geol. For. Stockh. Fork. 90, 479-480. COCKS, L. R. M. 1970. Silurian brachiopods of the Superfamily Plectambonitacea. Bull. Br. Mus. nat. Hist. Geol. 19, 141-203, pis. 1-17. 1982. The commoner brachiopods of the latest Ordovician of the Oslo- Asker district, Norway. Palaeontology, 25, 755-781, pis. 78-84. and fortey, r. a. 1982. Faunal evidence for oceanic separations in the Palaeozoic of Britain. J. geol. Soc. Lond. 139, 465-478. - 1988. Lower Palaeozoic facies and faunas round Gondwana. Spec. Publ. geol. Soc. Lond. (in press). fortey, r. a. 1985. Pelagic trilobites as an example of deducing the life habits of extinct arthropods. Trans. R. Soc. Edin. 76, 219-230. fursich, f. t. and hurst, j. m. 1974. Environmental factors determining the distribution of brachiopods. Palaeontology, 17, 879-900. harper, d. a. t. 1979. The environmental significance of some faunal changes in the Upper Ardmillan succession (upper Ordovician), Girvan, Scotland. Spec. Publ. geol. Soc. Lond. 8, 439-445. — 1980. The Brachiopod Foliomena fauna in the upper Ordovician Ballyvorgal Group of Slieve Bernagh, County Clare. J. Earth Sci. R. Dublin Soc. 2, 189-192. -1984. Brachiopods from the upper Ardmillan Succession (Ordovician) of the Girvan District, Scotland. Pt. 1. Palaeontogr. Soc. [ Monogr. ]. 1-78, pis. 1-11. havlicek, v. 1952. On the Ordovician representatives of the Family Plectambonitidae (Brachiopoda). Sb. ustred. Ust. geol. 19, 397-428, pis. 13-15. — 1967. Brachiopoda of the Suborder Strophomenidina in Czechoslovakia. Rozpr. ustred. Ust. geol. 235, 1-234, pis. 1-52. 1982. Ordovician in Bohemia: Development of the Prague Basin and its benthic communities. Sb. geol. Ved. Geol. 37, 103-136, pis. I -8. — and mergl, m. 1982. Deep water shelly fauna in the latest Kralovdvorian (upper Ordovician, Bohemia). Vest, ustred. Ust. geol. 57, 37-46, pis. 1-4. — and vanek, j. 1966. Biostratigraphy of the Ordovician of Bohemia. Ibid. 8, 7-68, pis. 1-16. ingham, j. k. 1966. The Ordovician rocks in the Cautley and Dent districts of Westmorland and Yorkshire. Proc. Yorks, geol. Soc. 35, 455-505, pis. 25-28. 1978. Geology of a continental margin 2: middle and late Ordovician transgression, Girvan. Geol. Jour. Spec. Iss. 10, 163-176. jaanusson, v. 1982. Ordovician in Vastergotland. In bruton, d. l. and williams, s. h. (eds.). Field excursion guide for IV International Symposium on the Ordovician System. Palaeont. Contr. Univ. Oslo, 279, 164-183. kielan, z. 1960. Upper Ordovician trilobites from Poland and some related forms from Bohemia and Scandinavia. Palaeont. pol. 11, 198 pp. koren, t. n., oradovskaya, m. m., pylma, l. y., sobolevskaya, r. f. and chugaeva, m. n. 1983. The Ordovician and Silurian boundary in the North east of the USSR, 1-192, pis. 1-48. Leningrad Nauka, Publ. Leningrad Branch. [In Russian.] kozlowski, r. 1929. Les brachiopodes gothlandiens de la Podolie Polonaise. Palaeont. pol. 1, 1-254, pis. 1-12. LU YAN-HAO, CHU, CHAO-LING, CHIEN YI-YUAN, ZHOU ZHI-YI, CHEN JUN-YUAN, LIU GENG-WU, YU WEN, CHEN XU and xu han-kui. 1976. Ordovician biostratigraphy and palaeozoogeography of China. Mem. Nanjing Inst. Geol. Palaeont. 7, 1-83, pis. 1 14. — and zhou zhi-yi. 1981. Early Upper Ordovician trilobites from the Nanjing Hills. Bull. Nanjing. Inst. Geol. Palaeont., Acad. sin. 3, 1-27, pis. 1-7. mu en-zhi, ge mei-yu, chen xu, ni yu-nan and lin yao-kun. 1980. New investigation of Ordovician strata of Southern Anhui. J. Stratigr. 4 (2), 81-86. -zhu zhao-ling, chen jun-yuan and rong jia-yu. 1979. Ordovician rocks in Southwest China, 108-154. In Carbonate rock stratigraphy of Southwest China. Beijing. nilsson, R. 1979. A boring through the Ordovician Silurian boundary in western Scania, South Sweden. Sver. geol. Unders. Afh. 766, 118. COCKS AND RONG: ORDOVICIAN FOLIOMENA FAUNA 67 percival, i. G. 1979. Ordovician plectambonitacean brachiopods from New South Wales. Alcheringa,3, 91 -116. price, d. 1981. Ashgill trilobite faunas from the Llyn Peninsula, North Wales, U.K. Geol. J. 16, 201 -216. Roberts, B. 1967. Succession and structure in the Llwyd Mawr Syncline, Caernarvonshire, North Wales. Ibid. 369-390. rong JIA-YU. 1984. Ecostratigraphic evidence of the Upper Ordovician regressive sequences and the effect of glaciation. J. Stratigr. 8, 19-29. and chen XU. 1987. Faunal differentiation and environmental patterns of late Ordovician (Ashgillian) in South China. Acta Palaeont. sin. 25 (in press). — and han nai-ren. 1986. Preliminary report on Upper Ordovician ( mid- Ashgillian ) brachiopods from Yushan, northeastern Jiangxi (Eastern China). Biostrat. Paleozoique , 4, 485-490, pi. I. roomusoks, a. 1981. Strophomenida of the Ordovician and Silurian of Estonia. Ill Genus Thaerodonta Wang. Eest. N.S.V. Tead. Akad. Toim. Geol. 30, 61-71. rukavishnikova, t. b. and sapelnikov, v. p. 1973. New taxa of late Ashgillian Pentameracea in Kazakhstan. Akad. Nauk SSSR , Ural. Nauch. Tsentr ., Trad. Inst. Geol. Geokhim. 99, 89-114. sapelnikov, v. p. and rukavishnikova, t. b. 1975. Upper Ordovician, Silurian and Lower Devonian Pentamerids of Kazakhstan. Akad. Nauk SSSR , Ural. Nauch. Tsentr. 1-227. sheehan, p. M. 1973. Brachiopods from the Jerrestad Mudstone (early Ashgillian, Ordovician) from a boring in Southern Sweden. Geol. Palaeont. 7, 59-76, pis. I 3. 1977. Ordovician and Silurian brachiopods from graptolitic shales and related deep-water argillaceous rocks. Lethaia , 10, 201-203. 1979. Swedish late Ordovician marine benthic assemblages and their bearing on brachiopod zoogeography. In gray, j. and boucot, a. j. (eds.). Historical Biogeography , Plate Tectonics , and the Changing Environment, 61-73. Oregon State University Press. and lesperance, p. j. 1 978. The occurrence of the Ordovician brachiopod Foliomena at Perce, Quebec. Can. J. Earth Sci. 15 (3), 454-458, pi. I. skidmore, w. b. and lesperance, p. j. 1981. Perce Area. The White Head Formation, Perce. In lesperance, p. j. (ed.). Subcommission on Silurian Stratigraphy, Ordovician Silurian Boundary Working Group. Field Meeting, Anticosti-Gaspe, Quebec 1981, Vol. 1: Guidebook, 31 40. williams, A. et al. 1965. In moore, r. c. (ed.). Treatise on Invertebrate Paleontology. Part H. Brachiopoda. Kansas, 927 pp. and wright, a. d. 1981. The Ordovician-Silurian boundary in the Garth area of southwest Powys, Wales. Geol. J. 16, 1-39. williams, s. h. 1986. Top Ordovician and Lowest Silurian of Dob’s Linn. Spec. Publ. geol. Soc. Lond. 20, 165-171. — and bruton, d. l. 1983. The Caradoc-Ashgill boundary in the central Oslo Region and associated graptolite faunas. Norsk geol. Tidsskr. 63, 147-191. yang sheng-wu. 1984. Ordovician tabulate coral assemblages of Britain and their zoogeographical relation- ships. Paleontogr. Am. 54, 169 173. L. R. M. COCKS Department of Palaeontology British Museum (Natural History) Cromwell Road London SW7 5BD Typescript received 8 October 1986 Revised typescript 9 January 1987 RONG jia-yu Nanjing Institute of Geology and Palaeontology Academia Sinica Chi-Ming-Ssu Nanjing China Note added in proof. Since this paper was written, R. A. Fortey has made a collection from a muddy limestone equivalent to the Setul Formation from an outcrop in La Ngu District, Satun Province, south-west Thailand, which contains trilobites of south Chinese aspect and can be dated as certainly close to the Caradoc-Ashgill boundary, and perhaps early Ashgill in age. From this limestone we have identified several specimens of Foliomena folium (including some small growth stages) and Christiania sp. and rather fewer specimens of Orbiculoidea sp. This is the first record of the Foliomena fauna from the Burma-Thai-Malaysian Peninsula and adjacent areas. A HERBACEOUS LYCOPHYTE FROM THE LOWER CARBONIFEROUS DRYBROOK SANDSTONE OF THE FOREST OF DEAN, GLOUCESTERSHIRE by N. P. ROWE Abstract. Leafy lycophyte shoots preserved as impressions and as material resembling fusain are described from a shale band in the Drybrook sandstone (Upper Visean), Puddlebrook, Forest of Dean. The consistently narrow stems with small terminal strobili bearing megaspore impressions place this material in Selaginellites Zeiller ( 1906). Fusainized material shows fine surface morphology and anatomy of the leaves and stem. The large number of specimens demonstrates a wide range of morphological variation in the position and shape of attached leaves. In particular, the impressions and fusainized preservation of laterally attached microphylls, which are orientated perpendicularly to the bedding plane, are often drastically altered from their original laminate structure. The leafy shoots show some similarity to those of Clwydia decussata Lacey (1962) and Archaeosigillaria kidstonii Krausel and Weyland (1949). These plants are also discussed in the light of recent approaches towards the interpretation of lycophyte impressions and compressions. The new material is assigned to S. resimus sp. nov. One of the major difficulties in reconstructing fossil lycophytes from compression assemblages is the inevitable fragmentary nature of the material and the problem in identifying specific characters common to large axes and the much smaller leafy shoots. A situation, therefore, may exist where narrow lycophyte axes from a given assemblage could represent either a herbaceous element or the unattached distal parts of an arborescent organism. For this reason relatively few truly herbaceous lycophytes are known from Carboniferous compression floras in spite of the abundance of small lycophyte shoots (Thomas 1967; Chaloner and Collinson 1975; Chaloner and Meyer-Berthaud 1983). This paper deals with a small leafy lycophyte with diminutive leaves and a stem width not exceeding 2 mm. Because the axes described here are only found as relatively short fragments the material is prone to the interpretative difficulties outlined above regarding its overall size. An argument is put forward suggesting a herbaceous habit. This is based on its consistently narrow stem at all levels of branching and also the presence of acutely curved portions of axis giving rise to terminal strobili. These are interpreted as representing upturned portions of the axis which arose from an otherwise sprawling shoot system. The lycophyte is preserved as impressions, compressions, and material resembling fusain. The impressions show a range of preservational variation which depended on the effects of compression on the axis and attached leaves, the orientation of the plant in the matrix, and on the extent of sediment accretion around the plant surface, and on the path of the plane of cleavage through the fossil (Thomas and Purdy 1982; Rex and Chaloner 1983; Grierson and Banks 1983; Edwards and Benedetto 1985). The fusain-like material corroborated much of the morphological detail obtained from the impressions and provided additional anatomical and fine morphological information. MATERIAL AND METHODS The locality at Hazel Hill, Puddlebrook is a disused quarry. The fossil plants occur in a laminated but poorly bedded shale with a blue/grey to buff coloration. The quality of preservation is highly variable and much of the material has been badly weathered by the movement of water through the shale. The most common type of [Palaeontology, Vol. 31, Part 1, 1988, pp. 69-83, pis. 10-12.| © The Palaeontological Association 70 PALAEONTOLOGY, VOLUME 31 text-fig. 1. Diagram illustrating the formation of the two principal types of impression surface of Selaginellites resimus as seen in hypothetical longitudinal section. A, the microphylls are at a wide angle to the axis at the time of deposition so sediment can penetrate in between the leaves and stem, and form an impression with the surface of the stem, b, the microphylls are at an acute angle to the stem and are actually overlapping and exclude sediment from the spaces between the stem surface and the laminae. In both A and B the organic component of the fossil has disappeared and the plane of cleavage has passed along the middle of the space originally occupied by the stem. The variation between the two types is not brought about so much by the position of the line of cleavage as by the position of the leaves at the time of deposition. preservation of this lycophyte is as impression material in which organic or compression material is absent. In their study of compressions of Haskinsia sagittata, Edwards and Benedetto (1985) draw attention to the fact that plant impressions exist with little or no organic matter in either counterpart. This is certainly the case with the majority of specimens from Puddlebrook. Two well-preserved specimens shown in Plate 10, figs. 1 and 3 illustrate what are probably the two principal configurations seen among the range of material collected, but which both lack organic material. In one, the median microphylls descend into the matrix and are visible as apertures (PI. 10, fig. 1 ), but in the other, surface impressions of the median laminae are clearly visible (PI. 10, fig. 3). In both cases the line of cleavage has passed through the middle of the space originally occupied by the axis (text-fig. 1, arrows). In the absence of any organic matter only impression surfaces remain and the differences between the two forms are believed to result from a difference in the angle of the leaves to the axis and the amount of sediment which accreted in the space between them at the time of deposition. In this way an impression of the stem surface is produced ( PI. 1 0, fig. 1 ; text-fig. 1 a) because the microphylls were perpendicular to the axis and sediment collected in between, whereas in Plate 10, fig. 3 and text-fig. 1b the median microphylls were at a more acute angle to the stem and effectively sealed off sediment from making contact with the stem surface. Using the terminology of Chaloner and Collinson (1975) and Grierson and Banks (1983) when both counterparts lack organic material these represent a double cleavage impression rather than a cleavage impression united with a cleavage compression. The two configurations displayed here corroborate Rex and Chaloner’s (1983) finding that the angle of the leaves to the stem at the time of deposition dramatically controls the appearance of the leafy shoot impression, which is seen here in lycophyte stems of only 2 mm diameter. Fusainized material is rare and only four specimens with leaves in attachment were found which were consolidated enough to withstand demineralization and observation with the SEM. Because of the close morphological similarity between the three-dimensional impressions and the fusainized axes, both types of EXPLANATION OF PLATE 10 Figs. 1-3. Selaginellites resimus sp. nov. 1, V. 62304, impression of axis with dichotomy, three vertical rows of leaves visible. Plane of cleavage has passed through the middle of the space originally occupied by the stem. Median microphyll impressions are visible as flattened letter Ws which descend into the matrix while marginal microphylls are visible as V-shaped grooves, x 4-5. 2, V.62303, branching axis where most of the stem is twisted at 45° relative to cleavage. Left branch is sharply recurved and partially defoliated. In this region are spheroidal depressions (arrowed) which are believed to represent the impressions formed around sporangia. Before the division of the axis the leaves are twisted at 45° to the bedding plane showing alternating arrangement of whole and partially hidden leaves, x 3. 3, V.62305, median microphylls are visible as impressions in surface view instead of descending into the matrix. Note the cordate to deltoid outline, and the attenuated distal tips of the vegetative leaves, x 4. PLATE 10 ROWE, Selaginellites 72 PALAEONTOLOGY, VOLUME 31 text-fig. 2. Selaginellites resimus sp. nov. a, V.62310a, fusainized axis with median microphyll in attachment having laminate profile, x 10. B, V. 62310b, marginal microphyll attached to fusainized axis with evidence of a considerable degree of distortion showing a broad vertical area of attachment, x 10. c, d, line-drawings of V.6231u and V.62316 respectively, x 10. e, diagram summarizing the effects of compression on an axis of S. resimus preserved as fusainized material. Only one median and one marginal microphyll are shown. preservation arc believed to be derived from the same plant species. Although reflectance studies have not been carried out, this material is referred to as fusain merely because of its superficial similarity to fusinite. In fusainized material the marginal microphylls are often distorted or compressed in a plane perpendicular to the surface of the lamina (text-fig. 2b, d, e). The direction of compression of the flattened microphyll in Plate 12, ROWE: CARBONIFEROUS HERBACEOUS LYCOPHYTE 73 tig. 1 is the same as that seen on the axis to which it is attached (PI. 12, fig. 2). There is a two-fold interest in this observation; first, the difference in susceptibility to compression between the median and marginal microphylls, and secondly, the fact that the material is composed of brittle fusain-like material and yet has been compressed quite substantially without total shattering of the cells visible in the cortex, or the complete breakage of the surface of the marginal microphyll. Text-fig. 2e summarizes the gross morphology of the stem and a median and marginal microphyll after compression. The median microphyll has retained its laminate structure (text-fig. 2a and c), whereas the marginal microphyll is perpendicularly flattened (text-fig. 2n, d; PI. 12, fig. 1). The behaviour of the median microphylls in Selaginellites resimus as seen in the fusainized material, differs considerably to that observed in spines of Sawdonia ornata (Chaloner et al. 1978) which were observed to have been flattened in a vertical plane, whereas those observed in the fusainized Puddlebrook material were only slightly compressed. This occurred in material which obviously had the potential to be deformed as seen by the distorted marginal microphylls on the same specimen. Unlike the observations on S. ornata , where spines on the median surfaces of the axis were distorted (Chaloner et al. 1978) it is the marginal leaves which are the most noticeably affected in Selaginellites resimus. There is evidence from the impression material that the median leaves, which appear as flattened letter Ws have been partially flattened, but this is in the opposite direction to that observed in the spines on the upper surface of Sawdonia. Unfortunately owing to the sometimes highly friable nature of the shale it was not always possible to retain part and counterpart or record the orientation of the fossil in the sediment. Some of the fusainized axes do not show any difference between median and marginal microphylls (PI. 11, fig. 6). Perhaps an explanation of this is the fact that the shale in which the plant fossils are preserved consists of irregular bands several millimetres thick comprising different sized sediment particles, and it is possible that this affected the compressibility of the matrix surrounding the plant material. The change in shape of the lateral appendages and the main axis in material that resembles fusain poses some doubt as to its formation and composition. It is inconceivable that brittle and highly fragile structures such as the microphylls and axes of this material could have been distorted and undergone such changes while in this state. Because of the compressed nature of the material, two hypotheses are possible concerning its origin. One is that the conversion to a fusain-like material took place after burial and compression, which would infer that conflagration by wild fire was not responsible for its formation. Alternatively, the organic matter might not represent fusinite sensu stricto but consist of some other maceral which preserves anatomical and fine morphological details. The latter explanation may be of some significance when charcoal-like material is observed in the geological record and popularly believed to have originated as a result of conflagration (Alvin 1974; Scott and Collinson 1978; Cope and Chaloner 1980). The plants were photographed under even or directional lighting. Fusainized material was either wholly or partially removed from the matrix with steel needles and then demineralized in 30% hydrofluoric acid for up to 10 min. Specimens were then washed thoroughly and dried at room temperature. Fusainized specimens were adhered to SEM stubs with silver dag which was applied when semi-dry, so that the fluid did not invade the specimen and ruin any fine detail. Careful application of silver dag and quite prolonged gold coating (up to 6 min) were necessary to reduce charging of this highly fragmentary material when examined by SEM. The specimens and preparations are housed in the Palaeontology Department of the British Museum of Natural History (BMNH, V.62303-V. 62330). The Lower Carboniferous Drybrook Sandstone assemblage at Puddlebrook has been the subject of several papers since Allen (1961) identified the lycophyte sporophyll Lepidostrobophyllum fimbriatum from there (Lele and Walton 1962; Thomas 1972; Thomas and Purdy 1982). The material described here was discovered during a recent reinvestigation. A direct, biostratigraphical age determination of the Puddlebrook assemblage is difficult because of the extreme rarity of animal remains and the poor preservation of miospores (Lele and Walton 1962). The plant-bearing shale outcrops on the western margin of the Wigpool syncline in the northernmost part of the Forest of Dean and is presumed to be continuous with an outcrop of Drybrook Sandstone on the eastern limb of the syncline at Plump Hill. The nearest age determination for the macroflora is based on miospore data collected from Plump Hill (Sullivan 1964). Because of the presence of Perotriletes tessellatus and Schultzospora (Sullivan 1964) and Carbaneuletes circularis (Spinner 1985), an Asbian/Holkerian age or a level equivalent to the TC zone of Clayton et al. (1977) is indicated. Correlation of the Puddlebrook locality with that at Plump Hill is substantiated by macrofossil remains of L. fimbriatum (Allen 1961) and impressions of fronds resembling Diplopteridium Walton, both of which are common elements at Puddlebrook. A poorly preserved spore assemblage from Puddlebrook includes the characteristic palynomorph Tetrapterites visensis (Sullivan and Hibbert 1964) which is known from a shale band at the Plump Hill locality and from the Upper Visean of the Carboniferous Limestone in the Menai region of north Wales. 74 PALAEONTOLOGY, VOLUME 31 DESCRIPTION Impressions/Compressions. The axes are 1 -5 to 2-0 mm in diameter and up to 7 5 mm wide including the attached leaves. The stems branch isotomously and usually every 2 to 4 cm (PI. 11, figs. 1-3). In such specimens the diameter of the axis does not change significantly from 2 0 mm over several divisions. The angle of branching ranges between 30° and 80° (PI. 10, figs. 1 and 2; PI. 1 1, figs. 1 -3). Three out of the four orthostichies are visible in most specimens (PI. 10, fig. 3). When the leafy shoots are arranged at 45° to the plane of cleavage a series of alternating, long and short, appendages is seen (PI. 10, fig. 2). The microphylls are 3-2 to 4-5 mm in length and when seen in surface view are cordate to deltoid in outline and terminate in fine pointed tips (PI. 10, fig. 3). In some cases the microphylls are not seen in surface view but are visible as slit-like apertures that resemble a flattened letter ‘W’ (PI. 10, fig. 1). These apertures are 1-8 mm wide and are sometimes surrounded by a weak ridge or groove. The apertures are interpreted as representing the impressions of leaf laminae which have since weathered away. The microphylls attached to the lateral margins of the stems indicate that the leaves were curved distally. They show a broad, vertical attachment (PI. 10, fig. 1) which is the same length as the distance between two vertically adjacent leaf bases in the median part of the stem. As was shown above this feature is interpreted as resulting from compression of the marginal laminae which are distorted from an originally laminate aspect. The terminal parts of some shoots have a different appearance from some of the more proximal parts of leafy axes (PI. 10, fig. 2, arrow; text-fig. 3a-f), and these are believed to be strobili. Two of the specimens consist of partially defoliated axes with additional circular to oval depressions (PI. 10, fig. 2; text-fig. 3b, d). A further specimen contains at least eight groups of rounded impressions and these are associated at the bases of closely arranged appendages in the distal portion of the axis (text-fig. 3a). Each depression is 0-9 to 1-8 mm in diameter and is subdivided into several oval or rounded triangular subunits. It is difficult to determine the arrangement and outline of the terminal appendages or sporophylls on this specimen as they lack the pointed distal tips characteristic of the more proximal leaves. This specimen represents an impression of a small terminal, compact strobilus, with the depressions representing the rounded impression surfaces of the sporangia at the sporophyll bases. This suggestion is supported by the specimen illustrated in text-fig. 3b-f. The specimen is complicated by the fact that two separate branches arise from the bend in the vegetative stem. The branch to the right side is attached to the proximal vegetative axis, and it is possible that the additional branchlet is either superimposed or represents part of a dichotomizing fertile region. Unlike the specimen in text-fig. 3a, many of the sporophylls have been shed but two or three are still attached to the slender axis and are visible in surface view (text-fig. 3c, e). The sporophyll is not pointed distally, and in addition to this, there is at least one circular impression of a megaspore which is flattened against the surface of the cordate sporophyll (text-fig. 3e, f). To the outside of this (arrow) there is a raised ridge of sediment which formed the impression surface around the megasporangium which is 1-6 mm in diameter. The uppermost surface is broken and reveals four empty spheroidal areas (text-fig. 3e, f) each of which is equivalent in size to the mineral cast of the megaspore, immediately above. This complex is interpreted as representing a tetrad of megaspores which formed a three-dimensional impression from which the organic matter has since disappeared. The minute mineral impression surfaces visible in these areas are complex and were presumably formed by mineral (Spicer 1977) or very fine sediment accretion around and inside the sporangium which even formed an impression surface of the presumably spiny megaspores. A very similar type of small-scale impression surface is seen around megaspores of L. fimbriatum from the same locality (Allen 1961). Fusainized material. One specimen, although having little fine structure preserved, shows sufficient gross morphology to compare very closely with the impressions described above (PI. 11, fig. 6). There is clearly little evidence to suggest any development of a leaf cushion or expanded leaf base. Instead the microphylls emerge from the axis as simple laminate structures. Although there was generally no cellular detail of the stem surface (PI. 1 1, figs. 4 and 6), in one specimen the structure of the cortex and the xylem was preserved. The cortex is between 250 EXPLANATION OF PLATE 1 1 Figs. 1-7. Selaginellites resimus sp. nov. 1, V.6231 1; 2, V. 62306; 3, V. 62312; all xO-75. Note equal dichotomies and the maintenance of a consistently narrow stem diameter at all levels of branching. 4 and 5, V.62310, oblique view of fusainized microphyll base with lens-shaped profile attached to main axis, x 30. 5, abaxial surface of microphyll close to junction with axis. Note the irregular surface with raised stomata (left) and the fractured surface, representing the hypodermal layer (right), x 200. 6, V. 62309, fragment of fusainized vegetative axis with laminate, deltoid to cordate leaves arranged in four vertical rows, x 14. 7, V.62310, group of stomata on abaxial surface of leaf with conspicuous rim or ridge perched above guard cells, x 550. PLATE 11 ROWE, Selaginellites F ROWE: CARBONIFEROUS HERBACEOUS LYCOPHYTE 77 to 350 /an wide, and encloses the xylem which is 400 /an in diameter (PI. 12, fig. 2). The cortical cells are oval to isodiametric in transverse view and 1 5 to 45 //m in diameter. There is little differentiation of the cortex into zones except a tendency for the smaller elements to be at the outside. The vascular tissue is rarely preserved and only fragmentary remnants of the tracheids are normally visible as oblique, transverse, and longitudinal fracture surfaces (PI. 12, figs. 3-5). The xylem is separated from the cortex by a narrow space. It is 400 /jm in diameter and composed ofscalariform tracheids (PI. 12, fig. 3). The tracheids apparently decrease in width from the inner to the outer part of the stele and range from 19 to 25 /an in diameter. The protoxylem is therefore probably exarch. The xylem is not sufficiently well preserved to determine whether protoxylem points were present, but the arrangement of some of the smaller protoxylem elements suggest this (PI. 12, fig. 3). The tracheids are at least 60 /mi long and possess scalariform ridges which are 3 /an thick. The wall material between the scalariform bars consists of a membraneous layer comprising four or five circular to oval pits which are approximately 1 to 2 /an in diameter and apparently confined to the tangential walls of the tracheids (PL 12, figs. 4 and 5). It is not known whether these different layers represent secondary wall development or both primary and secondary layers. High magnification of transversely fractured cell walls indicate that the thin, pitted wall layer appears as a series of broken pegs attached to a layer of wall material which is continuous with the thick, adjacent scalariform bar (PI. 12, fig. 4). The cell walls of adjacent tracheids are sometimes separated by a narrow groove (PI. 12, fig. 4, arrow). It is possible that this might represent the position of the middle lamella. Only one fusainized microphyll provided any evidence of a structure resembling a ligule or ligule pit but this is by no means certain (PI. 12, fig. 1, arrow B). Stomata are present on the abaxial surface of the leaves and have a frequency of 340 to 370 per mm2 (PI. 1 1, figs. 5 and 7). Of four specimens with cellular details, three ‘types’ of stomata were identified which are believed to represent preservational variations of the same structure (text-fig. 4). The stomata are 35 to 38 /mi long and 28 to 38 n m wide. In one specimen (text-fig. 4a) both guard cells are clearly visible, although the surrounding organic material is not well preserved. There is a broad stomatal aperture and an irregular ridge is present on the outer surface of both guard cells. In another specimen (text-fig. 4b), the stomata have a different appearance. There is a rim or ridge on the outer surface of the guard cells and j ust inside this are the inner faces of the guard cells which line the stomatal aperture. The best preserved type of stoma (text-fig. 4c) comprises a pair of guard cells which is raised above the surface of the leaf and partially overlain by a skirt of organic material which might represent cuticle. The stomata are closely arranged and are separated by narrow grooves which might indicate the position of modified subsidiary cells. Part of the abaxial surface of one microphyll shows details of the hypodermis (PL 11, figs. 4 and 5). On the right part of the leaf the outer surface layer of disorganized organic material has broken away to reveal a thick-walled hypodermis of randomly arranged, isodiametric to oval cells, 15 to 34 /an in diameter. The upper epidermis of the microphyll, as seen on a laterally distorted microphyll, consists of elongate cells that are aligned with the long axis of the microphyll (PL 12, fig. 1). The cells that are close to the central ridge are irregular in outline and approximately 20 to 28 /an in diameter. A transverse fracture through the base of a microphyll seen on one specimen (PL 12, fig. 6) shows several zones text-fig. 3. Selaginellites resimus sp. nov. a, V.62307, a vegetative axis is sharply curved and gives rise to a small compact strobilus. The sporophylls lack the attenuated distal tips characteristic of the vegetative leaves. Groups of spherical depressions along a central position of the strobilus represent the impressions of sporangia, x 4. b-f, V.62308, holotype. b, curved axis giving rise to one, possibly two, strobilar axes. The upper branch shows evidence of a blunt, cordate sporophyll associated with a complex, three-dimensional impression surface of a megaspore tetrad (arrowed), x 3 5. c, enlargement of area of strobilus arrowed in b. The broad sporophyll differs considerably from the vegetative leaves and is partially superimposed at its base by the complex impression surface of a megaspore tetrad, x 10. d, line-drawing of holotype showing arrangement and position of fertile areas in relation to vegetative part of axis, x 4-5. e, enlargement of sporophyll and megaspore tetrad impressions. The lower part of the tetrad impression consists of a mineral layer which covered the outer part of the sporangium and megaspores. Immediately above is a roughly circular impression surface with minute apertures which is interpreted as representing an impression surface of a spiny megaspore. Immediately to the left of this structure is an oval depression also assumed to represent the impression or indentation in the matrix of a megaspore, x 18. f, line-drawing of features seen in e, the megaspore impression (m), the depression in the matrix formed by a megaspore (d), and the cavity once containing megaspores and partially overlain by a thin wall of mineral matter (s), x 18. 78 PALAEONTOLOGY, VOLUME 31 ABC text-fig. 4. Preservational variation of stomatal complexes in fusainized Selaginellites resimus sp. nov. a, V.623 1 5, the pair of guard cells has a flattened profile and is sunk below the surface of the microphyll. An outer layer of material covers each guard cell which has an irregular inner edge (arrowed) surrounding the stomatal aperture, x 1400. b, V.62309, the stomatal complex is sunk below the surface of the leaf but the guard cells are not clearly visible. There is an oval hoop of material which terminates as a well-developed rim above the stomatal aperture, x 1400. c, V.623 10, the stoma is raised above the level of the leaf surface and consists of two guard cells which are contained beneath a skirt of material terminating in a rim (arrowed) above the stomatal aperture, x 1800. of cells. The hypodermis is two cells thick near the edges of the lamina and three to four cells deep closer to the midrib. The xylem consists of a single strand of about thirty tracheids (PI. 12, fig. 8), 8-15 /im in diameter. They possess scalariform to reticulate thickenings (PI. 12, fig. 7). In transverse section a series of peg-like outgrowths indicates that the tracheid wall was thin and pitted between the scalariform bars (PI. 12, fig. 7). The vascular strand of the leaf is 250 /mi in width. Immediately adaxial to the xylem, a number of thin-walled elements are visible which are arranged perpendicularly to the lamina (PL 12, fig. 6). Each element is about 75 /mi long and 20 /mi wide. This is interpreted as a zone of palisade parenchyma in the adaxial half of the microphyll. EXPLANATION OF PLATE 12 Figs. 1-8. Selaginellites resimus sp. nov. 1-5, V.623 106. 1, marginal microphyll showing evidence of compression in a direction perpendicular to the plane of the leaf resulting in the formation of an apparently broad vertical area of attachment to the axis. A transverse ridge (arrow A) indicates the original position of the lamina margin and this corresponds well with the impressions of lateral microphylls. A structure on the folded adaxial surface may represent a ligule pit (arrow B), x 45. 2, transverse fracture of axis showing the position of the cortex and xylem. Although the whole structure has been compressed the cortex is relatively complete and uncrushed, x 40. 3, transverse fracture surface of tracheids from main axis showing distribution of protoxylem elements towards the outer part of the xylem (lower right), x 500. 4, oblique view of tracheid of main xylem. The fractured part of the cell wall between the scalariform bars consists of a number of small pegs (arrow), x 2000. 5, longitudinal fracture surface of tracheid from main xylem with angular scalariform bars with thin pitted walls in between, x 1 500. 6-8, V.623 1 0 a. 6, transverse fracture surface of base of attached microphyll showing a layer of elongated palisade parenchyma above the leaf trace, x 220. 7, tracheids of leaf trace with scalariform to spiral thickenings of the cell wall, x 900. 8, leaf trace of attached microphyll consisting of approximately 30 tracheids, x 350. PLATE 12 ROWE, Selaginellites 80 PALAEONTOLOGY, VOLUME 31 SYSTEMATIC PALAEONTOLOGY Class LYCOPSIDA Order selaginellales Family selaginellaceae Genus selaginellites Zeiller (1906) Type species. Selaginellites suissei Zeiller (1906), Stephanian of Blanzy, France. Selaginellites resimus sp. nov. Plate 10, figs. 1-3; Plate 11, figs. 1-7; Plate 12, figs. 1-8; text-fig. 1a and b; text-fig. 2a-e; text-fig. 3a-f; text-fig. 4a-c Derivation of name. Latin resimus , bent backwards or upwards, referring to the upturned portion of the axis containing the strobilus. Holotype. V.62308, text-fig. 4b-f. Locality. Hazel Hill quarry, Puddlebrook, near Drybrook, Forest of Dean, Gloucestershire, Great Britain. Horizon. Shale band in the Drybrook Sandstone formation, Lower Carboniferous (Asbian-Holkerian). Diagnosis. Leafy shoots at least 5 cm long, 1 -5-2-0 mm in diameter, remaining narrow at all levels of branching. Microphylls deltoid with attenuated distal tips, 3-2-4- 5 mm long and 1 -8-2-0 mm wide, broadest at a point a third of the distance from the base, and arranged in four vertical orthostichies with a strict opposite and decussate arrangement. Stomata 35-38 gm long and 28-38 gm wide, frequency 340-380 per mm2. Stomatal aperture 16 gm long by 6-8 / e (o ® — <5 — — ACRITARCH SPECIMENS PER SAMPLE ® 5-19 irair© O 20-100 common • >100 very common GR APT OLITE HORIZON AGE Arenig Adelograptus, Clomograptus & Kiaerograptus earliest Arenig or latest T remadoc Psigraptus probably late Tremadoc T remadoc est.36m below Psigraptus & 160m above Rh. parabola (excluding early Tremadoc) 5 T 4 3 34 2 1 XIA 3 32 105, 1 51 104 MUX ER 10 o o text-fig. 2. Acritarch distribution at Erdaopuzi, Xiaoliaohuangdi, and Muxiantougou sections. Dimensions (based on eighty specimens). Vesicle diameter: 22 (45) 64 /on; length of processes: 2-0-7 0 ^m; vesicle wall thickness: 03-0-5 /an. Discussion. Playford and Martin (1984) showed that in the Ordovician assemblages from Western Australia the variation of R. palmata overlaps with that of R. pilata, the latter species being distinguished from the former by its more numerous, shorter processes and by the invariably psilate external surface of the vesicle. Although the specimens from Jilin Province sometimes show a scabrate to pilate ornamentation and, contrary to the original generic diagnosis, processes on the operculum were observed in 20% of cases, nevertheless they are attributed to R. pilata because of their large number of processes. The length of the latter is more constant than in the Australian specimens and is closer to the limit of variability between the two species. STRATIGRAPHIC AND PALAEOGEOGR APHICAL SIGNIFICANCE OF THE ACRITARCHS Bearing in mind the limitations of the present sampling, the most significant change in distribution of the acritarchs (text-fig. 2) appears to occur at the youngest level sampled at Muxiantougou (samples MUX-15 and MUX-105), 5-5 m below the first Adelograptus-Clonograptus , with Kiaerograptus , horizon. Aryballomorpha grootaertii and L. erdaopuziana were not found above this level, but R. pilata and R. cf. R. palmata enter and continue higher. A tentative limit between a probably late Tremadoc 124 PALAEONTOLOGY, VOLUME 31 age and a latest Tremadoc or earliest Arenig age is consequently proposed at the level where Rhopaliophora appears. (а) Aryballomorpha grootaertii Common in every sample from levels in or near the probably late Tremadoc Psigraptus horizons at Erdaopuzi, it is common also at Muxiantougou from about 36 0 m below the previous graptolite index level attributed to the Tremadoc (excluding early Tremadoc) to 5-5 m below the Adelograptus- Clonograptus , with Kiaerograptus , horizon, which is of latest Tremadoc or earliest Arenig age. Preliminary investigations in the southern Canadian Rocky Mountains, Alberta, indicate that Aryballomorpha grootaertii is variably abundant (Martin 1984) in the middle member of the Survey Peak Formation: these rocks formed part of Laurentia and their trilobites do not exhibit any clear affinities with those of north-east China (Zhou and Fortey 1986; W. T. Dean, pers. comm.). No graptolites have been recorded from these strata, which were assigned (Dean in Dean and Martin 1982) to trilobite letter zones D and E, proposed by Ross (1949, 1951) in the Tremadoc of Utah and Nevada, USA. Ross et al. (1982) correlated zones D and E with approximately the middle part of the Adelograptus tenellus-A. hunnebergensis fauna. According to Cooper (1979, p. 17) these two species occur in the Tremadoc Series of Shropshire, in the Transition Beds and in the overlying A. tenellus Biozone. The top of this fauna is succeeded (Ross et al. 1982) by the Tetragraptus approximatus Biozone, which conventionally marks the base of the Arenig Series in Great Britain, although its presence in the type Arenig area has not been proved (Fortey in Whittington et al. 1984, p. 21). According to Ross et al. (1982) letter zones D and E are equated in the Tremadoc with the uppermost part of the Cordylodus angulatus Biozone and the lowermost part of the Paltodus ( Drepanoistodus ) deltifer Biozone, in terms of the North Atlantic conodont biozones. (б) Lua erdaopuziana Usually common at Erdaopuzi and Muxiantougou in deposits in or near the Psigraptus horizons, at the former section it appears 5 8 m below the first of these horizons. At the latter section it is recognized from the Psigraptus horizon up to 5-5 m below the first Adelograptus-Clonograptus , with Kiaerograptus, horizon. Like Athabascaella penika , it is at present unknown elsewhere and both species are provisionally considered to be endemic to the northern border of the Sino-Korean platform from, probably, late Tremadoc to earliest Arenig. (c) Athabascaella penika, A. playfordii, and A. rossii All three are often present, though in variable abundance, at Erdaopuzi, Xiaoliaohuangdi, and Muxiantougou. At the last-named locality their lowest occurrence is 36 0 m below the Psigraptus level; their highest record is 100 cm to 2-0 m above the Adelograptus-Clonograptus, with Kiaerograptus, horizon at Xiaoliaohuangdi. Athabascaella playfordii and A. rossii, together with Aryballomorpha grootaertii, are known (Martin 1984) from the middle member of the Survey Peak Formation in Alberta, in strata without macrofossils located between letter zones D and E. (i d ) Rhopaliophora pilata and R. cf. R. palmata R. pilata is common at a level 5-5 m below the Adelograptus-Clonograptus, with Kiaerograptus, horizon at Muxiantougou; at Xiaoliaohuangdi it occurs in abundance from 3 0 m below the same graptolite horizon to 2-0 m above it. In the same samples from both sections R. cf. R. palmata is abundant or rare; its apparent scarcity may be due to selective destruction of the very delicate processes. In the Canning Basin, Western Australia (Combaz and Peniguel 1972; Playford and Martin 1984), R. pilata and R. palmata are usually common in the Willara, Goldwyer, and Nita formations; they are unrecorded below the upper Arenig but extend into the Llanvirn (or ?Llandeilo). Specimens identical with R. cf. R. palmata are present (Li, 1987; Martin, pers. obs.) in the Meitan Formation in northern Guizhou Province, south-west China, strata that formed part of the Yangtze MARTIN AND YIN: ORDOVICIAN ACRITARCHS 125 Platform (Huang et al. 1977) and are correlated with the upper part of the Didymograptus extensus Biozone, lower Arenig. Zhou and Fortey (1986, p. 169) emphasized that the greatest similarity between the trilobite faunas of the Sino-Korean and Yangtze platforms occurred in the Tremadoc and ended after the beginning of the Arenig. Rare fragments of chitinozoans found in all samples from Xiaoliaohuangdi, including that (XT A- 1 ) from 3 0 m below the Adelograptus-Clonograptus, with Kiaerograptus, horizon, do not at present provide any evidence for defining a boundary between probably late Tremadoc and latest Tremadoc or earliest Arenig deposits in the Hunjiang area. This group is absent from the Tremadoc of Quebec, Canada (Achab 1986), reported rarely from the Tremadoc in the Sahara (Combaz 1968; Poumot 1968), perhaps present in the Tremadoc of the Russian Platform (Umnova 1969), and is found in Hunneberg strata in Estonia (Grahn 1984). It appears to be particularly characteristic from the Arenig onwards, but this may reflect the greater number of papers devoted to Arenig and younger chitinozoans. Acknowledgements. We are indebted to Lin Yaokun and Zhang Junming (Nanjing Institute of Geology and Palaeontology, Academia Sinica), W. T. Dean (University College, Cardiff, UK), and B.-D. Erdtmann (Institut fiir Geologie und Palaontologie, Technische Universitat, Berlin, BRD) for permitting use of their personal communications concerning macrofossils and field observations in the Hunjiang area. C. R. Barnes (Memorial University of Newfoundland, Canada) commented on the age of a conodont sample from Xiaoliaohuangdi, and G. Playford (University of Queensland, Brisbane, Australia) read critically the manuscript. Li Jun (NIGPAS) loaned F. M. a palynological slide from Guizhou Province and made available information in press. At the IRScNB Brussels H. De Potter is thanked for processing the acritarchs, J. Cillis for operating the SEM, and G. Van der Veken for printing the photographs. REFERENCES achab, a. 1986. Assemblages de chitinozoaires dans l’Ordovicien inferieur de Test du Canada. Can. J. Earth Sci. 23, 682-695. barnes, c. R., norford, b. s. and skevington, d. 1981. The Ordovician System in Canada. Correlation chart and explanatory notes. Pub!. Intern. Union Geol. Sci. 8, 1-27. bergstrom, s. m. 1977. Early Paleozoic conodont biostratigraphy in the Atlantic Borderlands, 85-110. In swain, f. m. (ed.). Stratigraphic micropaleontology of Atlantic Basin and Borderlands. Elsevier, Amsterdam. Bradshaw, L. 1969. Conodonts from the Fort Pena Formation (Middle Ordovician), Marathon Basin, Texas. J. Paleont. 43, 1137-1168. bulman, o. m. b. 1954. Graptolite fauna of the Dictyonema shales of the Oslo region. Norsk Geol. Tidsskr. 33, 1-40. Chen junyuan (ed.). 1986. Aspects of Cambrian- Ordovician boundary in Dayangcha, China , 410 pp. China Prospect Publishing House, Beijing. — QIAN YIYUAN, LIN YAOKUN, ZHANG JUNMING, WANG ZHIHAO, YIN LEIMING and ERDTMANN, B.-D. 1985. Study on Cambrian-Ordovician boundary strata and its biota in Dayangcha , Hunjiang, Jilin , China, 138 pp. China Prospect Publishing House, Beijing. — teichert, c., zhou zhiyi, lin yaokun, wang zhihao and xu juntao. 1983. Faunal sequence across the Cambrian-Ordovician boundary in northern China and its international correlation. Geol. Palaeont. 17, 1-15. — ZHOU ZHIYI, LIN YAOKUN, WANG ZHIHAO, XU JUNTAO, ZHANG JINLIN, PENG YIJING and SHI XINZENG. 1985. The Ordovician of southern Jilin. J. Strat. 9, 21-27. [In Chinese.] combaz, a. 1968. Un microbios du Tremadocien dans un sondage d’Hassi-Messaoud. Act. Soc. linn. Bordeaux, 104B (1967), (29), 1-26. — and peniguel, G. 1972. Etude palynostratigraphique de l’Ordovicien dans quelques sondages du Bassin de Canning (Australie Occidentale). Bull. Cent. Rech. Pan, 6, 121-167. cooper, r. a. 1979. Sequence and correlation of Tremadoc graptolite assemblages. Alcheringa , 3, 7-19. correia, m. 1967. Relations possibles entre l’etat de conservation des elements figures de la matiere organique (microfossiles palyno-planctologiques) et l’existence de gisements d'hydrocarbures. Revue Inst. fr. Petrole, 22, 1285-1306. cowie, j. w., rushton, a. w. a. and Stubblefield, c. j. 1972. A correlation of Cambrian rocks in the British Isles. Geol. Soc. Lond. Spec. Rept. 2, 1-42. 126 PALAEONTOLOGY, VOLUME 31 dean, w. t. and martin, f. 1982. The sequence of trilobite faunas and acritarch microfloras at the Cambrian-Ordovician boundary, Wilcox Pass, Alberta, Canada, 131-140. In bassett, m. g. and dean, w. t. (eds.). The Cambrian-Ordovician boundary: sections, fossil distributions, and correlations. Nat. Mus. Wales, Geol. Ser. (3), Cardiff. deunff, j. 1955. Aremoricanium , genre nouveau d’hystrichospheres du Silurien breton. C. R. somm. Seanc. Soc. geol. Fr. (12), 227-229. and evitt, w. R. 1968. Tunisphaeridium, a new acritarch genus from the Silurian and Devonian. Stanf. Univ. Pubis. Geol. Sci. 12, 1-13. downie, c. 1973. Observations on the nature of acritarchs. Palaeontology, 16, 239-259. duan, jiye, an sulan and zhao da. 1986. Cambrian-Ordovician boundary and its interval biotas, southern Jilin, northeast China, 135 pp. Changchun College of Geology, Changchun. eichwald, e. j. 1840. U eber das silurische Schichtensystem in Esthland, 210 pp. St. Petersburg, [not seen.] eisenack, a. 1967. Axisphaeridium n.g., eine axialsymmetrische Hystrichosphare aus dem baltischen Ordovizium. Neues Jb. Geol. Palaont. Mh. 7, 398-400. erdtmann, b.-d. 1982. A reorganization and proposed phylogenic classification of planktic Tremadoc (early Ordovician) dendroid graptolites. Norsk geol. Tidsskr. 62, 121 144. — 1986. Review of lithofacies and graptolite-based biofacies of three critical Cambrian-Ordovician boundary stratotype sections, 374-391. In chen junyuan (ed.). Aspects of Cambrian-Ordovician boundary in Dayangcha , China. China Prospect Publishing House, Beijing. grahn, y. 1984. Ordovician chitinozoa from Tallinn, northern Estonia. Rev. Palaeobot. Palynol. 43, 5-31. huang, t. k., jen chishun, jiang chunfa, chang chimeng and xu ZHiQiN. 1977. An outline of the tectonic characteristics of China. Acta Geol. Sin. (2), 117-135. [In Chinese with English abstracts.] jackson, d. e. 1967. Psigraptus, a new graptolite genus from the Tremadoc of Yukon, Canada. Geol. Mag. 104, 317-321. 1974. Tremadoc graptolites from Yukon Territory, Canada. In rickards, r. b., jackson, d. e. and hughes, c. p. (eds.). Graptolite studies in honour of O. M. B. Bulman. Spec. Pap. Palaeontology, 13, 35-58. jacobson, s. R. and achab, a. 1985. Acritarch biostratigraphy of the Dicellograptus complanatus graptolite zone from the Vaureal Formation (Ashgillian), Anticosti Island, Quebec, Canada. Palynology, 9, 165-198. KUO hungchin, duan jiye and an sulan. 1982. Cambrian-Ordovician boundary in north China platform with descriptions of trilobites. Pap. Fourth Intern. Symp. Ordovician, 1982, 31 pp. Dept. Geol., Changchun College of Geology, Changchun. li jun. 1987. Ordovician acritarchs from the Meitan Formation of Guizhou Province, south-west China. Palaeontology, 30, 613-634. lin yaokun. 1986. A new planktonic graptolite fauna, 224-254. In chen junyuan (ed.). Aspects of Cambrian- Ordovician boundary in Dayangcha, China. China Prospect Publishing House, Beijing. lindstrom, m. 1971. Ordovician conodont-bearing sections in Sweden. A field trip guidebook. Symposium on conodont taxonomy. Marburg/Lahn, 4-18 Sept. 1971. linnarsson, j. g. c. 1871. Om nogra Forsteningar fran Sveriges och Norges Primordialzone. Ofvers. K. VetenskAkad. Forh. Stockh. 26 (6). [not seen.] loeblich, a. r. jr. and MacADAM, r. b. 1971. North American species of the Ordovician acritarch genus Aremoricanium. Palaeontographica, 8135, 41-47. martin, f. 1984. New Ordovician (Tremadoc) acritarch taxa from the middle member of the Survey Peak Formation at Wilcox Pass, southern Canadian Rocky Mountains, Alberta. Geol. Surv. Can., Curr. Res. A, Pap. 84-1 A, 441-448. playford, g. and martin, f. 1984. Ordovician acritarchs from the Canning Basin, Western Australia. Alcheringa, 8, 187-223. poumot, c. 1968. Amphorachitina, Ollachitina, Velachitina\ trois nouveaux genres de chitinozoaires de l’Erg oriental (Algerie-Tunisie). Bull. Cent. Rech. Pau, 2, 45-55. rickards, R. B. and st ait, b. A. 1984. Psigraptus, its classification, evolution and zooid. Alcheringa, 8, 101 111. ross, r. j. jr. 1949. Stratigraphy and trilobite faunal zones of the Garden City Formation, northeastern Utah. Am. J. Sci. 247, 472-491. 1951. Stratigraphy of the Garden City Formation in northeastern Utah, and its trilobite faunas. Bull. Peabody Mus. nat. Hist. 6, vi+ 161 pp. — ADLER, F. J., AMSDEN, T. W., BERGSTROM, D., BERGSTROM, S. M., CARTER, C., CHURKIN, M., CRESSMAN, E. A., DERBY, J. R., DUTRO, J. T. JR., ETHINGTON, R. L., FINNEY, S. C., FISCHER, D. W., FISCHER, J. H., HARRIS, A. G., HINTZE, L. F., KETNER, K. B., KOLATA, D. L., LANDING, E., NEUMAN, R. B., SWEET, W. C., POJETA, J. JR., POTTER, A. W., RADER, e. k., repetski, j. e., shaver, r. h., Thompson, t. l. and webers, G. f. 1982. The Ordovician System in the United States. Correlation chart and explanatory notes. Publ. Intern. Union Geol. Sci. 12, 1-73. MARTIN AND YIN: ORDOVICIAN ACRITARCHS 127 Sergeeva, s. p. 1963. Conodonts from the Lower Ordovician of the Leningrad region. Paleont. Zh. 2, 93-108. [In Russian.] Stubblefield, c. J. and bulman, o. m. B. 1927. The Shineton Shales of the Wrekin District: with notes on their development in other parts of Shropshire and Herefordshire. Q. J. geol. Soc. Loud. 83, 96-146. tjernvik, T. E. and johansson, J. v. 1980. Description of the upper portion of the drill-core from Finngrundet in the South Bothnian Bay. Bull. geol. Instn. Univ. Uppsala , ns 8, 173-204. umnova, n. i. 1969. Distribution of Chitinozoa in the Ordovician of the Russian Platform. Paleont. Zh. 3, 45-62. [In Russian.] wang, xiaofeng and erdtmann, b.-d. 1986. The earliest Ordovician graptolite sequence from Hunjiang, Jilin Province, China. Acta Geol. Sin. 3, 226-236. [In Chinese with English abstract.] WEBBY, B. D., VANDENBERG, A. H. M., COOPER, R. A., BANKS, M. R., BURRETT, C. F., HENDERSON, R. A., CLARKSON, P. D., hughes, c. p., laurie, j., stait, B., Thomson, m. r. a. and webers, G. f. 1981. The Ordovician System in Australia, New Zealand and Antarctica. Correlation chart and explanatory notes. Publ. Intern. Union geol. Sci. 6, 1-64. WHITTINGTON, H. B., DEAN, W. T., FORTEY, R. A., RICKARDS, R. B., RUSHTON, A. W. A. and WRIGHT, A. D. 1984. Definition of the Tremadoc Series and the series of the Ordovician System in Britain. Geol. Mag. 121, 17-23. yin leiming. 1985. Acritarchs, 101-112. In chbn junyuan, qian yiyuan et al. Study on Cambrian-Ordovician boundary and its biota in Dayangcha , Hunjiang, Jilin, China. China Prospect Publishing House, Beijing. — 1986. Acritarchs, 314-373. In chen junyuan (ed.). Aspects of Cambrian-Ordovician boundary in Dayangcha, China. China Prospect Publishing House, Beijing. zhang junming. 1986. Lithofacies sequence, 15-34. In chen junyuan (ed.). Ibid. zhang, z. M., liou, j. G. and Coleman, R. G. 1984. An outline of the plate tectonics of China. Bull. Geol. Soc. Ain. 95, 295-312. zhao xianglin and zhang shunxin. 1985. Reclined graptolites of the Xinchangian. Acta Changchun College Geol. 2, 13-24. [In Chinese with English abstract.] zhou zhiyi and fortey, r. a. 1986. Ordovician trilobites from north and northeast China. Palaeontographica, A 192, 157-210. F. MARTIN Institut Royal des Sciences Naturelles de Belgique Departement de Paleontologie rue Vautier, 29 B-1040 Bruxelles Belgium Typescript received 25 January 1987 Revised typescript received 6 April 1987 yin leiming Institute of Geology and Palaeontology Academia Sinica Chi-Ming-Ssu Nanjing People's Republic of China PALAEOCENE AND EOCENE MIXODONTIA (MAMMALIA, GL1RES) OF MONGOLIA AND CHINA by DEMBERLIYN DASHZEVEG and DONALD E. RUSSELL Abstract. Three new mixodont genera and species from the late Palaeocene of Mongolia, Khaychina elongata , Eomylus zhigdenensis , and Amar aleator are described; ‘ Mimotona' borealis (late Palaeocene of China) is referred to Eomylus. Also described is Mimotona lii n. sp. from the middle Palaeocene of China. The dentition of Gomphos elkema is more fully documented by new material, as is that of Rhombomylus and Eurymylus. Gomphos, Rhombomylus, and ‘ Matutinia ’ are compared and ‘M.’ nitidulus referred to Rhombomylus. The additional specimens and taxa contribute to our knowledge of a group that could be ancestral to rodents and lagomorphs. Mixodonts are frequently cited when the subject of the ancestry of rodents and lagomorphs is discussed (Sych 1971; Li 1977). Although this possibility has not received unanimous acclaim (Butler 1985), it is our belief that when the diversity of mixodonts is better known the transition between them and probably both rodents and lagomorphs will be demonstrable. This paper deals with new material that increases by 50% the number of described taxa in the group. The description of new mixodonts from the early Palaeogene of Mongolia has necessitated a review of the forms similarly endowed with a gliriform adaptation and with comparable cheek tooth morphology. "We have tried to provide here an essentially complete photographic documentation of the Mixodontia, not all of the previously described forms having been adequately illustrated. This paper is part of a series dealing with new discoveries in the late Palaeocene and early Eocene of Mongolia. Recent descriptions of the biostratigraphy and new faunal elements were made by Dashzeveg (1982a), Russell and Dashzeveg (1986), and Dashzeveg, Russell, and Flynn (1987). The Mongolian Mixodontia described in this article come from well-studied beds (see, for example, Badamgarav and Reshetov 1985) in three intermontane depressions in the southern part of the country, the Nemegt Basin, the Ulan-Nur Basin, and the Bugin-Tsav Basin (text-fig. 1). In the Nemegt Basin mixodonts and other vertebrates were collected from four quarries situated at different stratigraphic levels within the Naran-Bulak Beds, not taking into account surface finds (text-fig. 2). I. Tsagan-Khushu locality. Quarry 3, Zhigden Member: this new site, discovered (by D. D.) in 1984, is 300 m to the north of the classic Quarry 1. The bone-bearing lens of red sandy clay has a thickness of about 1 m and is situated in the upper part of the Zhigden Member. Jaws and teeth of mixodonts dominate in the assemblage; the new genera and species Eomylus zhigdenensis and Amar aleator were among these. The notoungulate Arctostylops macrodon and the tillodont Ernanodon sp. also occur here. II. Naran-Bulak locality. Quarry 1, Naran Member (alluvial deposits): this site was found near the landmark termed the North Sphinx during the Mongolian Palaeontological Expedition of the Academy of Sciences of the Soviet Union in 1948; its vertebrates are the most studied and its position at the base of the Naran Member is well established. The fossiliferous sediments are white quartz sands with intercalated lenses of gravel. Prodinoceras, Archaeolamda , and Pachyaena are the most important elements(see Flerov 1952, 1957; Kielan-Jaworowska 1968/1969; Dashzeveg 1976, 19826); a specimen referable to Eomylus zhigdenensis n. gen. n. sp. has also been found here. IPalaeontology, Vol. 31, Part 1, 1988, pp. 129 164.| © The Palaeontological Association 130 PALAEONTOLOGY, VOLUME 31 text-fig. 1. Sketch map of the south-central part of Mongolia showing the situation of 1, the Ulan-Nur Basin with the Gashato locality; 2, the Nemegt Basin with the localities of Naran-Bulak and Tsagan-Khushu; 3, the Bugin-Tsav Basin with the Khaychin-Ula locality. Modified from Gradzinski et al. 1968/1969. In the upper part of the white sands the sediment becomes more argillaceous and has furnished rare remains of small mammals: Arctostylops , Pseudictops , and the mixodont Eurymylus. III. Naran-Bulak locality. Quarry 2, Naran Member (lacustrine deposits): this richly fossiliferous site was discovered (by D. D.) during the Polish Palaeontological Expedition of 1964-1965 and is situated about 250 m east of the Naran-Bulak Quarry 1. The white sands and sandstones have yielded abundant remains of Arctostylopidae, Pseudictopidae, and mixodonts. The first author has also collected material referable to Prionessus , Oxyaena , ‘ Sinopa , and Dissacus (Dashzeveg 1977) and worked by screen-washing. text-fig. 2. Composite section of the Naran- Bulak Beds in the Nemegt Basin showing the stratigraphic position of the fossiliferous quarries, a, alluvial deposits, b , lacustrine deposits. DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 131 IV. Tsagan-Khushu locality. Quarry 1, Bumban Member: an extensive collection of fossil mammals was obtained here (by D. D.) by screen-washing. The fossils occur in lenses of sandy gravel, above the green clays of the Naran Member, which are 10-30 cm thick and contain numerous remains of small vertebrates. The material, referable to insectivores, primates, condylarths, and perissodactyls, as well as mixodonts, has been discussed in Dashzeveg ( 1 977, 1 979n, b), in Oashzeveg and McKenna (1977) and in Russell and Dashzeveg (1986). NEMEGT BASIN BUDIN-TSAV BASIN IS) a CQ 2*: c BUGIN 9-- z < oc < z Mbr. NEMEGT Fm. Khay china elongata AGUYT Mbr. BUMBAN Mbr. NARAN Mbr. ZHIGDEN Mbr. NEMEGT Fm. ULAN-NUR BASIN .Gomphos sp. Gomphos elkema^ Rhombomylus sp Eurymylus laficeps S \ Eomylus zhigdenensis S N .Eomylus zhigdenensis Amar aleator^ ? ^ Mbr.. HI IS) CD LU CD Mbr. IT f— nz IS) c zx: Mbr.I v: NEMEGT7 Fm. Gomphos elkema Gomphos elkema Gashaf o Fauna text-fig. 3. Biostratigraphic correlation between the Ulan-Nur Basin, the Nemegt Basin, and the Bugin-Tsav Basin. V. Tsagan-Khushu locality. Quarry 2, Bumban Member: this site, near the boundary between the Naran and Bumban Members, is 300 m south of Quarry 1. Brown sandstone and gravel lenses, 20 cm to 1 m in thickness, have produced a rich mammalian assemblage very similar to that from Quarry 1. As concerns mixodonts, Rhombomylus and Gomphos are known from here. Gomphos has also been discovered, as a surface find, on the southern side of Tsagan-Khushu. In the south-eastern part of the Ulan-Nur Basin is situated the important locality of Gashato. Its fauna and stratigraphy have been much discussed. According to the field research of the first author (D. D.) the mixodont E. laticeps comes from the red clay of the first member of the Khashat Beds (or Svita, in Soviet and Mongolian usage); it is found together with Arctostylops macrodon , A. iturus , and Prionessus lucifer. The fauna of the second and third members of the Khashat Beds is not yet known, except for G. elkema. Remains of the latter have also been found at the base of the third member. The material of Gomphos collected (by D. D. in 1978) from the second member is illustrated here (text-fig. 24). The known Gashato fauna of the first member is correlated with the lacustrine deposits of the Naran Member of the Naran-Bulak Beds in the Nemegt Basin and is considered to be late Palaeocene in age. Recently, G. elkema has been found in the Bumban Member of the Naran-Bulak Beds, which 132 PALAEONTOLOGY, VOLUME 31 are otherwise well dated as early Eocene by such forms as Hyopsodus, Homogalax, and numerous rodents. This implies that the second and third members of the Khashat Beds are approximate age equivalents of the upper part of the Naran-Bulak Beds. In the Bugin-Tsav Basin Palaeocene and Eocene deposits are exposed along its southern side where they form a series of cliffs extending over a distance of 20 km. The late Palaeocene sediments, named the Bugin Member by the first author, crop out in the region of Khaychin-Ula II (Khaychin I, according to Badamgarav and Reshetov 1985), and consist of light grey or greenish grey deposits of sandstones, clay, and gravelites unconformably overlying the late Cretaceous Nemegt Formation. The mixodont Khaychina elongata n. gen., n. sp. was found in the middle part of the Bugin Member, where Prodinoceras sp. and Archaeolambda trofimovi were also discovered. Based on the presence of the latter two taxa, the Bugin Member can be correlated with the Naran Member of the Naran-Bulak Beds in the Nemegt Basin. ABBREVIATIONS IVPP. Institute of Vertebrate Palaeontology and Palaeoanthropology, Beijing, Chinese People’s Republic. MgM. Prefix referring to collections in the Institute of Palaeobiology, Warsaw, Poland. PSS. Prefix referring to the collections of the Palaeontology and Stratigraphy Section, Institute of Geology, Academy of Sciences of the Mongolian People’s Republic, Ulan Bator. V. Prefix referring to the collections of the IVPP. Measurements are in millimetres. SYSTEMATIC PALAEONTOLOGY The Mixodontia were for many years confused with and included in the order Anagalida. We refer to the Mixodontia only those forms characterized by enlarged, apparently ever-growing incisors, a diastema, premolar loss, and a relatively short, deep mandible, in contrast to members of the Anagalida which have small incisors, little or no diastema, unreduced premolars, and an elongate, shallow mandible. This article is essentially a revision of the Mixodontia viewed in this concept. For the first described mixodont, Eurymylus, the family name Eurymylidae was proposed (Matthew et al. 1929). Nearly fifty years later, discoveries in China produced taxa that Li (1977) placed in the new family Mimotonidae. At the time he remarked that the family was tentatively established for the new forms, a fact that was noted by Bleefeld and McKenna (1985); the latter, basing their opinion on article 15 of the International Code of Zoological Nomenclature, declared the name not available. Since the name has been widely used for the past ten years, taxonomic stability is not particularly well served by this decision. Also, Dr Li (in Li & Ting 1985, p. 44) has referred to ‘the new family Mimotonidae’ that he created in 1977, obviously eliminating the tentativeness that was previously expressed. To date, the Eurymylidae and the Mimotonidae are the only recognized mixodont families. Cohort glires Linnaeus, 1758 Order mixodontia Sych 1971 Family eurymylidae Matthew, Granger and Simpson, 1929 Subfamily eurymylinae new Eomylus n. gen. Type species. Eomylus zhigdenensis n. sp. Referred species. E. borealis (Chow & Qi 1978). Age and distribution. Late Palaeocene, Mongolian People’s Republic and Chinese People’s Republic. Diagnosis. Dental formula: 7-0-2-3/1 7-0-2-3. Differs from Rhombomylus , Matutinia, Mimotona , Heomys, Eurymylus , Amur (n. gen.), and Gompbos by its short (anteroposteriorly), transversely DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 133 elongate upper molars and by the lack of expansion of the hypoconal shelf, and from all but Mimotona and Amar by the position of the hypocone lingual with respect to the protocone. Differs from Rhombomylus, Matutinia , Mimotona , and Heomys by the absence of a lingual groove extending vertically between the protocone and hypocone. Differs from Heomys by a greater degree of uni- lateral hypsodonty (enamel of crown low on one side of a tooth and higher on the other) and from Gomphos and Amar by the absence of a parastyle. Differs from Heomys and Eurymylus by the absence of a dorsoanteriorly oriented flange on the anterior facial surface of the jugal arch (but is similar to the condition in Mimotona; unknown in Rhombomylus , Gomphos , Amar, and Matutinia). It should be noted that the hypocone here is lingual only in the unworn condition. With considerable wear the protocone would become the more lingual. Also, concerning transverse elongation, since the upper molars in several of the enumerated genera tend to be pyramidal and to wear rapidly, the teeth in an older individual have an aspect of greater transverse elongation than do little or unworn teeth. This is well illustrated in Eurymylus where the specimens figured by Sych (1971) look anteroposteriorly short and wide transversely, but in the unworn specimen, PSS 20-162, figured here (text-fig. 5) the occlusal surface is subquadrate or even elongate anteroposteriorly. Etymology. Eos (Gk.), dawn; mylos (Gk.), grinder, and by extension, molar. Eomylus zhigdenensis n. sp. Text-figs. 4, 6, 7 Holotype. PSS 20-139, right maxillary fragment with M1/-M2/. Referred specimens. PSS 20-137, mandibular fragment with left M/1 -M/3; PSS 20-138, right mandibular fragment with M/1 -M/2. Locality and stratigraphic distribution. Tsagan-Khushu locality, Nemegt Basin, southern MPR; Zhigden Member of the Naran-Bulak Beds; late Palaeocene. Questionably referred specimen. PSS 20-133, left mandible with P/4-M/3, from the Naran-Bulak locality, Nemegt Basin, MPR; Naran Member of the Naran-Bulak Beds; late Palaeocene. All specimens are from the collections of the Department of Stratigraphy and Palaeontology, Institute of Geology, Academy of Sciences of the MPR, Ulan Bator. Diagnosis. As for genus, only species. Etymology. Zhigden (Mong.), gooseberry; from the name of the geologic member in which the type and referred specimens were found. Description. The type specimen, PSS 20-139 (text-fig. 4 a-c), is a maxillary fragment retaining only M 1/ and M2/, but the teeth are well preserved and very little worn. Remnants of anterior alveoli indicate that P4/ and P3/ were present, as was M3/. The molar teeth are notably narrow and transversely elongate. In posterior view, the protocone is of subequal height to the two labial cusps, but the hypoconal shelf, or posterior cingulum, is situated well below the summit of the protocone; no hypocone is present and the shelf is narrow. The pre- and postprotocristae form sharp crests extending between the protocone and the bases of the paracone and metacone; the preprotocrista is low just anterior to the protocone, forming an opening to the trigon basin, but the postprotocrista is high and strong throughout its length. Transverse chewing movements obviously followed the axis of this crest. A weak paraconule is present close to the lingual base of the paracone; the metaconule is more strongly developed and is winged, with an anterior crest curving to the lingual base of the metacone and a posterior crest extending to the posterior base of this cusp. There are no stylar cusps and no anterior cingulum. Labially, the crown height is very low, while lingually it is at least three times as high, thus manifesting unilateral hypsodonty. The lingual base of the molar teeth extends far beyond the level of the protocone, situating the latter at about 1/3 of the tooth’s width from its lingual border. Despite a marked curvature (in occlusal view) of the maxillary bordering the alveoli the molar teeth appear to have been aligned in a rather straight row and to have been all about the same size; the premolars are situated more lingually. On the basis of accordance in size and identical provenance, two partial mandibles are referred to this species, PSS 20-137, with the left M/1 - M/3 (text-fig. 6) and PSS 20-138, with the right M/1 -M/2 (text-fig. 4 d-f). Occlusion between the unworn right maxillary and the heavily worn right mandible does not give very satisfactory results, 134 PALAEONTOLOGY, VOLUME 31 text-fig. 4. Eomylus zhigdenensis n. gen., n. sp. a-c, PSS 20-139, holotype, right maxillary with M1/-M2/. a, occlusal view; £>, lingual view; c, labial view, d-f PSS 20-138, right mandibular fragment with M/l-M/2. d, occlusal view; e, labial view;/, lingual view. All views x 15; both specimens from the Zhigden Member of the Naran-Bulak Beds at Quarry III, Tsagan-Khushu, Mongolia; late Palaeocene. but does not entirely negate the possibility of their being conspecific. Most of the features of PSS 20-138 have been eliminated to the point of the two molars being represented only by the elongate oval surface of the trigonid and the larger (and wider) surface of the talonid. The posterior wall of the trigonid remains, however, and indicates the absence of propalinal movement. The incisor is preserved, extending posteriorly to below M/3; it is not known if a second incisor was present. From the curvature of the remaining part of the incisor and the position of the premolar alveoli it would seem that the diastema was very short. If PSS 20-138 is truly referable to Eomylus zhigdenensis it lends credence to the attribution of the latter to the DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 135 text-fig. 5. Eurymylus laticeps Matthew and Granger, 1925. a-c, PSS 20-162, right maxillary with DP4/-M3/. a, labial view; b , occlusal view; c, lingual view; note the definitive P4/ erupting below the DP4/. All views x 15. Specimen from the Naran Member of the Naran-Bulak Beds at Naran-Bulak locality, Mongolia; late Palaeocene. Mixodontia by the posterior extension of the incisor and by the apparent shortness of the mandible measured between P/3 and the anterior extension of this incisor. If PSS 20-138 and 20-137 can be considered as representative of E. zhigdenensis the latter differs (in its lower dentition) from Rhombomylus , Matutinia , Eurymylus , Gomphos, and Hypsimylus by the wider talonid on M/1 and M/2; from Mimotona (M. wana , M. robusta , and M. lii n. sp.), Gomphos , and Hypsimylus by having less unilateral hypsodonty; from Mimotona (M. wana , M. robusta , and M. lii n. sp.) and Gomphos by the greater separation of the protoconid and hypodonid; from ‘M.’ borealis by the absence of a deep groove between the hypoconulid and entoconid and by smaller size; from the mandible referred to Heomys (V 4322) by the M/1 and M/2 136 PALAEONTOLOGY, VOLUME 31 text-fig. 6. Eomylus zhigdenensis n. gen., n. sp. a-c, PSS 20-137, left mandible with M/l-M/3. a, occlusal view; b, labial view; c, lingual view. All views x 15. Specimen from the Zhigden Member of the Naran-Bulak Beds at Quarry III, Tsagan-Khushu, Mongolia; late Palaeocene. more subquadrate and the M/3 more elongate; and from Zagmys Dashzeveg et al. (1987) by a reduced paraconid. The three molars of PSS 20- 137 (text-fig. 6) are very well preserved and have undergone but little wear. There is no anterior cingulum; the paraconid is absent or perhaps barely indicated and crestiform on M/1. The protoconid, though the first to be reduced by wear (with the hypoconid), was probably nearly as high as the metaconid; a weak anterior crest extends from the protoconid to the anterior side of the metaconid and a posterior median crest of the protoconid is separated from a similar crest of the metaconid by a sharp, deep notch. The metaconid is the dominant cusp and is emphasized by a posterolingual crest extending vertically from its summit to its base. Difference in height of the trigonid with respect to the talonid is not great. On a fresh, unworn tooth the talonid basin is deep, with a groove in the bottom oriented obliquely, passing through the notch between the entoconid and the base of the metaconid and extending towards a notch between the hypoconid and the hypoconulid. With considerable wear, as in PSS 20-138, this obliquity becomes less apparent and the tooth takes the aspect of having undergone, principally, transverse wear. The cristid obliqua is inflated by a more or less crestiform mesoconid and attains the posterior surface of the trigonid below the protoconid-metaconid notch. In little or unworn condition the cristid obliqua is separated from the trigonid by a transverse groove. The hypoconid has a pinched (anteroposteriorly) aspect and extends notably beyond the protoconid in width; in M/3 it is subequal to the latter in width. In M/1 and M/2 the hypoconid is apparently connected by a low crest to the hypoconulid, but in M/3 the two cusps are separate. The hypoconulid is large, centrally located, and transversely elongate; it is possible that a small cuspule existed at its labial extremity (by which a connection was made with the hypoconid in M/1 and M/2), but it is largely removed by wear in the available material. The entoconid is high and situated close to the hypoconulid, to which it is connected by a high crest. In labial view the protoconid and DASH ZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 137 text-fig. 7. Eomylus zhigdenensis n. gen., n. sp. nov. a-c, PSS 20-133, left mandible with P/4-M/3. a, occlusal view; b , labial view; c, lingual view. All views x 1 5. Specimen from the Naran Member of the Naran-Bulak Beds at Quarry I, Naran-Bulak, Mongolia; late Palaeocene. hypoconid are rather widely separated and there is little or no evidence of unilateral hypsodonty. M/3 is the longest of the molars and also the narrowest. The mandible from the Naran Member of the Naran-Bulak Beds, PSS 20-133 (text-fig. 7), closely resembles those from the Zhigden Member; it is more complete but the teeth are considerably more worn than those of PSS 20-137. Its principal differences from the Zhigden specimens appear to be in the proportions of the molar teeth; M/1 and M/2 are slightly smaller and M/3 is notably longer with the talonid exceeding the width of the trigonid as in M/I and M/2. It is quite possible that these small differences reflect only intraspecific variation, although with a larger sample distinction of a form common to the Naran Member might be demonstrable. If the specimen is referable to Eomylus zhigdenensis (and in any case, it is closely related), it provides interesting details lacking in the Zhigden Member mandibles. For one thing, only a single incisor is present. The latter is separated from P/3 (represented only by the roots) by a short, sharply crested diastema. P/4 is well preserved and presents a 138 PALAEONTOLOGY, VOLUME 31 well-developed protoconid and metaconid with the latter being the higher. No paraconid is present but a strong paralophid closes a small trigonid basin anteriorly. The talonid is well basined (but open lingually) and much narrower than the trigonid. Two subequal cusps, aligned transversely, are situated centrally at the posterior extremity of the talonid. From what remains of the coronoid process it can be seen to arise well behind the M/3. L W L W L W L W PSS 20-139 Ml/ 1-4 3-2 M2/ 1-5 3-3 PSS 20-137 M/1 1-8 2-1 M/2 2-2 2-2 M/3 2-6 1-8 PSS 20-138 1-9 2-1 2-1 21 PSS 20-133 P/4 1-8 L5 1-6 20 2-0 20 2-8 20 Discussion. Even though none of the Zhigden specimens display the gliriform character of the anterior part of the dentition that characterizes mixodonts, comparison of the molar teeth provides adequate evidence to ensure referral to this group. Supplementary data from the Naran mandible (PSS 20-133) supports this attribution. The maxillary PSS 20-139 was chosen as type specimen because many of the taxa that have been referred to the Mixodontia are based on type material of the same nature ( Eurymylus , Matutinia, Rhombomylus, Mimotona , and Heomys). The relationships of Eomylus zhigdenensis are considered in detail below. Eomylus borealis (Chow and Qi, 1978) new combination Text-fig. 8 Synonym. Mimotona borealis Chow and Qi, 1978. Holotype. V 5531, right mandibular fragment with P/4-M/2; IVPP, Beijing, China. Referred material. This species is cited at Bayn Ulan, Nei Mongol (Zhai, pers. comm. 1981), but the material is undescribed. Locality and stratigraphic distribution (of the type). Nomogen locality, Nei Mongol, CPR; Nomogen Formation; late Palaeocene. Distinctive characters. Differs from E. zhigdenensis by its greater size and by the presence of a deep groove in the lower molars between the hypoconid and the hypoconulid. Discussion. E. borealis was first referred to the genus Mimotona, but it is immediately distinguished from the two other species, M. wana Li, 1977 and M. robust a Li, 1977, by the presence of only a single incisor. Further comparison shows that it differs fundamentally from these species also in the morphology of its molars. For example, in M. wana the protoconid and hypoconid in labial view form parallel columns that are close together and nearly vertical; in E. borealis these columns are considerably further apart and are oblique. Molar crown height in M. wana is moderate lingually but notably greater labially, representing a degree of unilateral hypsodonty; both sides of the molar crowns in E. borealis are of about the same height. In occlusal view the hypoconulid and entoconid in M. wana are strongly linked, while they are separated by a deep groove in E. borealis. In M. wana the P/3 has a strong protoconid and metaconid and even a rudimentary paraconid; a well-developed (for a P/3) talonid is also present. Of the P/3 in E. borealis only the root is preserved, but it appears too small to have supported the complicated sort of crown seen in M. wana. The teeth of the single specimen of M. robusta are rather heavily worn, but the labial columns formed by the protoconid and hypoconid appear to be more similar to those of M. wana than to those of M. borealis, that is to say, close together and nearly vertical. Like M. wana, M. robusta possesses two (lower, at least) incisors. Probably hypoconulid-entoconid relationships are also as in M. wana (strongly linked), reinforcing the affinity between it and M. wana and distinguishing the species from E. borealis. One can mention in passing that M. robusta differs principally from M. wana by a considerably greater size and by the proportions of P/4. To conclude, the observations enumerated above lead us to propose that the species borealis is better referred to Eomylus than it is to Mimotona. DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 139 text-fig. 8. Eomylus borealis (Chow and Qi, 1978). a-c , V 5531, holotype, right mandible with P/4 M/2, a, occlusal view; b, labial view; c, lingual view. All views x 15. Specimen from the Nomogen Formation at Nomogen, Nei Mongol, China; late Palaeocene. Amar n. gen. Type species. Amar aleator n. sp. Age and distribution. Late Palaeocene, MPR. Diagnosis. Differs from Eurymylus , Rhombomylus, Eomylus , Heomys , and Mimotona by the presence of a labial cingulum and a strong mesostyle on the upper molars, and by the absence of a vertical lingual groove between the protocone and hypocone; differs from Eomylus, Gomphos, and Mimotona by upper molars that are longer and less transversely elongate; differs from Gomphos by a stronger mesostyle and labial cingulum and (also from Mimotona) by less lingual hypsodonty; differs from Rhombomylus, Eurymylus, Gomphos, and Heomys by the position of the hypocone lingual with respect to the protocone. Etymology. Amar (Mong.), peace. Amar aleator n. sp. Holotype. PSS 20-161, right maxillary fragment with M1/-M2/. Locality and stratigraphic distribution. Tsagan-Khushu locality, Quarry 3, Nemegt Basin, southern MPR; Zhigden Member of the Naran-Bulak Beds; late Palaeocene. 140 PALAEONTOLOGY, VOLUME 31 text-fig. 9. Amur aleator n. gen., n. sp. a-c, PSS 20-161, holotype, right maxillary with M 1/-M2/. a , labial view; b, occlusal view; c, lingual view. All views x 15. Specimen from the Zhigden Member of the Naran- Bulak Beds at Quarry III, Tsagan-Khushu, Mongolia; late Palaeocene. DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 141 Diagnosis. As for genus, only species. Etymology. Aleator (L.), dice-player, referring to the uncertainty over its taxonomic position. Description. Only Ml/ and M2/ are preserved in PSS 20-161, but the presence of at least P4/ is indicated by alveoli, and of M3/by a wear facet on the posterior side of M2/and a partial alveolus. The two molars are notably subquadrate in their present state and even with wear would never become exaggeratedly transversely elongate. The strong preprotocrista (in which the presence of a paraconule is suggested by a widening of the crest) is continuous with the labial cingulum that itself joins the posterior cingulum. In M2/the labial cingulum is weak or briefly interrupted opposite the paracone and metacone. A prominent mesostyle is formed between the two cusps; the paracone is the larger of the two; a suggestion of a parastyle is present. A metaconule that is subequal in size to the metacone occurs immediately lingual to the latter. The protocone is as high as the labial cusps and even higher in M2/. The hypocone is situated only slightly below the summit of the protocone and is as far lingual as the latter in Ml/ and even further lingually in M2/. The hypoconal shelf is wide; there is no anterior cingulum. Lingual hypsodonty is well developed. Measurements. L W L W PSS 20-161 Ml/ 3-0 4-0 M2/ 3-3 4-5 Discussion. In the morphology of the labial region (cingulum, mesostyle) of its upper molars Amar resembles Gomphos, although in occlusal view the M1/-M2/ of the latter, with a labially protruding paracone, differ considerably. In lingual hypsodonty the upper molars of Amar do not equal that in Gomphos and little exceed, if any, the condition in Eomylus, Rhombomylus , and Eurymulus. The lingual slope of the molars is steeper than in Gomphos , but more sloping than in Mimotona and Heomys; it is about the same as in Rhombomylus , Eurymyhts and Eomylus. In the upper teeth of Amar we find no characters, not duplicated in the Eurymylidae, that indicate a particular relationship to lagomorphs, such as have been postulated for Mimotona and Gomphos. The similarity to Gomphos in the presence of a labial cingulum and a mesostyle is striking and could suggest affinity, but the upper molars otherwise differ markedly in the position of the hypocone, in the slope and inflation of the lingual part of the teeth, and in degree of hypsodonty. Pending discovery of more complete material, and referable lower jaws, we will consider Amar to be a member of the Eurymylidae and not a mimotonid. No contemporary taxa, known only from the lower dentition, possess teeth that could be compatible with the upper dentition that represents Amar. Family eurymylidae? Subfamily khaychininae new Khaychina n. gen. Type species. Khaychina elongata n. sp. Age and distribution. Late Palaeocene, MPR. Diagnosis. Dental formula 1-0-2-3. Differs from all described mixodonts by the length of the diastema between the incisor and P/3, by the proclivity of the incisor, and by the low height of the anterior part of the mandible. Differs from Rhombomylus , Matutinia , Eurymylus, Gomphos , and H ypsimylus by the wider talonid of M/1 -M/3; from Mimotona , Gomphos , and H ypsimylus by having no unilateral hypsodonty; from Eomylus by having lower crowned molars with a smaller hypoconulid; from the mandible referred to Heomys by molars with a wider talonid, narrower trigonid, and with the hypoconulid-entoconid being separated to the base of the crown; and from Zagmys by a reduced paraconid. Etymology. Khaychina , in allusion to the region that produced the type specimen, Khaychin-Ula, MPR. Khaychina elongata n. sp. Text-fig. 10 Holotype. PSS 30-3, left mandible with the incisor and M/1 -M/3. 142 PALAEONTOLOGY, VOLUME 31 d text-fig. 10. Khaychina elongata n. gen., n. sp. a~d, PSS 30-3, holotype, left mandible with incisor and M/l- M/3. a , occlusal view of molars; b, labial view; c, lingual view, a-c, x 15; d, labial view of mandible, x 7-5. Specimen from the Naran Member of the Naran-Bulak Beds at Khaychin-Ula I, Mongolia; late Palaeocene. DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 143 Locality and stratigraphic distribution. Khaychin-Ula I region, Bugin-Tsav Basin, southern MPR; Naran Member of the Naran-Bulak Beds; late Palaeocene. Diagnosis. As for genus, only species. Etymology. Elongata(L.), elongate, with reference to the length of the diastema between the incisor and the P/3 of the type specimen. Description. As noted above, one of the most striking characters of PSS 30-3 is the length of the diastema between the single, procumbent incisor and the P/3; it is slightly more than the length of the molar series. A common proportion in mixodonts is about half this length. Only the roots of P/3 and P/4 remain; they indicate that the P/3 was narrow (apparently about half the width of P/4) and supported by a single root. P/4 was prob- ably submolariform and nearly as big as M/1, although with a narrower talonid. The three molars increase in length from the first to the last; M/1 is subquadrate except for the bulge of the hypoconid. In all three the hypoconid extends very markedly labially. The trigonid is slightly damaged in M/1, but probably there was no paraconid developed; it is absent in the other molars. An anterior loph extends from the protoconid to the anterior side of the metaconid. The latter is the dominant cusp. The protoconid and hypoconid are rather close together but in labial view do not form the sort of columns seen in Mimotona or Gomphos. A large mesoconid exists on the cristid obliqua and extends well up on the posterior wall of the trigonid. The latter is strongly sloping but much of the inclination is due to wear. An oblique groove crosses the talonid basin from the notch between the metaconid and entoconid to another notch between the hypoconid and the hypoconulid. The latter is well separated from both the entoconid and the hypoconid and on M/3 forms a distinct third lobe. The teeth are low crowned, lower than those of any other described mixodont, including those of the mandible referred to Heomys. The jaw is low and elongate, particularly anteriorly; the coronoid process rises well behind M/3. Measurements. L W L W L W PSS 30-3 M/1 1-6 1-8 M/2 1-7 1-9 M/3 2-5 1-8 Discussion. K. elongata is a form characterized by extremes: it exceeds all other mixodonts in length of diastema, in incisor proclivity, and in relative width of M/1 and M/2. Its molars are very low crowned and lack unilateral hypsodonty. Attribution to any of the existing anagalidan or mixodont families would considerably expand the limits of that family. The question of reference to the Rodentia could even be raised, but primitive rodents do not generally have a distinct and isolated hypoconulid, nor a third lobe on M/3, and the trigonid basin of early forms opens posteriorly into the talonid basin through a notch between the protoconid and metaconid; the trigonid basin is closed in PSS 30-3. What we have here is a species that strongly diverges from all known mixodonts. Given the distinctions upon which some early rodent families have been proposed (for example, Yuomyidae as separate from Cocomyidae, in Dawson et al. 1984), it would be relatively easy to define a ‘Khaychinidae’. For the moment, we shall resist the temptation, knowing that many more mixodonts await discovery, among which might be morphological intermediates linking Khaychina more closely to the Eurymylidae. Hypsimylus beijingensis is known from a single specimen (text-fig. 11) comprised only of two lower teeth. One of the two has been identified by Zhai ( 1 977) as DP/4 and shows extreme hypsodonty. The following tooth is considered to be M/1 and is much less hypsodont, exhibiting a condition only slightly more than that seen in Eurymylus or Eomylus borealis and comparable to that in Rhombomylus. Concerning the identification of the anterior tooth, in the opinion of P. M. Butler (pers. comm., 1986), ‘to have a milk molar more hypsodont than the permanent molar would be most unusual, almost unique’. We will regard it as P/4. Apart from the hypsodonty (which is bilateral) of the P/4, there are no significant charac- ters indicating a relationship to the Mimotonidae. Until further evidence indicates differently, we will place it in the Eurymylidae, in a new subfamily Hypsimylinae, but its affinities are obscure. a text-fig. 1 1. H ypsimylus beijingensis Zhai, 1977. a-c , V 5242, holotype, left P/4 M/1, a, occlusal view; b, labial view; c, lingual view. All views x 15. Specimen from the Changxindian Formation at Changxindian, Beijing, China; late Eocene. DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 145 Family mimotonidae Li, 1977 Mimotona lii n. sp. Text-fig. 12 Synonym. Mimotona n. sp., in Li 1977. Holotype. V 4327, right mandublar fragment with P/3- M/3; IVPP, Beijing, China. Locality and stratigraphic distribution. Locality no. 71008, upper part of the Wanghudun Formation, Qianshan Basin, Anhui Province, CPR, middle Palaeocene. Diagnosis. Differs from M. wana by the (apparent) absence of a metaconid on P/3. Differs from both M. wana and M. robusta by a wider talonid on P/4 and by smaller size. Differs from the mandible referred to Heomys (V 4322) by the talonid of P/4 which is wider and (in labial view) of lesser height; by text-fig. 12. Mimotona lii n. sp. a-c, V 4327, holotype, right mandible with P/3-M/3. a , occlusal view; b , labial view; c, lingual view. All views x 15. Specimen from the Wanghudun Formation, Qianshan Basin, Anhui Province, China; middle Palaeocene. 146 PALAEONTOLOGY, VOLUME 31 text-fig. 1 3. Mimotona wana Li, 1977. a-c, V 4324, holotype, left maxillary with P3/-M3/. a, labial view; b , occlusal view; c, lingual view. All views x 15. Specimen from the Doumu Formation, Qianshan Basin, Anhui Province, China; late Palaeocene. the protoconid and hypoconid of the molars which (in labial view) are closer together and form parallel, subequal columns (the hypoconid is the bigger in M/1 -M/2 (labial view) of Heomysl , V 4322). Etymology. Named in honour of Dr Li Chuan-kuei, IVPP, Beijing, for his work on the possible ancestors of rodents and lagomorphs. Description. The P/3 is damaged but seems to have had a single anterior cusp; no details of the posterior part of the tooth are interpretable. P/4 is subquadrate, with the talonid exceeding in width the trigonid. The metaconid is slightly higher than the protoconid but the two are similar in dimensions; a faint cingulum connects them DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 147 text-fig. 14. Mimotona wana Li, 1977. a-c, V 4325, left mandible with P/3-M/3. a, occlusal view; b , labial view; c, lingual view. All views x 15. Specimen from the Doumu Formation, Qianshan Basin, Anhui Province, China; late Palaeocene. anteriorly. Although short, the talonid is broadly basined; situated at the posterior corners, the hypoconid and entoconid are of subequal size. If a hypoconulid was present, it was low, small, and crestiform. Very little distinguishes the molars of M. lii from those of M. wana (text-figs. 13 and 14) and M. robusta (text- fig. 15) except their proportions: M/1 and M/2 are relatively more elongate in M. lii and M/3 is less elongate than that of M. wana (it is unknown in M. robusta). The mandible of V 4327 is broken just anterior to the P/3. Consequently, the number of incisors is unknown. The enamel of all the teeth has suffered damage by chemical erosion and is pitted. Discussion. The specimen V 4327 was briefly mentioned by Li (1977), figured, and measured. Following Li, we refer it to Mimotona. The principal characters supporting this attribution are: the closeness in labial view of the parallel columns formed by the protoconid and hypoconid and the degree of unilateral hypsodonty seen. M. lii differs from the previously described species of Mimotona 148 PALAEONTOLOGY, VOLUME 31 text-fig. 15. Mimotona robust a Li, 1977. a-c, V 4329, holotype, right mandible with P/4- M/2, a, occlusal view; b, labial view; c, lingual view. All views x 15. Specimen from the Doumu Formation, Qianshan Basin, Anhui Province, China; late Palaeocene. essentially in premolar morphology. The proportions of the molars add an additional distinguishing element. DISCUSSION Our research on the interrelationships of the various forms that enter into the concept of Mixodontia has led us to compare closely Matutinia and Rhombomylus. The material we used for the latter was principally that referred to R. turpanensis (text-figs. 16-18); the maxillary (text-fig. 19) and mandible we had of R. laianensis seem to differ very little and the specimens are more worn. Following Zhai (1978) we consider V 4362 (P4/-M3/), V 4363 (I, P/3-M/3), V 4364 (P/4-M/2), and V 4365 (M/3) as representative of R. turpanensis. The lesser worn teeth of V 4364 (text-fig. 18) DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 149 text-fig. 16. Rhombomylus turpanensis Zhai, 1978. a-c, V 4362, lectotype, right maxillary with P4/-M3/. a, labial view; b, occlusal view; c, lingual view. All views x 10. Specimen from the Shisanjianfang Formation, Turpan Basin, Xinjiang Uygur Autonomous Region, China; early Eocene. display differences in morphology from those of V 4363, notably in P/4 (but the P/4 of R. laianensis (V 5 1 75) is virtually identical to the P/4 of the latter). The P/4 of Matutinia nitidulus (V 5360, the only specimen available with P/4) shows approximately intermediate characters (text-fig. 20). The cheek teeth in this species are slightly lower crowned than those of R. turpanensis , although we did not have specimens with identical stages of wear for the two forms. Other than this apparent difference in crown height, no diagnostic distinguishing features of generic value are evident from the lower teeth. From the single example (V 5359) of the upper dentition of M. nitidulus (text-fig. 21) that we were able to compare, a difference in shape of M3/ appears to be the major distinction separating this species from R. turpanensis ; difference in crown height seems negligible. The teeth in the holotype of the latter are rather heavily worn; the little worn specimen from Tsagan-Khushu (PSS 20-164; text- fig. 22) displays a cheek tooth morphology so similar that its attribution to Rhombomylus seems 150 PALAEONTOLOGY, VOLUME 31 subject to little doubt, but the M3/ is considerably longer than that of the type. As is commonly the case, the M3 appears to be particularly variable. We are unable to justify the generic distinction of Matutinia from Rhombomylus. While we retain the species nitidulus, further knowledge of its variability is necessary to ensure its validity. Rhombomylus is a highly variable form; Li (pers. comm., 1984) has informed us that there is some doubt concerning the separation of R. turpanensis and R. laianensis. The problem is under study. Given, then, this variability within Rhombomylus, and the numerous points of identity that link nitidulus and turpanensis , we feel that the former should be placed in the genus Rhombomylus (this decision was independently reached by McKenna, pers. comm. 1986). text-fig. 17. Rhombomylus turpanensis Zhai, 1978. a-c, V 4363, left mandible with P/3-M/3. a , occlusal view; b , labial view; c, lingual view. All views x 10. Specimen from the Shisanjianfang Formation, Turpan Basin, Xinjiang Uygur Autonomous Region, China; early Eocene. No type specimen was designated for R. turpanensis. The first specimen mentioned, a skull with an associated mandible (V 4361), is not figured. We shall designate the figured maxillary, V 4362, as the lectotype. If R. laianensis is shown to be conspecific with R. turpanensis a supplementary problem arises. Rhombomylus was described as a new genus by Zhai (1978) with the type species of turpanensis. But the description of R. laianensis was published (by Zhai et al.) in 1976 and thus has priority if it is truly a senior subjective synonym. DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 151 Isolated teeth and incomplete dentitions of Gomphos are found sometimes in deposits that contain similarly incomplete remains of Rhombomylus. As the teeth are about the same size we will present here a summary of their differences in order to facilitate identifications. The comparisons are based on the material referred to R. turpanensis cited above (text-figs. 16 18) and the specimens from Tsagan-Khushu (R. cf. turpanensis; text-fig. 22); for Gomphos we had (from Tsagan-Khushu) PSS 20- 1 63 (P/4- M/3; text-fig. 23), PSS 20- 1 32 ( M/2- M/3), PSS 20- 1 66 ( P4/- M 1/; text-fig. 25 ), PSS 20- 1 67 (M1/-M2/; text-fig. 25); and from Gashato (but not from the horizon of the holotype), PSS 33-1 1 (P/3, M/2-M/3; text-fig. 24). Upper teeth P4/ L W Ml/ L W M2/ L W M3/ L W Rhombomylus turpanensis (V 4362) 2-0 3-5 2-5 3-8 2-8 3-7 2-9 3-5 R. cf. turpanensis (PSS 20-164) 2-0 40 2-7 4-4 2-6 3-7 3-3 3-5 ‘ Matutinia ' nitidulus (V 5354) 20 3-6 2-5 4-2 2-4 3-4 2-4 ‘M.’ nitidulus (V 5359) 2-1 3-6 2-7 3-8 2-8 3-8 2-5 2-7 Gomphos elkema (PSS 20-167) 30 5-8 30 5-3 G. elkema (PSS 20-166) 31 5-5 2-9 5-7 Lower teeth P4/ L W Ml/ L W M2/ L W M3/ L W R. turpanensis (V 4363) 2-6 2-3 2-7 2-3 31 2-7 4-2 3-0 R. turpanensis (V 4364) 2-3 2-2 2-6 2-4 3-2 3 0 R. turpanensis (V 4365) 40 30 R. cf. turpanensis (PSS 20-169) 2-8 2-6 30 2-8 3-1 30 40 2-9 R. cf. turpanensis (PSS 20-165) 3-3 30 4-3 30 ‘M.’ nitidulus (V 5360) 2-6 2-3 2-9 2-5 31 2-7 3-5 2-6 G. elkema (PSS 20-163) 2-7 30 2-7 31 30 30 3-5 2-7 G. elkema (PSS 20-132) 3-0 3-0 4-1 3-2 G. elkema (PSS 33-11) P/3 2-5 1-9 2-9 31 3-5 3-0 G. elkema differs from R. turpanensis by the presence of two lower incisors (seen in PSS 20- 1 34 and PSS 20-98 from Tsagan-Khushu); by an elongate P/3, which is short and small in Rhombomylus; by a very molariform P/4 with a much wider talonid; the P/4 in Rhombomylus lacks the low situated paraconid; by the lower cheek teeth ( P/4-M/2; text-figs. 23 and 24) tending to be subquadrate and the M/3 being shorter with no enlarged third lobe (i.e. the hypoconulid is small); by the columns formed by the protoconid and hypoconid being very close together; and by a greater degree of unilateral hypsodonty. The upper teeth of Gomphos (text-fig. 25) differ from those of Rhombomylus by the subcircular contour of P4/ in occlusal view (it is more transverse in Rhombomylus); by the presence on P4/ of a single centro-labial cusp (four cusps in Rhombomylus); by the P4/-M2/ being greatly inflated lingually and more sloping with considerable unilateral hypsodonty; by the absence of a vertical groove between the protocone and hypocone; by the paracone being circular and cuspate, becoming lophlike only with advanced wear, and well separated from the anterior and labial cingula (in Rhombomylus there is no labial cingulum and the anterior one is closely appressed to the paracone); by the (apparently) stronger metaconule; and by the presence in the molars of a sort of mesostyle. None of the specimens available for this study retained all three upper molars. In the maxillary PSS 20-167 (text-fig. 25 d ), the more posterior of the two teeth preserved has an aspect that could permit its being regarded as M3/; in consequence its companion would be M2/. However, a contact wear facet on its posterior side confirms its identity as an M2/ and thus the tooth preceding it as Ml/. An M3/-like aspect for M2/ can also be seen in Rhombomylus. For most of the mixodont taxa very little information is available other than for dentitions, and these are often incomplete. For some of the genera, however, part of the maxillary is preserved which 152 PALAEONTOLOGY, VOLUME 31 includes the base of the jugal arch. Examination of this area reveals considerable variety, but in three specimens of Eurymylus laticeps a certain uniformity prevails. This suggests that generic diflferences may be expressed by this morphology as well as by the relationship of the position of the molars with respect to the anterior root of the jugal arch. text-fig. 18. Rhombomylus turpanensis Zhai, 1978. a, c,f, V 4364, left mandible with P/4 -M/2, a, occlusal view; c, labial view;/, lingual view, b , d , c, V 4365, right M/3, h, occlusal view; d, labial view; e , lingual view. All views x 10. Specimens from the Shisanjianfang Formation, Turpan Basin, Xinjiang Uygur Autonomous Region, China; early Eocene. As far as can be determined, Heomys and (to a lesser degree) Eurymylus are distinguished by the presence of a flange, or an abrupt widening of the base of the snout, that slopes in side view anterodorsally, much in the manner of early rodents. Viewed ventrally, the base of the jugal arch in these cases tends to be perpendicular to the tooth row (in Eurymylus) or slopes anteriorly with respect to the latter (in Heomys). In contrast, there does not seem to be a flange produced on the skulls of Mimotona , Eomylus , and Rhombomylus. The muzzle enlarges progressively toward the rear in what is doubtless a less specialized fashion. Within this general pattern, however, there is diversity; in Mimotona the base of the jugal arch is thick and massive, extending from the level of the anterior edge of M 1/ to the middle of M2/; in Eomylus the base of the arch is much more slender and occupies a position opposite only the posterior half of Ml/. It can be noted in passing that the jugal-maxillary DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 153 text-fig. 19. Rhombomylus laianensis Zhai et al ., 1976. a-c , V 5174, holotype, right maxillary with P3/-M3/. a, labial view; b, occlusal view; c, lingual view. All views x 10. Specimen from the Zhangshanji Formation at the Laian locality, Laian District, Anhui Province, China; early Eocene. suture is clearly preserved in both of these specimens (V 4324, M. wana, and PSS 20-139, E. zhigdenensis), as well as the ventral orbital rim. From what remains of the latter it appears likely that the orbit extended to the level of the anterior side of P3/ in Mimotona and well anterior to this level in Eomylus. The orbital situation in Eurymylus and Heomys seems similar to that in Eomylus. Orbital anterior extension in Rhombomylus is like that in Mimotona and the base of the jugal arch appears to be rather massive, as it is in that genus, but it extends from above P4/ to above Ml/, thus being situated more anteriorly. Summarizing the position of the base of the jugal arch with respect to the tooth row, the most anterior placement is probably that of Eleomys (above P4/ to above the anterior edge of M2/), which is approximated by that of Rhombomylus. Mimotona and Eurymylus follow (above the anterior side of Ml/ to above M2/), and the most posterior condition is that of Eomylus. 154 PALAEONTOLOGY, VOLUME 31 In early Asian rodents (for example, Cocomys lingchaensis, Tamquammys wilsoni , and Petrokozlovia notos ) the anterior edge of the base of the jugal arch falls opposite the level of the posterior side of P3/. In the early lagomorph Shamolagus, the same situation prevails; in Lushilcigus the jugal arch is slightly more posterior. text-fig. 20. Rhombomylus nitidulus (Li etal., 1979). a-c, V 5360, right mandible with P/3- M/3, a, occlusal view; b, labial view; c, lingual view. All views x 10. Specimen from the Limuping Formation, at the Lingcha locality, Hengyang Basin, Hunan Province, China; early Eocene. Continuing research on the structure of the incisor enamel makes it possible to remark that, where it is known in the Mixodontia, it can vary from one layer (as in modern lagomorphs) to a state approximating two (as in modern rodents). The enamel in Eurymylus seems to illustrate the weak differentiation of an outer layer and Heomys can be said to possess, more or less, two layers. Flynn (in Flynn, Russell and Dashzeveg 1987) noted the presence of only a single layer in Rhombomylus, Gomphos, and Zagmys; a single layer is also cited in Mimotona (and Mimolagus) by Li and Ting 1985. Given the fact that even early rodents ( Paramys , for example) do not always display two distinct layers within the enamel of their incisors, the precision of this character for taxonomic purpose is probably not great. It is quite likely that pre-rodents, like pre-lagomorphs, had incisors whose enamel was not differentiated into more than one layer. We have reviewed these taxa that fall into the category of neither rodents nor lagomorphs but possess the gliriform adaptation. The diversity already recorded and the lack of connecting links, not DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 155 only between the taxa in question but between them and their probable rodent and lagomorph descendants, makes it evident that we have barely begun to know the group. Classifying such an assemblage is challenging. A division can be made based on forms possessing two incisors in each tooth row, coupled with marked unilateral hypsodonty and parallel labial columns in the lower molars produced by the protoconid and hypoconid in close conjunction. text-fig. 21. Rhombomylus nitidulus (Li et al., 1979). a-c , V 5359, right maxillary with P3/-M3/. a , labial view; b, occlusal view; c, lingual view. All views x 10. Specimen from the Limuping Formation at the Lingcha locality, Hengyang Basin, Hunan Province, China; early Eocene. separated by a shallow groove. While this group, composed of M. wana, M. robusta , M. Hi, and G. elkema , is rather coherent, an opposing group, united essentially by the presence of a single incisor, little or no unilateral hypsodonty and widely separated protoconid and hypoconid, is more disparate. Eomylus zhigdenensis, E. borealis, Eurymylus laticeps , the species of Rhombomylus and H. orientalis constitute a loose association. More marginal is Khaychina oblongata and Zagmys; equally distinctive is Hypsimylus beijingensis. For the first group the family name Mimotonidae has been used and is often considered to represent either the first lagomorphs or those forms immediately preceding them. But lagomorph characters (as opposed to those anticipating rodents) that can be seen in available early mimotonid 156 PALAEONTOLOGY, VOLUME 31 text-fig. 22. Rhombomylus cf. turpanensis Zhai, 1978. a-c, PSS 20-164, left maxillary with P4/-M3/. a, labial view; b , occlusal view; c, lingual view. All views x 10. Specimen from the Bumban Member of the Naran-Bulak Beds at Quarry I, Tsagan-Khushu, Mongolia; early Eocene. specimens are few. The presence of two upper and lower incisors is an obvious distinction, but it is probably a primitive character quite likely shared with pre-rodents and is therefore not an irrefutable indication of one or the other lineage. Another clue that is invoked for lagomorph affinity is the unilateral hypsodonty that is particularly expressed in the upper molars of lagomorphs, and which is present in Mimolagus, an acknowledged member of the latter group according to Bleefeld and McKenna ( 1 985). Mimotona , in addition to having two incisors, also possesses the character of greater lingual than labial crown height; this difference in lingual versus labial crown height in the upper molars is more than in its contemporary Heomys, but not to a marked degree, and greater wear of the teeth in the Heomys specimen (V 4321) than in the maxillary of M. wana (V 4324) makes it seem less; the difference does exist, however. It is worthy of note that Dawson et al. (1984) exclude Heomys from the Rodentia in part because of the height of the tooth crowns. This character, then, is also not infallible. The condition (labial versus lingual crown height) in Heomys is approached by teeth DASHZEVEG AND RUSSELL: PALAEOGENE AND EOCENE MIXODONTIA 157 preserved in maxillaries of undescribed ctenodactyloid rodents from the early Eocene of Tsagan- Khushu. The other mixodont family, the Eurymylidae, is regarded here in a sufficiently elastic view to encompass the genera enumerated above; the division into subfamilies will probably be useful as more taxa become known. text-fig. 23. Gomphos elkema Shevyreva, 1975. a-c, PSS 20-163, right mandible with P/4-M/3. a, occlusal view; b, labial view; c, lingual view. All views x 10. Specimen from the Bumban Member of the Naran-Bulak Beds above Quarry II, Tsagan-Khushu, Mongolia; early Eocene. Our classification of the Mixodontia is as follows: Cohort Glires Linnaeus, 1758 Order Mixodontia Sych, 1971 Family Eurymylidae Matthew, Granger and Simpson, 1929 Subfamily Eurymylinae, new usage Eurymylus laticeps Matthew and Granger, 1925 Heoinys orient alis Li, 1977 Rhombomylus laianensis Zhai et al. , 1976 or R. turpanensis Zhai, 1978 158 PALAEONTOLOGY, VOLUME 31 text-fig. 24. Gomphos elkema Shevyreva, 1975. a-f PSS 33-11, left P/3, M/2- M/3. a , b , occlusal views; c, d. labial views; e,f, lingual views. All views x 10. Specimen from Bed 1 1 of ‘Svita T at the locality of Gashato, Ulan-Nur Basin, Mongolia; probably early Eocene. R. nitidulus (Li et al., 1979), new combination Eomyhts zhigdenensis n. gen., n. sp. E. borealis (Chow and Qi, 1978), new combination Amar aleator n. gen., n. sp. Subfamily Khaychininae nov. Khaychina elongata n. gen., n. sp. Subfamily Zagmyinae nov. Zagmys insolitus Dashzeveg et al. , 1987 Subfamily Hypsimylinae nov. Hypsimylus beijingensis Zhai, 1977 Family Mimotonidae Li, 1977 Mimotona wana Li, 1977 M. robusta Li, 1977 M. Hi , n. sp. Gomphos elkema Shevyreva, 1975. These new subfamilies are characterized by the single genus and species that each one includes. The D ASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 159 text-fig. 25. Gomphos elkema Shevyreva, 1975. a, c, e, PSS 20-166, left maxillary with P4/-MI/. a , labial view; c, occlusal view; e, lingual view. 6, d,f, PSS 20-167, right maxillary with M1/-M2/. h, labial view; c/, occlusal view; f lingual view. All views x 10. Both specimens from the Bumban Member of the Naran-Bulak Beds above Quarry II, Tsagan-Khushu, Mongolia; early Eocene. Eurymylinae is comprised of mixodont species showing none of the peculiarities (no proclive incisor with an exaggerated diastema, no strong paraconid, no excessive hypsodonty of P/4) that distinguish the members of the other subfamilies. Rodents, like lagomorphs, are probably derived from mixodont ancestors. As yet (and as usual) an ideal ancestral form is lacking, although Heomys is often cited as a borderline case. If it is rejected as a rodent by specialists (Hartenberger (1980), among others), its characters display a tantalizing resemblance to those of early Ctenodactylidae. 160 PALAEONTOLOGY, VOLUME 31 text-fig. 26. a, Heomys orientalis , V 4321, x4; modified from Li & Ting 1985. 6, Eurymylus laticeps , MgM 11/62, x 2-6; modified from Sych 1971. c, Rhombomylus sp„ V 5289, x 2; modified from Li & Ting 1985. d, Mimotona wana, V 4324 (reversed), x 3 8; modified from Li 1977. e, Eomylus zhigdenensis , PSS 20-139, x4. /, Rhombomylus nitidulus, V 5354 (reversed), x 2; modified from Li et al. 1979. Arrows indicate the infraorbital foramen. The lower jaw, V 4322 (text-fig. 28), that was referred to H. orientalis by Li (1977) but which apparently does not come from the same locality as the type partial skull, has badly worn teeth. Nevertheless, enough remains so that it can be said that, particularly in labial view, they are unlike those of M. wana , M. robust a , or M. lii. More resemblance to E. zhigdenensis is apparent (the P/4 is quite similar), although the M/3 of the latter is much larger. There is less similarity to E. borealis and still less to any of the other known mixodonts. If this lower jaw is not of Eleomys (but it could well be) it is different in any case from that of Mimotona , which was found in the same locality. For the reasons cited earlier that distinguish rodents, it cannot belong to that group. According to published information the type material of both M. orientalis and H. wana came from locality 71017 in the upper part of the Doumu Formation, Qianshan Basin, Anhui. An additional specimen, V 4326, with II/— 12/, was referred to M. wana but comes from the upper part of the The stratigraphic distribution of the members of the mixodont families late Eocene Elypsimylus beijingensis early Eocene Rhombomylus spp. (including Matutinia) Zagmys insolitus Gomphos elkema late Palaeocene Eurymylus laticeps Mimotona wana Heomys orientalis Eomylus zhigdenensis E. borealis Khay china elongata Amur aleator M. robusta middle Palaeocene M. lii EURYMYLIDAE MIMOTONIDAE 1 incisor, little or no unilateral hypso- 2 incisors, unilateral hypsodonty, proto- donty, protoconid and hypoconid widely conid and hypoconid close together and separated by a deep groove. separating shallow groove. DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 161 text-fig. 27. Heomys orient alis Li, 1977. a-c , V 4321, holotype, left maxillary with P3/-M3/. a, labial view; b, occlusal view; c, lingual view. All views x 15. Specimen from the Doumu Formation, Qianshan Basin, Anhui Province, China; late Palaeocene. underlying Wanghudun Formation. The determination of these incisors as belonging to M. warn, which would indicate the presence of the species in this lower level, seems rather dubious as none are associated with the type or referred material from the Doumu Formation. A recent effort made to decipher lagomorph origins is that of McKenna (1982) who sought illumination among the anagalids in the genera Huaiyangale and Hsiuannania. The former is stated by McKenna to be a non-lagomorph. Its upper molars, however, possess the same unilateral hypsodonty that is found in Mimotona , coupled with a broadly expanded postcingulum and a much more developed anterocingulum, very reminiscent of the condition in Hsiuannania , a genus that is included in McKenna’s concept of the Lagomorpha. Dentally (in the upper cheek teeth), Huaiyangale seems as good a lagomorph as does Hsiuannania and better than Mimotona. Based on the morphology of the anterior part of the lower jaw, however, it is clear that Huaiyangale is a member of the Anagalida 162 PALAEONTOLOGY, VOLUME 31 text-fig. 28. Heomysl sp. a-c, V 4322, right mandible with P/4-M/3. a, occlusal view; b , labial view; c, lingual view. All views x 15. Specimen questionably from the Doumu Formation, Quinshan Basin, Anhui Province, China; late Palaeocene. (since it completely lacks gliriform incisor adaptation), while Mimotona is a mixodontian. Hsiuan- nania possesses non-gliriform incisors, a canine and three premolars (in the lower jaw, V 4314, at least) and can no more be a mixodontian than is Huaiyangale. Cheek tooth morphology, then, is not enough, for on this basis there is no reason to exclude Huaiyangale from lagomorph affinity and tenuous reason to include Mimotona, wherein the anterocingulum is low placed and faint or absent — and thus not in the line of creatures that produced (eventually) a hypostria. Concerning cusp homologies in the upper molars of Eocene and later lagomorphs, we follow Lopez Martinez (1985) and Butler’s (1985) analyses, based on wear facets, in regarding the protocone as being lingually placed. Since the early fossil record of the groups in question is known to be extremely incomplete, designating the first possible lagomorph is difficult or even illusory. Mimotona could be a candidate for this position, but it might be more prudent to say only that it approaches the lagomorph condition. DASHZEVEG AND RUSSELL: PALAEOCENE AND EOCENE MIXODONTIA 163 We would not say that Hsiuannania shares this quality. The absence of gliriform adaptation would place it securely, in our opinion, in the Anagalida; it is contemporary with both Mimotona and Heomys. CONCLUSION The Mixodontia can be considered an evolutionary grade; it is thereby transitional. With an increase in discovered material it is conceivable that some day better established lineages would bring about its dissolution. However, in our opinion, none of the known forms can be properly classed as either an undoubted rodent or lagomorph. This is opposition to McKenna (1982) and Bleefeld and McKenna (1985) who regard Mimotona as already a lagomorph and the eurymylids as rodents. If one follows this reasoning it is necessary to add a totally new dimension to both Rodentia and Lagomorpha. At this stage in their unfolding history, and perhaps until the time when skulls and skeletons become available, the authors feel that the concept of Mixodontia is useful and unconfusing. The new taxa that we have been able to add to this group furnish a suggestion as to what its variety must have been in Asia during the later part of the Palaeocene and the early part of the Eocene. It follows that our present classification is only temporary. Of the five genera here placed in the Eurymylinae, Rhombomylus and Eurymylus seem to be the most prevalent and hence the most characteristic. Considerable mystery still shrouds Heomys , particularly as concerns its lower dentition; Eomylus is very distinctive and too little is known of Amur even to be certain of its placement in the Eurymylidae. Even more uncertainty applies to the familial situation of Khaychina and Zagmys. Hypsimylus has already been placed in the Mimotonidae (by Li and Ting 1985), but its unique dental morphology does not provide irrefutable evidence for such an attribution. Our concept of the Mimotonidae includes only Gomphos and the species of Mimotona. Acknowledgements. The authors express their appreciation to Dr J. J. Hooker and Dr P. M. Butler for having read the first version of the manuscript and for having greatly improved it by their comments and suggestions. Dr C.-k. Li furnished valuable information and casts during his stay in Paris in 1984. Other sharp resin casts were provided by Drs E. H. Lindsay and P. D. Gingerich and were indispensable for the comparative parts of this paper. The casts were also the basis of the SEM photographs taken by Madame C. Weber, the figures were drawn by Madame F. Pilard, and the manuscript typed by Mademoiselle L. Da. REFERENCES badamgarav, d. and reshetov, v. yu. 1985. Paleontologiya i stratigrafiya paleogena zaaltayskoy gobi (Palaeontology and stratigraphy of the Zaaltay Gobi). Ed. trofimov, b. a. Sovmest Sov.— Mong. Paleont. Eksped. Tr. 25, 1-104. [In Russian.] bleefeld, a. r. and mckenna, m. c. 1985. Skeletal integrity of Mimolagus rodens (Lagomorpha, Mammalia). Am. Mas. Novit. 2806, 1-5. butler, p. m. 1985. Homologies of Molar Cusps and Crests, and Their Bearing on Assessments of Rodent Phylogeny. In Evolutionary Relationships among Rodents. A Multidisciplinary Analysis. NATO Advanced Research Workshop (1984: Paris, France), 92, 381 402. chow, m.-c. and qi, t. 1978. Paleocene Mammalian Fossils from Nomogen Formation of Inner Mongolia. Vert. PalAsiatica , 16 (2), 77-85. dashzeveg, d. 1976. Novyye mezonikhidy (Condylarthra, Mesonychidae) iz paleogena Mongolii (New mesonychids (Condylarthra, Mesonychidae) from the Palaeogene of Mongolia). In kramarenko, n. n. (ed.). Paleontologiya i biostratigrafiya Mongolii, Sovmestnaya Sov. — Mong. Paleont. Eksped. 3, 14-31. [In Russian.] — 1977. O pervoy nakhodke Hyopsodus Leidy, 1870 (Mammalia, Condylarthra) v Mongol'skoy Narodnoy Respublike (On the first occurrence of Hyopsodus Leidy, 1870 (Mammalia, Condylarthra) in the People’s Republic of Mongolia). In barsbold, r. (ed.). Fauna, flora i biostratigrafiya mezozoya i kaynozoya Mongolii, Sovmestnaya Sov. — Mong. Nauchno-lssled. Geol. Eksped. 4, 7 13. [In Russian.] — 1979a. Nakhodka girakoteriya v Mongolii (Discovery of a hyracothere in Mongolia). Paleontol. Zh. 3, 108-113. [In Russian ] — 19796. Nakhodka Homogalax (Perissodactyla, Tapiroidea) v Mongolii i ego stratigraficheskoye znacheniye (A find of Homogalax (Perissodactyla, Tapiroidea) in Mongolia and its stratigraphic significance). Byulletin Moskovskogo Obshchestva Ispytateley Prirody, Otdel Geol. 54 (6), 105-11. [In Russian.] 164 PALAEONTOLOGY, VOLUME 31 dashzeveg, d. 1982a. La faune de Mammiferes du Paleogene inferieur de Naran-Bulak (Asie Centrale) et ses correlations avec l’Europe et l’Amerique du Nord. Bull. Soc. geol. Fr. (7), 24 (2), 275-281. — 1982 b. Reviziya prodinostseratin tsentral'noy Azii i severnoy Anieriki (Revision of the Prodinoceratinae of Central Asia and North America). Paleont. Zhurn. 1, 95-103. [In Russian.] — and mckenna, M. c. 1977. Tarsioid Primate from the early Tertiary of the Mongolian People’s Republic. Acta. Palaeont. Polonica , 22 (2), 119-137. — russell, d. e. and flynn, l. j. 1987. New Glires (Mammalia) from the early Eocene of the People’s Republic of Mongolia. Part I: Description and Systematics. Proc. K. ned. Akad. Wet. B90 (2), 133-142. dawson, m. r., li, c.-k. and qi, t. 1984. Eocene Ctenodactyloid Rodents (Mammalia) of Eastern and Central Asia. Carnegie Mus. Nat. Hist., Spec. Publ. 9, 138-150. flerov, K. K. 1952. Novyye Dinocerata iz Mongolii (New Dinocerata from Mongolia). Doklady Akad. Nauk SSSR, 86 (5), 1029-1032. [In Russian.] 1957. Dinotseraty Mongolii (Mongolian Dinocerata). Akad. Nauk, Trudy Paleont. Inst. 67, 1-82. [In Russian.] flynn, l. j., russell, d. e. and dashzeveg, d. 1987. New Glires (Mammalia) from the early Eocene of the People’s Republic of Mongolia. Part II: Enamel Microstructure. Proc. K. ned. Akad. Wet. B90 (2), 143-154. gradzinski, r., kazmierczak, j. and lefeld, j. 1968/1969. Geographical and geological data from the Polish- Mongolian Palaeontological Expeditions. In kielan-jaworowska, z. (ed.). Results Polish-Mongol. Palaeont. Exped. Part I. Palaeont. Pol. 19, 33-82. hartenberger, j.-l. 1980. Donnees et hypotheses sur la radiation initiale des rongeurs. In Mem. Jubilaire en Hommage a R. Lavocat. Palaeovertebrata, Mem. 4, 285-301. kielan-jaworowska, z. 1968/1969. Archaeolambda Flerov (Pantodonta) from the Paleocene of the Nemegt Basin, Gobi Desert. In kielan-jaworowska, z. (ed.). Results Polish- Mongolian Paleont. Exped. Part I. Palaeont. Pol. 19, 133-140. li, c.-K. 1977. Paleocene Eurymyloids (Anagalida, Mammalia) of Qianshan, Anhui. Vert. PalAsiatica, 15 (2), 103-118. — chiu, c.-s., yan, d.-f. and hsieh, s.-h. 1979. Notes on some Early Eocene Mammalian fossils of Hengtung, Hunan. Ibid. 17 (1), 71-80. — and ting, S.-Y. 1985. Possible Phylogenetic relationship of Asiatic Eurymylids and Rodents, with comments on Mimotonids. In Evolutionary Relationships among Rodents. A Multidisciplinary Analysis. NATO Advanced Research Workshop (1984: Paris, France), 92, 35-58. lopez martinez, n. 1985. Reconstruction of Ancestral Cranioskeletal Features in the Order Lagomorpha. In Evolutionary Relationships among Rodents. A Multidisciplinary Analysis. NATO Advanced Research Workshop (1984: Paris, France), 92, 151-190. mckenna, m. c. 1982. Lagomorph interrelationships. In Phylogenie et Paleobiogeographie. Livre jubilaire en l’honneur de Robert Hoffstetter. Geobios, mem. spec. 6, 213-224. matthew, w. d., granger, w. and simpson, g. G. 1929. Additions to the Fauna of the Gashato Formation of Mongolia. Am. Mus. Novit. 376, 1-12. russell, d. e. and dashzeveg, d. 1986. Early Eocene Insectivores (Mammalia) from the People’s Republic of Mongolia. Palaeontology, 29 (2), 269-291. sych, L. 1971. Mixodontia, a new Order of Mammalia from the Paleocene of Mongolia. Palaeont. Pol. 25, 147-158. zhai, r.-j. 1977. Supplementary Remarks on the Age of Changxindian Formation. Vert. PalAsiatica, 15 (3), 173-176. 1978. Two new Early Eocene Mammals from Sinkiang. Mem. Inst. Vert. Paleont. Paleoanthrop. 13, 102-106. [In Chinese.] bi, z.-G. and yu, z.-j. 1976. Stratigraphy of Eocene Zhangshanji Formation with Note on a New Species of Eurymylid Mammal. Vert. PalAsiatica, 14(2), 100-103. DEMBERLIYN DASHZEVEG Institute of Geology Academy of Sciences of the Mongolian People’s Republic Ulan Bator DONALD E. RUSSELL Typescript received 30 November 1986 Institut de Paleontologie, UA 12, CNRS Revised typescript received 27 January 1987 8 rue Buffon, F-75005 Paris CLASSIFICATION OF THE TRILOBITE SUBORDER ASAPHINA by r. a. fortey and B. d. e. chatterton Abstract. We present a new phylogenetic classification of trilobites which can be included in a ptychoparioid suborder Asaphina, considerably extending the range of families included in the group as compared with existing classifications. Much of the group is known to be united by the possession of a distinctive type of inflated and effaced larva termed the asaphoid protaspis. The morphology and occurrence of this kind of protaspis is reviewed. All of the group has a ventral median suture, except where it was secondarily lost through fusion of the free cheeks, and most morphological evidence is considered to favour a monophyletic origin for this structure. Relationships between families having such a suture are based on the analysis of morphology; however, stratigraphy is relevant to the determination of the sequencing of characters within a family and to the identification of minor character reversals and parallelisms which can be discounted in the higher level analysis. Two methods of analysis are used. One produces a cladogram based on our weighted assessment of the most important characters. The other, a computer-based analysis using the PAUP program, uses a much wider range of characters to produce two trees which are equally likely. There is generally good agreement between the different methods of analysis. As thus defined, Asaphina includes Cyclopygacea (comprising Cyclopygidae, Nileidae, and Taihungshaniidae), Asaphacea (Asaphidae and Ceratopygidae), Remopleuridacea, and Dikelo- cephalacea, together with some more primitive families which are more difficult to classify: Dikelokephalinidae, Pterocephaliidae, and Anomocaracea. We make a case that the Trinucleacea are linked to the Asaphina by more characters than to any other group. Trilobites included within the suborder are discussed family by family. The supposed olenid Hedinaspis should be included in Asaphacea; on the other hand, the Olenidae, which were included in Asaphina by Bergstrom (1973), are unrelated to the families considered here. The Asaphina was diverse from the mid-Cambrian until the end of the Ordovician, when the group was particularly vulnerable to extinction; this may have been connected with the planktic specialization of the asaphoid larva. High level classification of the trilobites is a long-standing problem. Most authors who have reviewed the subject (Henningsmoen 1951; Bergstrom 1973) have stated the principle that classification should be phylogenetically based, but the agreement stops there, foundering on exactly what characters are to be taken as phylogenetically significant. The plasticity of the trilobite exoskeleton, and the recurrence of certain types of adaptive morphology from more than one phylogenetic source, has made the framing of diagnoses above the family level difficult. There are exceptions; Phacopina, for example, with the unequivocal apomorphic character of the schizochroal eye. The problems are particularly acute when it comes to relating well-characterized Ordovician and younger families to those in the Cambrian. The Cambrian- Ordovician boundary remains a taxonomic one for many groups (for a list of those that crossed it see Fortey 1983). While it is true that the Ordovician sees the inception of new trilobite morphologies that have received familial recognition, it is also probable that they have Cambrian sister groups, and the identification of these is a necessary prerequisite for a phylogenetic classification which can be incorporated in the revision of the Treatise of Invertebrate Paleontology. Certainly this stratigraphic boundary is of little importance to the trilobite families discussed below. Our purpose here is to consider those families which may be classified in a suborder Asaphina, including perhaps one-fifth of the trilobites. The resulting classification differs in many ways from that used in the Treatise ( Moore 1959), both in its arrangement of families, and in the families assigned to it. The trilobite superfamily Cyclopygacea was discussed previously (Fortey 1981) and the details are not repeated here. We do not address the wider issue of whether the Asaphina merits ordinal status, as opposed to subordinal rank within an Order Ptychopariida. This will depend on a review of the other IPalaeontology, Vol. 31, Part 1, 1988, pp. 165-222, pis. 1 7— 19.| © The Palaeontological Association 166 PALAEONTOLOGY, VOLUME 31 ptychoparioid trilobites and an assessment of whether that group is or is not polyphyletic. However, the recognition of an Order Asaphida including only Asaphidae and Ceratopygidae, as used by Shergold and Sdzuy (1984) for example, neither addresses this issue nor includes several related families discussed here, and seems to us a taxonomic over-elevation on insufficient grounds. METHOD OF CONSTRUCTING CLASSIFICATION We have used a cladistic method here for the analysis and presentation of results. We have, however, used stratigraphic criteria for the identification of primitive character states within particular families, and have accepted similar criteria for tracing out segments of trees of particular groups. Such a combination of cladistic and stratigraphic methods is not acceptable to many cladists (Eldredge and Cracraft 1 980) nor is the use of cladistic analyses usual in analysing completely extinct groups. It does have the advantage of making clear exactly what characters are used in the definition of high level groups, and it frequently offers new insights into the phylogenetic arrangement of these families. Equally, it seems unwise to neglect the contribution that stratigraphy can make to an understanding of what happens to characters through time (Fortey and Jefferies 1982), although this kind of evidence tends to work better at low taxonomic levels. What we are trying to identify are synapomorphies linking accepted monophyletic units — derived characters which are considered likely to have appeared only once. We are also trying to identify autapomorphies useful for diagnosis at family level and above. Most of the families considered here are known from numerous species and have a relatively complete fossil record; hence we can use stratigraphic criteria to determine character polarity in some cases where there are ambiguities, and to observe the primitive morphology for an accepted family, which is of use in determining the most likely sister group in constructing the higher level classification. Our approach to construction of cladograms has been two-fold. We have first constructed a diagram of relationships incorporating those characters which we believe are of particular importance (i.e. weighted), especially those ontogenetic and axial characters which we discuss in detail below. This is shown as text-fig. 1. Then we constructed cladograms based on the PAUP (phylogenetic analysis using parsimony) program. This program uses a matrix of characters, which are not individually weighted, for which a polarity (primitive or derived) is assumed. Taxa are then coded for these characters, and the computer program selects from the universe of possible trees the most parsimonious tree (or trees) that can be permuted from the characters. The tree so produced is that which minimizes the number of character reversals or parallelisms. This method has previously been applied to a wholly fossil group by Forey (1987), to whom the reader is referred for technical details. Polarity of characters is determined by reference to an out-group, in our case ptychoparioid, and it was not difficult to assess polarity in this way without direct recourse to stratigraphy. We list the characters we have used in Table 2. This method allows for the inclusion and manipulation of many more characters than is possible by the more intuitive method of text-fig. 1. However, coding for characters has proved to be far from easy, and here we have found it necessary to take into account stratigraphic information/rom within accepted monophyletic families to detect such things as character reversals assumed to be at low taxonomic level, and usually concerning relatively trivial features. For example, lack of tuberculate sculpture is characteristic of a large group, Cyclopygacea + Asaphacea, but there is one genus, N orasaphus Fortey and Shergold, 1984, in which such sculpture is developed, and stratigraphic (as well as morphological) evidence indicates that this is a secondarily derived feature in this case. This genus is ignored; otherwise it would have to be coded entirely separately, to make a separate terminal taxon on the cladogram, which would make the process unwieldy. Advanced Remopleuridacea, Remopleurides and its allies, develop several peculiar autapomorphies, for example adaxial thoracic articulation; this is of use in defining a subgroup within remopleuri- daceans, but does not contribute to the larger analysis of relationships. In general, such features of within-group specialization are ignored, and in contentious cases a generalized, and stratigraphically early, representative of an accepted group forms the basis for the coding. This enables us to encode more characters than would be possible if we had to allow for minor parallelisms and reversals. Most FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 167 text-fig. 1. Cladogram of relationships of taxa included within Asaphina herein, based on weighting of characters which are considered particularly important in phylogeny, as discussed in text; some of these characters are sketched. A broad view of Anomocaracea is taken in this diagram. of the characters tabulated in Table 2 should be obvious from our definitions given there. A few characters, considered of special importance by us in text-fig. 1, are discussed in some detail below. An assumption we have been obliged to make is that when a character is known from a few (maybe only one) species in a family we assume that it applied generally to that group. For example, hypostomes have been assigned to very few Dikelokephalinidae and we have made our coding from those examples. For reasons discussed below we have not included Trinucleacea in the cladistic analyses, and this group is discussed separately towards the end of this paper. Character states which are regarded as primitive for the Asaphina are listed in Table 1. These characters have been determined from examination of a range of generalized Ptychopariina, and are those widely distributed through the various families of primitive ptychoparioids recognized in the present classification. Some of them are general for nearly all Asaphina as well and so do not reappear s 168 PALAEONTOLOGY, VOLUME 31 text-fig. 2. Cladogram produced from the PAUP computer analysis of the characters given in Table 2. Another equally parsimonious cladogram is shown in text-fig. 3. Note that text-figs. 2 and 3 resemble each other, apart from the position of Dikelokephalinidae and Pterocephaliidae. Pterocephaliidae and Anomocaridae are coded separately (cf. text-fig. 1) rather than included in ‘Anomocaracea’ sensu lato. Numbering of characters as in Table 2. Direction of character transformation for multistate characters is indicated. For symbols see text-fig. 3. FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 169 o> 8 o S 0 03 ■o 2 0 03 ~o co U) I 0 5 < c5 S 0 0 [O 03 0 0 ■D text-fig. 3. Another cladogram produced from the PAUP computer analysis of the characters given in Table 2, equally parsimonious with that shown on text-fig. 2. *, characters which are developed in parallel in more than one place on the cladogram; R, character reversals (i.e. advanced to primitive on Table 2); +, character transformations which we consider unlikely on other evidence (see text). in Table 2 (e.g. opisthoparian sutures and terrace ridges on the doublure), but it is as well to list these to clarify those characters that are considered in deducing asaphine relationships. However, since Ptychopariina itself is acknowledged as an unsatisfactory taxon a problem we cannot tackle £ Ptychopariina Dikelokephalinid ae Pterocephaliidae Remopleuridacea Dlkelocephalacea Anomocaridae Ceratopygidae Taihungshaniidae Nileidae Cyclopygidae 170 PALAEONTOLOGY, VOLUME 31 here the out-group to determine polarity on Table 2 has to be selected as an acceptable generalized form. We have used Ptychoparia striata (by definition the typical ptychoparioid) and Elrathia kingi as our reference. Characters weighted in text-fig. 1 Characters on which we place particular importance, and which appear as synapomorphies linking major groups, include two categories. 1. Character states present early in ontogeny. In many trilobites complex developmental changes occur throughout ontogeny; we use the principle that the morphology of the early growth stages tends to be of more use in determining relationships. For example, effacement of the dorsal furrows is a general phenomenon in the asaphine trilobites. Effacement proceeds progressively both within individual ontogenies and within phylo- genetic groups; it takes place repeatedly in unrelated families. The relationships of such trilobites are best judged from the immature forms in which dorsal furrows are still expressed. We argue below that a distinctive form of protaspis shared by many of the trilobites described here is also indicative of common ancestry. Characters initiated early on in ontogeny, whether or not they are subsequently lost during ontogeny or phylogeny, are regarded as likely to be of fundamental importance in demonstrating relationships, compared with characters 0 c 133 a o .c o > CL 0 0 ■0 c "0 -C a 0 O 0 h 0 0 O 0 ;u =5 0 a o E 0 cc. 0 0 O 0 0 .C a 0 o o 0 b 0 0 ;u 0 o o E o c < 0 0 "D 'o > a o 0 0 o 0 0 -u a 0 w < 0 0 T3 Q) iz 0 0 "D 'O) > a o o > O FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 171 table 1. General characters of ptychopariids, which are to be regarded as primitive when present in Asaphina. Cephalom Eyes with circum-ocular suture. Eye ridges present. Front end of palpebral lobes not reaching axial furrows. Mid-occipital tubercle (where tubercle present). Glabella tapering forwards or subquadrate. Three or four pairs of glabellar furrows (unless effaced) progressively shorter anteriorly; Is considerably the longest and directed inwards and backwards. Venter Hypostoma: ovoid middle body, narrow borders, pair of inward-backwardly directed middle furrows in posterior part of middle body. Rostral plate present; rostral sutures inwardly concave. Hypostomal condition natant. Thorax Thoracic segment number variable, and usually large. Simple diagonal pleural furrows. Unmodified articulation; fulcrum well removed from axial furrow. Facets unspecialized. Doublure narrow, with notches acting as vincular ‘stops’. Pygidium Shorter (sag.) than thorax or cephalon. Axis defined dorsally. which may be superficially more noticeable, but which are acquired later during ontogeny, or within the course of the diversification of a family. As an example, the median suture is present in Asaphidea, Remopleurididae, and primitive Nileidae and Cyclopygidae, but in the later species of the latter two families the cheeks become yoked as a single unit. For the determination of phylogenetic relationships far more importance is attached to the presence of the suture than to its subsequent loss, partly because it appears very early on in ontogeny as a discrete structure, and partly because its loss is demonstrably polyphyletic (such sutures are lost on occasion in the asaphids too, for example). 2. Structure of the cephalic axis. The structure of the glabella— its shape, disposition of furrows or muscle impressions and glabellar tubercle— has proved to be an important character linking some of the high level taxa. Again we often refer to the less derived structure in a family to identify relationships, rather than subsequent modifications which can disguise significant similarities; thus Cyclopygidae became almost wholly effaced in response to their pelagic habits, but early examples ( Prospectatrix ) clearly show a glabellar form comparing with that of primitive nileids, and indicating the common ancestry of the two families (Fortey 1981). The species showing the less derived state is also the stratigraphically earliest, which is encouraging, but it is not essential to use the stratigraphic criterion in this case because the primitive nature of the glabellar structure of Prospectatrix would still be apparent from comparison with the out-group. Placing emphasis on glabellar form is not unreasonable, because the glabellar segmentation is likely to be related to the insertion of muscles for the cephalic limbs, and hence with the feeding mode of the trilobite, which has been shown to be significant at high taxonomic level in arthropods in general (Manton 1964). Other characters, such as the thoracic structure and number of segments, or the incorporation of an anterior spinose segment in the pygidium, may come into play for the definition of families, as discussed below. If the relationships shown on text-figs. D3 are correct, spinose pygidia appeared several times in the Asaphina. Cladograms constructed on the basis that the appearance of spinose pygidia was a genuine synapomorphy are much less parsimonious than the ones shown, and result in unlikely distributions of the other characters; this 172 PALAEONTOLOGY, VOLUME 31 table 2. Characters used in the compilation of the cladogram using the PAUP program. 0, primitive condition; 1, 2, derived conditions. 1, glabellar shape: 0, tapering or parallel sided; 1, expands forwards; 2, expansion at SI. 2, position of palpebral lobe: 0, does not touch axial furrow anteriorly; 1, touches axial furrow anteriorly. 3, palpebral rim: 0, defined; 1, effaced; 2, inflated, deeply described by rim furrow. 4, hypostomal condition: 0, natant; 1, conterminant; 2, impendent. 5, occipital ring: 0, defined; 1, effaced. 6, occipital tubercle: 0, present; 1, absent. 7, preoccipital tubercle: 0, absent; 1, present. 8, ventral median suture: 0, absent; 1, present; 2, lost by fusion of cheeks. 9, rostral plate: 0, present; 1, absent. 10, anterior branches of facial sutures: 0, subparallel to divergent up to 30° to sag. line; 1, more strongly divergent. 11, glabellar furrows: 0, ptychoparioid type; 1, crescentic Is; 2, effaced abaxially; 3, entirely effaced. 12, protaspis type: 0, ptychoparioid type; 1, asaphoid type. 13, genal spines: 0, present; 1, absent. 14, posterior cephalic border furrow: 0, defined; 1, effaced. 15, thoracic facets: 0, 'ptychoparioid'; 1, petaloid (see text-fig. 13). 16, thoracic articulation: 0, first segment articulates at fulcrum at some distance from axial furrow; 1, first segment articulates at or very close to axial furrow. 1 7, pygidial spines: 0, marginal pygidial spines absent; 1, marginal pygidial spines anteriorly; 2, marginal pygidial spines along whole pygidial margin, conjoined at spine bases. 18, pygidial doublure: 0, narrow; 1, wide (arbitrary definition of narrow is where the width of doublure is one third, or less, the width of the pleural field inside doublure). 19, librigenal doublure: 0, narrow; 1, wide (particularly difficult to define objectively; our arbitrary definition of 'narrow' is where the doublure width is less than the width of the genal field inside doublure, to base of eye, at the anterior part of free cheek). 20, genal lateral border: 0, narrow, convex; 1, bevelled; 2, relatively wide and flattened, or gently convex; 3, obsolete (see text-fig. 6). 21, thoracic segment number: 0, group includes species with twelve or more segments; 1, group includes species with twelve to nine segments; 2, always eight segments; 3, group includes species with six to nine segments; 4, group includes species with six or fewer segments. 22, hypostomal outline: 0, elongate oval; 1, transverse. 23, hypostomal sculpture on middle body: 0, smooth, or fine pitting; 1, terrace ridges. 24, maculae and associated structures on hypostoma: 0, thin middle furrows; 1, smooth facets; 2, oval raised areas; 3, maculae lost. 25, hypostomal borders: 0, narrow; 1, wide. 26, bacculae/alae: 0, absent; 1, present, or on small growth stages, and lost in adult. 27, pygidial interpleural furrows: 0, present; 1, absent. 28, pygidial postaxial ridge: 0, absent; 1, present. 29, eye size: 0, medium (one quarter to one half glabellar length, which includes the occipital ring); 1, large (> half); 2, small (< quarter). 30, circumocular suture: 0, present; 1, absent. 31, eye ridges: 0, present; 1, absent; 2, not visible, because palpebral lobe touches axial furrow anteriorly. 32, eyes: 0, strip-like; 1, hypertrophied and inflated. 33, transglabellar glabellar furrows: 0, rarely present in group; 1, commonly present in group. 34, relationship of glabella to cephalic margin; 0, glabella does not reach furrow outlining marginal rim; 1, glabella reaches marginal rim or extends to cephalic margin. 35, course of dorsal sutures in front of glabella: 0, marginal; 1, supramarginal. 36, enrolment (after Bergstrom 1973): 0, not enrolled, or possibly spiral; 1, basket-and-lid; 2, sphaeroidal, or presumed to be sphaeroidal if specimen not known in enrolled condition (Bergstrom 1973 reported cylindrical enrolment in Remopleurides, but this is likely to have been secondarily derived from sphaeroidal in other remopleuridids). 37, pygidial length, excluding posterior spines; 0, shorter (sag.) than cephalon; 1, subequal to exceeding length of cephalon. FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 173 38, eye socle: 0, absent; 1, present (thin, wire-like); 2, present (band-like); see text-fig. 6. 39, postocular fixed cheek in relation to posterior border and border furrow: 0, postocular cheek includes long section of border furrow plus postocular genal field; 1, postocular fixed cheek consists almost entirely of posterior border. 40, surface sculpture: 0, often granulose/tuberculate; 1, never granulose/tuberculate (note: the only exceptions to this derived condition in Asaphacea are the genera Norasaphus Fortey and Shergold, 1984, and the sparsely granulose Ceratopyge, in both of which we regard tuberculation as secondarily derived). 41, frontal lobe of glabella: 0, broadly curved about mid-line; 1, more or less rectangular (see text-fig. 19). 42, genal spine: 0, posterior border furrow does not continue strongly into genal spine; 1, border furrows (especially continuation of posterior border furrow) curve backwards into base of genal spine (see text-fig. 7). supports the high taxonomic significance accorded to cephalic structure. Characters which appear poly- phyletically are often the result of heterochrony in development; such developmental changes may produce apparent ‘reversals’. As an example, genal spines are lost in Nileidae, but are primitively present in the whole Asaphina (as they are in the Trilobita as a whole); they secondarily reappear in a few nileid genera (such as Peraspis ), and here again stratigraphic criteria are of use in showing that this apparently ‘primitive’ condition is a derived one in this case. Comments on characters and taxa coded in text-figs. 2 and 3 Most of the characters used should be self-explanatory from Table 2, but a few notes are necessary to show how coding decisions shown in Table 3 were reached. The taxa employed were selected to minimize the number of terminal taxa used, which required certain assumptions that need to be stated, and were for the most part the same as on the weighted cladogram (text-fig. 1). However, ‘Anomocaracea’ was replaced by the family Anomocaridae because, as we discuss below, the assemblage of families included in this taxon are almost certainly para- if not polyphyletic. We also coded separately the family Pterocephaliidae, another taxon which in the generality of its features is typified more by retained plesiomorphic, ‘ptychoparioid’ characters than by obviously derived features. Nileidae and Cyclopygidae were coded for character 8 in state 2— loss of median sutures. The earliest species currently assigned to both families (Nileidae: PUitypeltoides croftiv, Cyclopygidae: Prospectatrix genatenta ) have median sutures, i.e. retain the primitive condition present in the sister group, Taihungshaniidae; both should really be coded separately, and on a strictly cladistic view their classification might pose problems. In other characters they appear typical of their respective families, and their ‘ancestral’ place in a stratigraphically determined phylogeny seems well established. To reduce the number of terminal taxa they are best omitted. Dikelocephalacea are used in the sense of Ludvigsen and Westrop (1983), to include Saukiidae, Dikelocephalidae, and Ptychaspididae. There is a good deal of information about in-group evolution table 3. Coding of characters in Table 2 used in the computer analysis of phylogeny of Asaphina. Character 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 31 32 33 34 35 36 37 38 39 40 41 42 1 . CYCLOPYGIDAE 111211121031111101134113001011210102100100 2. NILEIDAE 111211121131111101133111101011200102120100 3. TAIHUNGSHANIIDAE 11120I11103?001I11I2201110I0012001021?0100 4. CERATOPYGIDAE 00010111101?00?0110230?0010100000002000100 5. ASAPHIDAE 011101111011001001122011110001200012120100 6. DIKELOCEPHALACEA 202100011000000001120000000110101012120011 7. REMOPLEURIDACEA 21210001112100002101100, 2000111200012111000 8. DIKELOKEPHALINIDS 00010001100700001110101 101 101 ?0000 ?0?0 100000 9. ANOMOCARIDAE 0010??01102?0 0000011210???1001 O' 000002100100 10. PTEROCEPHALIIDAE 000000011 0 0 ?0 000001?20????0111000001?000000 11. 'PTYCHOPARIINA' 000000000000000000000000000000000000000000 174 PALAEONTOLOGY, VOLUME 31 a be text-fig. 5. Hypostomal attachment conditions, explaining new terminology introduced in this paper; illustrated by schematic ventral views of cephalic shield with hypostome in its life position (above), and by the corresponding section through the cephalic shield (below), a, natant hypostomal condition; hypostome is not attached at doublure, shown on a ptychoparioid with rostrum, b, conterminant hypostomal condition; doublure is docked with hypostoma, but retains position in front of glabella as in natant condition— hypostomal suture and preglabellar furrow correspond closely— shown here in Asaphina with median suture, c, impendent hypostomal condition; glabellar lobe expands forwards to cephalic margin so that its forward part is now underlain by cephalic doublure - hypostome is rigidly attached as in conterminant condition but has lost its relationship to the front of the glabella— here illustrated by species with fused cheeks. Note that the hypostomal condition is independent of the kind of ventral cephalic sutures, and the choice of illustrations is largely arbitrary, g, glabella; p, preglabellar field; b, cranidial border; d, cephalic doublure on mid-line; h, hypostome. in the Dikelocephalacea; the ptychaspidids include a number of advanced forms in which the characters are highly modified, but whose derivation from more generalized saukiid-like forms is well documented. These latter form the basis of coding. For a very few characters (e.g. hypostomes) there is a conflict between saukiids and dikelocephalids, and in this case we have taken the least specialized condition as that coded. Character 1. Glabellar shape. In many asaphids and one ceratopygid (Ceratopyge itself — see text-fig. 146) the glabella expands forwards. However, all stratigraphic and out-group evidence points to these forms being secondarily derived from species with a forward-tapering to parallel-sided glabella (see below), and for the group as a whole we are obliged to score this character as 0. Secondary glabellar expansion may be of use in within-group taxonomy. Character 3. This character is difficult to determine in some cases, especially Asaphidae. Crushed asaphid specimens can develop a false appearance of having palpebral rims. Some primitive asaphids and small growth stages show a feebly developed palpebral furrow defining a broad rim, while those species with upward-tilted palpebral lobes may also show a change in slope at the inner ends of the lobes which is not homologous with the palpebral furrow. Truly effaced palpebral furrows apply in the majority of asaphids, and this character is consequently scored 1. Character 4. The different conditions of hypostomal attachment are defined below (see text-figs. 5 and 22). Character 5. The occipital furrow is primitively present in all asaphid subfamilies, and its loss is secondary within-group; hence the character is scored 0 in this family. Character 7. The pre-occipital tubercle may attain a secondarily suboccipital position in certain asaphids, as we discuss below, but its homology with tubercles in other asaphids is certain. Character 8. A few, but not all, kainellids (Remopleuridacea) have lost the median suture (e.g. Palmer 1968, pi. 14, fig. 8) by fusion of the free cheeks; however, this is known to be a secondary condition which is not typical of the vast majority of remopleuridaceans— hence it is scored 1 for this taxon. FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 175 Character 12. The asaphoid protaspis type is discussed below. Protaspides of some groups are not known, and must be scored T. Hu (1971) attributed protaspides to a pterocephaliid (Dytremacephalus) and a ptychaspidid (Ptychaspis). However, several of Hu’s attributions to other taxa have been questioned or discounted (e.g. Evitt and Tripp 1977, p. 158) and for this reason we are reluctant to score these families definitely for protaspis type, and both have been recorded as ‘?\ Both are apparently of primitive type and their inclusion would not have significantly altered the most parsimonious cladogram. Character 13. Secondary derivation of genal spines in a nileid ( Peraspis ) from a species without has been demonstrated by Fortey (1975) from stratigraphic evidence. This reversal is discounted in coding this family for lacking genal spines. Conversely, genal spines are present in the vast majority of Asaphidae, and their secondary loss in a few genera is not reflected in the coding. We have not coded types of genal spine (broad, narrow, long, short, etc.) because of the difficulties of definition (even though most workers would probably describe asaphids as having 'broad’ genal spines compared with, say, remopleuridids). This also has the effect of removing from text-figs. 2 and 3 an autapomorphy of the Nileidae ('broadly rounded genal angle’) which figures on text-fig. 1. Character 15. See text-fig. 13 for explanation of petaloid thoracic facet. Character 16. This derived character is shown clearly for Nileus by Schrank (1972, pi. 10, fig. 1). For most Cyclopygacea the adaxial articulation applies to the first thoracic segment, while posterior segments have the fulcral point progressively removed from the axial furrow (e.g. the cyclopygid Degamella : see Fortey and Owens 1987, fig. 38). Character 17. Derived state 1 is where marginal spines are developed laterally or anterolaterally on the pygidium. Usually there is only one such pair of spines, but in some ceratopygids there are two pairs. Although this is not coded on Table 2 the marginal spines may not be strictly homologous. In ceratopygids the spinose margin is quite clearly an extension of the pygidial pleural segments, as if thoracic segments had been incorporated in the pygidium, whereas in Dikelokephalinidae and Taihungshaniidae the spines originate from the border and are wide enough to embrace more than one segment. This difference is acknowledged in text-fig. 1, where the ceratopygid pygidial spines comprise an autapomorphy of that group. In any case, these pygidial spines are different from the comb-like arrangement of remopleuridacean pygidial marginal spines which extend postaxially to conjoin at the mid-line, and comprise derived state 2. Character 20. The different character states are shown on text-fig. 6. Wide genal borders often vary between flat and gently convex in related taxa, and these two conditions have been included in one class. Genal border furrows in nileids are absent at least laterally and posteriorly. Their atavistic appearance in Peraspis is, like other features of that genus, a reversal. Character 21. There is clearly an overall reduction in thoracic segment number through time within the Asaphina, and the most advanced Cyclopygidae have the fewest, five. Within accepted monophyletic families the number is often reduced in later taxa, for example in Ceratopygidae the earliest Proceratopyge have nine segments while Tremadoc species of Dichelepyge have six. Nileids can have seven, eight, or probably nine text-fig. 6. Genal border structure, and characteristics of eye socle (s), to illustrate characters 20 and 38 (see Table 2). Diagrammatic sections through mid part of free cheek from eye (e) through border and doublure. a, primitive genal structure, with convex border forming a tube with doublure and socle not well developed. b , gently convex border with wide, reclined doublure (this structure often intergrades with flat border) and eye elevated on wide band-like socle, c, remopleuridid type with narrow, wire-like eye socle below flat visual surface, flat genal field, with narrow border furrow defining bevelled border. 176 PALAEONTOLOGY, VOLUME 31 text-fig. 7. Dikelocephalacean genal structure to illustrate character 42 (see Table 2). Posterior border furrow continues into basal part at least of genal spine, a , cheek of typical saukiid with defined, gently convex lateral border, b, in Dikelocephalidae lateral border is generally flattened, but characteristic furrow usually remains defined, c, primitive ptychoparioid condition. All approximately natural size. c segments. This variability makes for difficulty of coding. However, all asaphids have eight segments, and this character appears remarkably stable in that group— and it is given as an autapomorphy of the group on our weighted cladogram (text-fig. 1). The fact that this does not appear as an autapomorphy on the PAUP treatments is a result of the way we have coded the characters. An additional coding allowing 'variability in thoracic segment number within group' as primitive, and ‘stable thoracic segment number (8)’ as a derived state would certainly have appeared as an autapomorphy of Asaphidae. The ancestral cyclopygid Prospectatrix with seven thoracic segments has not been coded for the same reason as given under character 8. Character 24. Derived condition 2 refers to raised ‘macular’ areas on the hypostomes of Remopleuridacea, and is an autapomorphy of that group. While it is described from many Ordovician species there is no evidence to say whether it applied to the early species as well— the scoring is based on what evidence we have. Character 26. Further discussion of bacculae is given below under the section on Trinucleacea. Character 27. Loss of paired furrows on the pygidial pleural fields is a derived character. The remaining furrow is interpreted as pleural rather than interpleural, but this is difficult to prove. However, the resemblance of strong pygidial furrows to those pleural furrows on the thorax in Dikelokephalinidae and Taihungshaniidae is consistent with our interpretation of the homology (see Lu 1975, pi. 29, fig. 9; Courtessole et al. 1981, pi. 4, fig. 3). Character 30. This character is uncertain in Dikelokephalinidae and is recorded T. Loss of the circumocular suture and the routine presence of the eye attached to the free cheek evidently occurred more than once independently in Asaphina. Most figured dikelokephalinid cheeks apparently show no eye attached, but one, Hungioides figured by Lu ( 1975, pi. 29, figs. 14 and 1 5) probably shows the eye in place, in which case the loss of the circumocular suture may have happened yet again within the dikelokephalinid clade. Character 37. Several very derived remopleuridids, including Remopleurides itself, have relatively small pygidia; this is a character reversal within the group, and is not scored. Character 42. This character is shown on text-fig. 7. That the furrow passing into the genal spine is the a bed text-fig. 8. Hypostomes of higher Asaphina illustrating characters 22 and 23 (see Table 2); all these types have well-developed sculpture of terrace ridges on the middle body, a , generalized nileid hypostome; early Ordovician Poronileus (after Fortey 1975, pi. 13, fig. 8). b , presumed primitive hypostomal morphology of Asaphidae; Upper Cambrian Niobella (after Westergard 1939, pi. 2, fig. 2). c, loss of maculae in Cyclopygidae; early Ordovician Microparia (after Fortey and Owens 1987, fig. 45 b). d, Taihungshaniidae, showing similarity with Nileidae; early Ordovician Taihungshania (after Courtessole et al. 1981, pi. 4, fig. 7). Two to three times natural size. FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 177 continuation of the posterior border furrow, rather than the lateral border furrow, is shown by certain species in which the course of the latter is terminated before reaching the genal angle (e.g. Hoytaspis speciosa (Walcott) figured by Ludvigsen and Westrop 1983, pi. 14, fig. 6). The furrow is remarkable for the consistency of its presence even in such very derived Dikelocephalaceans as Eaptychaspis. COMPARATIVE DISCUSSION OF CLASSIFICATION Text-figs. 1 -3 show the several possibilities in producing a character-based classification of Asaphina. The two equally parsimonious trees resulting from the PAUP treatment share many points of similarity; the only significant difference between them concerns the placement of Dikelokephalinidae and Pterocephaliidae, both of them primitive groups which are problematic for several reasons, as we discuss below. Both also had a larger number of '?’ characters than other taxa. Only two trees is a robust result from PAUP; some published studies have had to cope with as many as forty equally parsimonious trees. The 'consensus tree’ between the two— essentially removing the problematic taxa to the lowest position as a trichotomy— has been deduced using the method of Adams (1972), and is shown in text-fig. 4. There is a good agreement also between the PAUP trees (especially the consensus tree) and the cladogram based on our assessment of the distribution of a smaller number of key characters (text-fig. 1); the ordering of taxa left to right is virtually the same. One difference is the appearance of Anomocaridae as the sister group of Ceratopygidae to Cyclopygidae in PAUP, whereas 'Anomo- caracea’ appears as one of the two most primitive groups on text-fig. 1. As we discuss below, Anomocaracea is not a satisfactory taxon, and whereas text-fig. 1 attempts to treat it as a whole (including pterocephaliids within it), the PAUP treatment uses Anomocaridae alone— which could well turn out to be the sister group of the asaphacean group. The only important difference is the hierarchical treatment of Asaphidae and Ceratopygidae; on text-fig. 1 Asaphacea (Ceratopygidae + Asaphidae) is a monophyletic group, whereas the PAUP treatment produces a paraphyletic group. The latter is not strictly permissible on a phylogenetic classification. The reason for the difference is the emphasis placed on the similarity of glabellar structure of less derived asaphids and ceratopygids in our weighted classification. Since all of the characters on PAUP characterizing ceratopygids are either developed in parallel with other asaphines or are reversals, it is not unequivocally defined. However, as mentioned above, character 17 (marginal pygidial spines) may not be homologous between groups, and the way these spines are developed in ceratopygids differs from both taihungshaniids and dikelokephalinids, i.e. the marginal pygidial spines may be a better autapo- morphy of Ceratopygidae than it appears in text-figs. 2 and 3. The majority of the characters separating Ceratopygidae from Asaphidae to Cyclopygidae are also parallel with those elsewhere on the cladogram, and some, particularly hypostomal characters, are poorly known for Ceratopygidae. Hence we believe that a more detailed treatment with more certain data on ceratopygid ventral structures may yet indicate a monophyletic Asaphidae + Ceratopygidae, as shown on text-fig. 1. The Asaphacea are retained in the discussion below in this sense. Asaphidae, Nileidae, and Taihungshaniidae come out as rather poorly characterized families on text-figs. 2 and 3. All three families have a coherent stratigraphic history. Typically ‘asaphid’ characteristics, such as the forked hypostome and panderian openings, are not developed throughout the family and are not present on the more primitive genera — hence these characters cannot be used as autapomorphies of Asaphidae. On the other hand, the fixed number of thoracic segments— eight — does not appear as an autapomorphy of the group because of the way the thoracic segment character was coded for PAUP. The same is true of the ‘broadly rounded genal angle' of Nileidae. It is, however, difficult to characterize Nileidae other than by retaining those characters shared with primitive cyclopygids, and by the lack of hypertrophied eyes. Even the autapomorphy of text-figs. 2 and 3 (divergent preocular sutures) does not apply to all nileids. More useful autapomorphies may emerge when more is known of the vincular structures developed on the librigenal doublure. Taihungshaniid marginal pygidial spines compare closely with those of Dikelokephalinidae, but there is no close 178 PALAEONTOLOGY, VOLUME 31 relationship between these two families; the group retains many primitive features compared with Cyclopygidae + Nileidae. MORPHOLOGICAL TERMS EMPLOYED IN SYSTEMATIC DISCUSSION For most morphological terms we follow the Treatise on Invertebrate Paleontology (Moore 1959). The terms ala (plural: alae) and baccula (plural bacculae) are used as defined by Fortey (1975, pp. 14-15). We coin three terms to describe the relationships of the hypostome to the cephalic doublure and glabella. These are important character states which have proved of use in defining high-level taxa in the Asaphina (see text-fig. 5). 1. Nat ant condition (Latin: ‘floating’). Hypostome not attached to cephalic doublure, which is relatively narrow. A true preglabellar field is present, not underlain by calcified cuticle. The hypostome is sited beneath the forward part of the glabella -and the area between it and the doublure was presumably covered by soft cuticle. Work in progress by R. A. F. shows that this is the primitive condition for ptychoparioid trilobites. 2. Conterminant condition (Latin: ‘coinciding’). Doublure extends backwards mesially as far as the preglabellar furrow but no further. Hypostome is docked against doublure, but still in the same relative position as in the natant condition, i.e. beneath forward part of glabella. Hence in this condition the hypostomal suture and the preglabellar furrow are spatially coincident. 3. Impendent condition (Latin: ‘overhanging’). Direct relationship between hypostomal position and glabella is lost. Glabella extends forwards so that medially the cephalic doublure underlies its forward part; as in the conterminant condition the hypostome abuts the cephalic doublure. Note that there is no direct relationship between the state of the ventral sutures (median, rostral plate, or fused) and the condition of hypostomal attachment— so, for example, fused cheeks can and do exist with natant hypostomal condition— the choice of suture pattern in text-fig. 5 was arbitrary. In the Ptychopariida as a whole the conterminant and impendent conditions were polyphyletically derived but are conservative enough to be of use in defining superfamilial taxa, e.g. Cyclopygacea below. The term 'petaloid thoracic facet’ (text-fig. 13) was defined by Fortey ( 1 987). It refers to a broad, subtriangular articulating facet on which there are terrace ridges— the upper terraces run more or less transversely while the lower ones are oblique in some cases. During enrolment the petaloid facet slides past the broad doublure of the preceding segment, which is also furnished with terrace ridges. ASAPHOID PROTASPIS TYPE, AND THE MEDIAN SUTURE Definition of asaphoid protaspis Protaspides have been assigned to Asaphidae ( Isotelus , Anataphrusf Nileidae (herein), Remopleuridi- dae ( Robergiella , Remopleurides), and Trinucleidae among the families that we would place in the Asaphina. These protaspides are similar enough to one another to warrant the term asaphoid protaspis for protaspides of this type. Asaphoid protaspides are spherical to ovoid in shape, with an enrolled rather than inturned doublure (text-fig. 9). One or more of three prominent pairs of submarginal, sharply pointed, conical spines project from the fused cranidium/protopygidium. The free cheeks are simple, without distinct visual surfaces, and without genal spines (except for minute spines in the latest protaspis stages of Isotelus , see Evitt 1961), and are free, fused to each other (R. eximius , see Whittington 1 959a), or fused to each other and the hypostome (smallest protaspides of Isotelus , see Chatterton 1980; and Cryptolithus tesselatus , herein). When free— later in ontogeny— they are always separated by a median connective suture, even when parts of their posterior portions may be separated by a small anterior protuberance of the hypostome ( Isotelus , see Chatterton 1980). No rostral plate is visible at any stage, in either a fused or a free state. The genal doublure extends far back under the protaspis, and almost joins posteromedially in Isotelus. The hypostome has up to nine (possibly the plesiomorphic number) sharp, elongate and conical marginal spines, and it covers most of the ventral surface of the protaspis (text-fig. 10.9- 11). There appear to be only two marginal spines in Remopleurides and four in Cryptolithus , described herein. In small protaspides the axial furrows are often shallow to indistinct, except for a pair of pits located at the forward limits of the axial furrows close to the anterior margin. In large, late stage protaspides a number of dorsal furrows may become distinct, including axial furrows and ring furrows, and the glabella may gain an independent convexity. The exoskeletal FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 179 text-fig. 9. Sagittal sections (a) and ventral views (ft) of four protaspides to show ‘enrolled’ (1), ‘incurved’ (2), and ‘inturned’ doublure (3, 4). For scale, see text-figs. 10 and 1 1. The open arrows point towards the protopygidial doublure. 1, Remopleurides sp. aff. R. eximius Whittington (Middle Ordovician, Edinburg Formation, Virginia). 2, Flexicalymene senarial (Conrad) (Middle Ordovician, Martinsburg Shale, Virginia). 3, Acanthopyge bifida Edgell (Lower Devonian, Receptaculites Limestone, nr. Yass, New South Wales). 4, Proetus talenti Chatterton (Lower Devonian, Warroo Limestone, nr. Yass, New South Wales). surface, where preserved, shows a fine pattern of polygonal ridges. No protaspides assigned to the Asaphina have either a sagittal glabellar furrow dividing the glabella into paired lobes, as in some Redlichiida, Ptychopariida, and Phacopida, or a pair of lobes (?palpebral) along the anterior margin on either side of the glabella, as in some Ptychopariida (PI. 17) and all Phacopida. The largest asaphoid protaspides are large for the Trilobita (see Pis. 1 7 and 1 8), exceeding 1 mm in length. These late stages have more distinct ridges and furrows, including axial furrows, occipital furrows, and axial ring furrows, and may even have a furrow that shows the junction between the cranidium and the protopygidium ( Remopleurides , see Whittington 1959a). Median connective and hypostomal sutures may develop during the protaspid period, as in Isotelus. Distinct, large (exsag.) palpebral lobes appear during the protaspid period (Asaphidae, Nileidae, and Remopleurididae). Minute librigenal spines appear near the back of the free cheeks of the largest protaspid growth stages of some species of Isotelus (see Evitt 1961, fig. 3), which may or may not be homologous with the genal spine of the adult. The first two pairs of prominent submarginal spines on the cranidium/protopygidium of the Remopleuridacea are cephalic and the third pair is protopygidial. This is based on demonstrable relationships in late growth stages of species of Remopleurides, and in particular the work of Whittington (1959a). It is more difficult to determine the location of the junction between the cranidium and protopygidium in relation to these spines in Asaphidae, Trinucleidae, and Nileidae, but we suppose that they are homologous with those in Remopleurides. It is difficult to consider the facial suture of the protaspides in the context of the terms proparian and opisthoparian since, in most cases, the exact location of the genal angle is not clearly recognizable. However, the late protaspid 180 PALAEONTOLOGY, VOLUME 31 stages of Remopleurides (Whittington 1959a, pi. 3, fig. 5; pi. 11, fig. 1; pi. 16, figs. 5 and 6) and Isotelus (Evitt 1961, pi. 1 17, fig. 19) are apparently opisthoparian. The sutures are definitely opisthoparian in known meraspid cranidia assigned to the Asaphina. Median suture The presence of a median suture (which is only secondarily lost by ankylosis in some groups, e.g. Trinucleacea and Nileidae) is considered an important apomorphic character of the Asaphina. There are two possibilities to account for its origin. It may have appeared as a result of a discrete and radical mutation; or it may have arisen as a result of the gradual disappearance of the rostral plate, associated with the ‘migration' of the connective sutures medially, eventually to meet as a single median suture. The lack of a rostrum at any growth stage of the known asaphoid protaspides favours the former hypothesis. Also, in those groups of trilobites where the rostral plate has been reduced to a very small size, as in some Proetida and Encrinuridae, the rostral plate is almost never lost, even when reduced to extremely narrow transverse dimensions (see, e.g. Ischyrotoma in Whittington 1963, pi. 7, fig. 13). Tripp (1962, pi. 67, fig. 12) illustrated the holotype cephalon of Encrinurus deomenos Tripp, 1962, from the Silurian of Anticosti Island, which apparently lacks a rostrum (a second specimen in the Yale University Collections was also stated to lack this feature); he suggested that the same might have applied to E. moe Mannil, 1958. Specimens of Anticosti encrinurids very similar to E. deomenos, from slightly lower horizons (Gun River, as opposed to Jupiter River Formation) than the types, apparently do have a rostrum. If this is substantiated it is the only example we know of conversion of rostrum to median suture within a monophyletic group of post-Cambrian trilobites. Other encrinurids which approach this condition are reconstructed with small, relict rostral plates, e.g. Physemataspis coopi Evitt and Tripp, 1977 and Encrinuroides torulatus Evitt and Tripp, 1977 (Evitt and Tripp 1977, figs. 6 and 13). Since all cheiruraceans have rostral plates early in their ontogeny it is likely that its loss occurred only late in ontogeny. In some proetids the rostral plate is so reduced in size that the posterior part of the connective sutures becomes medial. However, the small triangular rostrum is always present in this order. It remains overwhelmingly true that the rostral plate is a conservative structure. Evidence that may possibly be used to argue for the origin of the median suture from the EXPLANATION OF PLATE 17 Figs. 1-6. Remopleurides aft'. R. eximius Whittington, 1959, from the Edinburg Formation of Virginia, la, b, USNM 414581, ventral and ventrolateral views of protaspis with free cheeks and hypostome attached (slightly out of position), x 41. 2, USNM 414582, lateral view of protaspis with free cheeks attached, x 36. 3 a, b , USNM 414583, free cheeks, x 55. 4, USNM 414584, internal, dorsal view of free cheeks and hypostome, x 56. 5a, b, USNM 414585, lateral and anterior views of protaspis with free cheeks attached, x 34. 6, USNM 414586, ventral view of meraspid hypostome, x 34. Figs. 7, 9, 10, 13, 16-19. Bathyuriscusl sp. from a float block, probably from the Middle Cambrian Pika Formation, near Columbia Ice Fields, western Alberta. 7, UA 7750, stereo pair of small protaspis, x 62. 9, UA 7751, ventral view of small protaspis with cheeks, rostral plate, and hypostome attached, x 62. 10, UA 7752, ventral view of small protaspis with cheeks, rostral plate, and hypostome attached, x 62. 13, UA 7754, stereo pair, dorsal view of large protaspis, x 62. 16, UA 7753, dorsal view of largest protaspis stage, x 62. 17, UA 7755, ventral view of largest protaspis stage with free cheeks and rostral plate attached, x 62. 18, UA 7756, ventral view of incomplete largest protaspis stage with free cheeks, rostral plate, and hypostome attached, x 62. 19, US 7757, ventral view of small meraspid hypostome, x 62. Figs. 8, 11, 12, 14, 15. Spencellal sp. from the same float block of the Middle Cambrian Pika? Formation that contained the above specimens of Bathyuriscusl sp. 8, US 7758, stereo pair, dorsal view of small protaspis, x 62. 1 1, UA 7759, ventral view of small protaspis with one free cheek and hypostome/rostral plate attached (out of original position), x 62. 12, UA 7760, ventral view of small protaspis with free cheeks and hypostome/rostral plate attached (out of original position), x 62. 14, UA 7761, dorsal view of larger protaspis, x 62. 15, UA 7762, ventral view of larger protaspis (same stage as fig. 14) with free cheeks attached, x 62. PLATE 17 FORTEY and CE1ATTERTON, Remopleurides, Bathyuriscus ?, Spencella? 182 PALAEONTOLOGY, VOLUME 31 progressive reduction of the rostrum is the presence of a small, tongue-like extension on the front of the hypostomes of the protaspides of some species of Isotelus (Evitt 1961; Chatterton 1980, figs. 1-56, c, 2-12). It could be argued that this is a small rostral remnant fused to the hypostoma. The tongue-like extension disappears later in ontogeny. The few known ventral surfaces of Cambrian ptychopariine protaspides (text-fig. 10.6, 10.7; Palmer 1958) — which presumably represent the primitive condition— had rostral plates and separate connective sutures. In these ptychopariines the position of the connective sutures appears to vary ontogenetically in relation to the distal ends of the hypostomal suture, allowing the rostral plate to increase or decrease in width in relation to the rest of the cephalon (text-fig. 10.7b, 10.8b). Hence, if the asaphines had a ptychopariine ancestor, as is probable, the median suture either appeared as a discrete mutation early in ontogeny, or the rostral plate disappeared through medial reduction during ontogeny— and then this reduction was pushed back earlier in development during phylogeny. The latter is the more complex mechanism, of course, and it is difficult to see why the rostrum should not have been retained at small growth stages even if suppressed in the adult. Because there are also very few trilobites having median sutures which show any indication of a possible rostral homologue we favour the notion that the structure originated as a discrete mutation, although the difficulties of proving this important point are appreciated. For additional remarks see under Pterocephaliidae below. Metamorphosis A radical metamorphosis took place in asaphines following the protaspid period, usually between the protaspid and meraspid periods, but in some Remopleuridacea during the meraspid period. This produced a change from a globular form to a more flattened form. We consider that this change coincided with a change from a planktic to a benthic mode of life. There is some evidence to suggest that such metamorphoses in the Trilobita were associated with a greater than normal size difference between the relevant successive ecdyses (Chatterton 1980). In many cases the free cheeks have large genal spines after the metamorphosis — contrasting with small or no genal spines before the metamorphosis. At the same time the three pairs of submarginal conical spines disappeared (or became much reduced in size); the spines on the protopygidium pointed less ventrally and more to the posterior; the marginal spines on the hypostome showed a tendency to disappear or become reduced, text-fig. 10. Dorsal and ventral views of protaspides of different orders and suborders of trilobites, showing location of facial, hypostomal, rostral, and connective sutures. The specimen of Isotelus with hypothetically reconstructed appendages is based on Recent nauplius stages of crustaceans (see Fryer 1983). 1, dorsal (a), and ventral (b) views of second and largest protaspides of I. parvirugosus Chatterton and Ludvigsen, 1976 (after Chatterton 1980, fig. 3; pi. 2, figs. 6, 9-11, 16) (Middle Ordovician, Esbataottine Formation, Mackenzie Mountains, north-west Canada). 2, as fig. 1, with hypothetical appendages added. 3, dorsal (a) and ventral (b) views of small protaspis of I. parvirugosus (same source as fig. 1 ). 4, dorsal (a) and ventral (b) views of protaspis of Diacanthaspis cooperi Whittington, 1956 (his fig. 9a, b, and pi. 3, figs. 1, 2, 5, 6) (Middle Ordovician, Martinsburg Shale, Virginia). 5, dorsal (a), ventral (b), and lateral (c) views of smaller of two protaspides of Acanthopyge bifida Edgell (after Chatterton 1971, fig. 9a, b, d; pi. 7, figs. 1-4) (Lower Devonian, Receptaculites Limestone, nr. Yass, New South Wales). 6, smallest protaspid stage of Bathyuriscus ? sp. (source as figs. 7 and 8). 7, dorsal (a) and ventral (b) views of third and largest protaspid stage of Spencella (Middle Cambrian, ?Pika Formation, Canadian Rocky Mountains, nr. Columbia Ice Fields). 8, as fig. 7, dorsal (a) and ventral (b) views of smallest of three protaspid stages of Spencella. 9, dorsal (a) and ventral (b) views of protaspis of Dentaloscutellum hudsoni Chatterton, 1971 (after his fig. 4a, b; pi. 2, figs. 1-3; pi. 3, figs. 5 and 6) (locality as fig. 5). 10, dorsal (a) and ventral (b) views of largest of three protaspis stages of Flexicalymene senarial (Conrad), nov. herein (specimens coll. W. R. Evitt; locality as fig. 4; see also Whittington 19596 and Hu 1971). 1 1, same species, smallest of three protaspid stages in dorsal (a) and ventral (b) views (locality as fig. 4). 12, dorsal (a) and ventral (b) views of small protaspis of Pseudocybele nasuta Ross, 1951 (see Ross 1951a; drawings based on photographs by B. D. E. C. from specimens loaned to us by H. B. Whittington) (Lower Ordovician, Garden City Formation, Utah). 13, dorsal (a) and ventral (b) views of late protaspid stage of Proetus talenti Chatterton, 197 1 (after his fig. 15c, d; pi. 15, figs. 7 and 33; pi. 16, fig. 1) (Lower Devonian Warroo Limestone, nr. Yass, New South Wales). FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 183 ASAPHINA (Asaphidae) 3A 3B PTYCHOPARIIDA 8A 8B CALYMENINA ODONTOPLEURIDA SCALE 0.5 mm LICHIDA SCUTELLUINA PHACOPINA PROETIDA 184 PALAEONTOLOGY, VOLUME 31 especially the posteromedian spine; the hypostome covered a proportionately smaller part of the ventral surface, with shorter marginal projections and a greater space between hypostome and doublure laterally; the doublure changed from enrolled to inturned at the margin (terrace lines probably appear at this stage); furrows in the dorsal surface became more distinct. If the ontogenies of asaphids and nileids are contrasted with those of the remopleuridids, it may be seen that where the metamorphosis is delayed more mature characters appeared before the metamorphosis took place, and the subsequent changes were not as radical. The thin-shelled, globular asaphoid protaspides were poorly adapted to a benthic life. Their ventral surfaces were almost completely covered by the hypostoma, and there was very limited space for movement of the proximal parts of the appendages in the anterior and posterior directions because of the location of the marginal spines on the hypostome (e.g. text-fig. 11.9). In several taxa the posterior pair of submarginal spines projects ventrally, and would have dragged in the substrate if close to the sea-floor. We regard the asaphoid protaspis as certainly planktic, and the metamorphosis corresponding with a change in mode of life from planktic to benthic. Asaphoid protaspides of various trilobite families Nileid protaspides ( PI. 18). Nileid protaspides are described for the first time herein. Several different protaspides were obtained from low in the Profilbekken Member of the Valhallfonna Formation (Upper Arenig), northern Spitsbergen. Well-preserved, phosphatized trilobite larvae have been described from this formation by Fortey and Morris (1978). We are unable to associate particular protaspides generically, because several genera ( Nileus , Poronileus, Peraspis, see Fortey 1975) have been recovered from the stratigraphic interval that yielded them. There is more than one type of nileid protaspis, and more than one growth stage is present for each. The morphological characteristic that immediately distinguishes these nileid protaspides from all other protaspides is the presence of a keyhole-shaped re-entrant posteroventrally. The smaller protaspides are slightly wider and more spherical than later stages, which are ovoid in shape. All these protaspides are believed to have had three pairs of submarginal spines: an anterior pair projecting upwards and somewhat forwards and laterally, a median pair near the back of the protocranidium projecting nearly dorsolaterally, and a posterior pair projecting subventrally from a short distance in front of the keyhole-shaped re-entrant. text-fig. 11. Asaphoid protaspides contrasted with protaspides of a Middle Cambrian ptychoparioid, Spencella. 1, dorsal (a), ventral (b), and lateral (c), views of larger of at least two protaspides of a nileid (basal 50 m of Lower Ordovician Profilbekken Member, Ny Friesland, Spitsbergen; Nileus , Peraspis , and Poronileus species occur in this interval— see Fortey 1975); free cheeks and hypostome not shown. 2, dorsal (a) and ventral (b) views of smaller of two protaspid stages of nileid, probably same genus as fig. 1 (same locality as fig. 1). 3, dorsal (a) and ventral (b) views of incomplete nileid protaspis (missing parts hypothetically reconstructed, based on fig. 1), without free cheeks and hypostome (base of Profilbekken Member, as fig. 1, where Nileus, Peraspis, and Poronileus occur). 4, lateral (a) and dorsal (b) views of protaspis of Remopleurides caelatus Whittington, 1959a (after his pi. 3, figs. 2 and 4) (Middle Ordovician, lower Edinburg Formation, Virginia); without free cheeks or hypostome. 5, lateral view of first meraspid stage (MO) of R. caphyroides Whittington, 1959a (after his fig. 6); without free cheeks (locality as fig. 4). 6, lateral view of R. pattersoni Chatterton and Ludvigsen, 1976; without cheeks of hypostome (after Chatterton 1980, fig. 3) (Middle Ordovician, Esbataottine Formation, north-west Canada). 7, lateral (a), dorsal (b), and ventral (c) views of protaspis of R. sp. aff. R. eximius (Edinburgh Formation, Middle Ordovician, Virginia) (drawings taken from unpublished photographs ofW. R. Evitt); cheeks attached but no hypostome visible. 8, as fig. 7, a second specimen. 9, ventral (a), lateral (b), and anterior (c) views of large protaspis of Isotelus sp. (Middle Ordovician, Virginia) (after Evitt 1961, fig. 3). 10, ventral view of protaspis of /. sp. (locality as fig. 9) (after Evitt 1961, fig. 2c). 1 1, ventral (a) and lateral (b) views of small protaspis of /. sp. (locality as fig. 9) (after Evitt 1961, fig. 1a, c). 12, dorsal (a), ventral (b), and lateral (c) views of largest of three protaspid stages of Spencella sp. (Bathyuriscus-Elrathina Zone, Middle Cambrian, Pika Formation?, Canadian Rocky Mountains) (based on new photographs of undescribed material). 1 3, dorsal (a) ventral (b) and lateral (c) views of the smallest of three protaspid stages of Spencella (locality as fig. 12). 14, dorsal (a), ventral (b), and lateral (c) views of protaspis of Cryptolilluis (nov. herein) (horizon as fig. 7). FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 185 186 PALAEONTOLOGY, VOLUME 31 The doublure is narrow and enrolled rather than inturned. Very short lobes or spines may occur behind the large, posterolateral pair of spines, between those spines and the ‘keyhole’. The smaller protaspid growth stages have shallower axial furrows than larger ones, with a pair of more distinct subtriangular impressions at their forward ends near the anterior margin. Hypostomes and cheeks are not known in the protaspides of this family at present. The facial sutures do not extend as far back in these protaspides as they do in those of Asaphidae. The largest protaspis (PI. 18, fig. 1 1) is larger than ptychoparioid ones and indicates that the metamorphosis entailing the shift from planktic to benthic life may have been somewhat delayed; the earliest meraspid cranidia of nileids that we have recovered are like those of the holaspis in many features, and a ‘jump’ in size during metamorphosis is considered possible. Asaphid protaspides. Protaspides of Isotelus have been described by Evitt (1961), Hu (1971), Chatterton (1980), and Tripp and Evitt (1986). They have one or two pairs of submarginal conical spines on the cranidium/protopygidium, are subspherical, have nine pointed marginal spines on the hypostome, and initially have the hypostome fused to the free cheeks. The median connective suture appears in later stages in a single step, but often in front of a short, tongue-like extension of the hypostome, as discussed above. Dorsal furrows are absent in small growth stages, except for a pair of subtriangular pits near the anterior margin representing the limits of the axial furrows; such furrows appear in the largest protaspides. Genal spines are minute — and only known in the largest protaspid stage of a species of Isotelus (Evitt 1961). The facial sutures extend very far back, almost joining posteromedially on the ventral surface. Remopleuridid protaspides. These have been described by Ross (1951a and in Whittington 19596), Whittington (1959a), and Chatterton (1980) and are figured herein (PI. 17, figs. 1-6). Those described by Chatterton occur with common remopleuridid adults and abundant sclerites of both protaspid and adult stages of Isotelus— but they are as similar to asaphid protaspides as they are to most of the undoubted remopleuridid protaspides described by Whittington. The question arises whether these were correctly assigned to Remopleurides (which is found with them) or whether they belong to some asaphid taxon not represented by adult remains in the same beds. The protaspides described by Whittington are subspherical, have three large pairs of sharp, conical spines, and fused free cheeks; they are so curved in a sagittal plane that there is only a comparatively small part of the ventral surface not covered by either the cranidium/protopygidium or the free cheeks. Whittington ( 1959a, b) did not illustrate protaspides with hypostomes. Evitt (herein text-fig. 11.7) discovered protaspides of R. eximiusl Whittington, 1959 with pairs of fused cheeks attached. In one of these (PI. 17, fig. 1) the hypostome is attached and projects inside the protopygidium. In a number of remopleuridid protaspides additional small spines project subventrally between and behind the third large pair of submarginal spines on the cranidium/protopygidium. These spines are not preserved on the protaspides described by Chatterton (1980) or the specimens of R. eximiusl illustrated here. The posterior branch of the suture does not extend as far back on the ventral surface as it does in Isotelus. EXPLANATION OF PLATE 18 Figs. 1-11. Internal phosphatic moulds of the early growth stages of nileid trilobites from the lower part of the Proiilbekken Member, Valhallfonna Formation, Spitsbergen (with the exception of fig. 7, in which the original calcite shell material is preserved). The following nileid taxa occur in this stratigraphic interval (from Fortey 1975, fig. 1): Nileus orbiculatoides svalbardensis Fortey, 1975, N. porosus Fortey, 1975, Peraspis erugata Ross, 1970, Poronileusfistulosus Fortey, 1975, and P. isoteloides Fortey, 1975. 1, BMNH It20580, dorsal view offree cheek, x 62. 2, BMNH It20581, dorsal view of small cranidium, x 55. 3, BMNH It20582, dorsal view of small cranidium, x 42. 4, BMNH It20587, dorsal view of transitory pygidium, x 60. 5, BMNH B20589, dorsal view of larger protaspis, x47. 6, BMNH It20593, ventral view of partial larger protaspis, x 46. 7, BMNH It20590, dorsal view oflarger protaspis, x 54. 8, BMNH It20585, dorsal view of posterior portion of large protaspid stage, x 47. 9, BMNH It20592, dorsal view of partial smaller protaspis, x 75. 10, BMNH It20583, ventral view of hypostome, x 57-5. 11, BMNH It20581, ventral view of partial smaller protaspis, x 74. PLATE 18 FORTEY and CHATTERTON, nileid trilobites 188 PALAEONTOLOGY, VOLUME 31 Dorsal furrows and an independent convexity of the axis are present on large protaspides of this family. The metamorphosis accompanying the change from pelagic to benthic habits takes place in the meraspid period of several species of Remopleurides ( R . caphyroides, R. eximius, R. asperulus, R. plaesiourus, see Whittington 1 959a). Trinucleacean protaspides. The protaspides of the Trinucleacea are similar in many ways to those of the higher Asaphina on text-figs. 1 -3. Resemblances between them include the globose shape, the enrolled rather than inturned margin on the doublure, the presence of up to three pairs of submarginal conical spines on the cranidum/protopygidium, the fused free cheeks and hypostome, and the long, sharp marginal spines on the hypostome. Trinucleacean protaspides have been figured by Whittington (1959a: Cryptolithus, Tretaspis , Am pyx), Shaw (1968: Lonchodomas), and herein (PI. 19: Cryptolithus , Ampyxoides or Globampyx). Meraspid to holaspid growth stages have also been described for this superfamily by a number of other workers, including Hu (1971). We discuss below the additional reasons that suggest that Trinucleacea might be included in an enlarged concept of Asaphina. Trinucleacean protaspides differ from other asaphoid ones in the distinct convexity of the axis, the axial and other dorsal furrows being more distinct at the same size. The protaspides of this group are also comparatively small. The posterior pair of submarginal spines are very close to one another, and may be crossed (Whittington 1959a; Shaw 1968), and a second pair of spines may be present adjacent and anteroventral to this pair. This additional pair of small spines may be homologous with subsidiary marginal spines found between the large posterior pair of spines which occurs in some Remopleuridacea (see Whittington 1959a). The free cheeks of both protaspid and small meraspid stages (down to Meraspis degree 0) of the Trinucleacea are fused to one another to form a lower lamella, EXPLANATION OF PLATE 19 Figs. 1-9. Cryptolithus tesselatus Green, 1832 from the Lower Martinsburg Shale, about 1 km along Virginia secondary highway (at the base of the roadside outcrop along the north side of the road) from its intersection with Virginia secondary highway 732 (not quite as far along this road as Loc. 12 of Whittington 1959). The trilobites from this locality have been somewhat distorted as a result of stress. 1, USNM 414587, dorsal view of protaspis, x51. 2, USNM 414588, ventral view of protaspis, x46-5. 3, USNM 414589, ventral view of fused partial cheeks and hypostome, x 93. 4, USNM 414590, ventral view of partial protaspis with attached fused free cheeks and hypostome, x 62. 5, USNM 414591, dorsal view of fused free cheeks and hypostome (note posterior portions of free cheeks and hypostome are missing), x 93. 6, USNM 414592, ventral view of protaspis with attached fused free cheeks and hypostome (note marginal spines and posterior part of hypostome missing), x 78. 7, USNM 414593, ventral view of protaspis with attached fused free cheeks and hypostome (note long ventrolaterally directed anterior pair of spines on hypostome), x 77. 8, USNM 414594, dorsal view of protaspis, x 62. 9, USNM 414594, ventral view of protaspis, x 62. Fig. 10. Cybeloides sp., UA 7764, from strata of Late Ordovician Age in the Mackenzie Mountains (horizon AVI 54 of Nowlan et al. in press); ventral view of protaspis with attached free cheek, rostral plate, and hypostome, x 56. Figs. 11 and 12. Raphiophorid from the Lower Ordovician, lower part of the Profilbekken Member, Valhallfonna Formation, Spitsbergen Island. Ampyxoides inermis Fortey, 1975 and Globampyx trinucleoides Fortey, 1975 occur at this horizon (Fortey 1975, fig. 1). 11, BMNH It20596, ventral view of internal phosphatic mould of protaspis, x 108. 12, BMNH It20597, dorsal view of internal phosphatic mould, x 1 10. Figs. 13 and 14. Nileid from the Lower Ordovician, basal part of the Profilbekken Member, Valhallfonna Formation, Spitsbergen. The following nileids have been obtained from this interval: Nileus orbiculatoides svalbardensis Fortey, 1975, N. porosus Fortey, 1975, Peraspis erugata Ross, 1970, Poronileusfistulosus Fortey, 1975, and P. isoteloides Fortey, 1975 (see Fortey 1975, fig. 1). 13, BMNH H20598, ventral view of phosphatic mould of protaspis, x 50. 14, BMNH It20599, dorsal view of internal phosphatic mould of protaspis, x 55. Fig. 15. Pseudocybele nasuta Ross, 1951, BMNH It20594, from the Lower Ordovician Garden City Formation (Zone J), Loc. 13 (see Ross 19516), north-eastern Utah; ventral view of small protaspis stage (other later ones occur in the ontogeny of this species) with attached free cheeks, rostral plate (largely hidden under hypostome), and hypostome, x 90. PLATE 19 FORTEY and CHATTERTON, asaphine protaspides 190 PALAEONTOLOGY, VOLUME 31 with no connective suture visible. The protaspid cheeks are also fused to the hypostome, which has two pairs of marginal spines. The cheeks of small meraspid stages of other Asaphina are not fused to one another (although they may be fused to one another in early protaspid and late holaspid stages). No rostral plate is known for either group at any stage. The absence of connective suture(s) at any known stage in the Trinucleacea can be interpreted in more than one way. It could be the result of a complete absence of such a suture in the phylogenetic history of the group; this would set it apart from both the Asaphina as understood here, and from the Ptychopariida (even as an ‘ancestral’ group). Or it could be the result of retention by paedomorphosis of the larval lack of sutures; the same effect would be achieved by accelerated development with secondary fusion of the connective sutures. The earliest trilobite growth stage, the phaselus of Fortey and Morris (1978), lacks ventral sutures, and there is no reason in principle why this condition should not be fixed during subsequent development. The resemblance of early trinucleines such as Orometopus to asaphines and ptychopariids, including the presence of opisthoparian sutures, is such that it does seem probable that all belong within a monophyletic group — and makes the developmental explanation the more likely origin of the fused cheeks. The resemblance of the trinucleacean protaspis to other asaphoid protaspides is one line of evidence which inclines us to include the former group within the Asaphina. The crucial test of this will be the discovery of an early representative of undoubted Trinucleacea with an ontogeny complete enough to say whether or not the primitive condition for the group was with rostrum or median suture. This is not yet known, but we present a case below for including the Cambrian family Liostracinidae in the Trinucleacea. The former family does include species with a median suture, and if we are correct in our use of this as a synapomorphy of Asaphina, both it and Trinucleacea can be included within the Asaphina. Comparison of asaphoid protaspides with those of other groups Globular protaspides with three pairs of conical submarginal spines are known also in Calymenina (Apocalymene, Chatterton 1971; Flexicalymene , Hu 1971), Phacopida (e.g. Phacops, Chatterton 1971), and Cheirurina (Hyrokybe Chatterton and Perry, 1984; ISphaerexochus , Chatterton 1980). Late stage protaspides of these groups differ in most respects from those of the Asaphina as we understand it here. However, the presence of three pairs of marginal spines could be considered plesiomorphic on the basis of out-group comparisons; but there are doubts that these spines are really homologous between different groups of trilobites, as we discuss under Phacopida below. The conical marginal spines on the hypostome are probably a plesiomorphic character because they are present on the majority of protaspides of various families, including those from the Lower Cambrian. The number of marginal spines is usually seven or nine in the Ptychopariida and related orders of trilobites. There appear to be nine marginal spines on the hypostomes of many Asaphina (Evitt 1961; Chatterton 1980) and some Phacopida (Calymenina and Phacopina, Chatterton 1971; Cheirurina, Chatterton 1980). However, a globular protaspid form is not a plesiomorphic character, because it is not present in protaspides of primitive ptychoparioids, nor, to employ a stratigraphic criterion, in any known protaspis from the Lower Cambrian. The sporadic occurrence of such early protaspides in groups other than the Asaphina is presumably a parallelism associated with early planktic growth stages. Later development of the protaspis avoids confusion with the asaphoid morphology. The loss of one or two of the pairs of spines on the cranidium/protopygidium of these bulbous protaspides apparently occurred independently several times, and is not phylogenetically important (Remopleuridacea, Chatterton 1980 and herein; Trinucleacea, Whittington 1959a; Asaphidae, Evitt 1961). This feature is known to have varied within the single genera Isotelus and Remopleurides. The presence of additional spines on the protopygidium may prove to be an additional autapomorphy of the Remopleuridacea. Asaphoid protaspides compared with phacopid protaspides. Asaphoid protaspides differ from phacopid protaspides in lacking distinct axial and glabellar furrows, in lacking a sagittal furrow behind the frontal lobe of the glabella in all protaspid growth stages, and in lacking distinct lobes along the front of the cranidium on either side of the glabella. Evidence from late protaspid stages of phacopoids (B. D. E. C. unpublished) shows that all three pairs of prominent submarginal spines are FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 191 cephalic, and therefore not homologous with those of the Asaphina, and that protopygidial spines develop behind these three pairs (see text-figs. 10-11). Many phacopoid protaspides ( Pseudocybele , Encrinuroides, Encrinurus, Cybeloides, PhacopsT) have seven marginal spines on the hypostome compared with nine on the hypostomes of Asaphidae, four in the Trinucleacea, and two in the Remopleuridacea. However, calymenids usually have nine such spines on their protaspid hypostomes (Chatterton 1971; later growth stages show loss of the ninth, posteromedian spine, as shown by Whittington 1959a). Protaspis hypostome marginal spine number does not seem promising as a character for high level classification. Some of the smaller protaspides of members of the Phacopida, especially Calymenina, share characters with asaphoid protaspides that are considered to be convergent, as the result of their pelagic life habits. These include a bulbous shape and an enrolled rather than inturned doublure, together with a spinose hypostoma that covers most of the ventral surface. The presence of prominent anterolateral lobes on the cranidium, paired connective sutures, and distinct sagittal furrows on the glabella clearly distinguishes them from their asaphine functional equivalents. Asaphoid protaspides compared with bumastine protaspides. Chatterton (1980) assigned some small bulbous protaspides to the bumastine Failleana. These protaspides are similar, in some respects, to the asaphoid protaspis. However, they differ from the latter in having lateral connective sutures, a hypostome fused to what later becomes the rostral plate, a re-entrant posterior margin, and only one pair of submarginal spines. Asaphoid protaspides compared with primitive ptychoparioid protaspides. Dorsal and ventral views of protaspides from several families of generalized ptychoparioids have been described by Palmer (1958, Crassifimbra , Lower Cambrian; 1962a, Aphelaspis , Upper Cambrian), Hu (1971, Dunderbergia , Dytremacephalus, Upper Cambrian; 1986; Pachyaspis, Ehmaniellal), and one is added herein ( Spencella , Middle Cambrian, see PI. 17). They are all similar to one another, and are presumed to represent the primitive condition. All have a pair of connective sutures on either side of a transverse rostral plate, three pairs of conical submarginal spines on the cranidium/protopygidium, a slightly inflated discoid shape in the smallest protaspides, with a concave posterior margin. Some of the smallest protaspid stages have a sagittal furrow subdividing the median two-thirds of the glabella into four to six lobes. The protaspid hypostomes are spinose along their margins where known. Dorsal surfaces of ptychoparioid protaspides have been described for a much greater number of taxa, especially by Hu (1971), with a good account of olenid ontogeny (Upper Cambrian) by Whitworth (1970), but since this material is not silicified it is not possible to determine the courses of the ventral sutures, nor whether pairs of submarginal spines are present. Axial furrows are, however, distinct in ptychopariid protaspides, the glabella so defined being subparallel-sided to forward- expanding. Sagittal furrows are present in the glabellas of smallest protaspides of almost half the described ptychopariid taxa— these never continue on the frontal or occipital lobes, and usually disappear in the later protaspis. Similar sagittal furrows occur in small protaspides of redlichiids, and some phacopinids and calymenines. The occipital ring is often more inflated than the rest of the glabella. The proportion of the smallest protaspis (‘anaprotaspis’ of authors) occupied by the protopygidium is inconspicuous; and conspicuous, with furrows, ridges, and additional spines, in the later protaspis (‘metaprotaspis’). Palmer (1962a) considered the three pairs of submarginal spines that occur in each of Aphelaspis, Glaphyraspis, and Hardyoides to be homologous, and all to be cephalic. They are probably also homologues of the three pairs of submarginal spines of the phacopoid Pseudocybele, which are also cephalic (see text-fig. 10; Ross 19516). They are apparently not homologous with the three pairs of submarginal spines in protaspides of Remopleurides, only two of which have been considered cephalic (Whittington 1959a, pis. 3 and 10). Differences between asaphoid and ptychoparioid protaspides include: the more bulbous shape of the former; the asaphoid always lack sagittal furrows on the glabella; they also lack distinct (?palpebro-ocular) anterior cephalic ridges; at the same size axial furrows are shallower in asaphoid protaspides; where the sutural junction between the free cheeks can be recognized in asaphoid protaspides it is single and median, and this appears as a discrete structure early in ontogeny. 192 PALAEONTOLOGY, VOLUME 31 The most primitive families which we include in Asaphina do not necessarily have an asaphoid protaspis: this appears as a synapomorphy uniting Remopleurididae to Cyclopygidae on text-figs. 1-3. Unfortunately, there are few described ontogenies of Cambrian forms with a median suture in the adult. As we noted above, Hu (1971) described the ontogenies of three genera of Cambrian trilobites assigned to the Pterocephaliidae ( Dytremacephalus ), Anomocaridae (Glyphaspis), and Dikelo- cephalacea ( Ptychaspis ), respectively. Of these only the anomocarid could be compared with the asaphoid protaspis in its smaller stages; the other two resemble those of ptychoparioids. As doubt has been cast on some other ontogenetic series described by Hu we did not code these examples in the computer treatment of phylogeny. Glyphaspis and Ptychaspis are known from rock specimens only. However, it is perfectly possible that the asaphoid protaspis type was acquired after the inception of the median suture, as implied in text-fig. 1. If the distribution of characters shown in text-figs. 1-3 is correct, and Hu’s protaspis assignment to Ptychaspis is also correct, the primitive ptychoparioid protaspide form was retained (or achieved by reversal) in Dikelocephalacea. Clearly more information on early ontogeny of these Cambrian taxa is desirable. If we are correct in regarding protaspis morphology as an important criterion in classification, the question arises whether Asaphina should be used in a restricted sense to embrace only the families Remopleuridacea to Cyclopygidae on text-figs. 1-3. It could even be questioned whether Asaphina in this restricted sense and Ptychopariina should be classified in the same order. There are more differences between larval stages of these two groups than between Ptychopariina and Phacopida, which are ordinally separate. The Order Proetida Fortey and Owens, 1975 is also recognizable from protaspis characteristics. Small protaspis stages of Corynexochida resemble those of Ptychopariina, nor can protaspides of Redlichiida and Agnostida be regarded as more similar to those of Asaphina. However, it is probably the case that the resemblances between ptychopariid, corynexochid, redlichiid, and even phacopoid protaspides (such as the well-defined, subparallel-sided or forward-expanding glabella with defined furrows) are shared primitive (symplesiomorphic) characters, and as such should not be employed in the definition of taxa. Certainly, if we are correct in believing that the presence of a median suture is an indication of monophyly then the Asaphina so defined is likely to have included early members which had not yet acquired the asaphoid protaspis, and the more embracing definition of Asaphina is preferred here. However, if it is subsequently shown that the median suture is polyphyletically derived, then the protaspis character will presumably again achieve prominence in the definition of Asaphina. The general conservatism of protaspis morphology, when compared with wide variations in adult morphology, increases the level of confidence one can place in using major modifications from the primitive morphology in the definition of high level taxa. This notion is falsifiable from our cladograms, in that we predict that the asaphoid protaspis will be discovered for numerous species, for example of Cyclopygidae, for which ontogenies are not yet known. SYSTEMATIC DISCUSSION OF ASAPHINE SUPERFAMILIES Suborder asaphina Salter, 1864 emend. Diagnosis. Trilobites having a ventral median suture initiated early in ontogeny (and only lost where cheeks are secondarily fused); protaspis commonly globular and effaced dorsally, of asaphoid type. Pre-occipital glabellar tubercle present in advanced forms. General characters of the group may include retained primitive features: dorsal sutures opisthoparian; eyes usually moderately large to large (but some forms secondarily blind); thoracic articulation unspecialized (exception: advanced remopleuridids and later ceratopygids such as Dichelepyge}, subisopygous to macropygous, with marginal pygidial spines in many groups; doublure with terrace ridges. Superfamily asaphacea Burmeister, 1843 The superfamily Asaphacea is regarded as including two families: Asaphidae and Ceratopygidae; the latter family has been accorded superfamilial status in previous classifications (Moore 1959; FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 193 text-fig. 12. Characteristic structure of glabella of Asaphacea, showing pre-occipital tubercle between elongate (exsag.) Lx glabellar furrows, a , Hedinaspis , which has been classified as an olenacean (see text), b, Proceratopyge (Ceratopygidae). c, Niobella (Asaphidae). Approximately natural size. Bergstrom 1973; Shergold 1975). We exclude the families Nileidae and Taihungshaniidae from the superfamily, in which they have been previously incorporated. Thus defined, the superfamily ranges from the Middle Cambrian to the end of the Ordovician. Diagnosis. Asaphina with glabellar tubercle developed on pre-occipital segment; glabellar form primitively elongate (sag.) subparallel to tapering forwards, with defined occipital ring, and with curved, apostrophe-like pair of basal glabellar furrows isolated within glabella. Hypostome attached to doublure; hypostomal condition conterminant, rarely impendent. Discussion. The development of the glabellar tubercle, glabellar furrows, and occipital ring which are typical of the superfamily is clearly shown on the stratigraphically early ceratopygid Proceratopyge from the mid- to late Cambrian (Westergard 1947, pi. 2, figs. 1 and 2; Rushton 1978, pi. 26, fig. 4), and on such early asaphids as Promegalaspides (Westergard 1939, pi. 1, fig. 3 a) and Niobella (Tjernvik 1956, pi. 4, fig. 14). The characteristic arrangement with the glabellar tubercle lying between crescentic or apostrophe-shaped basal glabellar furrows some distance forward from the occipital ring is shown on text-fig. 1 2; no other trilobites of which we are aware have exactly this axial structure, which is our preferred evidence for common ancestry of the Ceratopygidae and Asaphidae. As noted above, the PAUP analysis recognizes Asaphidae + Ceratopygidae as a paraphyletic group, rather than a monophyletic group, on the totality of character distribution, and whether or not Asaphacea and Ceratopygacea are recognized as distinct superfamilies depends on which interpretation is preferred. For the moment we follow previous usage (e.g. Shergold and Sdzuy 1984) in including the two families Asaphidae and Ceratopygidae within a single taxon, here given the status of superfamily Asaphacea. In any case it is clear from all our phylogenetic analyses that none of the other families included within the Asaphacea in the Treatise by Moore (1959: Nileidae, Taihungshaniidae, Dikelokephalinidae, and Tsinaniidae) should be included in the same group. They have different axial cephalic and pygidial structure, and are accordingly excluded from the superfamily. The characteristic arrangement of cephalic furrows and median tubercle is found in the earlier Asaphacea. Stratigraphic evidence points to modifications to this arrangement, which are considered in the following paragraphs. Family asaphidae Burmeister, 1843 Diagnosis. Asaphaceans with eight thoracic segments; pygidial margin not incorporating anterior spinose segment; librigenal borders wide (or border furrow effaced); facial sutures supramarginal in front of glabella; later species often with forked hypostome and/or panderian openings; genal spine generally short, and wide at base. Discussion. The generalized glabellar structure is widespread among early asaphids, but the occipital ring tends to effacement even in late Cambrian forms. On Golasaphus Shergold, for example, the basic structure is visible on G. triquetrus (Shergold 1975, pi. 56, fig. 7), but the occipital furrow has become effaced on G. simus (Shergold 1975, pi. 55, fig. 2). Those asaphids classified in the subfamily Niobinae PALAEONTOLOGY, VOLUME 31 d e f g text-fig. 13. The petaloid thoracic facet (a and b ), typical of higher Asaphina, compared with facets of other trilobites (c-g). a, asaphid, Asaplms sp. BM 15430 (Middle Ordovician, Russia), x 8. b, nileid, Symphysurus palpebrosus , BM It20684 (Middle Ordovician, Sweden), x 8. c, granulose dalmanitacean facet, Dalmanites caudatus , BM I67a (Silurian), with postfacetal type of pleural furrow, x 8. d , primitive type, smooth, in the Silurian proetide Otarion diffraction , BM 13603, x 12. e,f primitive type in the Middle Cambrian ptychoparioid Elrathia\ e, BM It5396, dorsal surface, x 12;/, BM It5397, narrow doublure with terrace ridges primitive for Asaphina, x 15. g, granulose calymenacean facet, with epifacetal type of pleural furrow, Calymene sp., BM It20685 (Upper Ordovician, Anticosti Island), x 10. often maintain both the tubercle and the occipital ring in the same relative positions (e.g. Niobe, see Fortey 1975, pi. 5, fig. 2), but there is a tendency for the tubercle to migrate backwards to a nearly occipital position, a process carried furthest in Ogygiocarella. The pre-occipital origin of the tubercle is still betrayed by a marked backward curvature of the occipital furrow to form an embayment around the tubercle on the mid-line (Ogygiocciris, see Rushton and Hughes 1981, pi. 4, fig. 2). Niobines and ogygiocaridines retain an elongate glabella, but with the frontal lobe tending to enlarge; FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 195 text-fig. 14. Advanced asaphid (a) and ceratopygid (6) glabellas, showing retention of pre-occipital tubercle in more derived members of these families. a , Asaphus , BM I 5701, x 2. 6, Ceratopyge, BM It 1 2895, x 4. b bacculae are present on many species, but are apparently lacking on the primitive forms figured by Westergard (1939). Baccula-like structures are present on early asaphid growth stages. In Asaphinae the glabellar shape is much modified — becoming hourglass-shaped or expanding forwards, and often incorporating genal material within the axial structure (Fortey 1980); Asaphinae attain impendent hypostomal condition. Efifacement is common. The glabellar tubercle is still pre-occipital in asaphines, a fact revealed in those less eflfaced genera where the occipital ring is still defined dorsally ( Asaphus , see Neben and Krueger 1971, pi. 5, fig. 7; N orasaphus Fortey and Shergold, 1984; text-fig. 14 herein). The constancy of the pre-occipital tubercle within this diverse family lends support to considering it a feature of taxonomic importance. All asaphids have eight thoracic segments, a character stabilized within the Cambrian and invariant thereafter; other families of Asaphina are more variable in this regard. We do not consider the subfamily division of the Asaphidae further here. The forked hypostome of many species is typical of the family but cannot be used in its definition because it is unforked in primitive species; the hypostome invariably carried smooth maculae. Family ceratopygidae Linnarsson, 1869 Diagnosis. Asaphaceans with variable number of thoracic segments (nine to six); hypostomal condition conterminant; primitive asaphacean glabellar furrows retained throughout family; genal spines generally narrow and needle-like; cephalic rim narrow; pygidium typically incorporating one (or two) pairs of marginal spines which are extensions of pygidial segments. Discussion. Glabellar structure in ceratopygids is conservative compared with asaphids. The characteristic asaphacean glabellar structure is shown on all genera for which well-preserved material is known, e.g. Proceratopyge (Westergard 1947; Shergold 1982, pi. 16, fig. 9), Ceratopyge (text-fig. 146 herein; Flarrington and Leanza 1957, fig. 94.7), Dichelepyge (Fortey and Owens 1982, pi. 3, fig. i), Pseudohysterolenus (Harrington and Leanza 1957, fig. 98.2a), Hysterolenus (Lu and Lin 1984, pi. 17, fig. 3), and Diceratopyge (Peng 1984, pi. 5, fig. 16). Later ceratopygids, especially Ceratopyge itself, have the glabella modified to become forward-expanding and with a corresponding reduction in the preglabellar area. The family is best defined by the inclusion in the pygidium of one or more macropleural segments, although the narrow cephalic rims of most species are also characteristic (but see Haniwoides, Shergold 1980, fig. 37; text-fig. 15a herein). The primitive glabellar structure of ceratopygids seems to us to be indistinguishable from that of primitive asaphids (text-fig. 1 2). Rushton ( 1 983, text-fig. 66) has shown how the doublure on Proceratopyge curves backwards medially to reach 196 PALAEONTOLOGY, VOLUME 31 text-fig. 1 5. Cranidia of additional taxa here included in Asaphacea, showing diagnostic glabellar structure (cf. text-fig. 11). a, small cranidium of Haniwoides (Upper Cambrian, Australia) (after Shergold 1980, pi. 4), x 8. b , Macropyge, type species Macropyge chermi (Tremadoc) (new restoration), x4. the front of the glabella, and such curvature is reflected by a paradoublural line on the dorsal surface of the preglabellar field of some species (Westergard 1947, pi. 2, fig. 1; see Dikelokephalinidae below). Hypostomal condition conterminant throughout the family. As might be expected with the close relationship between ceratopygids and asaphids proposed here there are early members of the Asaphacea with intermediate combinations of characters which pose problems for classification. Shergold (1980, p. 86) noted the intermediate features of the subfamily Iwayaspidinae Kobayashi, 1962, between what he regarded as the superfamilies Asaphacea and Ceratopygacea. He placed this subfamily within the Ceratopygidae even though macropleural segments are lacking on the pygidium. Based on the morphology of the genus Cermatops Shergold, 1980, we agree with the assessment. In particular, the pygidial pleural segmentation is of a kind seen in other, macropleural ceratopygids in which the interpleural furrow encroaches closely upon the pleural furrow of the segments behind; the narrow and convex librigenal rim is also unlike the border of asaphids. OTHER ASAPHACEA HITHERTO CLASSIFIED IN DIFFERENT SUPERFAMILIES Subfamily macropyginae The Tremadoc genus Macropyge Stubblefield, 1927, has been compared with the Remopleurididae, and was classified there in the Treatise (Moore 1959) and elsewhere (e.g. Peng 1984). Recent descriptions of the cephalic morphology (Fortey and Owens 1982; Shergold and Sdzuy 1984) have shown that the cephalic features are not at all like those of remopleuridids, although Shergold and Sdzuy still tentatively classified Macropyge with the Remopleurididae. Kobayashi (1953) had previously proposed a separate family for its reception. It is clear, however, that the features of the cephalic axis are those of the Asaphacea as described here, with a pre-occipital tubercle well in advance of a clearly defined occipital ring (Fortey and Owens 1982, pi. 2, fig. h; Shergold and Sdzuy 1984, fig. 62; text-fig. 156 herein). Other features of the cephalon are like those of ceratopygids such as Haniwoides which lack pygidial spines. In fact the free cheeks, with their narrow rims and paradoublural ridge, are characteristically ceratopygid, and the palpebral lobes lack the inflated rims typical of remopleuridids; hence it is likely that the resemblance to the latter family is one of convergence only. If Macropyge is a ceratopygid, it is allied to those genera lacking pygidial spines placed in the subfamily Iwayaspidinae discussed above, and Macropyginae can be a senior name for that subfamily. Two other genera were assigned to the Macropyginae by Shergold and FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 197 text-fig. 1 6. Glabella of type species of Hedinaspis , H. regalis Troedsson, to show asaphine construction; Hedin Collection, Riksmuseet, Stockholm, Ar 47250, x 3. Sdzuy (1984): Promacropyge Lu, 1965 and Aksapyge Lisogor, 1977. The former is usually only regarded as a subgenus of Macropyge (e.g. Lu and Lin 1984), while the latter is doubtfully distinct from Promacropyge. In summary, Macropyge and its close relatives are accommodated as subfamily Macropyginae within the Ceratopygidae, together with those genera listed by Shergold (1980) as belonging to the subfamily Iwayaspidinae. To these we add the genus Tamdaspis Lisogor, 1977, which is extremely like Cermatops Shergold, 1980. The Macropyginae spans the Cambrian Ordovician boundary, as do the Ceratopyginae and Asaphidae. Genus hedinaspis Hedinaspis (type species H. regalis Troedsson, 1937) has been regarded as an olenacean, and has generally been assigned to the family Papyriaspididae. But the glabellar structure of the type species (Peng 1984, pi. 3, fig. 3a; text-fig. 16 herein) is of typical asaphacean appearance. Papyriaspis on the other hand has a typical generalized olenacean (or ptychoparioid) glabellar structure. There is no reason (Henningsmoen 1957, p. 20) to assume that Hedinaspis should be classified with the Olenacea, other than the presence of a large number of wide and narrow (exsag.) thoracic segments, a character associated with the poorly oxygenated olenid environment. The same kind of morphology has been produced in the same environment from several different phylogenetic origins, e.g. the Alsataspididae in the Trinucleacea, or the Aulacopleuridae among Proetida. Multiplication of narrow thoracic segments and their pleural extension is associated with multiplication and transverse extension of the respiratory exites and, as such, with the constraints of greater oxygen absorption in an oxygen-deficient environment. Hedinaspis should be classified either with the Ceratopygidae or the Asaphidae; the close resemblance of its cephalic features to those of Proceratopyge , together with the form of the pygidium, which is virtually identical to the immature pygidium of Haniwoides figured by Shergold (1980, pi. 33, fig. 4), indicate that the former is correct, i.e. Hedinaspis is an olenid-like ceratopygacean. Asiocephalus Palmer, 1968, from the Franconian of Alaska, is closely similar to Hedinaspis and should be classified with it. Superfamily cyclopygacea Raymond, 1925 The superfamily Cyclopygacea included only the family Cyclopygidae in the Treatise (Moore 1959). Here it is considered to include three families; Cyclopygidae, Nileidae, and Taihungshaniidae, the latter two having been previously classified with the Asaphacea. This rearrangement is strongly supported by derived characters in the cladograms, text-figs. 1 -3, and is consistent with stratigraphic evidence (see Fortey 1981). 198 PALAEONTOLOGY, VOLUME 31 text-fig. 17. Cephalic structure of the three families included in Cyclopygacea, showing similarity in glabella structure, a, the most primitive cyclopygid, Prospectatrix (Tremadoc, Shropshire), b, Taihungshaniidae, Taihungshanici (Arenig, France), c, Nileidae, Platypeltoides (Tremadoc, widespread). All approximately natural size. Diagnosis. Asaphina with glabellar lobe expanding forwards to cranidial margin (effaced in later cyclopygids); palpebral lobes, without distinct rims, touch axial furrows at anterior ends; hypostomal condition impendent; hypostome relatively transverse, often with tripartite posterior margin. Discussion. The glabellar structure of the superfamily is distinctive (text-fig. 17). The glabella is elongate (sag.), extending to the cranidial margin and expanding in width forwards, usually immediately in front of the palpebral lobes, which touch the axial furrows at their anterior ends. The fundamental similarity of glabellar structure of Cyclopygidae to that of Nileidae and Taihungshaniidae is revealed by the earliest known cyclopygid genus Prospectatrix Fortey, 1981, and by growth stages of Pricyclopyge (Rushton and Hughes 1981, pi. 2, figs. 20 and 23); later cyclopygids develop an entirely effaced median cephalic lobe, and other adaptations for pelagic life habits, which obscure their relationships outside the group. Clearly this glabellar structure is very different from that uniting the Asaphacea , and forms the basis for grouping these families in a different superfamily. Glabellar furrows are not usually incised. The median suture is present in Taihungshaniidae (Lu 1975, pi. 18, fig. 9) throughout their history; it is present in only the earliest nileids, such as Platypeltoides (Fortey and Owens 1982, pi. 2, fig. k), and early cyclopygids (Fortey 198 1, pi. I, fig. h). Through most of their Ordovician history nileids and cyclopygids have fused free cheeks, presumably resulting from ankylosis of the two halves of the doublure. Note that the cephalic doublure of the Cyclopygacea extends beneath the glabella, implying the impendent hypostomal condition, whereas on most asaphaceans it is conterminant. Sharply defined palpebral rims are lost in this group. On Cyclopygidae and Nileidae the occipital ring is effaced; it is weakly indicated on some Taihung- shaniidae ( Taihungshania omeishanensis , see Lu 1975, pi. 18, fig. 13). Because the Taihungshaniidae also retain the median suture, broad (tr.) postocular cheeks, and long genal spines, they are regarded as the more primitive member of the Cyclopygacea, and hence are shown as the sister group of the Cyclopygidae + Nileidae on the cladograms. Cyclopygaceans have five to eight (?nine) thoracic segments which may have a relatively broad doublure; adaxial articulation of the first thoracic segment is characteristic. The hypostoma is rigidly attached to the doublure. The glabellar tubercle of Cyclopygacea is clearly pre-occipital, usually being sited at a point opposite the hind end of the eyes, and far from the posterior margin of the glabella. Where cephalic muscle impressions are preserved the tubercle is near the SI pair and well forward from the occipital pair (e.g. Poronileus fistulosus, see Fortey 1975, pi. 16, fig. 6). The only exception of which we are aware is the superficially asaphid-like Peraspis lineolata (see Whittington 1965) in which the tubercle has assumed a sub-occipital position, which we believe to be a secondary backward migration, such as happens on ogygiocaridine asaphids of comparably low convexity. If the pre-occipital tubercles of Cyclopygacea and Asaphacea are homologous, then the two families should be regarded as sister groups because this appears to be a good synapomorphic character. The asaphoid protaspis has been proved in Nileidae (p. 184) and, from the similarity in FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 199 meraspis growth series of cyclopygids and nileids, there is no reason to doubt its occurrence throughout the superfamily. Family cyclopygidae Raymond. 1925 Diagnosis. Subisopygous Cyclopygacea with hypertrophied eyes; free cheeks fused on all but the earliest species; fixed cheeks and librigenal borders much reduced; five or six thoracic segments. The stratigraphically earliest form has seven thoracic segments. Hypostome (where known) has lost maculae. Discussion. Pelagic cyclopygaceans form a distinctive group. Their eyes are deep and usually inflated, unlike nileids, which may have eyes as long, but which are strip-like. The doublure of the free cheek of several cyclopygids ( Cyclopyge , Microparia , Degamella) has been shown to carry a series of vincular notches, and this may prove to be an additional diagnostic character of the family. The subfamily Ellipsotaphrinae Kobayashi and Hamada, 1971 has been a taxonomic problem. Ellipsotaphrus has very well-defined glabellar furrows, including what has been interpreted as an occipital furrow, and the presence of the latter should exclude it from the family, as did Fortey (1981). However, Fortey and Owens (1987) presented evidence that this ‘occipital’ furrow may be interpreted instead as pre-occipital, a conjoined pair of lp glabellar furrows. This would allow incorporation of Ellipsotaphrus within the Cyclopygidae, and such an interpretation is preferred here. However, incised glabellar furrows would represent a character reversal of some importance; we did not code this group into text-figs. 2 and 3. Ontogenetic information should resolve the homologies. Family nileidae Angelin, 1854 Diagnosis. Subisopygous Cyclopygacea without anterior spinose pygidial segment; border of free cheeks not reduced and usually with broadly rounded genal angle; median suture lacking in all but earliest representatives; cephalic and pygidial doublure wide; pygidium without strong pleural furrows (secondarily developed in a few forms); seven or eight (?nine) thoracic segments. Wide hypostome with broad posterior borders and weakly tripartite margin. Discussion. The cladistic analyses show that it is difficult to characterize Nileidae, other than by retention of characters which have been modified in Cyclopygidae. None the less it is a stratigraphically and morphologically coherent group. The relatively high divergence of the anterior branches of the facial sutures applies to virtually all species other than those in which the eyes have reached an extreme anterior position. Most nileids have large eyes, but there are several genera (e.g. Illaenopsis , Psilocephalinella ) which do not, and the characters of the eye cannot be incorporated into the diagnosis, although they are never hypertrophied in cyclopygid fashion. The majority of nileids lack genal spines, but they may be present on small growth stages even if absent in the adult, and their appearance in several genera (e.g. Homalopteon , Peraspis ) is regarded as a secondary result of paedomorphosis. Hence, we place loss of genal spines as an synapomorphy shared with Cyclopygidae. The same comment applies to the appearance of strong pygidial segmentation in the same genera. There is stratigraphic evidence of the secondary, probably paedomorphic derivation of these apparently primitive characters: a derivation of Peraspis from Symphysurus arcticus was proposed by Fortey (1975), while Homalopteon closely resembles the more usual nileid Barrandia which stratigraphically predates it. A subfamily classification of Nileidae was proposed by Courtessole and Fillet ( 1975), involving no less than seven subfamilies. A full critique is not possible here, except to state that such fine subdivision serves no purpose and is premature. For example, two of their proposed subfamilies, Lakaspidinae and Hemibarrandiinae, are based on genera which are not nileids; another, lllaenopsinae, is listed as including three genera, Illaenopsis , Borthaspidella, and Pseudobarrandia (proposed therein), the last named having the same type species as Rocykania Pribyl and Vanek, which is not mentioned, and which is in any case a subjective synonym of Illaenopsis itself (Fortey and Owens 1987). Of the seventeen genera listed for the family, five ( Macelloura , Lakaspis , Benthamaspis , Bumastides , and 200 PALAEONTOLOGY, VOLUME 31 Hemibarrandia) belong in other families according to the criteria of this paper, and were erroneously included in Nileidae by Courtessole and Pillet on the basis of general effacement, which is of no taxonomic importance. Furthermore, these authors failed to mention at least six validly proposed nileid genera ( Procephalops Whittard, Psilocephalinella Kobayashi, Homalopteon Salter, Eury- metopus Postlethwaite and Goodchild, Troedssonia Poletaeva, and Petrbokia Pribyl and Vanek), the inclusion of which would have blurred their supposed subfamilial characteristics. Subfamily symphysurininae One of us (Fortey 1983) has recently shown that the genus Symphysurina Walcott is not likely to be an asaphid. In the Treatise (Moore 1959) it is the eponymous genus of a supposed asaphid subfamily. However, since the Asaphidae consistently have eight thoracic segments the fact that Symphysurina can have nine casts doubt on its inclusion in that family; in addition, it has a series of vincular notches on the doublure of the free cheek unmatched on any asaphid genus. The glabellar tubercle does, however, appear to be pre-occipital, which places it within the Cyclopygacea or Asaphacea as understood here. The median suture is shared by primitive members of the Cyclopygacea, and, occurring as it does at the base of the Tremadoc, its presence in Symphysurina is only to be expected. We are uncertain how to classify it here. Cranidial and genal morphology is generally nileid-like, but the series of vincular notches can be matched in cyclopygids (Whittard 1 960, pi. 24, fig. 8). On this basis it can be considered as the sister group of both Nileidae and Cyclopygidae, which is tentatively adopted. Other genera classified with Symphysurina in the Symphysurininae in the Treatise appear to be conventional asaphids, and hence at present the subfamily Symphysurininae includes only the one genus. Family taihungshaniidae Sun, 1931 Diagnosis. Taihungshaniidae are cyclopygaceans carrying a pair of pygidial spines; rarely a second pair is developed. Pygidial interpleural furrows are lacking. They retain the median suture and usually have a clearly defined glabella showing typical cyclopygacean form. Discussion. From stratigraphic evidence, there is a trend in the group towards long pygidia with numerous segments, a trend reaching its maximum expression in T. multisegment at a Sheng (see Lu 1975, pi. 19, fig. 3) with at least twenty such segments. Again, stratigraphic evidence shows that this was not primitively the case: early taihungshaniids, such as Tungtzuella , have pygidia much like those of contemporary nileids but for the pygidial spines. Superfamily remopleuridacea Hawle and Corda, 1847 The asaphoid protaspis of remopleuridids has been described many times (see above), as has the presence of a median suture in later representatives of the group (e.g. Whittington 1959u), e.g. Remopleurides , Amphytrion, and Robergia. Neither the morphology not ontogeny of the earlier remopleuridaceans is as well known and comments upon these are more cautious. However, among earlier (Upper Cambrian Tremadoc) genera ventral median sutures have been described from Pseudokainella (Whitworth 1969), Menoparia (Ross 19516), and Elkanaspis (Ludvigsen 1982) and we assume its loss in certain kainellids is secondary. Remopleuridaceans do not have the derived glabellar structure of Cyclopygacea + Asaphacea, and also have an occipital tubercle positioned like that of other ptychoparioids. Diagnosis. Asaphina with spinose pygidia, spines flattened, united at their bases, and extending to mid-line. Glabella bulges in transverse width in front of occipital ring; narrow, wire-like eye socle; genal borders bevelled; palpebral rims inflated, deep rim furrows, extending to axial furrows. Discussion. The specializations of later members of the group makes the framing of a set of unequivocal uniting characters difficult. The free spinose tips on the pygidium, the bases of which are conjoined and flat and extend to the posterior mid-line, are particularly characteristic of FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 201 text-fig. 1 8. Remopleuridaceans from primitive (a) to advanced (c) types, showing change from inferred natant to impendent hypostomal condition. Smaller sketches show hypostomal position (stippled) in relation to extent of cephalic doublure (black), a, primitive (kainellid) natant type, Elkanaspis (Upper Cambrian; after Ludvigsen 1982, fig. 60). b , conterminant type, Menoparia (Tremadoc; after Ross 19516, pi. 20). c, impendent type, Remopleurides (mid to late Ordovician). Approximately natural size. remopleuridaceans. This character only fails to apply to the superfamily if loganellids are included, and see Auritamiidae ( Anomocaracea) below. Slit-like glabellar furrows are developed in later species, but many taxa are effaced. The outline of the axial furrows is characteristically bowed outwards at some point in front of the occipital ring, and usually at the level of the outer end of SI or the L2 glabellar lobe, a feature shared with generalized dikelocephalaceans. On later remopleuridids this feature dominates the glabellar outline (and the expanded glabella may incorporate the remnant genal area inside the palpebral lobes). Primitive and stratigraphically early remopleuridaceans are accommodated within the family Kainellidae ( sensu Shergold 1975, p. 158), and here the feature is subdued but still visible— see, for example, Kainella (Harrington and Leanza 1957, fig. 52.2), Richardsonella (Ludvigsen 1982, fig. 70n), Pseudokainella (Whitworth 1969, pi. 75, fig. 8), and Sigmakainella (Shergold 1975, pi. 31, fig. 2). It is preserved in otherwise aberrant remopleuridaceans, such as Apatokephalops (Lu 1975, pi. 5, fig. 1), and the pelagic, specialized forms Opipeuter (immature cranidium Fortey 1974, pi. 14, fig. 3) and Bohemilla (Fortey and Owens 1987). Remopleuridaceans have characteristically curved and inflated palpebral lobes, outlined by deep, narrow palpebral furrows, which run into the axial furrows (text-fig. 19). On earlier species these furrows circumscribe a small area of fixed cheek shaped like a crescent moon, which in Remopleurides and its allies becomes absorbed within the axial area. Genal borders are narrow and bevelled, sharply defined, and only effaced in some Remopleurides species. They are not broadly flattened as happens in Cyclopygacea and Asaphacea. Narrow genal spines are primitively present and almost invariably retained; on most species they are long, unlike Asaphacea. There are three different kinds of thoracic structure in the superfamily, but the presence of one or another does not define obvious subdivisions within the group. The presumably primitive thoracic structure shown by Pseudokainella (Whitworth 1969, pi. 75, fig. 7), or Kainella , with long falcate pleurae and diagonal pleural furrows, is of a generalized ptychoparioid type. Remopleurides itself has powerful adaxial articulation with reduced pleurae (Whittington 1959a), and similar morphology is seen also on Hypodicranotus , Opipeuter , and Bohemilla. Another modification of this is shown by Robergia (Cooper 1953) and Robergiella with a straight-sided thorax, wide (exsag.) pleurae with straight intersegmental boundaries. The cephalic doublure on later remopleuridids, such as Remopleurides , is broad and extends backwards beneath the glabellar tongue (text-fig. 18), with an impendent hypostomal condition comparable to that of Cyclopygidae and Nileidae, having the hypostome attached at the doublure. On earlier genera of the Apatokephalus type the doublure extends beneath the cranidial border as far as the preglabellar furrow but no farther in the conterminant condition, e.g. see Ross (19516, pi. 20). The hypostome was presumably still attached at the inner edge of the doublure. However, on kainellids with a broad preglabellar field the evidence from silicified free cheeks indicates that the doublure does not extend beneath the preglabellar field (e.g. Ludvigsen 1982, fig. 64d, o, n; text-fig. 18a 202 PALAEONTOLOGY, VOLUME 31 text-fig. 19. Remopleuridacean and Dikelocephalacean cranidia compared. 1, pre-occipital glabella expansion; 2, relation of palpebral furrow to axial furrow, u, dikelocephalacean, the generalized saukiid Prosaukia, BM 13869 (Upper Cambrian), x 4. b, typical remopleuridacean, Apatokephalus , BM It7300 (early Ordovician, Tremadoc), x 6. herein). If the hypostome on these forms lies beneath the front part of the glabella, it is hard to see how it could have been attached directly to the doublure, and the hypostomal condition was presumably natant. This is of interest because the same is true of other primitive asaphines, and is a retained character of Ptychopariina. These three ventral distinctions may afford a method of diagnosing remopleuridid subfamilies: Remopleurididae s.s. having impendent hypostomal condition; all Apatokephalinae conterminant; Kainellidae restricted to those forms with natant hypostomal condition. To avoid multiplication of terminal taxa on text-figs. 2 and 3 this character was arbitrarily coded as impendent there. That at least some kainellids may have lost the median suture to form yoked free cheeks is indicated by the Richardsonella sp. illustrated by Palmer (1968, pi. 14, fig. 8), and some Kainella spp.; as it is elsewhere in Asaphina, this condition is regarded as secondary. Constituent families of Remopleuridacea. Shergold (1975) included two families, Kainellidae and Remopleurididae, in the Remopleuridacea, the latter with three subfamilies, Remopleuridinae, Apatokephalinae, and Macropyginae. Of these, the Macropyginae has been assigned to the Asaphacea herein, as discussed above. The remopleuridid subfamily Richardsonellinae was employed in the Treatise (Moore 1959), and was equivalent to Kainellidae in Shergold’s usage. There are several additional families to be included. Two peculiar and specialized families of Ordovician pelagic trilobites are regarded as being remopleuridaceans: Bohemillidae Barrande, 1872 and Opipeuteridae Fortey, 1974. The earliest representative of the former has just been discovered (Fortey and Owens 1987), and a pygidium assigned for the first time, which is compatible with remopleuridacean affinities. Opipeuter is convergent with Bohemilla , but has a separate origin and may not be closely related. Lu (1975) proposed a family Loshanellidae for the reception of two Chinese genera Loshanella and Wanliangtingia. Glabellar furrows are effaced in these genera, and eyes are smaller than is usual in remopleuridaceans, but both (Lu 1975, pi. 5, figs. 18 and 20) show the glabellar shape typical of the superfamily. Zhou and Zhang (1978) placed Wanliangtingia in another new family, Apatokephalopsidae, together with Apatokephalops Lu, 1975 and Jiia Zhou and Zhang, 1978. Apatokephalops itself has a glabella of clear remopleuridacean shape, although unusually elongate (sag.). Ludvigsen ( 1 982) synonymized Apatokephalopsidae with Kainellidae. Clearly, the definition of FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 203 families and subfamilies within the Remopleuridacea requires a comparative assessment based on a critical evaluation of their shared characters, which is beyond the scope of our work. Provisionally, we can accept the following families within the Remopleuridacea: Remopleurididae, Kainellidae, Loshanellidae, Opipeuteridae, and Bohemillidae, possibly with the Hungaiidae, as discussed next. Loganellidae and Hungaiidae. These two families are included in the Remopleuridacea in the Treatise (Moore 1959). Shergold (1975) assigned both to Dikelocephalacea. Loganellids lack the median suture of Remopleuridacea but, since this can also be lost by ankylosis in other asaphines, this does not preclude its inclusion in the group. Fortey (1983, p. 198) pointed out that loganellids resemble olenids in several features, but that this could be convergence resulting from their adaptations to similar habitats, like Hedinaspis described above. Several other differences from olenids were noted, perhaps making their inclusion in that family improbable. For example, Levisella and Lauzonella have broad (sag.) cephalic doublures, which is not an olenacean characteristic (below). On the other hand, the lack of pygidial spines of loganellids, their often broad genal borders, and the debatable presence of the mid-glabellar expansion makes their inclusion in the Remopleuridacea as understood here problematic. By contrast, Hungaia conforms to the superfamily with the exception of having short genal spines, the glabella and pygidium being typical. Unless it can be shown that Ftungaiidae and Loganellidae should be classified together, we therefore favour the view that the former belong within the Remopleuridacea, and that the latter belong elsewhere, possibly with the Dikelocephalacea, following Shergold (1975). CAMBRIAN SUPERFAMILIES INCLUDED WITHIN ASAPHINA We have so far considered families and superfamilies sharing both median suture (unless lost by ankylosis) and asaphoid protaspis, which we claim as a monophyletic group. There are additionally several others which, we believe, belong to the same group, and are better classified, even with our present inadequate knowledge, with the Asaphina rather that in an indefinable suborder Ptycho- pariina. The presence of the median suture in these is a uniting character, and to disprove our classification it is necessary to show that this feature is capable of polyphyletic derivation. There is insufficient ontogenetic information on these groups to know whether they have the same protaspis type as the Asaphina described above. As shown on the cladistic analyses they share progressively more generalized ptychoparioid features. Superfamily dikelocephalacea Miller, 1889 emend. Ludvigsen and Westrop, 1983 Discussion. The concept of the Dikelocephalacea was reviewed by Ludvigsen and Westrop (1983), who included the families Dikelocephalidae, Saukiidae, and Ptychaspididae within it, hence uniting Dikelocephalacea with Ptychaspidacea of earlier authors. The presence of the median suture (see text-fig. 20a) was noted by them (also Ludvigsen and Westrop 1986, fig. 4f), which may be rarely secondarily lost by fusion, as in Cyclopygacea. These authors did not say whether they would include the families Pterocephaliidae, Housiidae, or Idahoiidae within the superfamily, as in the Treatise (Moore 1959). Both Pterocephalia and Housia (text-fig. 20c) have median sutures and should on this criterion be referred to Asaphina in our usage. Taking the restricted view of Dikelocephalacea, the group comes out as the sister group of Remopleuridacea on our cladistic analyses (text-figs. 1-3). The glabella form is characteristic, as noted by Ludvigsen and Westrop, typically truncate anteriorly and squat, with deep axial and glabellar furrows of which 1 p may fuse across the glabella. Most of the more primitive genera (e.g. Prosaukia ) may show a slight mid-glabellar expansion like that noted in the Remopleuridacea, which is evidence for supposing that these two superfamilies are more closely related than to other Asaphina. Much is known about the within-group evolution of dikelocephala- ceans, resulting in many departures from the generalized morphology, and mostly based on stratigraphic evidence from late Cambrian sequences in the North American platform; we shall not attempt to review this here. An important difference between the two superfamilies is in the palpebral lobes; in both superfamilies the lobes are inflated and well-defined, but in Dikelocephalacea they do 204 PALAEONTOLOGY, VOLUME 31 not run into the axial furrows anteriorly as described for remopleuridids. Instead, the lobes are rather sharply terminated even when they closely approach the glabella (text-fig. 19a), i.e. they retain the ptychoparioid condition. So far as is known, the cephalic doublure on dikelocephalaceans approaches the front of the glabella medially, and the hypostome is presumably attached at its inner edge beneath the frontal glabellar lobe, as is the case with most Asaphidae and Apatokephalus-group Remopleuridacea, i.e. the hypostomal condition is conterminant. Librigenal borders may be wide and convex, as in Prosaukia, or broad and flattened as in Dikelocephalus. Thoracic structure is of generalized ptychoparioid type. Pygidial structure is characteristic, with a convex axis having five or six axial rings and a rather sharp posterior termination, from which it is continued towards the margin as a postaxial ridge. Pygidial pleural and interpleural furrows are subequally incised (contrast dikelokephalinids) and extend on to the pygidial border, which is often flattened. On some forms the pygidial margin is spinose, but stratigraphic evidence shows that this is secondarily derived. Superfamily anomocaracea Poulson, 1927 Discussion. Herein are included Asaphina which show a combination of characters that are mostly those of generalized ptychoparioids; they probably include some of the most primitive members of the text-fig. 20. Free cheeks of Dikelocephalacea (a), Anomocaracea ( b ), and Housiidae (c) with doublure prepared to show median suture, a, Saukia separata Ulrich and Resser, USNM 84601 (Upper Cambrian), x 3. b , Anomocarioides limbatus (Angelin), Riksmuseet, Stockholm Ar 53023 (late Middle Cambrian, Andrarum Limestone, type loc., Scania, Sweden), x 4. c, Housia canadensis Walcott, USNM 5068 (Upper Cambrian, British Columbia), x 3. FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 205 suborder and may well include forms ancestral to those superfamilies discussed above (text-fig. 27). Anomocaracea sensu lato (including Anomocaridae and Auritamiidae at the least) plot out as a primitive group on text-fig. 1, and the group has no autapomorphies peculiar to it. The unsatisfactory nature of the taxon is shown by the fact that on text-figs. 2 and 3 the family Anomocaridae taken alone plots out as the sister group of Asaphacea (at the least). Clearly the phylogenetic position of Anomocaracea requires further scrutiny. The diagnosis given in the Treatise (Moore 1959) is scarcely diagnostic and would apply to most of the Asaphina. The presence of the median suture has been shown in Anomocare and Anomocarioides (see text-fig. 20 b), and Auritama (see Opik 1967); since the structure of the border in the superfamily is conservative, there seems no reason to doubt its presence throughout. Our contention is therefore that the median suture was acquired by a ptychoparioid, probably in the Middle Cambrian, and that its more generalized descendants are classified within the Anomocaracea. The primitive characters retained in the group include: 1 , the glabella is mostly gently tapering or parallel-sided with, where defined, three or four pairs of glabellar furrows which may be of usual ptychoparioid type; 2, facial sutures are only slightly divergent in front of the eyes, even in forms with a broad preglabellar field; 3, thoracic structure is unspecialized, with nine or more segments having simple falcate pleurae, moderately wide facets, and diagonal pleural furrows; 4, on some anomocaraceans the hypostomal condition is assuredly natant (e.g. A. limbatus (Angelin), as illustrated by Egorova et al. 1982, pi. 44, fig. 2), i.e. they retain the character of out-group ptychoparioids. Where the cephalic doublure becomes broader (A. novas Tchernysheva in Egorova et al. 1982, pi. 43, fig. 1 ) it could be interpreted as likely to have had conterminant hypostomal condition; however, on this species the paradoublural line still stops short of the front of the glabella, and we believe that an ‘advanced’ natant condition still pertained. On Auritama , a growth series of cranidia (Opik 1967, pi. 15, figs. 3-6) clearly shows a reduction in the width of the preglabellar field during ontogeny, and it is possible that this records the change from natant to conterminant hypostomal condition, although there is no proof that the hypostome actually docked against the cephalic doublure. The same problems apply to the Pterocephaliidae, discussed below. In any case, there is no doubt that the retention of natant hypostomal condition in anomocaraceans is primitive. Although pygidia are relatively large, there is no diagnostic characteristic which would distinguish them from those of asaphids (text-fig. 216) or Dikelokephalinidae (Opik 1967, pi. 31, fig. 3), or Remopleuridacea (Opik 1967, pi. 15, fig. 10). A more detailed phylogenetic analysis might result in the assignment of some anomocaraceans to any of these groups. Only the inclusion of more than four segments into a relatively large and effaced pygidium is presumably to be regarded as an ‘advanced’ character compared with typical Ptychopariina. In Anomocarioides the glabellar tubercle could be interpreted as lying in an immediately pre-occipital position (text-fig. 21a). It also appears to be a thinning of the exoskeleton (rather than an external tubercle) as has been described for certain nileids(Fortey and Clarkson 1976) and asaphids. If this is so, such forms record the transition into the asaphacean condition and provide further evidence for the paraphyletic, if not polyphyletic position of the superfamily, as interpreted in the tree (text-fig. 27). A strictly cladistic classification might place Anomocarioides in the Asaphacea. In other anomocaraceans the tubercle appears to be occipital. Palpebral lobes are long (exsag.), extending far back but not forwards to touch the glabella, often with subdued rims; their gentle curvature inwards at their posterior ends and presence of eye ridges probably afford a distinction from other asaphine superfamilies, but again these are primitive characters; they can be matched on certain species of Proceratopyge. Many, but certainly not all, genera assigned to Anomocaracea have bacculae adjacent to the basal glabellar lobes (text-fig. 21a). These are distinct swellings (see p. 212), which we have also noted in comparable position in ceratopygids, asaphids, and macropyginids. The presence of such bacculae, in small growth stages if not in adults, is probably a uniting character of higher Asaphina (further discussion under Trinucleacea). Further evidence of the artificial classification of the Anomocaracea is the structure of Auritama , and the family Auritamiidae was included in the Anomocaracea by Opik (1967). Auritama retains the unmodified glabella typical of primitive forms, and the eye does not approach the axial furrow 206 PALAEONTOLOGY, VOLUME 31 text-fig. 21. Anomocarioides limbatus (Angelin) (Middle Cambrian, Andrarum Limestone, Scania, Sweden), a, BM 1668, internal mould of cranidium showing bacculae and median glabellar tubercle (arrowed); note that, in relation to occipital furrows, this could be interpreted as pre-occipital (cf. Asaphacea), x 4. b , BMNH It20683, pygidium, x 3. anteriorly; however, it appears to have a narrow, wire-like eye socle, and to have lost the circumocular suture (Opik 1967, pi. 15, fig. 1), as well as having a deeply incised palpebral rim. These are derived characters of Dikelocephalacea + Remopleuridacea, while the spinose pygidium of Auritama is very like that of early remopleuridids. On a cladistic classification Auritamiidae should certainly be classified with the remopleurid-dikelocephalacean clade rather than retained in Anomocaracea on primitive characters. Family pterocephaliidae Kobayashi, 1935 Discussion. Pterocephaliids are an Upper Cambrian group of trilobites typifying the pterocephaliid biomere of North America (Palmer 1965). Palmer and others have included the Aphelaspidinae in the same family, and the presumption has been made that the pterocephaliids were derived from Aphelaspis itself (Robison 1964). Aphelaspis is described as having a rostral plate, and the transformation Aphelaspis to Pterocephaliidae is supposed to involve the loss of the rostral plate by ‘shrinkage’. Since we suppose that the median suture is a synapomorphy of Asaphina there are two possibilities with regard to the median suture in Pterocephaliidae: 1, Robison (1964, p. 520) is correct, and the appearance of median sutures in Pterocephaliidae is a parallelism, and the group is not classifiable in Asaphina— presumably the stratigraphic evidence would be applied here to indicate the derivation of this group later than Anomocaracea; 2, but if we are correct, the presence of the median suture as a synapomorphy indicates that the closest relatives of pterocephaliids are other Asaphina, and not Aphelaspis. To decide in favour of the former hypothesis we would have to find characters linking Aphelaspis and pterocephaliids which are of convincing phylogenetic significance. We cannot find such characters. Aphelaspis and pterocephaliids mostly share primitive ptychoparioid characters of no use in determining relationships. For example, Robison mentions similar axial structures: text-fig. 23 shows that the glabellar structure of Pteroceplialia is closely similar to that of Ptychoparia striata and hence by definition ptychoparioid; the fact that some species of Aphelaspis also have a comparable glabellar structure (A. camiro as figured by Rasetti 1965 is probably closest but other species are usually more effaced) does no more than indicate ultimate ptychoparioid ancestry. The pygidia of FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 207 C d text-fig. 22. Diagrammatic ventral views of cephala of Asaphina ( b-d ) compared with ptychoparioid out-group (a), showing development of fully conterminant condition, a , ptychopariine with natant hypostomal condition, b , median suture developed but natant hypostomal condition retained as in primitive, anomocaracean grade of Asaphina (pterocephaliids, anomocarids, auritamiids, etc.), c, advanced conterminant condition as in Asaphidae — hypostome strongly buttressed against doublure, and terrace ridges spread on to middle body. d, conterminant, by virtue of backward median curve of cephalic doublure, reflected in paradoublural line dorsally (primitive Ceratopygidae, and Dikelokephalinidae). some species of Aphelaspis have backward-curved pleural furrows and facets extending far posteriorly, which can be matched on some pterocephaliids. However, other Aphelaspis species (e.g. A. bridgei Rasetti, 1965) have pygidia of primitive ptychoparioid type, while some pterocephaliids (e.g. Pterocephalia constricta Palmer, 1965, pi. 8, fig. 9) have pygidia of anomocaroid form, so we cannot construe a convincing synapomorphy for pterocephaliids -Faphelaspids from pygidial structure alone. Backward-curved pygidial facets presumably relate to having posteriorly extended pleural tips on the thorax, which is a character developed in many trilobite families, and of little taxonomic importance. Transformation from aphelaspid to pterocephaliid would also require an increase in the width of the doublure at the expense of the preglabellar field. There is plenty of evidence to suggest that pterocephaliids combined the natant hypostomal condition with the median suture, as is the case with anomocaraceans. For example. Palmer (1968, pi. 16, fig. 15) illustrated the genal doublure of Strigit ambus! blepharina clearly showing that it must have fallen well short of the front of the glabella (as in text-fig. 22b) which is also shown by the paradoublural line on the cranidium. In this species the 208 PALAEONTOLOGY, VOLUME 31 text-fig. 23. Glabellas of a , the type species of Ptychoparia , P. striata (Emmrich) (Ptychopariidae), USNM 61492, x 3-5, and 6, Pterocephalia , P. sanctisabae Roemer (Pteroceplialiidae), BM It4807, x 5, to show fundamental similarity in structure— the latter is essentially primitive ptychoparioid. doublure extends well inside the cranidial border, whereas on aphelaspids there is no example known to us in which the cephalic doublure extends beyond the marginal rim, even though the rim itself varies in width. A similar paradoublural line to that in SP blepharina is shown also by P. concava Palmer (1960, pi. 9, fig. 1; 1965, pi. 17, fig. 6), Cenmolimbus (e.g. Palmer 1960, text-fig. 18), and Sigmocheilus compressus Palmer (1968, pi. 8, figs. 22-24), and these are also reasonably supposed to have combined a natant hypostomal condition with a relatively wide cephalic doublure carrying a median suture. This is a grade of organization characteristic of early Asaphina, although it cannot be used to classify a group within Asaphina because it is a combination of an advanced character with a retained primitive one: however, it is no coincidence that this grade is found in many Cambrian species. In fact, it is difficult to prove whether any pterocephaliid achieved the conterminant hypostomal condition — the only candidate is probably P. sanctisabae Roemer in which the cranidial paradoublural line and preglabellar furrow appear nearly (but perhaps not quite, see Palmer 1960, fig. 19) to coincide; since Palmer (1965, p. 20) shows an ancestor-descendant relationship between P. concava , which was natant, and P. sanctisabae , it is perhaps likely that the hypostome of the latter never really docked against the doublure. The cladistic analysis on text-figs. 2 and 3 shows the Pterocephaliidae in a basal relationship to the rest of the tree (see consensus tree, text-fig. 4), with Dikelokephalinidae. Housiidae have been closely associated with pterocephaliids, indeed they were regarded as a pterocephaliid subfamily by Palmer (1965); they also have a median suture (text-fig. 20c). Similar arguments to those explained above would also include housiids in Asaphina. Housiidae, Pterocephaliidae, and Idahoiidae were included in the superfamily Dikelocephalacea in the Treatise (Moore 1959); Ludvigsen and Westrop (1983) implicitly excluded them from their revised concept of this superfamily. Most of those characters they have are retained primitive characters shared with a ptychoparioid out-group. For example Palmer’s (1965, p. 57) ‘diagnosis’ of Pterocephaliidae would apply to most ptychoparioids. Hence it is difficult to assign them to a superfamily; we retain them in an admittedly paraphyletic Anomocaracea until more detailed phylogenetic analysis is carried out. Too little is known of ventral structures in Idahoiidae to be confident about their placement in Asaphina. Provisionally we suggest that this family is a paraphyletic group related to Dikelocephalacea + Remopleuridacea. In summary, Anomocaracea show a combination of characters that indicates both its possible ancestral position relative to other asaphine groups, and that its closest non-asaphine relatives are FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 209 ptychoparioids. Further work may show that its constituent families should be distributed between other superfamilies. Knowledge of the morphogenesis of anomocarids, and particularly of the protaspis, is highly desirable to help resolve these relationships further. Not surprisingly with such a group it is difficult to formulate a set of its own diagnostic characters, because those that it has are either primitive or shared by one or more descendant. As a grade, the combination of natant hypostomal condition with median suture is present on many of these mid- to late Cambrian trilobites. Their inclusion in Asaphina does depend on the recognition of the median suture as a monophyletic character; no compelling morphological evidence to the contrary has been found. PROBLEM ASAPHINA, AND FURTHER RELEVANT TRILOBITE FAMILIES We briefly consider here further families which are certainly Asaphina, but difficult to classify, or may prove to belong within the group, but of which knowledge is inadequate to be sure. Family dikelokephalinidae Kobayashi, 1936 Discussion. This distinctive family has been classified with the Asaphacea in the Treatise (Moore 1959) and with the Dikelocephalacea by Opik (1967). Although it is usually described as Ordovician, Opik (1967) named an Upper Cambrian genus, Nomadinis, from the Mindyallan of Queensland which we believe he correctly referred to the family; like other families in the Asaphina stratigraphic criteria are not primarily relevant to its taxonomic position. The glabellar tubercle in Dikelokephalinidae is occipital, which is unlike Asaphacea as understood here. The distinctive autapomorphy of the group is the presence of prominent semicircular bacculae adjacent to the basal glabellar lobes. They are present wherever preservation is good (see, for example, a variety of forms figured by Lu 1975) and are often somewhat depressed rather than inflated, i.e. they have the appearance of true alae. They may not be homologous with the ‘bacculae’ of other Asaphina, and they appear as a parallelism on the cladogram, text-fig. 3. The cephalic doublure is broad, reaching the front of the glabella, and the paradoublural line along which this happens is preserved as a pair of concave-backward curves transversely crossing the mid-part of the preocular fixed cheeks (Lu 1975, pi. 27, fig. 1; Fortey and Shergold 1984, pi. 44, fig. 2). This is an extreme development of the outline of paradoublural fine shown in some primitive asaphaceans such as Proceratopyge (Westergard 1947, pi. 2, fig. 1; see text-fig. 24). The problem with classifying Dikelokephalinidae is that other characters are too general to be diagnostic. This is again reflected in the basal position of the family on the cladistic analyses. The resemblance to Dikelocephalacea is superficial, being largely that dikelokephalinids also have large text-fig. 24. Dikelokephalinid cephalic structure and comparable forms, a, Proceratopyge (Upper Cambrian), showing form of cephalic doublure reflected in paradoublural line on preglabellar area, b, the dikelokephalinid Hungioides (early Ordovician; after Fortey and Shergold 1984, pi. 44, fig. 2), showing similarity of paradoublural line to Proceratopyge, presumed to reflect similar course of doublure on venter, c, Chelidonocephalus (latest Middle Cambrian; after Wittke 1984, pi. 3), again showing similar preglabellar structure to a and b. 210 PALAEONTOLOGY, VOLUME 31 fan-like pygidia and correspondingly extended cephalic borders. None of the critical dikelocephala- cean characters described above are present. Glabellar structure of Dikelokephalinidae, especially of Nomadinis , is of the generalized kind (tapering glabella; lateral glabellar furrows of ptychoparioid type), although on later forms the IS furrow tends to bifurcate, and the furrows become pit-like, isolated within the glabella. The majority of Dikelokephalinidae have anterior pygidial spines giving them a superficial pygidial similarity to taihungshaniids. This is already seen to be a polyphyletic character in Asaphina, and in any case a few dikelokephalinids do not have such spines (see, for example, Warendia Gilbert-Tomlinson in Hill et al, 1969). Palpebral lobes are typically strongly curved, but do not reach the axial furrow anteriorly in remopleuridacean fashion. Pygidial interpleural furrows are absent (contrast Dikelocephalacea) or at most extremely feeble. The best recourse seems to be to temporarily classify Dikelokephalinidae with the Anomocaracea, along with other families lacking derived characters to link them with other groups. As noted above bacculae are present in many anomocaraceans, and now that the record of Dikelokephalinidae extends to the Upper Cambrian there is no particular objection to this on stratigraphic grounds. ‘Transitional’ forms are to be found in the literature, e.g. Paracoosia kingi Wittke, 1984, from the Upper Cambrian of Iran, was assigned to the Anomocaridae by that author, but resembles a dikelokephalinid in all features but the lack of bacculae. A cephalic border structure much like that of Dikelokephalina is present on the Middle-Upper Cambrian genus Chelidonocephalus King (see Wittke 1984, pi. 3; text-fig. 24c herein). The so-called ‘false border furrow’ on this genus is identical to the paradoublural line on dikelokephalinids. If this is an important homologue it suggests that the subfamily Chelidonocephalinae Wittke, 1984, belongs within the Asaphina, and possibly as the sister group of Dikelokephalinidae as understood here. Superfamily leiostegiacea Discussion. Some leiostegiaceans show a glabellar structure generally like that of Cyclopygacea as defined here, with a pestle-shaped glabella extending forwards to the border or to the cephalic margin (. Lloydia , Annamitella ). The Cyclopygacea are generally effaced but it is reasonable to suppose that its relatives were less so. Some leiostegiaceans also developed spinose pygidia comparable with those of ceratopygids. Little is known about the ventral cephalic structures ofleiostegiids. However, Jell (1985, pi. 22, fig. 3) has figured the genal doublure of Leiostegium which apparently terminates in a rostral suture. This indicates that the leiostegiaceans are not Asaphina, and that the similarities to the group are parallelisms. It would be as well to confirm the presence of the rostral plate on other leiostegiaceans, in case Jell’s specimen is broken rather than suturally bounded. Family catillicephalidae Raymond, 1938 ( partim ) Family isocolidae Angelin, 1854 Discussion. The family Catillicephalidae includes the only well-documented example of the presence of the rostral plate and the median suture within the same supposed family. Catillicephala itself has a rostral plate (Rasetti 1954), whereas Acheilus has a median suture (Rasetti 1954; Fortey 1983). If we are correct in our interpretation of the median suture as indicating a monophyletic origin of the Asaphina this means that the Catillicephalidae is likely to be a polyphyletic taxon. In fact, there is little in common between Catillicephala and Acheilus other than small size and a convex (tr.) glabella: for example Catillicephala has a subcircular glabella without furrows and a remnant cranidial border, whereas Acheilus lacks a cranidial border and has a forward-expanding glabella with three pairs of glabellar furrows. In our opinion the true relationships of these genera have yet to be proved. Acheilus is clearly related to several other genera: Triarthropsis, Acheilops, and Calculites. Fortey (1983) suggested that these in turn were related to the Ordovician family Isocolidae, for which a median suture has been demonstrated (Whittington 1963). Such isocolids are generally more similar to Acheilus than to Catillicephala. However, it is difficult to accommodate the Isocolidae within any of the superfamilies of Asaphina described above. Whether they prove to be a truly independent lineage, or whether they are related to some as yet unidentified asaphine taxon remains to be seen. For the FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 211 moment we classify them as Asaphina incertae superfamiliae and include Acheilus and its allies in the same family. Because Acheilus has a forward-expanding glabella, palpebral lobes that touch the axial furrows, and is likely to have had an impendent hypostomal condition, it is likely that its relationships within Asaphina will prove to be with Cyclopygacea. Superfamily trinucleacea Hawle and Corda, 1847 emend. The Trinucleina are at present regarded as a suborder of Ptychopariida, but for reasons which are considered below, we consider the group as a superfamily in this work. In Treatise usage the taxon includes seven families: Trinucleidae, Raphiophoridae, Hapalopleuridae, Endymioniidae, Orometopidae, Dionididae, and Alsataspididae. Myindidae Hupe, 1955 should also be referred to the group, and there is now stratigraphic evidence to suggest that Myindella and Myinda are close relatives of the hapalopleurid Araiopleura (Rushton 1982, p. 57) which is a clear indication that separate family status for Myindidae is not necessary. Fortey and Shergold (1984) regarded Hapalopleuridae and Orometopidae as synonymous, representing the more primitive Trinucleacea retaining wide free cheeks and in most cases probably eyes. Endymioniidae have been regarded as a subfamily of Raphiophoridae. Hence, a modern view of Trinucleacea would include perhaps five families. Regardless of internal classification, so far as we are aware nobody has challenged the Trinucleacea as a monophyletic group: the convex and pyriform glabella alone may be a sufficient uniting character to support this. There are other autapomorphies: the long and narrow adaxial part of thoracic pleurae; the triangular pygidium with very narrow doublure; the basket-and-lid (Bergstrom 1973) style of enrolment. Dionididae and Trinucleidae have their own unique fringe structure— but although such structures adequately define the families they do not help with assessing the relationships of the trinucleaceans to other trilobites. Although Trinucleacea is an accepted, phylogenetically based taxon, there has been little discussion of the relationships of the group to other trilobites. Their inclusion in Ptychopariida in the Treatise (Moore 1 959) is presumably based on the presence of opisthoparian sutures in the least derived forms (Orometopidae, including Hapalopleuridae). The eventual status of the group as superfamily or suborder does depend on an assessment of its relationships to other major groups. Trinucleacea is one of those groups which apparently appear in the stratigraphic record just below the base of the Tremadoc (Rushton 1982). Clearly, it must have had Cambrian sister taxa but, as in other cases, the Cambrian-Ordovician boundary has been interpreted as a taxonomic one as well. We present a case here that the Trinucleacea should be classified with the Asaphina. This is based on several lines of evidence: 1, the resemblance of the protaspides of the group to the asaphoid protaspis (see above); 2, the presence of a pre-occipital glabellar tubercle in many trinucleaceans; 3, the identification of stratigraphically intergrading trinucleaceans across the Cambrian-Ordovician boundary; and 4, the identification of a reasonable candidate for the Cambrian sister group which shows the cephalic median suture. All Ordovician Trinucleacea of which we are aware have the free cheeks fused together ventrally as a single unit — which becomes extensively modified to the lower lamella of those trinucleaceans having fringes. As in other cases discussed above, this fusion is secondary, and it is obviously crucial to an assessment of relationships to discover whether the fused condition resulted from the incorporation of a rostral plate (as in olenids) or from the loss of a median suture (as in nileids). Primitive morphology of Trinucleacea may be exemplified by Orometopus. Text-fig. 25 compares cephalic morphology of Orometopus with the Upper Cambrian genus Liostracina. Opik (1967, pi. 35, figs. 4 and 5) clearly figured a ventral median suture on the latter, which means, if our contention of a monophyletic origin for this character is correct, that it should be included within Asaphina in our terms. The other resemblances between Orometopus and Liostracina are compelling and include: transversely inflated glabella; eye size and position; presence of inflated bacculae; and triangular, transverse pygidium. The main differences are all primitive characters for Ptychopariida in Liostracina , and hence of no importance in assessing relationships. However, they merit discussion because they may be considered significant by those workers favouring classification on the basis of overall similarity. 212 PALAEONTOLOGY, VOLUME 31 b text-fig. 25. Comparative reconstructions of a primitive trinucleacean (a) and the late Middle Cambrian Liostracinidae (b), to show similarity apart from retained primitive characters of the latter. a, Orometopus (Tremadoc; new reconstruction), fused cheeks are indicated by ‘broken' doublure, b , Liostracina , cranidium and cheek of L. nolens (after Opik 1967, pi. 35, figs. 1 and 4), with undoubted median suture, and pygidium of same genus (after Chu 1959, pi. 1, fig. 31). Both x 8 approx. They are: 1, tapering glabella— this is present in all primitive Asaphina, secondary derivation of forward-expanding glabella within accepted monophyletic groups of Asaphina has been demonstra- ted for Ceratopygidae ( Ceratopyge ) and Asaphidae (most Asaphinae) above, and there seems no special reason to regard this difference between Liostracina and other Trinculeacea as more significant; 2, presence of preglabellar field — again, a preglabellar field is present in Ptychopariida and primitive Asaphina, such as kainellid remopleuridaceans, and this character does not define derived groups; it is connected with 3, natant hypostomal condition— the relatively narrow width of the cephalic doublure indicates that the hypostome was not attached at the doublure in Liostracina ; this is also a characteristic of primitive Asaphina, and is a retained primitive character from the ptychoparioid state (cf. kainellids, pterocephaliids, and most if not all anomocaraceans). In their review of Trinucleidae, Hughes et al. (1975, p. 541) remarked of the hypostome that ‘it appears to have had no sutural union with the inner margin of the lower lamella and it may well have been suspended entirely by the unsclerotized (sic) ventral cuticle’ which is a description of the natant hypostomal condition. The natant condition may account for the rarity of definite assignments of hypostomes to trinucleids, as is also true of ptychopariids. During enrolment of ‘basket-and-lid’ (Bergstrdm 1973) type the pygidium is tucked well beneath the inner edge of the fringe (Hughes et al. 1975, p. 545) and it would be difficult to imagine how the hypostome could be attached in the usual conterminant fashion at the inner edge of the doublure in a horizontal orientation. However, we have examined specimens of Raphiophoridae in which the hypostome is attached to the inner edge of the doublure in the conterminant position but with the hypostome tucked away up inside the frontal lobe of the glabella in a manner analogous to that in certain illaenids. It seems possible that for many Trinucleacea the natant condition was retained into the Ordovician when many of their con- temporaneous Asaphina were conterminant or impendent, but that the conterminant condition was acquired by at least some raphiophorids. In any case there is nothing incompatible between the hypostomal attachment mode of Liostracina and that of Trinucleacea. The noticeable bacculae adjacent to the base of the glabella in both Orometopus and Liostracina are also present in other primitive Asaphina, such as anomocarids, many ceratopygids, and all FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 213 dikelokephalinids. Niobine asaphids also have bacculae, as do many small asaphid growth stages. Hughes et al. (1975) indicated their presence as primitive in the Trinucleidae. They are present in small growth stages of trinucleids and raphiophorids (Whittington 1959u; Hughes et al. 1975, p. 589) even when lost in the adult. Since they are not present in generalized ptychoparioids (nor apparently in meraspides of this group) it is tempting to assume that their presence is a synapomorphy of higher Asaphina. Their absence in Remopleuridacea and Cyclopygacea would then be considered a secondary loss. However, we are reluctant to be too definite about this, because there are other monophyletic groups outside the Asaphina in which the presence of bacculae is of low phylogenetic significance; for example, bacculae appear in the one genus Carolinites in the family Telephinidae, or are present in some species only of the genus Shumardia. text-fig. 26. Trinucleaceans crossing the Cambrian-Ordovician boundary, a, Jegorovaia (Upper Cambrian; after Lu 1974, pi. 2). b, Araiopleura (Tremadoc; after Dean 1970, pi. 1, fig. 11). Both x8 approx. Although there is a considerable stratigraphic separation between Liostracina and those families usually included in Trinucleacea we do not regard this as evidence for their taxonomic separation. There are late Cambrian trilobites which conform to Trinucleacea which have been placed in the separate family Jegorovaiidae. They are not well known. However, Lu (1974) has already synonymized this family with Hapalopleuridae. Text-fig. 26 shows comparative reconstructions of Jegorovaia and the early hapalopleurid A. beothuk Dean; apart from the more anteriorly directed eye ridges of the former it would be difficult to place these forms in different genera, let alone families. Both have preglabellar fields, like Liostracina ; both also have the ovoid shape of thorax + pygidium which is typical of Trinucleacea as a whole. ft is difficult to place the Trinucleacea within the classification scheme of the Asaphina as a whole, tf we are correct in assigning the group here, its status should be reduced to that of superfamily, compatible with the other groups we have recognized. The preglabellar tubercle would suggest a sister group relationship with Asaphacea + Cyclopygacea. However, if stratigraphic evidence is to be believed, the early members of the group do not have a preglabellar tubercle— but they do have an occipital tubercle. 0. aridos Bulman and Rushton, 1973 assuredly has a pre-occipital tubercle, while other forms assigned to Hapalopleuridae by Harrington and Leanza( 1957) certainly have an occipital tubercle ( Hapalopleura clavata) or are ambiguous in this regard (A. reticulata , their fig. 112.2c). Liostracina has an occipital tubercle. This evidence could be taken as showing the parallel derivation of the pre-occipital tubercle in what we would call trinucleaceans. The alternative would be to assume monophyletic derivation of the pre-occipital tubercle, thereby removing from the group a number of forms which resemble trinucleaceans in far more characters than the tubercle alone. We prefer the 214 PALAEONTOLOGY, VOLUME 31 former course. This means that it becomes exceedingly difficult to rank the Trinucleacea on a cladogram, because the remaining characters are either autapomorphies of the group (e.g. pyriform glabella; adaxial extension of thoracic segment; triangular, transverse pygidium) or primitive ptychoparioid characters (natant hypostomal condition in some forms; having genal spines; narrow cephalic doublure, etc.) of no service in determining relationships. The definition of dorsal furrows in the trinucleid and raphiophorid protaspides is primitive compared with their effacement on higher Asaphina, as is the lack of terrace ridges on the known hypostomes of trinucleaceans. Hence the group has to be placed among the less derived Asaphina and probably as an unresolved trichotomy with ‘Anomocaracea’, itself a paraphyletic group. In any case it is clear that if we wish to include the Trinucleacea within Asaphina, its taxonomic status should be that of superfamily, like the other major groups included, rather than suborder, as at present. Finally, a brief comment is given on what may seem the rather bold inclusion of Cambrian forms within what is regarded as an ‘Ordovician’ group. It has become almost axiomatic to treat resemblances between Cambrian trilobites of disparate ages— let alone Cambrian and Ordovician trilobites — as if they were likely to be the product of convergence rather than indicating phylogenetic relationships: ‘every student of Cambrian trilobites knows that genera widely separated in time and space, therefore unlikely to be closely related, may appear almost identical’ (Rasetti 1972, p. 44). This assumption of parallelism quickly becomes self-fulfilling and non-testable, because every occurrence from a different horizon or different ‘space’ (How are these defined? By distance in kilometres? By inferred palaeogeography? Or by the author’s predelictions?) becomes subject a priori to different taxonomy, whether or not the morphology of the trilobites might indicate that they are ‘closely related’. In our analysis of Asaphina as a whole, and the Trinucleina discussed here in particular, we prefer the opposing view that definable derived characters should form the basis of classification. If what we consider as characters indicating monophyly are regarded by another author as capable of polyphyletic development, then the burden of proof is upon the critic to demonstrate their independent origin. Hence for the Trinucleacea Liostracina and the rest of the Trincleacea are linked by glabellar, sutural, and pygidial characters, and differ only in retained primitive characters of the former, which are irrelevant other than for inferring ultimate ‘ptychoparioid’ ancestry. The median suture of the primitive form, and protaspis characters of the later ones, indicate to us that the group belongs within Asaphina. While we cannot claim that the list of uniting characters is a long one, they can be clearly stated, and our hypothesis of relationships can be disproved by demonstrating that the characters we cite are polyphyletically derived. This seems to us to be an advance on the present classification, in which Trinucleacea are unrelated to any other group, other than vaguely placed as a subgroup of Ptychopariida for unspecified reasons. DISCUSSION OF PREVIOUS CLASSIFICATIONS IN RELATION TO THAT PROPOSED HERE The classification given here (summary in Table 4) differs from that used in the Treatise (Moore 1959) and from the newer classification of Bergstrom (1973). The reasons for our view of Asaphina have been given in detail above, but Bergstrom’s classification in particular deserves consideration so that our reasons for differing from it can be explicitly listed. 1. Bergstrom placed Ceratopygidae and Asaphidae in two different suborders of his Order Redlichiida. We consider them certainly closely related, and favour the view that they are sister groups in Asaphacea. It is difficult to see any critical characters in Bergstrom’s diagnosis which might serve as a basis to falsify our classification. Most of the characters mentioned in his diagnosis of Ceratopygacea either apply also to his diagnosis of Asaphacea (and of Asaphina) or do not apply to the families under consideration. Certainly there are no synapomorphies mentioned linking asaphids and ceratopygids to their respective supposed superfamilies which would compare with the distinctive glabellar structure that is one of the stronger pieces of evidence for their alliance in our classification. FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 215 table 4. Summary of classification of Asaphina as reviewed here, to family level. Suborder asaphina Salter, 1864 emend. Superfamily asaphacea Burmeister, 1843 Family asaphidae Burmeister, 1843 Family ceratopygidae Linnarsson, 1869 [includes macropygidae Kobayashi, 1953] Hedinaspis, Asiocephalus incertae familiae Superfamily cyclopygacea Raymond, 1925 Family cyclopygidae Raymond, 1925 Family nileidae Angelin, 1854 Family taihungshaniidae Sun, 1931 Superfamily remopleuridacea Hawle and Corda, 1847 Family remopleurididae Hawle and Corda, 1847 Family hungaiidae Raymond, 1924 Family bohemillidae Barrande, 1872 Family opipeuteridae Fortey, 1974 Superfamily dikelocephalacea Miller, 1889 emend. Ludvigsen and Westrop, 1983 Family dikelocephalidae Miller, 1889 Family saukiidae Ulrich and Resser, 1930 Family ptychaspididae Raymond, 1924 [includes eurekiidae Hupe, 1953] Superfamily trinucleacea Hawle and Corda, 1847, emend. Family trinucleidae Hawle and Corda, 1847 Family raphiophoridae Angelin, 1854 [includes endymioniidae Raymond, 1924] Family orometopidae Hupe, 1955 (includes myindidae Hupe, 1955 and hapalopleuridae Harrington and Leanza, 1957) Family dionididae Giirich, 1907 Family alsataspididae Turner, 1940 Family liostracinidae Raymond, 1937 (?part only) Superfamily anomocaracea Poulsen, 1927 (paraphyletic) Family anomocaridae Poulsen, 1927 Family dikelokephalinidae Kobayashi, 1936 Family pterocephaliidae Kobayashi, 1935 [probably includes housiidae] Family auritamiidae Opik, 1967 ? Family idahoiidae Lochman, 1956 Incertae superfamiliae isocolidae Angelin, 1854 catillicephalidae Raymond, 1938 (part only) 2. Although Bergstrom placed Asaphacea and Remopleuridacea within Asaphina, and included in the former some of the families considered to belong in Asaphina here, he also included Olenacea in the same suborder. We disagree with this placement of the Olenacea for the following reasons: 1, although most olenids have the free cheeks fused as a single piece— i.e. there is no evidence for either median suture or rostral plate — it is clear that the earliest olenids of the genus Olenus (O. wahlenbergi, see Rushton 1983) have a rostral plate and not a median suture; this allies the Olenidae with Ptychopariina and militates against an origin in Asaphina. 2, the diagnostic olenacean characters listed by Bergstrom are: a, ’hypostome probably disconnected from doublure’— this is probably a primitive ptychoparioid character, and as such is not diagnostic; b , ‘genal spines needle-like’— this may be an autapomorphy for Olenidae, but suggests no connection with any Asaphina. 3, olenids all maintain a narrow cephalic doublure medially beneath the cranidial border, even in forms with a relatively large pygidium. This is fundamentally different from Asaphina, in which a wide cephalic doublure is present medially in all but a few of the most primitive forms, and is invariable in those species with a large pygidium. 216 PALAEONTOLOGY, VOLUME 31 EVOLUTIONARY HISTORY OF ASAPHINA We can briefly summarize the account of character distribution and phylogeny in the tree shown as text-fig. 27. No attempt is made to do more than sketch the varied stratigraphically based histories of within-group evolution. We have included the Trinucleacea in this diagram. The stratigraphic record of most groups is reasonably complete, the most important ‘break’ being the early history of Cyclopygacea. This break may be filled when the relationships of Catillicephalidae are resolved. The origin of the group was presumably in the Middle Cambrian. The tree shows the reason for the problems with defining ‘Anomocaracea’ sensu lato , which forms a root group from which the various derived taxa originated. The Dikelokephalinidae alone of this plexus survived beyond the Cambrian. The Cambrian-Ordovician boundary closely coincides with the end of the dikelocephalacean clade, but otherwise the influence of this horizon on the group does not seem to have been profound. This is reflected by the persistence of individual genera (e.g. Niobella) through sections spanning the boundary. However, three of the surviving clades, Ceratopygidae, Dikelokephalinidae, and Taihungshaniidae, did not survive beyond the Middle Ordovician. Asaphidae, Nileidae, Cyclo- pygidae, Remopleurididae, and Trinucleacea continued as diverse elements of Ordovician faunas until the end of the System. During this long Cambrian-Ordovician interval Asaphina diversified into many habitats. Some taxa are particularly characteristic of cratonic sedimentary environments (some Asaphidae and Dikelokephalinidae, and perhaps most Dikelocephalacea). Pelagic habits probably arose on four occasions within the group (Bohemillidae with remopleuridid ancestry, Cyclopygidae, Parabarrandia, and Girvanopyge). Of these, the cyclopygids were mesopelagic and are only found in relatively exterior (off craton) facies. At least some asaphids and ceratopygids, and some aberrant small-eyed nileids, were benthic forms inhabiting deep-water sites. Trinucleids were slow moving benthos occupying various water depths, apparently on muddy substrates. Other asaphines are familiar components of limestone biofacies. Asaphina were not, in general, typical of reef-like habitats (illaenid-cheirurid trilobite biofacies), although some raphiophorids ( Lonchodomas ), asaphids ( Anataphrus ), and rare nileids (Afi'/eus) have been reported from faunas of this kind. We may conclude that the group was diverse and successful within almost the whole range of marine habitats adopted by the trilobites as a whole. Many stratigraphic case studies show that the Trinucleacea in particular evolved rapidly (Hughes et al. 1975), as did Asaphidae. The same is not true of Cyclopygidae, which have generic ranges extending from Arenig to Ashgill, and extraordinarily conservative morphology in a range of genera after what was presumably a rapid late Tremadoc to early Arenig radiation (Fortey and Owens 1987). Regardless of such differences, virtually the entire Asaphina was extinguished at the Ordovician- Silurian boundary. Available evidence suggests that a range of genera of the various families persisted into the last, and very short, Ordovician Stage, the Hirnantian, and so the extinction event has to be considered of some magnitude. The only survivor is the raphiophorid Raphiophorus , which persisted into the later Silurian. The Asaphina (particularly if we include Trinucleacea and isocolids) is the group much the most affected by the end-Ordovician event; other groups at family level which did not survive include agnostids, dimeropygids, together with the last deep-shelf olenid, and the pelagic telephinids. Why the Asaphina proved so vulnerable to the end Ordovician extinction is an interesting question. Clearly it was not because they had become too specialized in their habitat requirements, and vulnerable to major perturbations, because they had as wide a range of adaptations as any other trilobite group. Although some of the non-asaphine genera which survived the event were reef dwellers, others were not, and the scarcity of Asaphina adapted to this habitat does not seem an adequate explanation. Because members of the group were capable of both rapid evolution in the manner of trinucleids, or exceptional stasis, as in cyclopygids, it is difficult to invoke any explanation requiring differential response to rapid environmental change. The group were also distributed world-wide, and so we cannot account for their disappearance as coinciding with the removal of some palaeobiogeographic ‘province’. We are left with the fact that the very different groups of Asaphina which survived to the end of the Ordovician all had the asaphoid larva, which we have argued above FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 217 PTYCHOPARIINA text-fig. 27. Summary in the form of an evolutionary tree of the Asaphina, as understood in this work, with sketches of some primitive and advanced examples known from stratigraphic evidence within given clades. The Anomocaracea is here shown to include the plexus of primitive forms, and is not a natural group. Only one trinucleacean genus, Raphiophorus, continues beyond the end of the Ordovician. Reconstructions of exoskeletons are only very approximately to scale. > 3 3 z 5 1 O o 218 PALAEONTOLOGY, VOLUME 31 was well-adapted for planktic— possibly epiplanktic— habits. Could it be that the extinction of the group was related to a change in oceanic circulation which rendered this larval type particularly vulnerable? A widespread anoxic event has been quoted at the very end of the Ordovician on other evidence, and the planktic graptolites were also seriously affected at the end of the Ordovician. Although this is speculation, it is the only explanation we can offer as to why this great group of trilobites declined suddenly after 100 million years of successful history. REFERENCES adams, e. n. 1972. Consensus techniques and the comparison of taxonomic trees. Syst. Zool. 21, 390-397. angelin, N. p. 1 854. Palaeontologica Scandinavica , Part 1, Crustacea formationis trartsitionis , fasc. 2, 21 92, Lund. barrande, J. 1872. Systeme Silurien du centre de la Boheme: lere partie. Supplement au Vol. 1, 647 pp., 37 pis. Prague. Bergstrom, j. 1973. Organisation, life and systematics of trilobites. Fossils Strata , 2, 1-69. bulman, o. m. b. and rushton, a. w. a. 1973. Tremadoc faunas from boreholes in Central England. Bull. geol. Surv. Gt. Br. 43, 1-40, pis. 1-7. burmeister, H. 1843. Die organisation der Trilobiten . . . &c, xii+ 147 pp., 6 pis. Berlin. chatterton, b. d. e. 1971. Taxonomy and ontogeny of Siluro-Devonian trilobites from near Yass, New South Wales. Palaeontographica , Abt. A, 137, 1-108, pis. 1-24. — 1980. Ontogenetic studies of Middle Ordovician trilobites from the Esbataottine Formation, Mackenzie Mountains, Canada. Ibid. 171, 1-74, pis. 1-19. — and perry, d. g. 1984. Silurian cheirurid trilobites from the Mackenzie Mountains, northwestern Canada. Ibid. 184, 1-78, pis. 1-35. CHU, chao-ling. 1959. Trilobites from the Kushan Formation of north and northeastern China. Mem. Inst. Paleont., 2, 128 pp„ 7 pis. [In Chinese.] cooper, b. n. 1953. Trilobites from the Lower Champlainian formations of the Appalachian Valley. Mem. geol. Soc. Am. 55, 1-69, pis. 1-19. courtessole, r. and pillet, j. 1975. Contribution a l’etude des faunes de l’Ordovicien inferieur de la Montagne Noire: les Eulominae et les Nileidae. Annls Soc. geol. Nord , 95, 251-272. — and vizcaino, d. 1981. Nouvelles Donnees sur la biostratigraphie de l’Ordovicien Inferieur de la Montagne Noire: revision des Taihungshaniidae, de Megistaspis ( Ekeraspis ) et d’ Asaphopsoides (Trilobites). Mem. Soc. Etud. Scient. Aude. 1-32, 11 pis. dean, w. t. 1970. A new Lower Ordovician trilobite faunule from Random Island, Eastern Newfoundland. Geol. Surv. Pap. Can. 70-19, vi + 10 pp. EGOROVA, L. I., SHABANOV, YU. YA., PEGEL, T. V., SAVITSKY, V. E., SUCHOV, S. S. and TCHERNYSHEVA, N. E. 1982. Maya Stage of the Type Locality (Middle Cambrian of the Siberian Platform). Trudy Mezhvedom. Strat. Kom. CCCP , 8, 1-146, pis. 1-64. eldredge, N. and cracraft, J. 1980. Phylogenetic patterns and the evolutionary process , 348 pp. Columbia University Press, New York. evitt, w. r. 1961. Early ontogeny in the trilobite family Asaphidae. J. Paleont. 35, 986-995, pis. 117 and 1 18. — and tripp, r. p. 1977. Silicified Middle Ordovician trilobites from the families Encrinuridae and Staurocephalidae. Palaeontographica , Abt. A, 157, 109-174, 24 pis. and Whittington, H. B. 1963. The exoskeleton of Flexicalymene (Trilobita). J. Paleont. 27, 49-55, pis. 9 and 10. forey, p. L. 1 987. The Downtonian ostracoderm Sclerodus Agassiz (Osteostraci: T remataspididae). Bull. Br. Mus. nat. Hist. (Geol.), 41, 1-29. fortey, r. a. 1974. A new pelagic trilobite from the Ordovician of Spitsbergen, Ireland and Utah. Palaeontology, 17, 111-124, pis. 13 and 14. 1975. The Ordovician trilobites of Spitsbergen. II. Asaphidae, Nileidae, Raphiophoridae and Telephinidae of the Valhallfonna Formation. Skr. norsk Polarinst. 162, 1-125, pis. 1-41. 1980. Basilicus tyrannus (Murchison) and the glabellar structure of asaphid trilobites. Bull. Br. Mus. nat. Hist. (Geol.), 34, 255-264. — 1981. Prospectatrix genatenta (Stubblefield) and the trilobite superfamily Cyclopygacea. Geol. Mag. 118, 603-614. 1983. Cambrian Ordovician boundary trilobites from western Newfoundland and their phylogenetic significance. Spec. Pap. Palaeont. 30, 179-211, pis. 23-27. FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 219 1987. The type species of the Ordovician trilobite Symphysurus: systematics, functional morphology, and terrace ridges. Palaeont. Z. 60, 255-275. — and clarkson, E. N. K. 1976. The function of the glabellar ‘tubercle’ in N ileus and other trilobites. Lethaia,9, 101 106. and jefferies, r. p. s. 1982. Fossils and phylogeny: a compromise approach, 197-234. In joysey, k. a. and Friday, a. E. (eds. ). Problems of phylogenetic reconstruction. Academic Press, London. — and morris, s. f. 1978. Discovery of nauplius-like trilobite larvae. Palaeontology , 21, 823-833, pi. 94. and owens, R. m. 1975. Proetida — a new order of trilobites. Fossils Strata , 4, 227-239. — 1982. Trilobita. In owens, r. m., fortey, r. a., cope, j. c. w., rushton, a. w. a. and bassett, m. g. Tremadoc faunas from the Carmarthen district. South Wales. Geol. Mag. 119, 1-38, pis. 1-8. — 1987. The Arenig Series in South Wales. Bull. Br. Mus. nat. Hist. (Geol ), 41, 69-303. and shergold, j. h. 1984. Early Ordovician trilobites from the Nora Formation, central Australia. Palaeontology, 27, 315-366, pis. 38-46. fryer, g. 1983. Functional ontogenetic changes in Branchinecta ferox (Milne-Edwards) (Crustacea: Anostraca). Phil. Trans. R. Soc., ser. B, 303, 229 343, pis. 1-12. gurich, G. 1907. Versuch einer Neueinteilung der Trilobiten. Zentbl. Miner. Geol. Paldont. 1907, 129-133. Harrington, h. j. and leanza, a. f. 1957. Ordovician trilobites of Argentina, 276 pp. University of Kansas Press, Lawrence. hawle, i. and corda, a. j. c. 1847. Prodrom einer monographic der bohmischen trilobiten, 167 pp., 7 pis. Prague. henningsmoen, g. 1951. Remarks on the classification of trilobites. Norsk geol. Tiddskr. 29, 174-217. — 1957. The trilobite family Olenidae. Skr. norske Vid. Akad. Oslo, 1, Mat.-Nat. Kl. 1, 1-303. hill, d., playford, G. and woods, J. T. 1969. Ordovician and Silurian fossils of Queensland, 18 pp. Queensland Palaeontographical Society, Brisbane. hu, c.-H. 1971. Ontogeny and sexual dimorphism of Lower Palaeozoic Trilobita. Palaeontogr. am. 44, 29-155, pis. 7-26. — 1986. Ontogenetic development of Cambrian trilobites from British Columbia and Alberta, Canada (Part 2). J. Taiwan Mus. 39, 1-44, pis. 10-19. hughes, c. p., ingham, j. k. and addison, r. 1975. The morphology, classification and evolution of the Trinucleidae (Trilobita). Phil. Trans. R. Soc., ser. B, 272, 537-607. hupe, p. 1953. Classification des trilobites (Part I). Annls Paleont. 39, 61 168. 1955. Classification des trilobites (Part 2). Ibid. 41, 91-325. jell, p. A. 1985. Tremadoc trilobites of the Digger Island Formation, Waratah Bay, Victoria. Mem. Natn. Mus. Viet. 46, 53-88, pis. 19-33. kobayashi, T. 1935. The Cambro-Ordovician formations of South Chosen. Palaeontology, Part 3. Cambrian faunas of South Chosen with a special study on the Cambrian trilobite genera and families. J. Fac. Sci. Tokyo Univ., sec. 2, 4, 49-344. 1936. Three contributions to the Cambro-Ordovician faunas. I. The Dikelokephalininae (nov.), its distribution, migration and evolution. Jap. J. Geol. Geogr. 13, 163-178. 1953. On the Kainellidae. Ibid. 23, 37-61. 1962. The Cambro-Ordovician formations and faunas of South Korea. Part IX. Palaeontology VIII. The Machari fauna. J. Fac. Sci. Tokyo Univ., Ser. 2, 14, 1-152, pis. 1-12. and hamada, t. 1971. Contributions to the geology and palaeontology of southeast Asia. 78. A cyclopygid bearing faunule discovered in Malaya with a note on the Cyclopygidae. Geol. Palaeont. SE Asia, 8, 1-18." lake, p. 1907. A monograph of the British Cambrian trilobites. Palaeontogr. Soc. [Monor/r.], Pt. 2, 29-48, pis. 3 and 4. linnarsson, j. g. o. 1869. Om vestergotlands cambriska och siluriska aflagringar. K. svenska VetenskAkad. Handl. 7, 1-89, pis. 1-2. lisogor, k. a. 1977. Trilobita. In Zhuravleva, i. t. and rozova, a. v. (eds.). Biostratigraphy and fauna of the Upper Cambrian boundary beds; new data from the asiatic regions of the USSR. Trudy Inst. Geol. Geofiz. sib. Otd. 313, 355 pp., 31 pis. [In Russian.] lochman, c. 1956. Stratigraphy and paleogeography of the Elliptocephala asaphoides strata in Cambridge and Hoosick quadrangles. New York. Bull. geol. Soc. Am. 67, 1331 1396, pis. I 10. lu yanhao. 1965. In lu yanhao, zhang wentan et al. Fossils of each group of China: Chinese trilobites. Science Press, Beijing. [In Chinese ] -1974. Bioenvironmental control hypothesis and its application to the Cambrian biostratigraphy and palaeozoogeography. Mem. Inst. Geol. Paleont., Nanjing, 5, 27-110, pis. 1-4. [In Chinese.] 220 PALAEONTOLOGY, VOLUME 31 lu yanhao. 1975. Ordovician trilobite faunas of central and southwestern China. Palaeont. sin. 152, (ns) 1 1 B, i-vi + 1-261 [Chinese], 265-463 [English], 50 pis. -and lin huanling. 1984. Late Cambrian and earliest Ordovician trilobites of Jiangshan-Changshan area, Zhejiang, 45-143, pis. 1- 19. In Stratigraphy and Palaeontology of Systemic Boundaries in China, Cambrian Ordovician Boundary, 1. Anhui Science and Technology Publishing House. ludvigsen, r. 1982. Upper Cambrian and Lower Ordovician trilobite biostratigraphy of the Rabbitkettle Formation, western District of Mackenzie. Contr. Life Sci. Div. R. Ont. Mus. 134, 1 1 88. — and westrop, s. 1983. Franconian trilobites of New York State. Mem. N.Y. St. Mus. 23, 1-44, 19 pis. — 1986. Classification of the Late Cambrian trilobite Idiomesus Raymond. Can. J. Earth Sci. 23, 300-307. mannil, r. 1958. Trilobites of the families Cheiruridae and Encrinuridae from Estonia. Eesti NSV Tead. Akad. Geol. Inst. Uurimused, 3, 165-212, 8 pis. [In Russian, with Estonian and English summaries.] manton, s. m. 1964. Mandibular mechanisms and the evolution of arthropods. Phil. Trans. R. Soc., ser. B, 247, 1-183. miller, s. A. 1889. North American geology for the use of amateurs, students and scientists, 664 pp. Cincinnati. moore, R. E. (ed.). 1959. Treatise on Invertebrate Paleontology. Part 0. Arthropoda 1. Geological Society of America and University of Kansas Press, New York and Lawrence, Kansas. neben, w. and krueger, h. h. 1971. Fossilien ordovischer Geschiebe. Staringia, 1, v + 49 pis. nowlan, G. s., mccracken, A. d. and chatterton, b. d. e. In press. Condonts from Ordovician-Silurian boundary strata, Whittaker Formation, Mackenzie Mountains, Northwest Territories, Canada. Bull. geol. Surv. Can. 373. OPIK, A. A. 1967. The Mindyallan fauna of northwestern Queensland. Bull. Bur. Miner. Resour. Geol. Geophys. Aust. 74, vol. 1, 404 pp.; vol. 2, 166 pp., 61 pis. palmer, A. r. 1958. Morphology and ontogeny of a Lower Cambrian ptychoparioid trilobite from Nevada. J. Paleont. 32, 154 170, pis. 25 and 26. — 1960. Triiobites from the Upper Cambrian Dunderberg Shale, Eureka District, Nevada. Prof Pap. US geol. Surv. 334-C, 109 pp., 1 1 pis. — 1962a. Glyptagnostus and associated trilobites in the United States. Ibid. 374-F, 49 pp., 6 pis. — 1962b. Comparative ontogeny of some opisthoparian, gonatoparian and proparian Upper Cambrian trilobites. J. Paleont. 36, 87-96, pi. 19. — 1965. Trilobites of the late Cambrian pterocephaliid biomere in the Great Basin, western United States. Prof. Pap. US geol. Surv. 493, 105 pp., 23 pis. 1968. Cambrian trilobites of east central Alaska. Ibid. 559-B, 1 15 pp., 15 pis. peng shanchi. 1984. Cambrian-Ordovician boundary in the Cili-Taoyan border area, northwestern Hunan, with descriptions of relative trilobites, 285- 405, pis. 1-18. In Stratigraphy and Palaeontology of Systemic Boundaries in China: Cambrian-Ordovician boundary. Anhui Science and Technology Publishing House. poulsen, c. 1927. The Cambrian, Ozarkian and Canadian faunas of Northwest Greenland. Meddr. Gronland. 70, 237-343, pis. 14-21. rasetti, f. 1954. Phytogeny of the Cambrian trilobite family Catillicephalidae and the ontogeny of Welleraspis. J. Paleont. 28, 599-612. — 1963. Middle Cambrian ptychoparioid trilobites from the conglomerates of Quebec. Ibid. 37, 575-594, pis. 66 70. — 1965. Upper Cambrian trilobites of northeastern Tennessee. Smithson, misc. Colins, 148(3), 1-127, pis. 1-21. — 1972. Cambrian trilobite faunas of Sardinia. Atti Accad. Naz. Lincei Memorie, ser. VIII, sez. 2a, 1 1, 1-100, pis. 1-19. Raymond, p. E. 1924. New Upper Cambrian and Lower Ordovician trilobites from Vermont. Proc. Boston Soc. nat. Hist. 37, 389-466, pis. 12-14. — 1925. Some trilobites of the Lower Middle Ordovician of eastern North America. Bull. Mus. comp. Zool. Harv. 67, 1-180, pis. 1-10. 1938. Corrections and emendations (to vol. 38, printed after the last part). Bull. geol. Soc. Am. 38, xv [separate pagination], robison, r. a. 1964. Late Middle Cambrian faunas from western Utah. J. Paleont. 38, 510-566, pis. 79-92. — and pantoja-alor, J. 1968. Tremadocian trilobites from the Nochixtlan region, Oaxaca, Mexico. Ibid. 42, 767-800, pis. 97-104. ross, r. j. 1951a. The ontogenies of three Garden City (Early Ordovician) trilobites. Ibid. 25, 578-586, pis. 81-84. 1951b. Stratigraphy of the Garden City Formation in northeastern Utah and its trilobite faunas. Bull. Peabody Mus. nat. Hist. 6, 1-155, 36 pis. — 1953. Additional Garden City (Early Ordovician) trilobites. J. Paleont. 27, 633-646. FORTEY AND CHATTERTON: TRILOBITE CLASSIFICATION 221 rushton, a. w. a. 1978. Fossils from the Middle-Upper Cambrian transition in the Nuneaton district. Palaeontology, 21, 245-283, pis. 24-26. -1982. The biostratigraphy and correlation of the Merioneth-Tremadoc Series boundary in North Wales, 41-59. In bassett, m. G. and dean, w. t. (eds.). The Camhrian-Ordovician boundary . . ,&c. National Museum of Wales, Cardiff, Geological Series No. 3. — 1983. Trilobites from the Upper Cambrian Olenus Zone in central England. Spec. Pap. Palaeont. 30, 107-139, pis. 14-19. — and hughes, c. p. 1981. Trilobites from the Great Paxton Borehole. Geol. Mag. 118, 623-646, 6 pis. salter, J. w. 1 864. A monograph of the British trilobites from the Cambrian, Silurian and Devonian Formations. Palaeontogr. Soc. [ Monogr.~\ , part I, 1 -80, pis. 1 6. schrank, E. 1972. Nileus- arten (Trilobita) aus Geschieben des Tremadoc bis tieferen Caradoc. Ber. dt. Ges. geol. (TTss., ser. A, 17, 351 -375. shaw, F. c. 1968. Early Middle Ordovician Chazy trilobites of New York. Mem. N.Y. St. Mas. 17, 1 163, pis. 1-24. shergold, J. H. 1972. Late Upper Cambrian trilobites from the Gola Beds, western Queensland. Bull. Bur. Miner. Resour. Geol. Geophys. Aust. 112, 1-89, 19 pis. — 1975. Late Cambrian and early Ordovician trilobites from the Burke River Structural belt, western Queensland, Australia. Ibid. 153, vol. 1 (text), 251 pp.; vol. 2 (plates), 1-58. — 1980. Late Cambrian trilobites from the Chatsworth Limestone, western Queensland. Ibid. 186, 1-111, pis. 1-35. — 1982. Idamean (Late Cambrian) trilobites, Burke River Structural Belt, western Queensland. Ibid. 187, 1-69, pis. 1 17. — and sdzuy, k. 1984. Cambrian and early Tremadocian trilobites from Sultan Dag, central Turkey. Senckenberg. leth. 65, 53-135. Snajdr, m. 1975. On the ontogeny of Bohemian representatives of the genus Parabarrandia Prantl et Pribyl (Trilobita). Vest, ustred. Ust. geol. 50, 241-244, pis. 1 and 2. Stubblefield, c. J. 1927. In Stubblefield, c. J. and bulman, o. m. b. The Shineton Shales of the Wrekin District, with notes on their development in other parts of Shropshire and Herefordshire. Q. Jl geol. Soc. Lond. 83, 96-146, pis. 3-5. SUN, Y. c. 1931. Ordovician trilobites of central and southern China. Palaeont. sin. ser. B, 7, 1-47, pis. 1-3. tiernvik, t. 1956. On the early Ordovician of Sweden: stratigraphy and fauna. Bull. geol. Instn. Univ. Uppsala , 36, 107-284, 11 pis. tripp, r. p. 1962. The Silurian trilobite Encrinurus punctatus (Wahlenberg), and allied species. Palaeontology , 5, 460-477, pis. 65-68. — and evitt, w. r. 1986. Silicified trilobites of the family Asaphidae from the Middle Ordovician of Virginia. Ibid. 29, 705-724, pis. 54-57. troedsson, g. t. 1937. On the Cambro-Ordovician faunas of western Qurugtagh, eastern T’ien Shan. Palaeont. sin. (ns) B 2, 1-74, pis. 1-10. turner, f. e. 1940. Alsataspis bakeri , a new lower Ordovician trilobite. J. Paleont. 14, 516-518. ulrich, e. o. and resser, c. e. 1930. The Cambrian of the upper Mississippi Valley. Part 1. Trilobita; Dicelocephalinae and Osceolinae. Bull. publ. Mus. Milwaukee, 12, 1-122. westergard, a. h. 1939. On Swedish Cambrian Asaphidae. Sver. geol. Unders. Afh., Ser. C, 421 [Arsbok 33 (1939) No. 1], 1 11, 3 pis. — 1947. Supplementary notes on the Upper Cambrian trilobites of Sweden. Ibid. 489, 1-34, 3 pis. whittard, w. f. 1960. The Ordovician trilobites of the Shelve Inlier, west Shropshire. Palaeontogr. Soc. [. Monogr.l part IV, 117-162, pis. 16-21. Whittington, h. b. 1956. Silicified Middle Ordovician Odontopleuridae. Bull. Mus. comp. Zool. Harv. 114, 155-288, pis. 1-24. 1959 a. Silicified Middle Ordovician trilobites: Remopleurididae, Trinucleidae, Raphiophoridae, Endy- mioniidae. Ibid. 121, 369-496, pis. 1-36. — 19596. Ontogeny of Trilobita, Pp. 0127-0144 In moore, r. c. (ed.). Treatise on Invertebrate Paleontology. Part O, Arthropoda 1. Geological Society of America and University of Kansas Press, New York and Lawrence, Kansas. — 1963. Middle Ordovician trilobites from Lower Head, western Newfoundland. Bull. Mus. comp. Zool. Harv. 129, 1-118, 36 pis. — 1965. Trilobites of the Ordovician Table Head Formation, western Newfoundland. Ibid. 132, 275-442, 68 pis. 222 PALAEONTOLOGY, VOLUME 31 Whittington, H. b. and evitt, w. r. 1954. Silicified Middle Ordovician Trilobites. Mem.geol. Soc. Am. 59, 1 137, pis. 1-33. whitworth, p. 1969. The Tremadoc trilobite Pseudokainella impar (Salter). Palaeontology, 12, 406-413, pi. 75. — 1970. Ontogeny of the Upper Cambrian trilobite Leptoplastus crassicornis (Westergaard) from Sweden. Ibid. 13, 100-111, pis. 22-24. wittke, h. w. 1984. Middle and Upper Cambrian trilobites from Iran: their taxonomy, stratigraphy and significance for provincialism. Palaeontographica, Abt. A, 183, 91 161, II pis. ZHOU zhiyi and zhang jinlin. 1978. Cambrian-Ordovician boundary of the Tangshan area with descriptions of the related trilobite fauna. Acta palaeont. sin. 17, 1-26, pis. 1-4. R. A. FORTEY British Museum (Natural History) Cromwell Road London SW7 5BD Typescript received 20 December 1986 Revised typescript received 15 May 1987 B. D. E. CHATTERTON Department of Geology University of Alberta Edmonton, Alberta Canada T6G 2E3 OPEN NOMENCLATURE by PETER BENGTSON Abstract. Open nomenclature plays an important role in taxonomic decisions by palaeontologists, but usage and interpretation of the signs employed vary considerably. Prevailing fashion seems to favour aff. to indicate affinity of a potentially new, as yet undescribed species with a known species, whereas cf. and ? indicate uncertainty. Use of aff., cf., and ? for different degrees of uncertainty, as recommended by some workers, leads to instability in interpretation. Abbreviated taxonomic expressions such as ‘ Trichiurus cf. lepturus ’ are un- ambiguous and are to be preferred to ‘ Trichiurus cf. T. lepturus'. Careful, judicious use of open nomenclature is to be encouraged and should be covered by the International Code of Zoological Nomenclature. A set of recommendations is given. In a recent article Lucas (1986) discusses the use of the qualifiers aff. and cf. in taxonomy, and more specifically their proper position in taxonomic names. Such expressions are usually termed ‘open nomenclature’. Lucas refers to the International Code of Zoological Nomenclature (International Commission on Zoological Nomenclature, 1985) to support his contention that the common procedure of inserting aff. and cf. between the generic name and the specific name is syntactically incorrect. However, since the Code makes no provision for cases of open nomenclature, his arguments lack authority and arguably credibility. Nevertheless Lucas’s article alerts us to the fact that the status of open nomenclature is neglected, perhaps needing a set of rules or recommendations by the International Commission on Zoological Nomenclature. Although I was inspired to write this article by Lucas’s note, I have been concerned for some time about the inconsistent approaches to open nomenclature. These circumstances lead to fluctuating interpretations of taxonomic statements which, in turn, may impede scientific communication. This article aims to deal with the core of the subject of open nomenclature, on which ideally there should be no argument. For more detailed discussions regarding special cases and more arcane expressions in open nomenclature, the reader is referred to Richter (1948) and Matthews (1973). OPEN NOMENCLATURE Use of open nomenclature is the procedure by which a taxonomist comments upon the identity of a specimen that cannot be readily or securely determined. The procedure is more common in palaeontology than in neontology, a fact that, of course, stems from the incompleteness of most palaeontological material. Uncertainty or the provisional status of a taxonomic identification may be expressed in prose, such as ‘probably Agenus aspecies ’, but is more often codified through the use of qualifiers such as aff., cf., or ?. Richter (1943, pp. 34-40; 1948, pp. 45-52) treated the subject of open nomenclature in detail, and Matthews (1973) is essentially a translation of Richter’s work. Richter emphasized the need for open nomenclature: should a specimen be too hastily referred to a known species or genus, taxonomic information may be concealed or distorted. If on the other hand the specimen is left without any attempt at identification, potentially useful information may be left in limbo. In order to derive the maximum benefit from any specimen with a minimum of distortion of information, open nomenclature is an essential tool in the taxonomist’s repertory. As expressed by Richter (1948) (and Matthews 1973), open nomenclature is an especially perspicacious form of nomenclature. Careful and judicious use of open nomenclature reflects scientific honesty; its use is not a sign of weakness or lack of confidence, contrary to the opinion of some taxonomists. [Palaeontology, Vol. 31, Part 1, 1988, pp. 223-227.] © The Palaeontological Association 224 PALAEONTOLOGY, VOLUME 31 THE SIGNS IN OPEN NOMENCLATURE By far the most common signs used are aff., cf., ?, and sp. The question of the position of aff. and cf. within the binomen (Lucas 1986) is trivial in comparison with the problem of vacillating interpretations of the signs. Since the International Code of Zoological Nomenclature refrains from any reference to open nomenclature, taxonomists are left to their own subjective interpretations of what the signs stand for in each individual case. However, since neither Richter’s (1948) nor Matthews’s ( 1973) works seem to have been widely accepted by our community, I suggest that the time is ripe for a more stringent approach to open nomenclature. My impression is that the following usage prevails amont palaeontologists: aff. (or n. sp., aff., or sp. nov., aff.) preceding a species-group name indicates that the specimen(s) is considered a new, previously undescribed species or subspecies. The material is insufficient for formal description and naming of a new taxon, but the specimen(s) can be most closely related to the species or subspecies following the qualifier. Thus, aff. does not necessarily involve uncertainty. Some workers make a distinction between aff. and n. sp., aff., using aff. alone to signal that the specimen(s) differs clearly from the holotype but may still fall within the limits of variation of the species (e.g. Kennedy and Hancock 1971, p. 437). ‘. . i.e. quotation marks, around a genus-group name indicate that the species is thought to belong to a new genus (or subgenus) related to the named genus, but the material available is insufficient for the formal erection of a new genus. (Obviously, an aff. in front of the genus-group name will convey the same message but is for some reason little used.) Quotation marks around a generic name are also used to indicate that the name is obsolete (cf. Jeppsson and Merrill 1982). cf. preceding a species-group name (rarely a genus-group name) indicates that the determination is uncertain, the reason for which may be poor preservation of the material studied or that the determination is provisional. ? overlaps partly in usage with cf., although the former is less frequently used for provisional determinations. sp. (and ssp.) indicates that the specimen cannot be related to any established species (or subspecies) or that specific identification has not (yet) been attempted. These signs cover the majority of situations where open nomenclature is required. Other, less commonly used expressions, such as sp. imdet., sp. A, ex gr., are in most cases self-explanatory. Current usage as discussed above differs fundamentally from Lucas’s (1986) opinion that aff. and cf. express different degrees of uncertainty. If uncertainty is involved, it is rather a matter of different kinds of uncertainty. The reason for aff. and cf. being less commonly used for genus-group names can be sought in the differences in definition of species and genera, respectively. The inherently greater uncertainty in the genus concept does not normally call for yet further qualification. A question mark, which should be placed after the generic name (cf. Kornicker 1979), is sufficient for most situations. Incidentally, cf. stands for confer , not conformis (Lucas 1986), which means ‘compare to\ rather than ‘compare with’. The difference may appear academic, but is worth considering. The wording ‘compare to’ expresses a possible identity, which is what most taxonomists have in mind when they use cf., whereas ‘compare with’ rather implies a distinction (cf. Fowler 1982, pp. 99-100), thus approaching aff. in meaning. A comparison of current usage with that recommended by Matthews (1973) also shows some differences. Apparently quoting Richter (1948), Matthews states that cf. implies only a possibility of comparison with the named species, whereas ? means that attribution to the species is possible but cannot be thought certain. I take this to mean that cf. is meant to express greater uncertainty than a ?. It is interesting to note that in the first edition of Richter’s book (Richter 1943, pp. 37-38) the reverse practice was recommended. At that time Richter considered cf. to mean that attribution to the species BENGTSON: OPEN NOMENCLATURE 225 is probable but uncertain, whereas a ? would mean that attribution is improbable but possible. In Richter’s original view, then, cf. would express less uncertainty than a ?, i.e. exactly the opposite of the recommendation given in his second edition of the book. The change was explained and said to have been made to achieve conformity with prevailing usage in the literature (Richter 1948, p. 49). This illustrates how codification of degree of uncertainty is bound to lead to varying and unstable usage, so it is only natural that cf. and ? are considered synonymous in open nomenclature by a majority of taxonomists. If both signs are to be retained in open nomenclature, the differences in kinds of uncertainty should be emphasized rather than an ill-defined difference in degree of uncertainty. SYNTAX What then is the ’proper syntax’ when using the signs in open nomenclature? Lucas ( 1986) refers to the International Code of Zoological Nomenclature in advocating that expressions like ‘ Trichiurus cf. lepturus ’ are incorrect, and that the correct syntax is ‘ Trichiurus cf. T. lepturus'. His reference to the Code is curious, since it is not concerned with open nomenclature and thus cannot prohibit the insertion of cf. or all', between the generic name and the specific name. What is recommended (not prescribed) by the Code is that names of a former generic association should be given as a supplementary piece of information within parentheses rather than interpolated between the generic and specific names (Recommendation 6a), as often occurs. The reason is presumably that an interpolated generic name, even in square brackets, might be mistaken for a subgeneric name, the interpolation of which is the correct procedure. I agree entirely with Lucas (1986) that taxonomic nomenclature ‘should be as precise and unambiguous as possible’. But in the case of an interpolated aff. or ef. no confusion is possible, and this is, of course, why many palaeontologists prefer the shorter, more convenient construction. To write ‘ Trichiurus cf. T. lepturus ’ instead of "Trichiurus cf. lepturus' contributes nothing to clarity. The latter expression conveys in an unambiguous way the message that the author considers the specimen in question to be ‘probably or possibly the species lepturus, although there is not enough material to be sure, but if it is lepturus it should be referred to the genus Trichiurus'. Lucas (1986) also is correct in stating that ‘the phrase “cf. lepturus" in “ Trichiurus cf. lepturus" does not actually mean that the specimen(s) in question should be compared to Trichiurus lepturus', but his motives for saying so are contorted. The meaning of the expression is that the specimen(s) should be compared to the species lepturus, which the author refers to the genus Trichiurus, i.e. the uncertainty lies at the specific level, not at the generic level. Lucas continues: ‘That different species can have the same specific name . . . underscores the fact that a species is identified by a binomen, not by just its specific name.’ It is exactly the opposite: a species is identified by its specific name (and its author), not by a binomen. This is why it is important to include authors’ names (and year) in taxonomic nomenclature, as emphasized by Richter (1948) and Matthews (1973) in the statement ‘These two names (species + author) make up a nomenclatural entity, which nothing should be allowed to divide.’ Although the inclusion of authors’ names is left optional by the Code in key positions a species-group name should never be cited without its author. By following this practice the problems of homonymy can be practically eliminated, since there are few cases where an author during the course of the same year has given the same specific name to closely related taxa. Since a species is identified by its objectively defined species name, not by any subjective binomial combination, the expression ‘ Trichiurus cf. T. lepturus' is, strictly speaking incorrect syntax. It implies that the specimen(s) should be compared to only those specimens that have been described under the name T. lepturus, which is hardly what is meant. Abbreviation of taxonomic names, as discussed above, is often practical in applied palaeontology. For example, a biostratigraphic zone may be referred to as ‘the plenus Zone’, when it is understood that the ‘A. plenus Zone’ or the ‘ Actinocamax plenus Zone’ is intended. 226 PALAEONTOLOGY, VOLUME 31 CONCLUSIONS The fact that usage varies considerably, that the recommendations of Richter (1948) (and Matthews 1973) have not been universally adopted, and the fact that time and ink are expended on discussing the meaning of the signs used in open nomenclature is a strong motive for the International Commission on Zoological Nomenclature to consider issuing rules or recommendations on open nomenclature. Such recommendations may not put an end to discussions on the matter but they will provide a nomenclatural pillar to lean on. Analysis of current usage and the reasoning outlined above impels me to formulate a set of recommendations. These are presented here for discussion, and hopefully as a first step towards a formal proposal on open nomenclature to the Commission. As suitable fora for discussion of the matter I suggest, for example, the Palaeontological Association Circular , the Lethaia ‘Seminar’, or the section 'Points of view’ in Systematic Zoology. I shall, of course, also be glad to receive comments by letter. RECOMMENDATIONS aff. relates a new, undescribed taxon to a named taxon: e.g. aff. Agenus aspecies (for a new genus), Agenus aff. aspecies (for a new species), aff. Agenus aff. aspecies (for both a new species and a new genus). cf. indicates that the identification is provisional: e.g. cf. Agenus aspecies (for a provisionally assigned genus), Agenus cf. aspecies (for a provisionally identified species), cf. Agenus cf. aspecies (for both a provisionally assigned genus and a provisionally identified species). ? indicates that the identification is uncertain: e.g. Agenus ? aspecies (genus uncertain), Agenus aspeciesl (species uncertain). Agenusl aspeciesl (both genus and species uncertain). sp. (or ssp.) indicates that specific identification is impossible or has not been attempted, n. sp. (or n. ssp.) that the species (or subspecies) belongs to a new species and cannot be associated with any known species. ‘. . .’ indicates that the name is obsolete in the immediate context of systematic interest: e.g. ‘ Agenus ’ aspecies (generic name obsolete), Agenus ‘ aspecies ’ (specific name obsolete), ' Agenus aspecies' (both generic and specific name obsolete). These rules are intended to cover the great majority of situations where full identification is not possible. As is the case today, aff. and cf. will probably continue to be less commonly applied to genus-group names. Acknowledgements. I thank William T. Dean (University College, Cardiff) for constructive criticism and advice. The English was kindly improved by Simon Conway Morris (University of Cambridge). The work was supported by Swedish Natural Science Research Council (NFR) grant No. G-GU 3475. REFERENCES fowler, H. w. 1982. Fowler's Modern English Usage. 2nd edn„ reprinted with corrections, 725 pp. Oxford University Press, Oxford. international commission on zoological nomenclature 1985. International Code of Zoological Nomen- clature, 338 pp. International Trust for Zoological Nomenclature, London. jeppsson, L. and Merrill, G. k. 1982. How best to designate obsolete taxonomic names and concepts: examples among conodonts. J. Paleont. 56, 1489-1493. Kennedy, w. j. and Hancock, J. M. 1971. Mantelliceras saxbii, and the horizon of the Martimpreyi Zone in the Cenomanian of England. Palaeontology , 14, 437-454. kornicker, l. s. 1979. The question mark in taxonomic literature. J. Paleont. 53, p. 761. lucas, s. c. 1986. Proper syntax when using aff. and cf. in taxonomic statements. J. Vert. Paleont. 6, p. 202. BENGTSON: OPEN NOMENCLATURE 227 Matthews, s. c. 1973. Notes on open nomenclature and on synonymy lists. Palaeontology , 16, 713-719. richter, R. 1943. Einfiihrung in die Zoologische Nomenklatura 154 pp. Senckenbergische Naturforschende Gesellschaft, Frankfurt. - 1948. Einfiihrung in die Zoologische Nomenclatur. 2nd edn., 252 pp. Kramer, Frankfurt. Typescript received 12 December 1986 Revised typescript received 20 April 1987 PETER BENGTSON Paleontologiska museet Box 558 S-751 22 Uppsala Sweden Note added in proof. After the completion of the manuscript, two notes have appeared commenting on Lucas’s (1986) article. Zidek ( 1987) advocates that cf. and aff. are synonymous and used for tentative identifications, and that their meaning equals that of a question mark. Estes (1987), on the other hand, maintains that aff. indicates a greater degree of confidence than cf. Both these opinions differ from prevailing usage, as discussed in the present article; this further underscores the need for standardization of the signs in open nomenclature. REFERENCES estes, R. 1987. Lucas and Zidek on taxonomic syntax. J. Vert. Paleont. 7, p. 101. zidek, j. 1987. Response to Lucas on syntax in taxonomic statements. Ibid. 100-101 A REINTERPRETATION OF THE ARENIG CRINOID RAMSEYOCRINUS by JOHN C. W. COPE Abstract. A new specimen of the early Arenig inadunate crinoid Ramseyocrinus shows unequivocally that the genus had a cup composed of four basal and live radial plates, the latter supporting five arms. There are no anal plates in the dorsal cup. The separate familial status of the genus is confirmed and emended diagnoses of the family, and genus, and a revised description of the species R. cambriensis are given. Ramseyocrinus was proposed by Bates (1968) for Dendrocrinus cambriensis Hicks from the early Arenig of Ramsey Island, south-west Dyfed. Bates’ description of material in the National Museum of Wales allowed him to clarify certain aspects of the morphology of this early crinoid and established several features which Ramsbottom’s earlier description (1961) had not specified. Thus, Bates demonstrated that Ramseyocrinus had a four-lobed stem and suggested that the dorsal cup was composed of probably three basal and four radial plates together with an infer-radianal at the dorsal end of the anal sac. Donovan (1984) showed that the four-lobed stem of Ramseyocrinus was tetrameric proximally and suggested, contrary to Bates’ interpretation, that three radial plates supported four fixed brachials and an anal plate. Donovan considered that basal plates were absent or hidden by the stem attachment and on the basis of the unique plating he proposed a new family, the Ramseyocrinidae, for the genus (Donovan 1984). Recent work in the Carmarthen area has yielded a single specimen of Ramseyocrinus from the Bolahaul Member of the Ogof Hen Formation in the Roman Road section, Carmarthen, first described by Murchison (1839) and recently revised by Fortey and Owens (1978). The horizon is the same as that from which the holotype was obtained on Ramsey Island some 70 km to the west and the associated fauna is similar. The Carmarthen Ramseyocrinus is, however, the best preserved specimen of the species yet available and shows clearly that: (i) above the stem is a circlet of four plates (basals of Bates 1968, radials of Donovan 1984); (ii) above these four plates is a circlet of five equal plates each of which supports an arm (see text-fig. 1); (iii) there is no anal sac arising from the dorsal cup. This interpretation is clearly at variance with those previously produced and in an attempt to clarify the morphology of the species, all the existing material in the National Museum of Wales, the British Museum Natural History, and Manchester Museum was re-examined. The conclusions were that: (i) the Carmarthen specimen was certainly conspecific with R. cambriensis ; (ii) there was no morphological discrepancy between the specimens which could not be explained in the light of the new specimen; (iii) that this new specimen showed that R. cambriensis shared many characters with R. vizcainoi Ubaghs from the Arenig of the Montagne Noire (Languedoc, France). The first apparent disparity which had to be resolved was the anal tube, reported first by Bates in specimen NMW 29.308G.296 (Bates 1968, pi. 76, fig. 2). In this figure the ‘anal tube’ projects out towards the reader. Comparisons with the Carmarthen specimen would require this ‘anal tube’ to be an arm and the original specimen from which Bates' latex cast was made was restudied. Upon (Palaeontology, Vol. 31, Part 1, 1988, pp. 229-235.( © The Palaeontological Association 230 PALAEONTOLOGY, VOLUME 31 text-fig. 1. Camera lucida drawings of the dorsal cup and lower brachial plates of latex moulds of Ramseyocrinus cambriensis. Carpenter ray notation based on anus being as in text-fig. 3. a, NMW 86.93G.la showing, from left to right B, A, and E radials. The arm showing food grooves is in the C ray. b , NMW 86.93G. lb showing, from left to right, D, C, and B radials. Note that radial D, supported by two basals, is shorter than the other radial plates. The arms showing food grooves are in the B and A rays. examination it is apparent that the ‘anal tube’ disappears into the matrix and the length of the ‘tube’ corresponds to the depth to which the latex has penetrated the mould and does not reflect the true length of the appendage. In order to ascertain the nature of this appendage the reverse of the specimen was excavated initially in the hope of exposing either a fifth arm or an anal tube. In the event the specimen proved to be so compressed that when the shale thickness had been reduced to 3 mm or so there was real danger of damage to the specimen. At this stage the specimen was X-rayed by means of a Faxitron X-ray imager and the resulting print showed a Y-shaped structure, clearly a branching arm. There is thus no longer any discrepancy between the morphology of that specimen and the Carmarthen specimen concerning the anal tube. The fourth basal plate of the Carmarthen specimen can be explained by loss of an inter-basal sutural ridge on the limonitic mould of Bates’ specimen through damage. COPE: REINTERPRETATION OF RAMSEYOCRINUS 231 a b TEXT-FIG. 2. Photographs of latex moulds of Ramseyocrinus cambriensis from the Bolahaul Member, Ogof Hen Formation, Roman Road Section, Pensarn, Carmarthen, South Wales, a, crown NMW 86.93G.la showing arms, from left to right, B; A (branching and with food grooves); adoral view of C; E, crossing over D to form branches with food grooves at extreme right of figure; and D, branching with food grooves. Note circular cluster of plates between C and D arms at level of lBr8 plates, x 5. b, crown of counterpart, NMW 86.93G.16. Food grooves are seen in the B and A rays and the arms in D and C rays are seen to branch. Note plate between arms D and C at level of lBr4 plates, and cluster of plates between these arms at level of lBr8 plates, x 5. Another fine example of Ramseyocrinus is housed in the collections of Manchester Museum (L. 12360). This specimen shows five appendages, one of which was interpreted by Donovan (1984, p. 624) as the anal tube. Again in this specimen the ‘anal tube' disappears into the matrix. In this case it has not been possible to develop the specimen far enough with safety to see if this appendage is branched and therefore an arm, but the rest of the features are identical to the Carmarthen specimen and to Bates’ material, so that a similar organization can be safely assumed. 232 PALAEONTOLOGY, VOLUME 31 text-fig. 3. Camera lucida drawings taken from the specimens of the auxiliary plates visible between two arms, here interpreted as anal plates, a, NMW 86.93G.lu. b, NMW 86.93G.16. Orientation of the cup Having shown that the ‘anal tube’ of Ramseyocrinus is a fifth arm, there is now some doubt over the orientation of the cup. However, the Carmarthen specimen does show a mass of very small plates between two arms, apparently originating from a single large plate visible at the level of one of the fourth primibrachs. This series of plates is terminated distally by a roughly circular arrangement of plates at the level of the eighth primibrachs and can be seen in text-fig. 2 between two arms at that level (see also text-fig. 3). This series of plates could well be the anal series, as no anal plates appear in the dorsal cup and no other plates occur between any of the other arms. If one takes the interray containing these plates as the posterior interray (CD) then the plates can be identified as in text-fig. 4a. The number and size of plates bears a remarkable similarity to that obtaining in R. vizcainoi Ubaghs, 1983. Ubaghs (1983, text-figs. I and h) identified the smallest radial plate as supporting an anal sac, but admitted that this appendage could be a fifth arm (1983, p. 49). If one allows this, and redesignates the plates of Ubaghs’ species, R. vizcainoi is seen to be virtually identical in plating of the dorsal cup to R. camhriensis (see text-fig. 4b). In his description of the former, Ubaghs (1983, p. 52) further comments on a mass of small irregular granules between the arms at the posterior of the crown; these could well correspond with the presumed anal plates of the Carmarthen specimen. The small plates figured by COPE: REINTERPRETATION OF RAMSEYOCRINUS 233 R ADI ALS B ASALS FIXED BRACHIALS ANAL PLATES text-fig. 4. Varying interpretations of the dorsal cup of Ramseyocrinus. a, R. cambriensis as interpreted herein; b , R. vizcainoi after Ubaghs 1983, text-fig. 11h; c, R. cambriensis after Donovan 1984, text-fig. lc; d , R. cambriensis after Bates’ 1968 description and Donovan 1984, text-fig. Id. Bates (1968, p. 76, fig. 2) remarked on by Ubaghs (1983, p. 52) are very much more widespread on the specimens (part and counterpart) than the figure implies and cannot be identified solely with the anal plates. Plate homology Ramsbottom (1961), Bates (1968), and Ubaghs (1983) termed the lowest visible plates of the dorsal cup basals, and the circlet above these the radials. In contrast, Donovan (1983) considered that, as the radials were directly supported by the basals in the earlier interpretations, the basals were radial in position and so could not be considered as other than themselves radial plates, as they were the lowest plates of the cup directly in line with the arms. In his interpretation (1984, pp. 624-625) three radial plates supported four fixed brachial plates, the basal plates being absent or hidden beneath the stem attachment (text-fig. 4c). A further interpretation was given by Moore et al. (1978, p. T554) who suggested that basal plates supported split radial plates (infer- and super-radials) in each ray. This reading interprets the lowest brachials as a fixed ‘super-radial’. [Those unfamiliar with British stratigraphy should note that the species does not occur in the Tremadoc as recorded by Moore et al. (1978, p. T554) since the rocks from which it was described are in fact of Arenig age (Pringle 1930).] The stem attachment is shown very clearly in the Carmarthen specimen and it appears unlikely that another circlet of plates lies beneath the lowest visible plates. In addition, as the anal tube in Donovan’s interpretation is now shown to be an arm, the lowest plate in line with the arm in the D ray (labelled C by Donovan) must be a radial plate; the other four plates in this circlet must also therefore be radial plates, and the plates below are basals. In this interpretation the cup has four basal and five 234 PALAEONTOLOGY, VOLUME 31 radial plates. No brachial plates are fixed; the Carmarthen specimen shows clear articulation facets on the radial plates (text-fig. 1). The suprageneric position of Ramseyocrinus With its tetrameric stem, four basal plates, five radials, and five arms, Ramseyocrinus is unique. There is a plane of bilateral symmetry through the cup plating passing from the AB interray through the D ray. Of the various types of symmetry of inadunate crinoids figured by Ubaghs (1978, p. T62), Ramseyocrinus comes closest to the heterocrinid type, but does not have split radials. The symmetry pattern of Ramseyocrinus is considered to have arisen from the amalgamation of a tetrameric stem with the pentamerous symmetry of the arms, and is not considered a useful taxonomic pointer. It was because Ramseyocrinus was believed to have four arms and an anal tube that the genus was included in the family Eustenocrinidae by Moore et al. (1978). However, Donovan (1984, pp. 624 and 626) listed the ways in which Ramseyocrinus differed from eustenocrinid genera, including the fact that all the latter have compound radials, various different arrangements of the anal plates, and different stems. He concluded that the Ramseyocrinus was so different from the eustenocrinids that he erected a new family, the Ramseyocrinidae to accommodate it. Since the morphology of Ramseyocrinus is now known in more detail, a revised diagnosis of the family and only known genus is required. Class crinoidea J. S. Miller, 1821 Subclass inadunata Wachsmuth and Springer, 1881 Order disparida Moore and Laudon, 1943 Family ramseyocrinidae Donovan, 1984 Emended diagnosis. Monocyclic inadunate crinoids with cylindrical cup as wide as proximal column. Cup symmetrical in AB-D plane. Four basal plates supporting five radial plates; anal plates not present in dorsal cup. Five arms. Proximal stem quadripartite; distal stem holomeric, tetragonal. Genus ramseyocrinus Bates, 1968 Type species (by original designation). Dendrocrinus cambriensis Hicks 1873. Diagnosis. As for family Ramseyocrinidae. Ramseyocrinus cambriensis (Hicks 1873) Text-fig. 2 1873 Dendrocrinus cambriensis Hicks, p. 51, pi. 4, figs. 17-20. 1960 Iocrinusl cambriensis (Hicks) Ramsbottom, pp. 5-6, pi. 3, figs. 9- 1 1. 1968 Ramseyocrinus cambriensis (Hicks) Bates, pp. 406-409, pi. 76, figs. 1-5. 1984 Ramseyocrinus cambriensis (Hicks) Donovan, pp. 627-629, text-fig. 3. Material , horizon , and localities. The addition to the specimens listed by Donovan (1984, p. 627) is NMW 86.93G.la and lb (part and counterpart) from the Bolahaul Member of the Ogof Hen Formation, Moridunian Stage of the Arenig (Fortey and Owens 1987) at the Roman Road, Pensarn, Carmarthen (GR.SN 41361915). Description. The description by Donovan (1984, p. 627) is very full and the new specimen adds nothing to his description of stem and arm structure. The cup consists of four basal plates and five radial plates. The sutures between the four basal plates are each located over one of the lobes of the tetrameric stem and the lower plate sutures follow closely the upper margin of the stem. It appears unlikely that there are further plates of the dorsal cup hidden at this point. The basal plates are of two sizes; small basals are present in the A and B rays and are joined respectively to two large basals which occupy the E and half of D ray, and C and half of D ray. The radial plates in the A, B, C, and E rays are equal in size, that of the D ray is shorter and, unlike the other radial plates, is supported by two basals. The arms are not seen in their entirety in the Carmarthen specimen. All the brachial plates are free. The number of primibrachs is estimated at nine or ten; in the A ray there are at least eight visible and in the B ray six are visible. Arms in all rays have some preserved secundibrachs; there are at least fifteen of COPE: REINTERPRETATION OF R.4MSEYOCRINUS 235 these in ray D. No tertibrachs or further brachial subdivisions are seen in the Carmarthen specimen, although Donovan ( 1 984, p. 627) recorded pentibrachs. Food grooves are well defined in the arms (see text-fig. 2). Between the C and D rays is a series of plates interpreted as the anal series. The lowest plate of this group visible is at the level of the 1 Br4 plate between the C and D arms (see text-figs. 2 and 3); this plate is succeeded by a mass of small granular material, clearly visible on the shale of the specimen but less apparent on latex casts. Adorally this granular area is terminated by a roughly circular arrangement of some seven or eight larger platelets (text-figs. 2 and 3). Discussion. R. cambriensis is similar in many respects to R. vizcainoi Ubaghs from the Arenig of the Montagne Noire, which has a similar arrangement of four basal and five radial plates, one of which is shorter than the others (this assumes that the fifth appendage is an arm, not an anal tube). There are slight differences in the proportion of the plates, those of the dorsal cup are, in the French species, somewhat taller. The short radial plate here interpreted as in ray D is only marginally shorter in the French specimen. The plating of the arms differs too with only six to seven primibrachs and seven or eight secundibrachs in the French species. Clearly the two species are closely related. Acknowledgements. I would like to thank Drs S. K. Donovan, C. R. C. Paul, and A. B. Smith for helpful discussion and acknowledge assistance of Drs M. G. Bassett, R. M. C. Eagar, and R. P. S. Jefferies for the loan of specimens in their care. REFERENCES bates, d. e. b. 1968. On Dendrocrinus cambriensis Flicks, the earliest known crinoid. Palaeontology , 11, 406-409, pi. 76. donovan, s. K. 1984. Ramseyocrinus and Ristnacrinus from the Ordovician of Britain. Ibid. 27, 623-634. fortey, r. a. and Owens, r. m. 1978. Early Ordovician (Arenig) stratigraphy and faunas of the Carmarthen district, south-west Wales. Bull. Br. Mus. nat. Hist. ( Geo! .). 30, 225-294, pis. 1-11. 1987. The Arenig Series in South Wales. Ibid. 41, 69-307. hicks, m. 1873. On the Tremadoc rocks in the neighbourhood of St. David’s, South Wales, and their fossil contents. Q. Jl geol. Soc. Lond. 29, 29-52, pis. 3-5. miller, j. s. 1821. A natural history of the Crinoidea or lily-shaped animals , with observations on the genera Asteria , Euryale, Comatula and Marsupites, viii+ 150 pp., 50 pis. Bryan and Co., Bristol. moore, r. c. and laudon, l. r. 1943. Evolution and classification of Paleozoic crinoids. Spec. Pap. geol. Soc. Am. 46, 1 53 pp., 14 pis. — lane, n. g., strimple, h. l., sprinkle, j. and fay, r. o., 1978. Inadunata. In moore, r. c. and teichert, c. (eds). Treatise on invertebrate paleontology. Part T. Echinodermata 2 (2), T520-T759. Geological Society of America and University of Kansas Press. Murchison, r. l. 1839. The Silurian System, xxxii + 768 pp., 37 pis. London. pringle, j. 1930. The geology of Ramsey Island (Pembrokeshire). Proc. Geol. Tss. 41, 1-31, pis. 1-3. ramsbottom, w. h. c. 1961. A monograph of the British Ordovician Crinoidea. Palaeontogr. Soc. [Monopr.], 1-37, pis. 1-8. ubaghs, G. 1978. Skeletal morphology of fossil crinoids. In moore, r. c. and teichert, c. (eds.). Treatise on invertebrate paleontology. Part T. Echinodermata 2 ( 1 ), T58-T216. Geological Society of America and University of Kansas Press. — 1983. Notes sur les Echinoderms de l’Ordovicien inferieur de la Montagne Noire (France). In courtessole, r., marek, l., pillet, j., ubaghs, g. and vizcaino, d. Calymenina, Echinodermata et Hyolitha de l’Ordovicien inferieur de la Montagne Noire (France Meridionale). Mem. Soc. Etud. Sci. Aude, 33-55, pis. 8-10. wachsmuth, c. and springer, f. 1881. Revision of the Palaeocrinoidea, part II. Family Sphaerocrinidae, with the subfamilies Platycrinidae, Rhodocrinidae and Actinocrinidae. Proc. Acad. nat. Sci. Philad. 175-411, pis. 17-19. Typescript received 26 February 1987 Revised typescript 28 May 1987 JOHN C. W. COPE Department of Geology University College Swansea SA2 8PP J 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. Dianne Edwards, Department of Plant Science, University College, P.O. Box 78, Cardiff CF1 1XL, who will supply detailed instructions for authors on request (these are published in Palaeontology 1985, 28, pp. 793-800). 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 1988 is £37-50 (U.S. $66) for Institutional Members, and £ 1 8-50 (U.S. 132) 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 1-19 are still in print and are available (post free) together with those listed below: 20. (for 1977): Fossil Priapulid Worms, by s. conway morris. 155 pp., 99 text-figs., 30 plates. Price £16 (U.S. $24). 21. (for 1978): Devonian Ammonoids from the Appalachians and their bearing on International Zonation and Correlation, by m. r. house. 70 pp., 12 text-figs., 10 plates. Price £12 (U.S. $18). 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. $38). 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. $45). 24. (for 1980): Dinoflagellate Cysts and Acritarchs from the Eocene of Southern England, by j. p. bujak, c. downie, g. l. eaton and g. l. williams. 100 pp., 24 text-figs., 22 plates. Price £15 (U.S. $23). 25. (for 1980): Stereom Microstructure of the Echinoid Test, by a. b. smith. 81 pp., 20 text-figs., 23 plates. Price £15 (U.S. $23). 26. (for 1981): The Fine Structure of Graptolite Periderm, by p. r. crowther. 1 19 pp., 37 text-figs., 20 plates. Price £25 (U.S. $38). 27. (for 1981): Late Devonian Acritarchs from the Carnarvon Basin, Western Australia, by g. playford and r. s. dring. IS pp., 10 text-figs., 19 plates. Price £15 (U.S. $23). 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. $38). 29. (for 1982): Fossil Cichlid Fish of Africa, by j. a. h. van couvering. 103 pp., 35 text-figs., 10 plates. Price £30 (U.S. $45). 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. $60). 31 . (for 1984): Systematic palaeontology and stratigraphic distribution of ammonite faunas of the French Coniacian, by w. j. Kennedy. 160 pp., 42 text-figs., 33 plates. Price £25 (U.S. $38). 32. (for 1984): Autecology of Silurian organisms. Edited by m. g. bassett and j. d. lawson. 295 pp., 75 text-figs., 13 plates. Price £40 (U.S. $60). 33. (for 1 985): Evolutionary Case Histories from the Fossil Record. Edited by j. c. w. cope and p. w. skelton. 202 pp., 80 text- figs., 4 plates. Price £30 (U.S. $45). 34. (for 1985): Review of the upper Silurian and lower Devonian articulate brachiopods of Podolia, by o. i. Nikiforova, t. l. modzalevskaya and m. g. bassett. 66 pp., 6 text-figs., 16 plates. Price £10 (U.S. $15). 35. (for 1986): Studies in palaeobotany and palynology in honour of N. F. Hughes. Edited by d. j. batten and d. e. g. BRIGGS. 178 pp., 29 plates. Price £30 (U.S. $50). 36. (for 1986): Campanian and Maastrichtian ammonites from northern Aquitaine, France, by w. j. Kennedy. 145 pp., 43 text-figs., 23 plates. Price £20 (U.S. $35). 37. (for 1987): Biology and revised systematics of some late Mesozoic stromatoporoids, by rachel wood. 89 pp., 31 text- figs., 1 plates. Price £20 (U.S. $35). 38. (for 1987): Taxonomy, evolution, and biostratigraphy of late Triassic-early Jurassic calcareous nannofossils, by p. r. bown. 118 pp., 19 text-figs., 15 plates. Price £30 (U.S. $50). Field Guides to Fossils 1. (1983): Fossil Plants of the London Clay, by m. e. colijnson. 121 pp., 242 text-figs. Price £7-95 (U.S. $12). Other Publications 1982. Atlas of the Burgess Shale. Edited by s. conway morris. 31 pp., 24 plates. Price £20 (U.S. $30). 1985. Atlas of Invertebrate Macrofossils. Edited by j. w. Murray. Published by Longman in collaboration with the Palaeontological Association, xiii + 241 pp. Price £13-95. Available in the USA from Halsted Press at U.S. $24.95. © The Palaeontological Association, 1988 Palaeontology VOLUME 31 • PART 1 CONTENTS Preservation of fish in the Cretaceous Santana formation of Brazil D. M. MARTILL 1 Middle Cretaceous wood from the Nanushuk Group, central North Slope, Alaska J. T. PARRISH and R. A. SPICER 19 Analysis of heteromorph ammonoids by differential geometry T. OKAMOTO ' 35 A review of the late Ordovician Foliomena brachiopod fauna with new data from China, Wales, and Poland l. r. M. cocks and rong jia-yu 53 A herbaceous lycophyte from the Lower Carboniferous Dry- brook Sandstone of the Forest of Dean, Gloucestershire N. P. rowe 69 The braincase of the anthracosaur Archeria crassidisca with comments on the interrelationships of primitive tetrapods j. a. clack and R. holmes 85 Early Ordovician acritarchs from southern Jilin Province, north-east China F. MARTIN and YIN LEI MING 109 Palaeocene and Eocene Mixodontia (Mammalia, Glires) of Mongolia and China D. DASHZEVEG and D. E. RUSSELL 129 Classification of the trilobite suborder Asaphina R. A. FORTEY and B. D. E. CHATTERTON 165 Open nomenclature P. BENGTSON 223 A reinterpretation of the Arenig crinoid Ramseyocrinus j. c. w. cope 229 Printed in Great Britain at the University Press, Oxford issn 0031-0239 THE PALAEONTOLOGICAL ASSOCIATION The Association was founded in 1957 to promote research in palaeontology and its allied sciences. COUNCIL 1988-1989 President : Dr J. D. Hudson, Department of Geology, University of Leicester, Leicester LEI 7RH Vice-Presidents: Dr L. B. Halstead, Department of Geology, University of Reading, Reading RG6 2AB Dr P. W. Skelton, Department of Earth Sciences, The Open University, Milton Keynes MK7 6AA Treasurer : Dr M. E. Collinson, Department of Biology, King’s College, London W8 7AH Membership Treasurer: Dr H. A. Armstrong, Department of Geology, University of Newcastle, Newcastle upon Tyne NE1 7RU Institutional Membership Treasurer: Dr A. W. Owen, Department of Geology, University of Dundee, Dundee DD1 4HN Secretary: Dr P. Wallace, The Croft Barn, Church Street, East Hendred, Oxon 0X12 8LA Circular Reporter: Dr D. Palmer, Department of Geology, Trinity College, Dublin 2 Marketing Manager: Dr V. P. Wright, Department of Geology, University of Bristol, Bristol BS8 1RJ Public Relations Officer: Dr M. J. Benton, Department of Geology, The Queen’s University of Belfast, Belfast BT5 6FB Editors Dr M. J. Benton, Department of Geology, The Queen’s University of Belfast, Belfast BT5 6FB Dr J. E. Dalingwater, Department of Environmental Biology, University of Manchester, Manchester Ml 3 9PL Dr D. Edwards, Department of Plant Sciences, University College, Cardiff CF1 1XL Dr C. R. C. Paul, Department of Earth Sciences, University of Liverpool, Liverpool L69 3BX Dr P. A. Selden, Department of Extra-Mural Studies, University of Manchester, Manchester M13 9PL Dr P. D. Taylor, Department of Palaeontology, British Museum (Natural History), London SW7 5BD Other Members Dr J. A. Crame, Cambridge Dr C. Hill, London Dr G. B. Curry, Glasgow Dr E. A. Jarzembowski, Brighton Professor B. M. Funnell, Norwich Dr R. A. Spicer, London 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. J. Cuffey, Department of Geology, Pennsylvania State University, Pennsylvania 16802 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 1988 are: Institutional membership Ordinary membership . Student membership Retired membership £50-00 (U.S. $79) £21-00 (U.S. $38) £11-50 (U.S. $20) £10-50 (U.S. $19) There is no admission fee. Correspondence concerned with Institutional Membership should be addressed to Dr A. W. Owen, Department of Geology, The University, Dundee DD1 4HN. 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 H. A. Armstrong, Department of Geology, University of Newcastle, Newcastle upon Tyne NE1 7RU. Subscriptions cover one calendar year and are due each January; they should be sent to the Membership Treasurer. All members who join for 1988 will receive Palaeontology, Volume 31, Parts 1-4. Back numbers still in print may be ordered from Basil Blackwell, Journals Department, 108 Cowley Road, Oxford OX4 1JF, England. Cover: The brachiopod Meristina obtusa (J. de C. Sowerby, 1823), a life position assemblage from the Much Wenlock Limestone Formation, Abberley Hills, Hereford (Specimen no. BB52671, x 1). Photograph by Harry Taylor of the British Museum (Natural History) Photographic Studio. DISARTICULATED BIVALVE SHELLS AS SUBSTRATES FOR ENCRUSTATION BY THE BRYOZOAN CRIBRILINA PUNCTURATA IN THE PLIO - PLEISTOCENE RED CRAG OF EASTERN ENGLAND by J. D. D. BISHOP Abstract. Cribrilina puncturata (Wood, 1844) forms small, patch-like colonies encrusting the concave (almost invariably lower) surface of disarticulated bivalve mollusc shells in the Plio-Pleistocene Red Crag of eastern England. The species is restricted to the deeper central part of each shell, away from the margin. Peripheral abrasion of the shell’s surface, which might have removed marginal settlement, is rejected as the major agent in producing the observed distribution. A number of larval behaviour mechanisms that might have been responsible are therefore considered. All are discounted except one: that the larva crept up the slope of the shell's inner surface towards the highest point (geonegative movement) before fixation. This hypothesis seems to explain many details of the settlement pattern. C. puncturata apparently exploited the shell’s major concavity as a refuge from physical disturbance in a high-energy environment, and its larval settlement behaviour appears to have been specialized for concavo-convex (bivalve) substrates. A second category of refuge, minor local concavities anywhere on the shell’s surface, was occupied during the vulnerable early astogenetic stages of two other bryozoan species, both with a runner-like colony morphology. Disarticulated bivalve shells on an unconsolidated sea-bed offer a restricted area for potential encrustation, and are often isolated from other solid substrates (including other shells). Shells of the same species and size may be regarded as natural replicate settlement panels, and topographically equivalent areas can be defined on conspecific shells of different size. The bivalve assemblage as a whole may include a wide variety of sizes and shapes. Shells therefore present considerable scope for study as small, discrete, and sometimes short-lived substrates that have relevance to the ecological concepts of spatial refuges (Jackson 1977; Buss 1979) and habitat islands or isolated patches (Schoener 1974; Osman 1977; Keough 1984). This was recognized by Miller and Alvis (1986). In addition to their small size and potential isolation, most shells have another notable feature, bidirectional curvature, which has a strong influence on their hydrodynamic behaviour (Allen 1984). The significance of this concavo-convex shape for a potential encruster, particularly in relation to physical disturbance, is the underlying theme of the present paper. The lower, sand-wave facies of the Plio-Pleistocene Red Crag is exposed on the coast and at numerous inland localities in a region of north-east Essex and south-east Suffolk approximately bounded by Walton-on-the-Naze, Ipswich, Woodbridge, and Chillesford. It consists of poorly consolidated, coarse (but often poorly sorted) shelly sands, showing large-scale tabular cross-bedding. The sand-waves had amplitudes of up to 5 m (commonly 2-3-5 m), and were the product of a flow regime that generated net sediment transport roughly towards the south-west in perhaps 15-25 m of water (Dixon 1979). Boatman (1973) and Bridges (1982) considered the local current regime to have been tidally dominated, while Dixon (1979) attributed only secondary sedimentary structures in this facies to tidal action. The sand-wave facies often contains abundant disarticulated bivalve mollusc shells, which may be fairly evenly distributed within the deposit or may form more or less distinct bands (in which IPalaeontology, Vol. 31, Part 2, 1988, pp. 237-253.| © The Palaeontological Association 238 PALAEONTOLOGY. VOLUME 31 text-fig. 1. a, Cribrilina puncturata (Wood, 1844), BMNH D56116, ancestrula (with proximal parts of periancestrular zooids); Red Crag, Brightwell, approx, x 250. b, BMNH D56997, successive episodes of colonization of the same region of a shell by bryozoans; Red Crag, Newbourn (different exposure from that referred to as ‘Newbourn’ in text), x 58. The oval pits were etched in the shell surface by a colony of Electro sp., subsequently detached; the etchings were then partially overgrown by a tubuliporine cyclostome; a cheilostome, Walantiostoma sp. (ancestrula indicated by lower arrow), settled on the cyclostome, probably after its death since the ancestrula partially covered an already abraded peristome; finally, a second cheilostome, probably also Walantiostoma sp. (ancestrula only, indicated by upper arrow), apparently settled within a damaged zooid of the first (although the established first colony may possibly have overgrown the ancestrula). c, E. crustulenta (Pallas, 1766), British Geological Survey ZK 2658; Red Crag, Butley Priory, x 82. Calcified opercula, attached to the frontal membrane in life, are preserved in situ (or slightly displaced) at the distal (top) ends of the zooids. d, C. puncturata , BMNH D55856, basal remains of ancestrula after abrasion; Red Crag, Shottisham, approx, x 210. imbrication sometimes occurs). The shells are almost all in convex-up orientation. They are often encrusted by bryozoans on their concave (inner, almost invariably lower) surface; similar encrustation of the convex surface is much rarer. The Bryozoa are quite diverse, with at least thirty-five species recognized by the present author from the Red Crag as a whole. Preservation is often very good, although colonies may be extremely delicate and require careful cleaning. At most exposures, the BISHOP: BRYOZOAN SETTLEMENT PATTERN 239 encrusters are dominated by inconspicuous species forming small, scattered colonies. Perhaps for this reason, the bryozoan fauna is poorly documented. Cribrilina puncturata (Wood, 1844) is the most widespread and, at many localities, the most abundant cheilostome bryozoan encrusting Red Crag bivalves. It is the only cheilostome to encrust shells in any numbers in the relatively impoverished assemblage of the Red Crag of Butley. The colonies are generally small, rarely having more than fifty autozooids, but produced ovicells (skeletal chambers in which embryos were brooded) very early in colony development. As is typical for Bryozoa, the colony was founded by an ancestrula (text-fig. 1a), a single zooid arising in all known cases from the metamorphosis of a mobile larva. The ancestrula of C. puncturata is morphologically distinct from subsequent zooids, which were produced by a process of budding to yield a patch-like colony of contiguous zooids. C. puncturata is not known to occur outside the Red Crag. In this paper, the pattern of its settlement on the concave surface of bivalve shells is described, and possible explanations for this pattern discussed. The behavioural (or other) mechanism by which a settling larva came to be on the underside of a shell is not considered. In the material studied, only about 25% of C. puncturata colonies are in contact with another bryozoan. The Bryozoa as a whole generally cover less than 5% of the area of the concave shell surface. It is furthermore apparent that the preserved colonies were not all alive at the same time— a particular shell may show several successive episodes of encrustation (text-fig. 1b). Other encrusting organisms are rare or absent (although serpulid worms and barnacles may be reasonably abundant at Red Crag sites other than those studied here). Whilst the presence of soft-bodied encrusters cannot be discounted, it seems unlikely that biological interactions in the form of competition for space were important on the shells studied. The settlement patterns observed are accordingly discussed primarily in terms of the rigorous physical conditions of the sand-wave environment. MATERIAL AND METHODS The encrusted fossil shells studied came from three Suffolk pits, at NGR TM 250430, TM 275434, and TM 3 1 8457, which are referred to below simply as Bright well, Newbourn, and Shottisham, respectively. The material from Brightwell and Newbourn is referable to the lower, sand-wave facies of the Red Crag described above. The shells from Shottisham are from trough-bedded sands somewhat intermediate in character between the lower facies and the upper, nearshore/onshore facies of Dixon ( 1 979). Shells from the channel deposits above the trough-bedded sands at Shottisham were not included in the study. Four taxa of shells are considered: Glycymeris glycymeris (Linnaeus, 1758), Macoma obliqua (J. Sowerby, 1817), M. praetenuis (Leathes in Woodward, 1833), and Spisula sp(p). A range of shell shapes was present in the Spisula material available, representing a number of nominal species and subspecies that could not be confidently distinguished. To restrict the degree of variation (in order to ensure that comparison between shells was reasonably valid), the investigation was limited to relatively equilateral Spisula shells with oval (rather than markedly triangular) outlines; the shells were selected on this basis without a priori knowledge of the settlement of C. puncturata on them. G. glycymeris and Spisula sp(p). are referred to below by genus. Glycymeris from Brightwell were collected by P. G. Cambridge and S. A. Clark and presented to the British Museum (Natural History) in 1964 and 1965. The material studied here was selected from a larger series to give a set of complete shells with undamaged inner surfaces, covering the desired size ranges. There was no a priori knowledge of the positions of C. puncturata settlement on the shells at the time of selection. Spisula , Glycymeris, M. obliqua, and M. praetenuis from Shottisham, and Spisula from Newbourn were collected during 1 983 and 1 984 by the author and colleagues. All suitable specimens of the relevant mollusc species from the Shottisham and Newbourn series were included in the analysis. Only 2-3% of shells at these two localities were preserved in concave-up orientation. The encrusted shells on which settlement was studied are detailed below, with their registration numbers in the Department of Palaeontology, British Museum (Natural History): Spisula: Newbourn, n = 49 (D55670, 55671, 55679, 55682, 55684 55687, 55693, 55694, 56912, 56913, 56915, 56921, 56923, 56934, 56949, 56950, 56978, 56981, 57437-57465), mean height = 16-7 mm (standard deviation, a = 1-3 mm); Shottisham, n = 27 (D55920, 55934, 55935, 55941, 55944, 55945, 55947-55949, 55953-55958, 56066, 56094, 56095, 56097, 56100, 56102, 56105-56108, 56110, 57730), mean ht. = 14-7 mm ( M. praetenuis. In Glycymeris and Spisula , the left and right valves are almost exact mirror images of one another, apart from details of the hinge line. However, in the two species of Macoma considered here, the curvature of the shells differ. The posterior portions of both valves, viewed dorsally, are deflected to the right (although the overall concavo-convexity and the outlines of the left and right valves in two dimensions are similar). When laid convex-up on a flat surface, the margin of the left valve is therefore raised at the umbo and along much of the ventral margin; that of the right valve is raised at the posterior corner and along the anterodorsal margin. The ancestrula of each colony represents the position of larval fixation and metamorphosis (‘settlement’ below). In colonies where the origin was not preserved, settlement was mapped immediately proximal to the astogenetically youngest (most proximal) part of the remaining portion. For each species of shell, a standard outline was taken from one left valve and one right valve of typical shape near the middle of the size range of the BISHOP: BRYOZOAN SETTLEMENT PATTERN 241 shells from which settlement was mapped. The position of settlement of C. punctwata , and the isoclinic lines, were plotted on to the standard outline of the appropriate species of shell of the correct handedness using a zoom binocular microscope fitted with drawing-tube attachment and with its optical axis perpendicular to the glass lid of the observation box. The standard shell outline and the image of an actual shell were aligned as follows: the umbos were superimposed, and the outlines placed so that any discrepancy between the respective dorsal margins was evenly shared between the anterior and posterior sides; the magnification was adjusted so that the ventral margins were congruent or, in the event of slight differences in ventral shape, so that the overlap of one outline by the other in one place was compensated elsewhere on the ventral margin. In this way, details of several shells could be mapped on to a common outline. Repeated mapping of the same shell indicated an acceptable level of precision. Shell depth contours were transferred from xerographic reductions of the graph paper plots (see above) on to the standard shell outlines by the same system. Kuiper’s two-sample test, as described by Batschelet (1981), was used to evaluate statistically the differences between the settlement patterns on contrasting categories of shell of the same species. (The test is a development of the Kolmogorov-Smirnov test for application to circular distributions, and gives results independent of the chosen zero direction.) The plane delimited by the standard shell outline was divided into seventy-two 5° sectors radiating from its centre of area. Settlement points were counted within each sector for both categories of shell, then Kuiper’s test was applied to the counts. Two patterns were considered to be significantly different if the resulting value of the test statistic, Vn-m (where n and m were the respective numbers of points of the two categories), corresponded to a probability, p, < 0 05. THE PATTERN OF SETTLEMENT When the positions of C. puncturata ancestrulae were mapped from several shells on to a common outline, definite settlement patches were apparent for each of the four taxa of shell studied (text-fig. 2). Settlement was almost always away from the margin, within the deeper part of the shell. In Spisula , settlement was tightly packed within the 06 depth contour, and indeed mostly fell within the 08 depth contour; patterns of settlement on the left and right valves were not significantly different (Newbourn: V33, 52 = 0198, p > 0 50. Shottisham: V22 32 = 0 362, 0-50 > p > 0-20). In Glycymeris, settlement was mostly within the 0-6 depth contour, and distributions of ancestrulae were again similar on the left and right valves (Brightwell: V65,71 = 0T27, p > 0-50. Shottisham: V18 56 = 0-161, p > 0-50). Settlement on M. praetenuis was also almost restricted to the area within the 0-6 depth contour, but in this case the spatial distribution differed between left and right valves (Shottisham: V88 99 = 0-314, 0 005 > p > 0 002), with settlement on the right valve being generally slightly nearer to the anterior margin. M. obliqua showed a similar pattern, with dense settlement within the 0 6 depth contour but with the exact pattern differing between left and right valves (Shottisham: V57 112 = 0 306, 0 02 > p > 0 01); the main settlement patch of the right valve was nearer to the anterior margin, and small secondary patches associated with the adductor muscle scars were also apparent on this valve. POSSIBLE EXPLANATIONS FOR THE OBSERVED PATTERN Abrasion It might be argued that settlement of C. puncturata is indiscriminate, and that the pattern described above is produced by abrasion of peripheral encrustation before fossilization, leaving only colonies on the deeper, more protected part of the shell. This pattern of peripheral abrasion was observed in the differential removal of Indian ink from Glycymeris shells abraded by Red Crag sediment during mechanical agitation under water in ajar. Furthermore, discrepancies between the settlement patches of C. puncturata on the differently curved left and right valves of Macoma spp. might be expected under this hypothesis. However, the following observations suggest that the hypothesis should be rejected: a. The umbonal region of the inner surface of the shell is the most strongly concave (text-fig. 4c) and is thus relatively well protected from abrasion. It nevertheless has very sparse recorded settlement of C. puncturata (for instance, text-fig. 2). b. On Glycymeris from Brightwell, colony portions of Electro crustulenta (Pallas, 1766) and Phylactellipora sp. are occasionally seen near the shell margin in positions unprotected from abrasion 242 PALAEONTOLOGY, VOLUME 31 by any local concavity. These species are considerably less common on the shells than C. puncturata (and no more robust than it), and the persistence of recognizable remains indicates that at least traces of C. puncturata would be expected in marginal regions of the shell if colonies had ever been present there. c. The preservation of C. puncturata colonies is often very good, with no sign of abrasion. Other encrusting species on the shells have erect or semi-erect structures such as long peristomes (in tubuliporine cyclostomes) and oral spines (in Hemicyclopora sp.) which often remain intact. This suggests that the most recent encrustation of many Red Crag shells was not exposed to any wear, i.e. that many shells were not transported subsequent to their last colonization. Perhaps the best evidence of this is the preservation in situ of the calcified opercula of E. crustulenta, which are unsupported in the absence of soft tissue (text-fig. lc). These are seen at many localities, including Brightwell and Shottisham. It therefore seems that at least some shells would reveal random settlement by C. puncturata if such a pattern existed before abrasion. This has not been observed. Avoidance of light Once under a shell, larvae might show a behavioural response (such as skototaxis, low photokinesis, or negative phototaxis) taking them away from any source of light. Since light could only filter into the space under a complete convex-up shell around its margins, settlement in the central part of the shell might result. It is possible to test this idea using the fact that some of the shells studied have borings, presumably produced by naticid gastropods. The holes are in a characteristic place for a particular species of shell, and are relatively large (text-fig. 3a, c). The penetration of light through a hole in the shell would be expected to affect the settlement pattern of C. puncturata under the present hypothesis. Spisula shells are bored close to the umbo (text-fig. 3a). Displacement of the settlement patch towards the ventral margin of the shell might be predicted. In fact (text-fig. 3b) the position of settlement is not affected (Newbourn: V26, 59 = 0-246, p > 0-50. Shottisham: V10 44 = 0-350, p > 0-50). M. praetenuis shells are bored some distance ventral to the umbo (text-fig. 3c). Again (text-fig. 3d), no disruption of the pattern of settlement is apparent (Shottisham, left valves: V34 65 = 0-221, p > 0-50; only one bored, encrusted right valve available). A behavioural response to light does not, therefore, seem to be important in determining the final position of settlement. The possibility that the position of settlement is mediated by a behavioural response to current patterns under the shell (or to pressure differentials associated with the current) can similarly be discounted because the presence of borings in the shell might also be expected to modify or disrupt the resulting pattern. Avoidance of sediment Settlement might occur only on regions of the undersurface of the shell free of sediment. Since contact with the sediment would be expected around the margin of a convex-up shell, a pattern of settlement broadly similar to that observed might result. However: a. Under a hypothesis of simple sediment avoidance (i.e. without the requirement for a threshold distance above the sediment) settlement very close to the margin would occur on shells that did not text-fig. 2. Settlement of Cribrilina puncturata mapped on to standard outlines of four taxa of shell from the Red Crag, with shell depth contours; all outlines are of left valve (anterior margin to the right) except B, right valve; settlement on left valve shown as dots, on right valve as crosses; contours of left valve shown as large dashes, of right valve as small dashes, a-c, Spisula from Newbourn (30 left valves, 19 right), with 0-8 depth contours; in c the settlement and contour of the right valve have been transferred, after left-right reversal, on to the outline of the left valve. In d-f, the details of the right valve have similarly been transferred to the left valve. D, Glycymeris from Brightwell (7 left valves, 9 right), with 0-6 depth contours. E, Macoma obliqua from Shottisham (14 left valves, 21 right), with 0-6 depth contours. F, M. praetenuis from Shottisham (22 left valves, 24 right), with 0-6 depth contours. BISHOP: BRYOZOAN SETTLEMENT PATTERN 243 244 PALAEONTOLOGY, VOLUME 31 text-fig. 3. Settlement of Cribrilina puncturata on bored and unbored shells, a, b, Spisula from Newbourn, data from left and right valves shown together on outline of left valve (11 bored shells, 38 unbored). c, D, Macoma praetenuis from Shottisham, left valves only (5 bored shells, 1 7 unbored). a, c, outlines of borings, drawn to scale for individual shells, b, d, settlement (dots, bored shells; crosses, unbored). sink into the sand at all, possibly because they were in contact with another shell or large sediment particle (such as a shell fragment). Such settlement in C. puncturata is not observed. b. If a threshold distance above the sediment were required for settlement to occur, a greater proportion of the inner surface of large (deeper) shells would be encrusted than of small (shallower) shells. This is not observed. In text-fig. 4a, settlement on large and small Glycymeris shells from Brightwell is compared. (Medium sized shells were ignored.) The settlement patches are not significantly different (V41 95 = 0226, 0-50 > p > 0 20), being almost completely restricted to the area within the 0 6 depth contour in both shell sizes. The position of the contour relative to the shell margin remains much the same as the shell grows (text-fig. 4b), reflecting the almost isometric growth in Glycymeris noted by Thomas (1975). However, the contour is at an absolute vertical distance of about 6-5-7 mm from the margin in large shells but about 4 mm in small shells. The greatest depth of the small shells is in fact only about 6-5-7 mm. It seems extremely unlikely, under the hypothesis of a threshold distance, that a compensatory factor, such as greater penetration of the sediment by larger shells, could produce the observed congruity of settlement patches. It appears instead that the BISHOP: BRYOZOAN SETTLEMENT PATTERN 245 text-fig. 4. a, settlement of Cribrilina puncturata on small (height 27-31 mm) shells (dots) and large (height 43-47 mm) shells (crosses) of Glycymeris from Brightwell; data from left and right valves shown together on outline of left valve, with 06 shell depth contour. B, 06 depth contour for three right valves of Glycymeris , shown on common shell outline. Small dots, 29 mm high shell from Brightwell; medium dots, 39 mm high shell from Walton-on-the-Naze; large dots, 44 mm high shell from Brightwell. (Dots do not represent individual data points in b.) c, D, sections of relatively unworn Glycymeris shells from Walton-on-the-Naze. Approximate position of highest point on inside of shell at various angles of tilt indicated (figures in degrees); *— indicates position of 06 depth contour. settlement pattern relates not to any absolute dimension of the shells hut to their shape , which remains relatively constant as they grow, giving an almost identical distribution of ancestrulae on shells of different size. Most convex-up shells in situ in the Red Crag are in any case full of sand, and recent observations indicate that at least some present-day bryozoans can live on surfaces that are in contact with or buried in sand (see Discussion). Settlement at highest point Larvae might settle at or near the highest part of the inside of the shell. If the shells were all laying on a horizontal surface, clumped settlement at the same place in each shell (the very deepest point) might be expected. However, little of the sea-bed in a sand-wave system is horizontal. Shells resting on a sloping sediment surface may be correspondingly inclined from the horizontal. Thus, Salazar-Jimenez et al. (1982, p. 580) observed shells inclined parallel to bedding in steeply dipping stoss laminae. Even on a horizontal surface, shells may be tilted by imbrication or local sediment scouring. It should also be borne in mind that several shells of each species were scored to produce the observed settlement patterns, and that each individual shell could show several different episodes of encrustation, possibly with movement of the shell between episodes as the sand-wave system migrated. To predict the 246 PALAEONTOLOGY, VOLUME 31 text-fig. 5. A, 30° (outer group) and 15° (inner) isoclinic lines for 5 left valves of Glycymeris from Brightwell (heights 27-46 mm), plotted on a common shell outline, b, summary 30° (outer) and 1 5° (inner) isoclinic lines for 5 left (L) and 5 right (R) valves of Glycymeris from Brightwell, shown on outline of left valve, c, D, settlement of Cribrilina puncturata at Shottisham, with 30° and 15° isoclinic lines averaged for both left and right valves; settlement on left (dots) and right valves (crosses) shown on outline of left valve, c, on Glycymeris (8 left valves, 5 right). D, on Spisula (15 left valves, 12 right); 30° isoclinic line incomplete anteriorly where it passed to the margin of a number of individual shells. pattern of settlement that would be expected in these circumstances under a hypothesis of highest-point settlement, it is first necessary to consider the shape of the bivalve shell. Sections through a Glycymeris valve are shown in text-fig. 4c, d. The section from the umbo to the ventral margin (text-fig. 4c) resembles part of a logarithmic spiral (as noted by Thomas 1975), with the curvature decreasing towards the ventral margin. As the shell is tilted, the highest point inside it will move. Very high angles of tilt are required to displace the highest point into the umbo, whereas tilting BISHOP: BRYOZOAN SETTLEMENT PATTERN 247 text-fig. 6. Settlement of Cribrilina puncturata on four taxa of Red Crag shells (settlement data the same as in text-fig. 1 ), with isoclinic lines; settlement on left (dots) and right valves (crosses) shown on outline of left valve. A, Spisula from Newbourn, and B, Glycymeris from Brightwell; both with 30° (outer) and 15° (inner) isoclinic lines averaged for left and right valves, c, Macoma obliqua from Shottisham, with 30° (outer) and 15° (inner) isoclinic lines shown separately for left and right valves; lines for right valve are further to the right (more anterior); see text for explanation of incompleteness of 30° lines, d, M. praetenuis from Shottisham, with 1 5° isoclinic lines shown separately for left and right valves; line for right valve is further to the right (more anterior). in the opposite direction rapidly moves it towards the ventral margin. The depth contours, used up till now to describe the settlement patterns, are not symmetrical with respect to these angles of tilt. For instance, the 0-6 depth contour in Glycymeris passes nearer to the umbo but further from the ventral margin than the 30° tilt points (text-fig. 4c). The anterior to posterior section (text-fig. 4d) is more symmetrical, but has steps at the edge of the buttressed anterior and posterior adductor muscle scars that block the movement of the highest point toward the margin of the shell as it is progressively tilted. (In the other shell taxa considered, the adductor muscle scars are not raised on low buttresses as in Glycymeris , but form more or less distinct depressions.) An isoclinic line (see Methods section) describes the movement of the highest point on the inside of the convex-up shell as it is rotated at a particular angle of tilt. It follows that, under a hypothesis of 248 PALAEONTOLOGY, VOLUME 31 highest-point settlement, the isoclinic line for a particular angle will delimit the predicted settlement patch on shells that may be tilted in any direction by that angle or less at the time of settlement. Text-fig. 5a shows the 15° and 30° isoclinic lines for five left valves of Glycymeris from Brightwell plotted on to a common outline. The lines are similar for different sizes of shell. The position of the 30° line is relatively uniform between shells in the dorsal half of the shell, running close to the edges of the adductor muscle scars, but varies between shells somewhat towards the ventral margin, where curvature is relatively weak. In text-fig. 5b, single lines summarize those for individual left valves and individual right valves. The lines for left and right valves agree closely. In text-figs. 5c and 6b, the summary lines for left and right valves of Glycymeris have been combined. At both localities studied, the area delimited by the 30° isoclinic line agrees well with the observed settlement patch, but the region within the 15° isoclinic line does not seem to be associated with particularly dense settle- ment. In contrast, the densest settlement on Spisula (in which the isoclinic lines were again con- gruent between left and right valves) is restricted to the area within the 15° isoclinic line at both localities studied (text-figs. 5d and 6a). The proportion of settlement inside the 15° line is significantly lower for Glycymeris from Shottisham than for Spisula from the same locality (text-fig. 5c cf. d) (Fisher’s 2x2 exact test, Z = 2-54, 0 02 > p > 0002 in two-tailed test; settlement on right and left valves combined). The 30° isoclinic lines of M. obliqua were interrupted in the shallowly curved posterior and posteroventral sectors of all five right valves, and in the anteroventral sector of three of the left valves (i.e. the ball rolled to the edge of the shell in these cases). The 30° summary lines shown in text-fig. 6c are therefore incomplete, but where both are plotted the line for the right valve is somewhat anterior to that for the left valve. A similar relative displacement is seen very clearly in the 15° isoclinic lines, which roughly delimit the non-coincident zones of densest settlement on left and right valves. Particularly in large examples, the adductor muscle scars of M. obliqua constitute significant secondary concavities within the shell. In the case of the right valve, there are recognizable patches of settlement associated with these regions, giving a large overall spread of settlement. However, little or no settlement is associated with the scars of the left valve. When M. praetenuis shells were tilted at 30°, the ball moved to the edge of the shell in a high proportion of cases, particularly around the anterior and posterior margins. This obviously reflects the low concavo-convexity of M. praetenuis shells, and no 30° isoclinic lines are shown in text-fig. 6d. The 15° summary isoclinic lines for left and right valves are not concurrent (text-fig. 6d), the line for right valves showing a relative anterior displacement corresponding to the similar displacement of settlement noted previously. It therefore appears that highest-point settlement could produce differences in settlement between right and left valves of Macoma spp. very similar to those observed. DISCUSSION Settlement vs. recruitment Keough and Downes (1982) issued a strong caveat for those seeking to infer the larval settlement behaviour of sessile marine invertebrates from the observed spatial distribution of recruits to the attached juvenile phase. They argued that the initial settlement pattern could be subject to strong modification by differential mortality in the period between settlement and observation, even if that period were reasonably brief. In the present study, it was possible to recognize isolated ancestrulae of C. puncturata early in the second (histogenic) phase of metamorphosis (as defined by Zimmer and Woollacott 1977), at a stage in which only the most basal parts of the vertical zooecial walls were calcified. (An abraded condition superficially similar to this early stage is shown in text-fig. Id.) Nielsen (1981) reported that calcification was underway 12 hours after settlement in two cheilostome species from California. Cook (1985) observed development of the ancestrula in a variety of Ghanaian cheilostomes, and reported that calcification could be observed as little as 2 hours after settlement; the ancestrula was complete after 24 to 48 hours. The possibility of significant mortality during the BISHOP: BRYOZOAN SETTLEMENT PATTERN 249 brief first (morphogenetic) phase of metamorphosis cannot be discounted in the present study. Nevertheless, it is considered probable that the time period between larval fixation and the deposition of recognizable calcification by the developing ancestrula of C. puncturata was sufficiently short for the true settlement pattern to have remained substantially unaltered by differential mortality. Larval behaviour Having encountered an appropriate substrate and stopped swimming, gymnolaemate larvae generally creep or glide over its surface for some time by ciliary action. They eventually undergo fixation and metamorphosis at a chosen point or swim off, presumably in search of another potential settlement site (Cook 1985, and references therein). As stated in the Introduction, the mechanism by which the larvae of C. puncturata reached the underside of shells is not considered here. However, it may be noted in this context that some bryozoan larvae, after a swimming phase, are able to burrow in sand (Cook and Chimonides 1985). Once under a suitable convex-up bivalve shell, it is considered very probable that larvae of C. puncturata crept up the slope of the lower surface before metamorphosis. Settlement at or near the highest point on the undersurface of the shell appears to offer the best explanation of the observed distribution of ancestrulae. The following observations support this hypothesis: 1, settlement very close to the umbo of a shell would not be expected, and it is rarely observed; 2, the extent of the settlement patch is similar in small and large shells, in proportion to the total area of the shell, as would be predicted from the proposed behaviour; 3, no major disruption of the pattern is anticipated in bored shells, and none is seen; and 4, this behaviour could produce the observed difference in settlement patterns between the slightly asymmetrical left and right valves of Macoma spp. Might a gravity response by the creeping larva be involved in the production of the observed pattern in C. puncturata ? Research on larval behaviour of Recent encrusting marine cheilostome Bryozoa has almost exclusively involved swimming movements of littoral and shallow sublittoral species. Whilst phototactic responses in these have been reported frequently, geotaxis has rarely been mentioned. Negative geotaxis may be partly responsible for the observed settlement of E. crustulenta predominantly on the underside of hard substrates in the Baltic (Silcn and Jansson 1972). Ryland (1974, 1977) suggested, on the basis of observations by earlier authors, that initial geonegativity was shown by two species of Bugula; he noted that the role of gravity responses in free-swimming bryozoan larvae remained little understood, and had rarely been distinguished from associated light responses. Nielsen (1981) considered that an initial geonegative response, as well as photopositivity, was present in swimming larvae of 'Hippodiplosia' insculpta (Hincks, 1882). Pires and Woollacott (1983) demonstrated a true gravity response in larvae of B. stolonifera Ryland, 1960 and light-independent upward swimming in response to an unknown cue in B. neritina (Linnaeus, 1758). They pointed out that, in the context of free-swimming invertebrate larvae, the term geotaxis had sometimes been applied in a broad sense to any light-independent vertical movement, and that apparent geotaxis might arise as a response to temperature gradients, hydrostatic pressure, or geomagnetic fields, as well as directly to gravity. Depth gradients of hydrostatic pressure or temperature seem very unlikely to have been involved in the orientation of creeping C. puncturata larvae since the vertical distance between the centre and margin of a shell is very small (several mm at most). The possibility of a response to the vertical component of the local geomagnetic field cannot be excluded, although a true gravity response seems more probable. Larvae creeping to the edge of a highly tilted shell presumably moved off in search of another shell, since settlement at the edge of a shell has not been observed. It is perhaps worth noting that the location of the highest point on the concave surface of a shell by a larva moving according to any of the vertical stimuli suggested here would probably not be very accurate. This is because the component of the stimulus acting in the relevant direction for orientation (i.e. up and down the slope of the shell’s surface) would decrease as the highest point was approached, being proportional to the sine of the angle of inclination of the local surface to the horizontal. Thus settlement was probably only approximately at the highest point. The observed spread of settlement 250 PALAEONTOLOGY, VOLUME 31 within an isoclinic line may have been partly caused by this inaccuracy in addition to any tilting of the shell at the time of colonization. Under the hypothesis of highest-point settlement, two possible explanations may be put forward for the greater concentration of settlement within the 15° isoclinic line in Spisula compared with Glycymeris (text-fig. 5c cf. d). Glycymeris may, on average, have been more tilted than Spisula at the time of settlement. No evidence is available to confirm or disprove this suggestion. Alternatively, the larva of C. puncturata may have been able to locate the highest point more accurately in Spisula than in Glycymeris. Two factors suggest that this may have been the case. First, the Spisula shells studied are considerably smaller than those of Glycymeris , so that the distance to be travelled by the larva to the highest point would have been much shorter. Second, the inner surface of the shells studied is in general considerably smoother in Spisula than in Glycymeris. The upward movement of the larva in response to the true slope of the shell would therefore have been less disrupted by local surface microrelief in Spisula. ( Glycymeris has very durable shells, and most Red Crag examples are considerably worn and pitted, except in the cliff exposure at Walton-on-the-Naze, Essex.) It should be mentioned that a similar difference between Spisula and Glycymeris might be predicted under a hypothesis of peripheral abrasion after random settlement, since abrasion would be expected to affect a greater proportion of the surface in smaller shells. Reasons for discounting abrasion as the major cause of the observed distribution of C. puncturata have been given above. In addition, the broad settlement patch on the relatively small shells of M. praetenuis (text-fig. 6d) is contrary to the pattern expected from peripheral abrasion. It is in agreement with the prediction for highest-point settlement, since the shells of this species are relatively weakly concavo-convex. Disarticulated bivalve shells as substrates for encrustation Potential problems for an organism encrusting a small solid substrate on a particulate sea-bed include: 1, abrasion by current-entrained sediment particles; 2, crushing when the substrate is transported or changes attitude; 3, clogging of feeding or respiratory structures by suspended or settling sediment, particularly fine particles; and 4, deep burial of the substrate, curtailing the supply of resources from the water column. It seems possible to make a number of general statements concerning the behaviour of single concavo-convex mollusc (or brachiopod) shell substrates on a sandy sea-bed that may relate to these dangers. Shells in the convex-up position are much more hydrodynamically stable than when resting concave-up (Hall 1843, p. 52; Sorby 1908, p. 189; confirmed by numerous subsequent workers). Concave-up shells are relatively easily transported across sand by currents, but may readily invert to the convex-up position, stopping transport, if tripped by an obstacle or an irregularity of the sediment surface (Brenchley and Newall 1970). Conversely, convex-up shells may flip to the concave-up position at the crest of a passing sand ripple (Menard and Boucot 1951, p. 148). This occurs more easily with relatively small shells (Clifton and Boggs 1970). Stationary shells in either attitude may also topple into their own current scour marks and thus rest at high angles to the general sediment surface. This is often a prelude to burial (Johnson 1957; Brenchley and Newall 1970). Shells washed into sediment traps such as burrows may exhibit predominantly concave-up or near-vertical orientation (Salazar-Jimenez et al. 1982). The convex surface of a shell in either the convex-up or concave-up position may be abraded by passing grains when there is a current capable of entraining sediment particles. Sediment scour may occur locally around the shell even in currents incapable of mobilizing the sediment elsewhere (Johnson 1957). The fate of the concave surface will depend on the position of the shell. Sediment winnowed from eddies in the lee of a concave-up shell may accumulate on its upper surface ( Brenchley and Newall 1970). In contrast, when the shell is convex-up, the concave surface is unaffected by sediment falling from local eddies (or settling when the current slackens), and is also relatively protected from abrasion by current-entrained particles. Settlement on the concave surface of a convex-up disarticulated shell would therefore seem to minimize the chances of mechanical damage or clogging for an encruster in an environment with significant sediment transport. Certain disadvantages appear to offset this protection: restricted water BISHOP: BRYOZOAN SETTLEMENT PATTERN 251 text-fig. 7. Settlement of Phylactellipora sp. (dots) and Electro crustulenta (crosses) on Glycymeris from Brightwell, data from left and right valves shown together on outline of left valve, with 06 shell depth contour. movement would be expected compared with the upper surface of the substrate, and there may be very limited space (if any) between the lower surface of the shell and the sediment. However, various observations strongly suggest that Recent marine bryozoans can sometimes grow and reproduce buried in the superficial layers of sand (Cook 1985; Hakansson and Winston 1985; Cook and Chimonides 1985). It must be presumed that they are able to feed successfully by extending their lophophores (delicate ciliated tentacle crowns by which Bryozoa suspension-feed) into the interstitial spaces of the sand immediately adjacent to the colony, provided that the sediment is sufficiently coarse. It seems probable that the supply of resources, such as food and oxygen, from the water column would be progressively reduced with increasing depth of burial, so that survival would no longer be possible below a certain level. Geonegative movement during selection of the site of fixation seems particularly appropriate for a colonizer of the underside of markedly concavo-convex disarticulated bivalves in a current-swept sandy environment. It would result in settlement in the relatively protected central region away from the margins of the shell. Settlement on an excessively tilted shell (which might be in a sediment trap or in the process of burial by scouring) would be avoided if the shell’s attitude were such that its highest point was somewhere on its ventral or lateral margins. If, however, the umbonal region of a highly tilted shell were uppermost, settlement on or near this very concave (and hence relatively protected) area of the undersurface would result. Settlement on shells in the unstable concave-up position would be precluded. It therefore seems that geonegative movement immediately prior to fixation could minimize a colony’s subsequent exposure to shell substrate transport or deep burial within the sand-wave system, and reduce mechanical damage to the colony during those episodes of shell overturning and transport that did occur. This perhaps accounts in part for the relative success of C. puncturata in the harsh Red Crag environment. However, it should be noted that the expected resultant colonization is in precisely the area most likely to be affected by settling sediment on a shell that was subsequently overturned. The behaviour mechanism inferred here would therefore only be advantageous in environments where a colonized shell, having then been inverted to the concave-up position, would be unlikely to remain in that position for a significant period of time before reverting to the hydrodynamically stable convex-up configuration. The very clear preponderance of convex-up shells in Red Crag exposures suggests that this was the case in the Red Crag sea. The small, patch-like colony of C. puncturata completed its life history in the area immediately adjacent to the point of larval settlement. Any mechanical damage or clogging was therefore likely to be colony-wide. It appears that the deepest central region of the undersurface of convex-up shells was 252 PALAEONTOLOGY, VOLUME 31 treated by C. puncturata as a temporally and spatially predictable refuge (in the sense of Buss 1979) from physically mediated disturbance in a high-energy environment, exploited by the bryozoan through specialized settlement behaviour. The Recent pterobranch hemichordate Rhabdopleura compact a Hincks, 1880, which also forms minute encrusting colonies, was reported by Stebbing (1970) to be restricted to a microhabitat apparently very similar to that noted here for C. puncturata. Stebbing studied material dredged from a coarse to very coarse sand bottom at depths of c. 2 1 -24 m oft' south Devon, England. R. compact a, and many other encrusting species of the associated fauna, was found only on the concave surface of disarticulated bivalve shells (mostly Glycymeris). (The precise distribution of settlement was not recorded.) Observations by SCUBA diving in the same area revealed that shells were scattered on an undulating sea-bed, and that all disarticulated valves were lying in the convex-up position. In contrast to C. puncturata , colony growth in Phylactellipora sp. and E. crustulenta on Red Crag shells was strongly directional with periodic branching (runner-like morphology in the sense of Buss 1979). A large colony therefore reached various parts of the shell distant from the point of settlement. These two species commonly occurred as colony fragments dissociated from any recognizable colony origin. Survival after the early growth stages may have involved regrowth of undamaged fragments after partial colony mortality caused by episodes of physical disturbance. When most of the colony was covered by an accumulation of fine sediment, unaffected parts that were able to feed may have exported metabolites to neighbouring zooids. Phylactellipora sp. and E. crustulenta had rugophilic settlement behaviour that resulted in the ancestrula nestling in a local minor irregularity anywhere on the shell’s (lower ?) surface (text-fig. 7). Such sites were found amongst the dentition of the cardinal plate, along the edges of adductor muscle scars, at the margin of pre-existing encrustation of the shell, and in any surface pitting caused by abrasion or bioerosion. Through their settlement behaviour, these two species therefore utilized local minor irregularities as predictable refuges for early astogeny, but they treated refuges for later colony growth as spatially unpredictable (in the sense of Buss 1979), each colony spreading out to minimize the probability of total colony mortality. This settlement and growth strategy is appropriate for colonization of a variety of hard substrates in addition to disarticulated shells, and is relevant to the survival of biological competition for space as well as of physically mediated disturbance (Buss 1979; Jackson 1979). Acknowledgements. P. D. Taylor, P. L. Cook, P. J. Chimonides, and B. Okamura read the manuscript and suggested several improvements, as did the anonymous referees. I also wish to thank K. M. Shaw and N. Goldman for help with Kuiper’s test. I am grateful to P. R. Crowther for drawing my attention to Stebbing’s work on Rhabdopleura. REFERENCES allen, j. r. l. 1984. Experiments on the settling, overturning and entrainment of bivalve shells and related models. Sedimentology , 31, 227-250. batschelet, E. 1981. Circular statistics in biology, xvi + 371 pp. Academic Press, London. boatman, a. r. c. 1973. Sedimentary characteristics of the Red Crag. In rose, j. and turner, c. (eds.). Easter Field Meeting 1973. Clacton. Quaternary Res. Tss., Field Guide, pages unnumbered. brenchley, p. j. and newall, G. 1970. Flume experiments on the orientation and transport of models and shell valves. Palaeogeogr. Palaeoclimat. Palaeoecol. 7, 185-220. bridges, p. h. 1982. Ancient offshore tidal deposits. In stride, a. h. (ed.). Offshore tidal sands. Processes and deposits, 172-192. Chapman and Hall, London and New York. buss, L. w. 1979. Habitat selection, directional growth and spatial refuges: why colonial animals have more hiding places. In larwood, g. and rosen, b. r. (eds.). Biology and systematics of colonial organisms, 459-497. Academic Press, London, New York, and San Francisco. Clifton, H. E. and boggs, s. 1970. Concave-up pelecypod (Psephidia) shells in shallow marine sand. Elk River Beds, Southwestern Oregon. J. sedim. Petrol. 40, 888-897. cook, p. L. 1985. Bryozoa from Ghana— a preliminary survey. Annls Mus. r. Afr. cent., Sci. zool. 238, 1-315. and chimonides, p. j. 1985. Larval settlement and early astogeny of Parmularia (Cheilostomata). In nielsen, C. and larwood, g. p. (eds.). Bryozoa: Ordovician to Recent, 71-78. Olsen and Olsen, Fredensborg. BISHOP: BRYOZOAN SETTLEMENT PATTERN 253 dixon, R. G. 1979. Sedimentary facies in the Red Crag (Lower Pleistocene, East Anglia). Proc. Geol. Ass. 90, 117-132. hakansson, E. and winston, j. E. 1985. Interstitial bryozoans: unexpected life forms in a high energy environment. In nielsen, c. and larwood, g. p. (eds.). Bryozoa: Ordovician to Recent, 125-134. Olsen and Olsen, Fredensborg. hall, j. 1843. Part IV. Comprising the survey of the fourth geological district. Nat. Hist. NY, Geol. 4, i-xxii + 1 -683. jackson, J. b. c. 1977. Habitat area, colonization, and development of epibenthic community structure. In keegan, b. F., o ceidigh, p. and boaden, p. J. s. (eds.). Biology of benthic organisms, 349-358. Pergamon Press, Oxford. 1979. Morphological strategies of sessile animals. In larwood, g. and rosen, b. r. (eds.). Biology and systematics of colonial organisms, 499-555. Academic Press, London, New York, and San Francisco. Johnson, r. G. 1957. Experiments on the burial of shells. J. Geol. 65, 527-535. keough, m. j. 1984. Effects of patch size on the abundance of sessile marine invertebrates. Ecology, 65, 423-437. — and downes, b. j. 1982. Recruitment of marine invertebrates: the role of active larval choices and early mortality. Oecologia, 54, 348-352. menard, H. w. and boucot, a. J. 1951. Experiments on the movement of shells by water. Am. J. Sci. 249, 131-151. miller, w. and alvis, l. m. 1986. Temporal change as an aspect of biogenic shell utilization and damage. Pleistocene of North Carolina, U.S.A. Palaeogeogr. Palaeoclimat. Palaeoecol. 56, 197-215. nielsen, c. 1981. On morphology and reproduction of ‘ Hippodiplosia ' insculpta and Fenestrulina malusii (Bryozoa, Cheilostomata). Ophelia, 20, 91-125. osman, r. w. 1977. The establishment and development of a marine epifaunal community. Ecol. Monogr. 47, 37-63. pires, a. and woollacott, r. m. 1983. A direct and active influence of gravity on the behavior of a marine invertebrate larva. Science, NY. 220, 731-733. ryland, j. s. 1974. Behaviour, settlement and metamorphosis of bryozoan larvae: a review. Thalassia jugosl. 10, 239-262. 1977. Taxes and tropisms of bryozoans. In woollacott, r. m. and zimmer, r. l. (eds.). Biology of bryozoans, 41 1 -436. Academic Press, New York, San Francisco, and London. salazar-jimenez, a., frey, R. w. and Howard, j. d. 1982. Concavity orientations of bivalve shells in estuarine and nearshore shelf sediments, Georgia. J. sedim. Petrol. 52, 565-586. schoener, a. 1974. Experimental zoogeography: colonization of marine mini-islands. Am. Nat. 108, 715-738. silen, l. and jansson, b.-o. 1972. Occurrence of Electra crust ulenta (Bryozoa) in relation to light. Oikos, 23, 59-62. sorby, h. c. 1908. On the application of quantitative methods to the study of the structure and history of rocks. Q. Jl geol. Soc. Lond. 64, 171-233. stebbing, a. r. d. 1970. The status and ecology of Rhabdopleura compacta (Hemichordata) from Plymouth. .7. mar. biol. Ass. UK. 50, 209-221. thomas, r. d. k. 1975. Functional morphology, ecology, and evolutionary conservatism in the Glycymerididae (Bivalvia). Palaeontology, 18, 217-254. zimmer, R. L. and woollacott, R. m. 1977. Metamorphosis, ancestrulae and coloniality in bryozoan life cycles. In woollacott, r. m. and zimmer, r. l. (eds.). Biology of bryozoans, 91-142. Academic Press, New York, San Francisco, and London. Typescript received 9 March 1987 Revised typescript received 15 May 1987 j. d. d. bishop Departments of Palaeontology and Zoology British Museum (Natural History) Cromwell Road, London SW7 5BD FISH TRAILS IN THE UPPER CARBONIFEROUS OF SOUTH-WEST ENGLAND by ROGER HIGGS Abstract. The ichnogenus Undichna Anderson, 1976, consisting of wavy horizontal grooves representing the drag marks of fish fins, is reported from the lacustrine Bude Formation. Two of the three species erected by Anderson are present. Two additional types of trail share strong similarities with Anderson’s established species, but exhibit new features which warrant the erection of two new species, namely U. britannica and U . consulca. The new features necessitate modification of the original generic diagnosis. U. consulca includes a broad, shallow furrow, suggesting that the fish swam with its belly brushing the sediment. This behaviour would have been impossible unless the pectoral fins were mounted abnormally high, to avoid fouling the sediment. Of four species of fossil fish previously known from the Bude Formation, one ( Cornuboniscus budensis White, 1939) is remarkable for its high pectoral fins. It is suggested that the high fins were an evolutionary adaptation which enabled the fish to hug the bottom in search of food. This food may have included xiphosurid crabs, whose trackways ( Kouphichnium ) are intimately associated with the fish trails. Of the four species of fossil fish found in the Bude Formation, two cannot be correlated with any of the trails; this suggests that the two species in question were mid- to surface-water feeders. None of the four known Bude fish species is morphologically suitable to have produced U. britannica , suggesting that a fifth species awaits discovery. The ichnogenus Undichna was proposed by Anderson (1976) for various combinations of sinusoidal and/or ‘scolloped’ waves; these were observed on parting surfaces in flaggy siltstones, and were interpreted as the drag marks of fish fins. Of the three species of Undichna erected by Anderson, two are reported here from the Bude Formation of south-west England (text-fig. 1 ). In addition, the Bude Formation has yielded two morphologically similar, but as yet unclassified, types of trace, for which two new ichnospecies are proposed. All grid references (GR) given in this paper refer to Sheet SS of the UK National Grid. GEOLOGICAL SETTING The Bude Formation consists of about 1300 m of Westphalian A-C mudstones and sharp-based, very-fine sandstones (Higgs 1986n, b , and in prep.). Deposition is thought to have taken place on a broad shelf in an equatorial, foreland-basin lake, named ‘Lake Bude’ by Higgs (1986u). The sandstones are interpreted as turbidites, deposited by river-fed underflows during storm-floods; many show evidence for simultaneous wave action (Higgs 1986u). The shelf lay on the northern side of the lake, and passed southward into a flysch trough formed by thrust-loading in front of the N-advancing Variscan orogenic front. Unconformably overlying the Bude Formation are the post-orogenic continental deposits of the Stephanian-Triassic New Red Sandstone (Laming 1982). The Bude Formation is laterally equivalent to the (Lower and Middle) Coal Measures lying on the stable foreland immediately to the north, in central and northern Britain (Ramsbottom et al. 1978). The Bude Formation shows a dm-m scale cyclicity, whereby two facies (FI and F2) alternate. FI consists of dark grey mudstone with sparse, thin (up to 20 cm) turbidites. F2 is coarser (shallower?) and consists of light grey silty mudstone/muddy siltstone with thicker (up to 40 cm) and often amalgamated (up to 10 m) turbidites. Body fossils, apart from three marine horizons (cm) with goniatites and pelagic bivalves, are limited to rare fish and Crustacea in FI. Undichna has only been found in F2. Body fossils, trace fossils, and mudstone C/S ratios (Berner and Raiswell 1984) suggest IPalaeontology, Vol. 31, Part 2, 1988, pp. 255 272. | © The Palaeontological Association 256 PALAEONTOLOGY, VOLUME 31 text-fig. 1. Map showing present-day outcrop of the Bude Formation (after Edmonds et al. 1975). Exposure is essentially limited to the coast, which is characterized by continuous cliffs and wave-cut platform. that FI and F2 were deposited in brackish water and fresh water, respectively (Higgs 1986a, 6, and in prep.). The implied salinity fluctuations in Lake Bude can be explained in terms of a low-lying sill which was intermittently overtopped by sea water (cf. the Bosporus (Scholten 1974)). The supposed fish trails were first described by King (1965), who interpreted them as xiphosurid (king-crab) mating traces, as discussed below. The trails are exposed on parting surfaces in parallel-laminated (varved?) muddy siltstones (F2). This flaggy lithology is rare; the trails have only been observed at two stratigraphic levels, in each case in a 2-3 m muddy siltstone unit (King 1965). In both cases, Undichna is intimately associated with xiphosurid trackways ( Kouphichnium; see King 1965, Goldring and Seilacher 1971, and Higgs 19866); no other fossils are present in the units concerned. The fact that fish trails and actual fish appear to be mutually exclusive (i.e. have only been found in F2 and FI, respectively) is probably an artifact resulting from (1) preferential exposure offish trails in F2, due to the presence of flaggy lithologies, and (2) preferential preservation of fish in FI, due to anoxic bottom conditions (Higgs 19866). SYSTEMATIC PALAEONTOLOGY The three species of Undichna erected by Anderson are all characterized by horizontal grooves in the form of waves with a regular wavelength. Two of these species are present in the Bude Formation. In addition, there are two other types of trace which are here assigned to the genus Undichna on the basis of regular wavy grooves; however, these traces are considered sufficiently different to justify the erection of two new species. Location of specimens. All specimens mentioned in the paper are housed in the palaeontological collections of the University Museum, Oxford, except two specimens in the Geology Department of the University of Reading (UR 14403 and UR 14404). HIGGS: CARBONIFEROUS FISH TRAILS 257 Ichnogenus undichna Anderson, 1976 Diagnosis {emended). The genus includes those trace fossils comprising a set of horizontal waves (incised grooves) with a common wavelength and alignment. The waves may or may not be accompanied by a straight, continuous furrow, upon which they are superimposed. Individual waves may be continuous, or the troughs/crests may be preferentially absent or preferentially present. There may be as many as nine waves in a set; commonly, there are only two, and in some cases just one. Waves occur as (1) parallel pairs, (2) non-parallel pairs which are (a) intertwined or (£>) separate, and (3) unpaired waves. These wave types occur in a variety of combinations. The traces are impressions (and corresponding moulds) on parting surfaces in flaggy lithologies. Undichna bina Anderson, 1976 Text-fig. 2 (trail a) Description. Only one specimen has been observed in the present study. The trail consists of a pair of sinuous grooves a constant distance (2-8 cm) apart. U. bina is the least complicated of the three species of Undichna erected by Anderson. Interpretation. Anderson (1976) interpreted U. bina as the engravings of the pelvic fins (e.g. text-fig. 3) of a fish. The pectoral fins cannot be responsible: because they occur near the head, where the amplitude of undulation during swimming is practically zero (Bainbridge 1962), the pectoral fins would not leave a sinuous trail. The absence of waves attributable to the anal and caudal fins may be due to the ‘undertrack fallout’ effect described by Goldring and Seilacher (1971), whereby the ‘copies’ of a surface trace impressed through the surficial sediment become increasingly simple with depth due to the ‘fallout’ of less deeply impressed elements of the trace. Alternatively, these other fins may have been held clear of the sediment. In either case, the implication is that the pelvic fins protruded further below the body of the fish than all other appendages. Undichna britannica nov. ichnosp. Text-figs. 2 and 4 1965 undesignated xiphosurid ‘nuptial embrace’ trails, King, fig. 1. 1970 undesignated fish trail, Fliri et al ., fig. 9 f. 1971 undesignated fish trails, Fliri et al., fig. 8. 1976 1 Undichna sp., Anderson, pi. 54, fig. 5. 1984 Undichna Isimplicitas , Archer and Maples, fig. 7e. Types. Holotype: specimen E.3841b, from a 2 m flaggy siltstone unit at the base of the cliff at GR (2017 0751), near Bude, Cornwall. Paratype: E.3842a, same horizon and locality as holotype. The siltstone unit occurs 4 m stratigraphically below a prominent ‘marker shale’ (FI), 5 m thick, known as the Saturday’s Pit Shale (Freshney et al. 1979). Diagnosis. The trace consists of a pair of sharply incised, intertwined grooves, each groove having the form of a sinusoidal or slightly asymmetrical sinusoidal wave. Description. The waves are of equal wavelength, but are out of phase, the phase difference ranging up to one half-cycle. One wave is of rather greater amplitude than the other. The larger (‘outer’) wave, whose lateral extremities are commonly faint, invariably cuts the smaller (‘inner’) wave. Rarely, an extra pair of discontinuous grooves is visible; these are sub-parallel to the inner wave, and lie one on each side of it, confined to the ‘inside’ of each bend (see fig. 1 of King 1965). Comparison. There is more than one pair of continuous waves in U. insolentia. In U. bina the two waves are always parallel. U. simplicitas usually consists of an odd number of waves. U . constdca has an associated furrow. 258 PALAEONTOLOGY, VOLUME 31 text-fig. 2 (a, line drawing, and b, photograph). Rock slab E.3841, showing: E. 3841a, Undichna bina\ E. 3841b, U. britannica, holotype; E.3841c, U. consulca , holotype; E.3841d, probable U. consulca. Note that, in specimen d, the only evidence for the sinusoidal grooves are the faint dimples bordering the right-hand side of the trail. The trails are in concave relief. From GR (2017 0751). Natural size. HIGGS: CARBONIFEROUS FISH TRAILS 259 mm m i fam , M. .■/ ; 260 PALAEONTOLOGY, VOLUME 31 text-fig. 3. Proposed correlation between fish fins and inscribed waves for: a, Undichna simplicitas (after Anderson (1976); note that the trail shown is a slightly asymmetrical one); B, U. britaimica. See text for details. Dimensions. Measurements (to the nearest 05 cm) of five representative Bude Formation specimens are as follows. The body-length calculation is discussed below. Amplitude of Amplitude of Phase Calculated Wavelength outer wave inner wave difference body length (cm) (cm) (cm) (cm) (cm) E.3841b 160 6-5 30 2-0 260 E.3842a 13-0 4-0 2-5 1-5 160 Text-fig. 4, trail b 12 0 30 F5 1-5 120 U R 1 4404 1 140 30 2-0 2-0 120 Latex peel2 42 0 120 4-5 2-5 48-0 1 A line-drawing of this specimen is figured by King (1965, fig. 1), but his scale is incorrect, and should be one-third shorter. 2 Taken from the type locality by Dr R. Goldring, and stored at the University of Reading. Remarks. U. britannica was considered ‘problematic’ by Anderson (1976, p. 407), and illustrated under the caption " Wndichna sp.’ (her pi. 54, fig. 5); she opted to leave it unclassified, remarking on the uncertainty about its interpretation (i.e. fish versus mating crabs (see below)). In all of the observed Bude Formation specimens, the two intertwined waves are only slightly out of phase. In contrast, the specimen illustrated by Anderson (1976, pi. 54, fig. 5) shows the two waves about 180° out of phase. HIGGS: CARBONIFEROUS FISH TRAILS 261 The overall trace is usually straight or gently curved, and extends for at least a metre or two (i.e. to the extremities of the exposure). In one case, a trace was observed to undergo a sudden sharp turn, doubling back on itself to make an angle of about 60°. In the Bude Formation, U. britannica occurs in association (on the same parting surface) with U. consulca and Kouphichnium sp., the three types of trace cutting across one another (text-fig. 4). Any one parting surface typically shows five to ten U. britannica individuals per square metre. Interpretation. King (1965) argued that the out-of-phase sinusoidal waves here named U. britannica were formed by the dragging tails of a male-female pair of xiphosurids locked in a mobile ‘nuptial embrace’ ( King 1 965, fig. 1 ), with the male clinging to the back of the female as she walked toward the shoreline to lay her eggs. The implication is that the female for some reason walked a sinuous course. By way of a modern analogy. King cited Caster’s (1938) review of the behaviour of present-day king-crabs, which, during the mating season, habitually ‘seek the shore in pairs, the male often hanging onto the tail of the female’ (Caster 1938, p. 22). However, nowhere in his account does Caster indicate that the female walks a sinuous path during this seasonal migration. In fact, it seems most unlikely that the female, laden not only with her eggs but also with her mate, would follow anything other than a straight course. The fact that normal xiphosurid trackways (i.e. Kouphichnium) are straight to gently curved (e.g. text-figs. 4 and 8) indicates that there is no physiological reason why a xiphosurid should have followed a sinuous course. Hence, King’s mating-crab model is unsound, in that for a xiphosurid to meander would surely have been a waste of energy. A more logical explanation for the sinusoidal grooves is that they were produced by an organism, or part of an organism, which was biomechanically compelled to follow a sinuous course (e.g. the posterior fins of a fish). Two additional difficulties with King’s model are as follows: 1. The out-of-phase waves are never superimposed upon xiphosurid footprints (Kouphichnium). This is hard to reconcile with King’s proposal that the female crab was walking toward the shore. King ascribes the lack of footprints to the undertrack-fallout effect. However, this argument is unsatisfactory, since one would expect to see associated footprints in at least some cases, representing different undertrack levels. 2. The observation that the trails occasionally do a ‘U-turn’ (see above) is incompatible with King’s suggestion that the trails reflect a seasonal migration to the spawning ground. It is submitted, therefore, that U. britannica is made not by xiphosurids, but by a fish swimming with its anal fin and caudal fin in contact with the substrate (text-fig. 3). The outer wave represents the caudal fin, since it cuts through (i.e. is ‘later’ than) the inner wave; its greater amplitude reflects the fact that the amplitude of undulation in a swimming fish increases posteriorly (Bainbridge 1962). Clearly, the anal and caudal fins of the fish in question must have projected lower than either ( 1 ) the pelvic fins, or (2) the pectoral fins, since the trail lacks any continuous, non-interfering wave pairs such as these paired fins would have produced (cf. text-fig. 3a). The discontinuous ‘side’ waves sometimes seen in U. britannica (text-fig. 3b) are thought to be due to the pelvic fins scratching the sediment briefly and alternately as the fish swam. An interesting parallel is found in Bainbridge’s (1962) laboratory observations of modern bream, dace, and goldfish: ‘during swimming a certain amount of rolling about the antero-posterior axis has been observed’ (Bainbridge 1962, p. 44). The rolling motion is ‘rather rarely observed’ and ‘seems most apparent during slower swimming’ (Bainbridge 1962, p. 45); this suggests that V . britannica specimens with and without ‘side’ waves are not necessarily the product of two different species of fish, but may instead be due to a single species which under certain circumstances swam with a slight roll. It is possible to estimate the size of the fish responsible for U. britannica (cf. Anderson 1970). Bainbridge (1962) found, in his observations of modern fish, that the amplitude of the tail beat is approximately equal to one quarter of the body length; from this relationship comes the equation ‘Body length ~ 4 x caudal wave amplitude’. The table of dimensions given above includes an estimate of body length thus obtained. Assuming that Bainbridge’s relationship for modern fish is grossly applicable to the ancient, the fish responsible for U. britannica ranged in length between about 10 and 50 cm. 262 PALAEONTOLOGY, VOLUME 31 text-fig. 4 (a, line drawing, and b, photograph). Field view of a parting surface showing: a, b, Undichna britannica; c, d, U. consulca; e, f, Kouphichnium sp. The latter consists of pusher imprints and dragged-out walking-leg impressions (line drawing shows schematic pusher imprints only). Strata right way up, younging toward viewer. Note that various additional fragmentary specimens of U. britannica are visible. Cliff base at type locality of U. britannica and U. consulca , GR (2017 0751). Concave relief, x 0-5. Undichtxa consulca nov. ichnosp. Text-figs. 2, 4, 5 Derivation of name. Con (form of cum, Latin) = with; sulcus (Latin, from Greek) = a furrow. Types. Holotype: specimen E.3841c, same horizon and locality as the V . britannica holotype. Paratypes: E.3841e, E.3843, both same horizon and locality as holotype; E.3838a, same horizon, cliff base at GR (2016 0895). Diagnosis. A composite trace consisting of a simple, unornamented, very shallow furrow, upon which is superimposed, throughout its length, a pair of intertwined grooves, each groove having HIGGS. CARBONIFEROUS FISH TRAILS 263 the form of a sinusoidal or slightly asymmetrical sinusoidal wave. The grooves may be either sharp or blurred. Description. The furrow is straight to gently curving over distances of dm to m. It is smoothly concave-upward in transverse profile, and is symmetrically flanked by an identical pair of subtle, sharp-crested ridges. The inner face of each ridge merges smoothly downward, with declining gradient, into the floor of the furrow. The outer face decreases gradually in angle of dip away from the furrow, merging smoothly with the adjacent undisturbed "plain’ within a distance of a few mm. The grooves are of equal wavelength, slightly out of phase, and of slightly differing amplitude (cf. the intertwined grooves of U. britannica). The higher amplitude (‘outer’) wave invariably cuts the lower amplitude (‘inner’) wave. The grooves are sharply incised in some cases; in others, they have a blurred or ‘washed out’ appearance (text-figs. 2 (trail c) and 5a, b). In some specimens, the lateral extremities (i.e. the crests and troughs) of the sinusoidal waves are missing, so that the grooves are reduced to a series of paired ‘flick marks’ inclined in alternating directions down the length of the trail (text-figs. 4 and 5a, b). The amplitude of the outer wave exceeds 264 PALAEONTOLOGY, VOLUME 31 text-fig. 5. Undiclma consulca , paratypes, natural size, a, undersurface of rock slab E.3841, showing U. consulca specimen E.3841e. Note that the lateral extremities of the (blurred) sinusoidal waves are missing. Convex relief. Locality as in text-fig. 2 (same slab), b , rock slab E.3838, showing specimen E. 3838a. Part of a Kouphichnium specimen (E. 3838b) is also visible, at the extreme left, running bottom right to top left. Concave relief. From GR (2016 0895). c, undersurface of rock slab E.3843. Note that (i) the sinusoidal grooves are sharply incised; (ii) both grooves suddenly terminate toward the left, suggesting that the trail-maker abruptly stopped; and (iii) the grooves are ‘off-centre’ with respect to the furrow. Convex relief. Locality as in text-fig. 2. the width of the furrow; the same is not always true of the inner wave. The waves may either straddle the furrow symmetrically, the outer wave overlapping both of the marginal ridges, or they may be displaced to one side, straddling one ridge but not the other (text-figs. 2 (trail c) and 5c). Both the furrow and the waves must be present to allow positive identification as U. consulca. Comparison. The furrow immediately distinguishes U. consulca from the other species of Undiclma. Even if the grooves are missing or very faint, the furrow could not be confused with other gutter-like ichnogenera: Gordia forms complex looping patterns, and Scolicia is heavily ornamented. A shallow furrow occurs at the highest undertrack level of Kouphichnium, but is accompanied by a blurred central (telson) groove (e.g. Goldring and Seilacher 1971, fig. 2). Dimensions. The maximum depth of the furrow (i.e. the elevation difference between the base of the furrow and the crests of the flanking ridges) is I mm. The elevation of the ridges, relative to the adjacent ‘plain’, never exceeds HIGGS: CARBONIFEROUS FISH TRAILS 265 text-fig. 6. Proposed origin of Undichna consulca. The furrow is produced by the fish’s belly dragging in the surficial mud, while the two sinusoidal grooves are inscribed by the fish’s anal fin and caudal fin (cf. text-fig. 3b). 0-5 mm. Other dimensions, from seven representative specimens, are as follows (measurements to the nearest 05 cm, except furrow width): E.3841c E.3841e E.3843 E.3838a E. 3842b UR 14403 Latex peel1 Wavelength outer wave inner wave furrow body length (cm) (cm) (cm) (mm) (cm) 4-5 2-5 1-5 24 10-0 4- 0 2-0 E5 9 8-0 7-0 2-0 1-5 18 80 2-0 1-5 1-5 7 60 5- 0 2-0 1-5 16 8-0 4- 0 2-5 2-0 15 100 5- 5 2-5 ? 12 100 1 Same peel as in table of dimensions for U. britannica. Amplitude of Amplitude of Width of Calculated Remarks. See the section on U. britannica for remarks about associated trace fossils. The density of U. consulca specimens on individual parting surfaces is typically between five and ten per square metre. Interpretation. By analogy with U. britannica , the two sinusoidal grooves of U. consulca are interpreted as the drag marks of the caudal fin and anal fin of a fish swimming in contact with the substrate (text-fig. 6). Specimens in which the grooves are blurred are probably surface traces, as opposed to undertraces; this is because surface traces tend to be 'less distinct . . . because they often become blurred by collapse and water action’ (Goldring and Seilacher 1971, p. 428). Where the lateral extremities of the waves are missing, the possible explanations are: (1) they were not formed, perhaps because the fish swept its tail slightly upward at the end of each tail beat; or (2) they collapsed or were washed out. An important question is whether the furrow and the sinusoidal grooves were formed simultaneously by a single animal (i.e. a fish), or whether the grooves were formed significantly later by a fish which was following a pre-existing furrow in the hope of finding (and eating?) the furrow-maker. The second of these two alternatives is highly unlikely, because (1) furrows are always accompanied by grooves, and (2) there is a definite correlation between furrow width and groove dimensions (text-fig. 7). Considering the anatomy of a fish, the furrow is thought to represent the impression of a fish’s belly dragging through the uppermost millimetre or two of the surficial sediment (text-fig. 6). Material displaced sideways by the fish was heaped up to form the low ridges marginal to the furrow. The furrow was then overprinted by the sinusoidal grooves, representing the drag marks of the anal fin and the caudal fin. The fact that the supposed belly impression is not sinusoidal reflects the fact that the 266 PALAEONTOLOGY, VOLUME 31 text-fig. 7. Plot of furrow width versus (outer) groove amplitude for seven representative specimens of Undichna consulca (see table of dimensions in text). Note the positive correlation. A better correlation might have been obtained but for the fact that (1) the amplitude values were only measured to the nearest 0-5 cm, and (2) the furrow width of individual specimens varies slightly, depending on the undertrack level sampled (see text). amplitude of undulation in a swimming fish decreases anteriorly, reaching zero (or almost zero) near the back of the head (Bainbridge 1962). Those specimens of U. consulca in which the grooves are off-centre with respect to the furrow are possibly due to a fish swimming across a current. Examples of U. consulca on modern sea- or lake-floors appear to be unknown. The nearest morphological analogue may be the ‘snout-marks’ illustrated by Stanley (1971, fig. 4), consisting of slightly curving leveed furrows formed by fish feeding in a snout-down posture. Unlike U. consulca, however, these modern furrows appear to be relatively short (cm-dm), and there is no sign of any sinusoidal grooves. These modern traces are formed by fish rooting for food in the mud (Stanley 1971; Marshall and Bourne 1964); in contrast, the U. consulca fish is thought to have been coasting along the bottom in search of epifaunal prey. The body length of the U. consulca fish, based on the caudal-wave amplitude (see the section on U. britannica , and the table of dimensions for U. consulca), ranged from 5 to 10 cm. Given the similarity between U. britannica on the one hand, and the sinusoidal grooves of U. consulca on the other, the question arises whether the former is simply the undertrail of the latter. This is thought unlikely, because the measurements presented in the tables above suggest that the U. britannica fish was substantially larger than the U. consulca fish (note that the measured specimens were selected at random). HIGGS: CARBONIFEROUS FISH TRAILS 267 With regard to the ‘fallout sequence’ in U. consulca, there appear to be two separate cases: (1) in specimens in which the grooves are symmetrically disposed over the furrow (e.g. text-fig. 5a, b), the grooves presumably fall out last, since they incise even the deepest part of the furrow; (2) in the remaining specimens, whose grooves are asymmetrically disposed, the furrow and the grooves appear to fall out at approximately the same level (e.g. text-fig. 2, trail d), suggesting that the fish swam in a more ‘tail-up’ attitude in such cases. Undichna simplicitas Anderson, 1976 Text-fig. 8 Description. In its most complete form, U. simplicitas consists of one (inner) pair of parallel sinusoidal waves, one (outer) pair of non-parallel scolloped waves, and a single unpaired wave (text-fig. 3a). However, it is not uncommon to find the unpaired wave alone (Anderson 1976). Solitary unpaired waves occur in the Bude Formation (text-fig. 8), and are accordingly assigned to U. simplicitas. Interpretation. The unpaired wave of U. simplicitas was probably formed by one or the other of two unpaired fins which most fish carry on their undersides, namely the anal fin or the caudal fin. Anderson (1976) favoured the caudal fin, for unspecified reasons (text-fig. 3a). Where only a solitary wave is encountered, it is reasonable to infer that either the anal fin or the caudal fin protruded lower than any other part of the body. It is impossible to say which fin, anal or caudal, was responsible, because the anal fin is the lowest in some genera, while the caudal fin is the lowest in others (see the many fish restorations in Traquair (1877-1914), Woodward (1891), Romer (1966), and Miles (1971)). A case could be made for erecting a separate new ichnospecies to accommodate all occurrences of unpaired waves, for two reasons. First, solitary waves are morphologically very different from ‘complete’ specimens of U. simplicitas. Secondly, in assigning solitary waves to U. simplicitas , Anderson is implying that such a wave represents the undertrail of a more complete specimen of U. simplicitas', in other words, it was produced by the same type of fish; however, this is an unreasonable assumption, since an unpaired wave could equally likely be the undertrail of U. britannica or U. consulca', alternatively, the solitary wave may not be an undertrail at all. Before any new ichnospecies is erected, however, a restudy of Anderson’s (1970, 1976) material should be undertaken, in order to (1) define the full range of variability in the unpaired waves, and (2) nominate type specimens. THE IDENTITY OF THE TRACE MAKERS The purpose of this section is to identify which, if any, of the four species of fossil fish discovered in the Bude Formation could have made the four types of fish trail described above. The four fish are: Acanthodes wardi (Woodward 1891, Part 2, fig. 1), Cornuboniscus budensis (text-fig. 9), Elonichthys aitkeni (Traquair 1877-1914, pi. 16), and Rhabdoderma elegans (Forey 1981, text-fig. 9). U. bina As mentioned earlier, the parallel sinusoidal waves of U. bina were probably made by a fish whose pelvic fin-tips were lower than any of the other fin-tips. This condition is not satisfied by any of the above fish, with the possible exception of E. aitkeni. (Unfortunately, the two specimens of E. aitkeni illustrated by Traquair, including the type specimen, are contorted, and the fin relationships are therefore not visible.) If E. aitkeni is not responsible for U. bina, then some other chondrostean fish could be responsible, since this group includes many genera with ‘low’ pelvic fins. Alternatively, among the other six groups of Upper Carboniferous fishes (Miles 1971), certain crossopterygians, dipnoans, elasmobranchs, and holocephalans are likewise of suitable morphology to have produced U. bina. U. britannica If the intertwined grooves of U. britannica were produced by the anal and caudal fins of a fish, as inferred earlier, then the fish must have swum with its anal and caudal fin-tips lower than any other 268 PALAEONTOLOGY, VOLUME 31 text-fig. 8 (a, line drawing, and b, photograph). Rock slab E.3837, showing: E.3837a, U. simplicitas ; E. 3837b, possible U. simplicitas; E. 3837c, Kouphichnium sp. The latter consists of the following elements: (i) an outer set of regularly spaced genal-spine impressions; (ii) a central telson groove; and (iii) blurred walking-leg and pusher imprints. This Kouphichnium specimen represents a shallower undertrack level than trails e and f in text-fig. 4 (see text). Convex relief. From GR (2017 0751). Natural size. fin-tips. The pelvic fins, responsible for the discontinuous ‘side’ grooves, evidently did not extend so far. Based on these characteristics, neither A. wardi nor R. elegans could have produced U. britannica. The morphology of C. budensis seems to be approximately correct (text-fig. 9); however, the largest of twenty-five specimens examined by White (1939) is only 7 cm long (allowing for its missing tail), substantially shorter than the 10-50 cm fish inferred from the measurements of U. britannica. Finally, the morphological suitability of E. aitkeni is unknown, due to the imperfect condition of the type specimen; however, this fish may also have been too small, the maximum length being about 18 cm according to Woodward (1891). It seems likely, therefore, that U. britannica was made by a fish whose fossilized remains have yet to be found in the Bude Formation. Based on the morphology deduced above, only one of the seven groups of Carboniferous fishes, namely the chondrosteans, includes suitably shaped genera. How- ever, few chondrosteans exceed 20 cm in length. One genus which is of the correct morphology and size for U. britannica is Acrolepis (e.g. Traquair 1877-1914, pi. 25, fig. 7). Specimens of Acrolepis examined by Traquair range from 9 cm (excluding the missing head) to 65 cm. Acrolepis is also of the correct age, ranging from Lower Carboniferous to Upper Permian (Romer 1966). The possibility exists, therefore, that Acrolepis will one day be discovered in the Bude Formation. U. consulca The furrow and the intertwined grooves of U. consulca permit the following deductions concerning the morphology of the fish responsible: (1) the anal and caudal fins (responsible for the grooves) must have protruded lower than the fish’s belly (responsible for the furrow), since the grooves are incised into the furrow; (2) the absence of any ‘paired and parallel’ waves attributable to the pelvic fins implies that these fins were high enough to remain clear of the sediment; and (3) the pectoral fins must have HIGGS: CARBONIFEROUS FISH TRAILS 269 been inserted high up the flank of the body, otherwise they would have fouled the bottom whenever the fish’s belly was touching the sediment. The last of these anatomical deductions provides an important clue to the identity of the fish in question. One of the fossil fish in the Bude Formation, C. budensis , is characterized by very unusual pectoral fins, mounted high on the body (text-fig. 9); in fact, the pectorals are so unusual that White (1939) erected a new genus, of which C. budensis is the only known species; furthermore, this fish is unknown outside the Bude Formation. White (1939, p. 52) remarked that The most striking feature of this little fish is the form of the pectoral fin’: unlike normal palaeoniscoid (suborder) fishes, in which the pectoral fins were inserted low down the flank and held more-or-less horizontally (White 1939; Westoll 1944), the pectorals of C. budensis are inserted vertically, high up the flank (text-fig. 9). Normal palaeoniscoids, therefore, would have been unable to engage in ‘belly-skimming’ without dragging their pectoral fins; in contrast, the pectorals of C. budensis would have been well clear of the sediment. Westoll (1944) argued that White’s reconstruction of the pectoral fins of C. budensis (as in text-fig. 9) is incorrect, since if held vertically instead of horizontally, the pectorals could not have fulfilled their usual role of acting as hydrofoils to counterbalance the tail-lift induced by the heterocercal tail. However, Westoll was neglecting the possibility of a belly-skimming mode of feeding, in which hydrofoil-type pectorals might actually be disadvantageous. With regard to other morphological features of C. budensis , the anal and caudal fin-tips were suitably positioned to have produced the intertwined grooves of U. consulca (text-fig. 9). However, there is a potential difficulty with the pelvic fins: in the position shown (text-fig. 9), they would inevitably have dragged if the anal fin, caudal fin, and belly were in contact with the sediment, yet there is no record of this in U. consulca. The solution to the problem may be that the pelvic fins actually protruded outward more than is shown in White’s reconstruction; indeed, Westoll (1944, p. 85) stated that ‘Probably the drooping position in which they (the pelvic fins) are customarily restored in palaeoniscids (family) is misleading’ (bracketed words added). Furthermore, the pelvic fins of C. budensis ‘are placed well to the rear’ (White 1939, p. 43); in this position, the pelvics are less likely to have contacted the sediment than are those of normal palaeoniscids. As well as having the correct morphology, C. budensis is the correct size to have produced U. consulca. The range in length of 4-7 cm for the twenty-five C. budensis specimens examined by White 270 PALAEONTOLOGY, VOLUME 31 Dorsal fin ( 1939) (allowing for incomplete specimens) compares favourably with the range of 5- 10 cm calculated above from the dimensions of U. consulca. Moreover, C. budensis has been found in close stratigraphic proximity to U. consulca: the fish occurs only in the Saturday’s Pit Shale (Freshney et al. 1979), which lies just 4 m above the U. consulca- bearing unit at the type locality. Hence, there is a considerable body of evidence to suggest that C. budensis, or an undiscovered close relative, is the U. consulca fish. This conclusion is strengthened by the fact that no other known Upper Carboniferous fish had a suitable morphology. It is interesting to speculate on what the belly-skimming fish’s diet might have been. The fact that the furrow of U. consulca is flanked by ridges suggests that the fish was pushing the mud aside, rather than filtering it to extract contained invertebrates or plant detritus. Two additional factors militate against the idea that the food was contained in the mud: (1) there are no burrows in F2, suggesting that the mud lacked an infauna; and (2) C. budensis is unlikely to have favoured plant detritus, since it has sharp, conical teeth and a wide gape, suggesting a carnivorous diet. The evidence suggests that the fish was preying on live epifaunal organisms; of these, the only indication consists of xiphosurid trackways ( Kouphichnium ). The trackways are only 1 to 3 cm across; hence the xiphosurids might have been small enough for the (5-10 cm) fish to tackle. U. simplicitas It was suggested above that the solitary wave, here assigned to U. simplicitas, was produced by a fish whose caudal fin or anal fin extended lower than any other part of the body. Therefore, of the four fish species discovered in the Bude Formation, A. wardi and R. elegans can be discounted. The correct morphology is shown by C. budensis, but this fish was less than 10 cm long, and therefore could only have produced the smaller of the two U. simplicitas specimens visible in text-fig. 8. (Applying the body-length calculation discussed earlier, the amplitude of the larger trace (min. 5 cm) suggests a fish at least 20 cm long.) Whether or not E. aitkeni could have produced an unpaired wave is uncertain, owing to the poor condition of the type specimen (Traquair 1877-1914). Acrolepis, proposed earlier as the likely maker of U. britannica, is of suitable shape and size to have produced the larger of the two U. simplicitas specimens. Thus, Acrolepis may have been responsible for both (the larger) U . simplicitas and for U. britannica; this is consistent with the idea, advanced earlier, that solitary-wave specimens of U. simplicitas could, in some cases, represent the undertrail of U. britannica. Similarly, the deduction that C. budensis could be responsible for both (the smaller) U. HIGGS: CARBONIFEROUS FISH TRAILS 271 simplicitas and for U . consulca is consistent with the idea that the former is potentially the undertrail of the latter. Absence of trails corresponding to two of the Bade Formation fish From the foregoing discussion, it is apparent that A. wardi and R. elegans (and possibly E. aitkeni ) cannot be matched to any of the four types of trail described here. In the case of A. wardi , the lack of corresponding trails is not surprising, since acanthodians were probably mid- to surface-water feeders throughout their history ( Miles 1971). R. elegans , which is a coelacanth, may or may not have swum in contact with the bottom; however, any trails which it happened to produce might be other than sinuous, since the modern coelacanth ( Latimeria ) appears to scull with its paired fins instead of waving its tail (Mackenzie 1987). Salinity preference of Bade Formation fish It has been shown that C. budensis was probably the maker of U. consulca. This suggests that C. budensis was tolerant of both brackish- and fresh water, since the fish and its trails have been found in brackish facies (FI) and in fresh facies (F2) (respectively; see Introduction). Evolution of endemic fishes in Lake Bude As discussed earlier, C. budensis appears to have evolved special features to facilitate a belly-skimming mode of life. The fact that this fish has never been found outside the Bude Formation suggests that it may have been endemic to Lake Bude. Analogous circumstances are found in modern lakes, where reproductive isolation is common (e.g. Beadle 1981), giving rise to endemic fishes that are specialized to take advantage of the local conditions. Acknowledgements. I thank Drs P. C. Jackson and W. J. Kennedy for their comments on an early draft of this manuscript. I am especially grateful to Dr Roland Goldring, who provided access to the Reading specimens of Undichna , and who discussed the paper at great length with me, leading to substantial improvements. Drs Peter Forey and Sally Young of the British Museum kindly supplied references to fish illustrations. This study represents part of my doctoral research at the University of Oxford; my thanks to Dr H. G. Reading for his stimulating supervision. The work was carried out during tenure of a British Petroleum Studentship and a UK government Overseas Research Student Award; I gratefully acknowledge this support. REFERENCES anderson, a. 1970. An analysis of supposed fish trails from interglacial sediments in the Dwyka Series, near Vryheid, Natal. Proc. Pap. 2nd I.U.G.S. Gondwana Symp .. South Africa , 637-647. CSIR, Pretoria. 1976. Fish trails from the early Permian of South Africa. Palaeontology. 19, 397-409. archer, a. w. and maples, c. G. 1984. Trace-fossil distribution across a marine-to-nonmarine gradient in the Pennsylvanian of southwestern Indiana. J. Paleont. 58, 448-466. bainbridge, r. 1962. Caudal fin and body movement in the propulsion of some fish. J. exp. Biol 40, 23-56. beadle, L. c. 1981. The inland waters of tropical Africa. 475 pp. Longman, London. berner, r. a. and raiswell, r. 1984. C/S method for distinguishing freshwater from marine sedimentary rocks. Geology , 12, 365-368. caster, k. e. 1938. A restudy of the tracks of Paramphibius. J. Paleont. 12, 3-60. edmonds, E. A., mckeown, M. c. and williams, M. 1975. British regional geology: south-west England. 136 pp. HMSO, London. fliri, f., bortenschlager, s., felber, h., heissel, w., hilscher, h. and resch, w. 1970. Der Banderton von Baumkirchen (Inntal, Tirol): eine neue Schliisselstelle zur Kenntnis der Wiirm-vereisung der Alpen. Zeitschr.f. Gletscherkunde u. Glazialgeol. 6, 5-35. — hilscher, h. and markgraf, v. 1971. Weitere untersuchungen zur Chronologie der Alpinen Vereisung (Banderton von Baumkirchen, Inntal, Nordtirol). Ibid. 7, 5-24. forey, p. l. 1981. The coelacanth Rhabdoderma in the Carboniferous of the British Isles. Palaeontology, 24, 203-229. 272 PALAEONTOLOGY, VOLUME 31 freshney, e. c., edmonds, e. a., taylor, r. t. and williams, b. j. 1979. Geology of the country around Bude and Bradworthy. Mem. geol. Surv. GB , 62 pp. HMSO, London. goldring, r. and seilacher, a. 1971. Limulid undertracks and their sedimentological implications. Neues Jb. Geol. Paldont. Abh. 137, 422-442. higgs, r. 1986a. ‘Lake Bude’ (early Westphalian, SW England): storm-dominated siliciclastic shelf sedimentation in an equatorial lake. Proc. Ussher Soc. 6, 417 418. 19866. A facies analysis of the Bude Formation (Lower Westphalian), SW England. D.Phil. thesis (unpublished). University of Oxford. In prep. The Bude Formation (Lower Westphalian, SW England): storm-dominated siliciclastic shelf sedimentation in a giant equatorial lake. Sediment ology. king, a. f. 1965. Xiphosurid trails from the Upper Carboniferous of Bude, north Cornwall. Proc. geol. Soc. London. 1626, 162-165. laming, d. j. c. 1982. The New Red Sandstone. In durrance, e. m. and laming, d. j. c. (eds.). The geology of Devon. 148-178. University of Exeter. Mackenzie, d. 1987. ‘Living fossils’ come alive on him. New Scientist, 113 (1547), 20. marshall, N. b. and bourne, D. w. 1964. A photographic survey of benthic fishes in the Red Sea and Gulf of Aden. Bull. Mus. comp. Zool., Harvard, 132, 223-244. miles, R. s. 1971. Palaeozoic fishes, 2nd edition (extensively revised from the first edition of J. A. Moy-Thomas, 1939), 259 pp. Chapman and Hall, London. RAMSBOTTOM, W. H. C., CALVER, M. A., EAGAR, R. M. C., HODSON, F., HOLLIDAY, D. W., STUBBLEFIELD, C. J. and wilson, R. b. 1978. A correlation of Silesian rocks in the British Isles. Spec. Rep. geol. Soc. London , 10, 81pp. romer, a. s. 1966. Vertebrate paleontology, 468 pp. University of Chicago Press. scholten, R. 1974. Role of the Bosporus in Black Sea chemistry and sedimentation. Mem. Am. Assoc, petroleum Geol. 20, 115-126. Stanley, d. j. 1971. Fish-produced markings on the outer continental margin east of the Middle Atlantic States. J. sedim. Petrol. 41, 159-170. stormer, L. 1955. Merostomata. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part P. Arthropoda 2, P4 P41. Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas. traquair, r. h. 1877-1914. The ganoid fishes of the British Carboniferous formations. Part 1, Palaeoniscidae. Mon. pal. Soc. 186 pp. London. westoll, T. s. 1944. The Haplolepidae, a new family of late Carboniferous bony fishes. Bull. Am. Mus. nat. Hist. 83, 1-122. white, E. i. 1939. A new type of palaeoniscoid fish, with remarks on the evolution of the actinopterygian pectoral tins. Proc. zool. Soc. London, B109, 41-61. woodward, a. s. 1891. Catalogue of the fossil fishes in the British Museum (Natural History ), Part 2. London. ROGER HIGGS Department of Earth Sciences University of Oxford Parks Road, Oxford 0X1 3PR, UK Present address: Geological Survey of Canada Pacific Geoscience Centre 9860 West Saanich Road Sidney, B.C. V8L 4B2 Canada Typescript received 14 October 1986 Revised typescript 1 June 1987 Note added in proof With regard to the section on p. 27 1 entitled ‘Absence of trails . . .’, a paper by Fricke et al. on locomotion of the modern coelacanth ( Latimeria ) appeared in 1987 (Nature, 329, 331 333). These authors observed Latimeria frequently resting, but not swimming, in contact with the bottom. Assuming that ancient coelacanths, including R. elegans, behaved similarly, they are unlikely to have produced trails. THE OLDEST FRESHWATER DECAPOD CRUSTACEAN, FROM THE TRIASSIC OF ARIZONA by GARY L. MILLER and SIDNEY R. ASH Abstract. The oldest known freshwater crayfish is described here from a nearly complete specimen found in the Chinle Formation of Late Triassic (Late Carnian) age in Petrified Forest National Park, Arizona, USA. It differs significantly from other Triassic crayfish and is placed in Enoploclytia porteri sp. nov. in the family Erymidae (Malacostraca, Decapoda). Except for clam shrimps (Eubranchiopoda) and notostracans E. porteri sp. nov. is the only crustacean that has been described from the non-marine Upper Triassic strata of North America. The fossil record of crayfish (Malacostraca, Decapoda) is rather sparse and consists primarily of isolated appendage fragments (Schram 1986). Definite early Mesozoic non-marine forms were unknown until the nearly complete specimen described here was collected from the Chinle Formation of Late Trassic Age in Petrified Forest National Park, Arizona, USA (text-fig. 1a, b). Although the fossil (text-figs. 2a, b and 3) is similar to fossil shrimp of the family Penaeidae, it is distinguishable as a crayfish and is assigned to the genus Enoploclytia , in the family Erymidae. The specimen differs in significant features (e.g. long, slender first pleopods, lack of carapace and chelae ornamentation— see reconstruction in text-fig. 4) from other species of Enoploclytia and is placed in the new species porteri. Crustacea are generally scarce in the non-marine Upper Triassic strata of North America. The most abundant are clam shrimps (Eubranchiopoda) which have been described from the Chinle Formation in New Mexico (Tasch 1978) and the Newark supergroup of eastern North America (Bock 1953; Olsen in Bain and Harvey 1977; Olsen et al. 1978). Notostracans have also been described recently from the Newark Supergroup of eastern North America (Gore 1986). Olsen (in Bain and Harvey 1977) has reported the occurrence of a supposed crayfish in the Newark Supergroup in North Carolina. However, that fossil has never been described and the drawings published by Olsen are not clear enough to evaluate satisfactorily. The fossil described here has been deposited in the natural history collections at Petrified Forest National Park (PEFO), Arizona. GEOLOGICAL SETTING The Chinle Formation is widely distributed in the Colorado Plateau area of the south-western United States where it ranges up to about 400 m in thickness. It is composed principally of structureless variegated mudstone and many relatively thin, discontinuous beds of grey and tan sandstone and conglomerate together with minor amounts of non-marine limestone (Stewart et al. 1972). The formation was deposited in a broad basin by streams and in lakes during the Carnian and Norian Stages of the Late Triassic (Ash et al. 1986). Although about a dozen lithologically distinctive members have been recognized in the formation only the Petrified Forest and Owl Rock Members are exposed in the Petrified Forest (Billingsley 1985). The Chinle is estimated to be about 350 m thick in the vicinity of the Petrified Forest (Stewart et al. 1972). Many types of fossils occur in the Chinle Formation including enormous quantities of petrified wood and large numbers of leaves, cones, and palynomorphs (Ash 19746; Litwin 1985). Invertebrate fossils, including gastropods and bivalves, horseshoe crab trackways, insect remains, and clam shrimps are also found in the formation (Breed 1972). Vertebrate remains are locally abundant and consist of the remains of many types of fish, several species of amphibians, and many taxa of reptiles | Palaeontology, Vol. 31, Part 2, 1988, pp. 273-279.| © The Palaeontological Association 274 PALAEONTOLOGY, VOLUME 31 text-fig. 1. a, map of Petrified Forest National Park, Arizona showing the location of the fossil localities and geographic features mentioned in the text. B, composite stratigraphic section of the lower part of the Petrified Forest Member of the Chinle Formation in the vicinity of fossil localities PF1 and PF5. including dinosaurs (R. A. Long, pers. comm., 1985). Recent investigations show that the largest known concentration of all types of fossils known from the Chinle Formation are found in Petrified Forest National Park. Locality The crayfish described here was collected from a locality about 5-5 km south-east of the Puerco River in Petrified Forest National Park, Arizona (text-fig. 1a). The locality (PF5 of this report) is situated in some low hills about 120 m east of the main park road in the SW1/4, NW1/4, sec. 23, T. 18 N., R. 24 E. and is the same as the University of California Museum of Paleontology (UCMP) fossil plant locality P3901-4 (Daugherty 1941). It is about 100 m east and 2 m stratigraphically higher than the principal fossil leaf locality (locality PF1 of this report and UCMP fossil plant locality P3901-1) but in the same MILLER AND ASH: TRIASSIC FRESHWATER DECAPOD 275 bed of structureless grey mudstone (text-fig. 1b). Locality PF5 is about 55 m below the Sonsela Sandstone Bed in the lower part of the Petrified Forest Member. Several thin beds of the Newspaper Sandstone Bed are present on the slope just to the west of the crayfish locality and another is about one metre above the locality. Locality PF5 is in the Dinophyton floral zone of Ash (1980) which correlates with the Late Carnian Stage of the Upper Trassic (Ash et al. 1986). Many fossils of several types occur scattered throughout the bed of grey mudstone at locality PF5, including the carapaces of an unidentified clam shrimp, beetle elytra, a number of plant megafossils, and the crayfish. Daugherty (1941) reported the occurrence of three fossil plants at the locality: Dadoxylon chaneyi , Equisitites sp., and Lycostrobus chinleana. Additional plant fossils have been collected from the locality by the second author including Neocalamites sp. and Zamites powellii. Three deeply weathered stumps of trees that resemble Araucarioxylon arizonicum are exposed a few metres to the east of the locality in the same bed of grey mudstone. MATERIAL The description is based on a single, fairly complete, laterally compressed specimen which is preserved as part and counterpart that are more or less separated along the mid-sagittal plane into right and left halves (text-figs. 2a, b and 3). As shown in the figures, most of the left and right first pereiopods and right pereiopods two to four are preserved. The telson is fragmentary however, and the uropods are absent. The fossil does not appear to have been altered appreciably since burial. Fine details are visible as a consequence of the fine-grained nature of the rock in which it is embedded. The fossil is described and reconstructed here as an astacid crayfish (text-fig. 4). SYSTEMATIC PALAEONTOLOGY Class MALACOSTRACA Order decapoda Infraorder astacidea Latreille, 1803 Family erymidae Van Straelen, 1924 Subfamily eryminae Van Straelen, 1924 Genus enoploclytia McCoy, 1849 Enoploclytia porteri sp. nov. Text-figs. 2a, b and 3 Type specimen. Holotype PEFO 2991. Late Carnian, lower part of the Petrified Forest Member of the Chinle Formation in Petrified Forest National Park, Arizona. Derivation of name. The trivial name honours Mr David Porter who discovered the only known specimen of the species. Diagnosis. Specimen small (length 2-5 cm, from most posterior part of telson to anterior tip of rostrum; total length 3 8 cm, from most posterior part of telson to anterior tip of first pereiopod). First pereiopod long (protopod length L8 mm, endopod length 15-7 mm), slender with narrow chelae (left 5 4 mm, right 5 6 mm). Chelae width does not exceed that of distal portion of leg. Pereiopods two and three with small chelae (average length 1-2 mm), chelae toothless. Chelae of pereiopod two slightly larger than chelae of pereiopod three. Pereiopods four and five lacking chelae. Pereiopods two to five are about the same length, 9-8- 10-5 mm (10 6, 9 8, 10T, 10-5 mm total length legs two to five respectively). Cephalothorax subcylindrical. Carapace length 111 mm (tip of rostrum to most posterior part), depth 5-3 mm at deepest. Carapace unornamented, rostrum spiny, well developed (length 3 0 mm). Cervical, postcervical, and antennar grooves present. Abdomen length 10 7 mm. Pleopods slender, equal in length. Abdominal segments with well-developed pleura. First abdominal segment approximately half the width of the second and somewhat shorter. text-fig. 2. Enoploclytia porteri sp. nov. from the Upper Triassic of Petrified Forest National Park, Arizona, USA. a, right half, x 3. b, left half, x 3. text-fig. 3. Outline drawing of the right half of the fossil showing the slender first chelae (fc), segment of antenna (sa), spiny rostrum (sr), antennar groove (ag), cervical groove (eg), first and second abdominal segments (as), and lobed abdominal tergites (at). MILLER AND ASH: TRIASSIC FRESHWATER DECAPOD 277 text-fig. 4. Reconstruction of Enoploclytia porteri sp. nov. Note that the hist pleopods are slender and the second and third chelae are about the same size, a, dorsal view, b, lateral view. Affinities. The astacid crayfish share some important characteristics with fossil shrimp of the family Penaeidae such as well-developed pleura, a spiny rostrum, and long slender chelae. However, the new specimen differs significantly from penaeids in several important respects. In astacids, as in the fossil described here (text-fig. 4), the first pereiopods are chelate whereas in most penaeids the second or third pereiopods bear the chelae. In penaeid groups where the first pereiopod is chelate, it is small (e.g. Antrimpos , see Glaessner 1 969) or there are other significant diagnostic characteristics present (e.g. the distinctive rostrum of Bylgia , see Glaessner 1969). Moreover, with respect to other characteristics of the pereiopods, the most distinguishable group of penaeids (Aeger, Schram 1986; Glaessner 1969), and also some members of less cohesive penaeid groups (e.g. Antrimpos , see Glaessner 1969) bear distinct spines on the largest (and other) pereiopods. No such spines are visible in the fossil described here (text-figs. 2a, b and 3). Furthermore, the first and second abdominal segments of penaeid shrimp are generally equal in width and height (Glaessner 1969). In the new fossil, the first segment is distinctly narrower and shorter than the second segment; the typical astacid condition. Thus, it is concluded that the fossil most probably represents an astacid crayfish. The placement of Enoploclytia porteri sp. nov. within the family Erymidae is based primarily on the presence of cervical and postcervical grooves, and the subcylindrical carapace (cf. text-figs. 2a and 3). Within the Erymidae, the Eryminae is distinguished from the Clytiopsinae (in which Forster 1967 places only Clytiopsis and Paraclytiopsis) by the presence of an intercalated plate and large chelae on the first pereiopods. Also the Eryminae generally have better developed rostra, and may exhibit ornamentation on the carapace and pereiopods. Although the presence of an intercalated plate cannot be observed in this fossil it is placed in the Eryminae because of the longer first chelae, and the presence of a well-developed rostrum. E. porteri sp. nov. is excluded from the eryrnid genus Eryma because of the shorter rostrum and stouter first chelae of that genus. 278 PALAEONTOLOGY, VOLUME 31 DISCUSSION The early history of the decapod crustaceans is poorly known. The earliest may be Palaeopalaemon newberryi (Schram et al. 1978) from the Upper Devonian of North America, although Felgenhauer and Abele (1983) have an alternate view of the systematic affinities of that fossil (see also Brooks 1 969). The earliest post-Devonian decapods known from North America are the marine species Pseudoglyphea mulleri (Van Straelen 1936) from the Upper Triassic of Nevada and the undescribed, fragmentary fossil which has been attributed to a crayfish from the Upper Triassic Newark Supergroup in North Carolina (Olson in Bain and Harvey 1977). Thus Enoploclytia porteri sp. nov. is the oldest described freshwater decapod, predating Enoploclytia from the ?Middle Jurassic of Europe, Erymastacus from the Lower Jurassic of Canada, and Eryma from the Lower Jurassic of Europe (Glaessner 1969). Zoogeographically the new species is of interest because of its location relative to the supposed centre of origin of modern North American crayfish. The largest group of them, the Cambarinae (Cambaridae), are thought to have originated in central Mexico in the Late Cretaceous (Ortman 1905). Ancestors of modern Procambarus and Cambarellus are thought to have migrated from Mexico to the mid-southern United States (Pennak 1978). The Eryminae extended into the Cretaceous of North America and other areas (Glaessner 1969; Schram 1986). The extent and nature of the ecological interactions of these groups would be of interest. Enoploclytia porteri sp. nov. occurred in strata that are thought to represent overbank deposits. However, a stream deposit, the Newspaper Sandstone Bed, is near by. Probably the species was a stream dweller (in the stream in which the Newspaper Sandstone Bed was deposited?) and the specimen was washed into the overbank area during a flood where it was buried almost immediately before the specimen could disassociate. Unfortunately, the comparative morphology of the fossil actually reveals little about its natural history so it is not possible to confirm the supposition that the species was a stream dweller. Modern crayfish exhibit extreme diversity in habitat and food preferences and considerable morphological variation exists among species within similar habitats (e.g. the size of the first chelae of modern stream forms varies considerably (Pennak 1978)). Acknowledgements. We are grateful to M r David Porter who collected the specimen and generously presented the fossil to the authors for description. The second author extends his thanks to Superintendent Edward Gastelum for allowing him to work in Petrified Forest National Park. We thank Dr Gale Bishop of Georgia Southern College for reviewing an early draft of this paper. REFERENCES ash, s. r. 1974«. Notes on the Chinle Formation (Upper Triassic) in east-central Arizona. In ash, s. r. (ed.). Guidebook , Devonian , Permian , and Triassic plant localities, east-central Arizona, 40-42. Paleobotanical Section, Botanical Society of America. 19746. The Upper Triassic Chinle flora of Petrified Forest National Park, Arizona. Ibid. 43-48. 1980. Upper Triassic floral zones of North America. In dilcher, d. l. and taylor, t. m. (eds.). Biostratigraphy of fossil plants, 153-270. Dowden, Hutchinson and Ross, Inc., Stroudsburg, Pennsylvania. litwin, r. and long, r. a. 1986. Biostratigraphic correlation of the Chinle Fm. (Late Triassic) on the Colorado Plateau, a progress report (abs.). Abstr. Progs geol. Soc. Am. 18, 338-339. bain, g. l. and harvey, b. w. 1977. Field guide to the geology of the Durham Triassic basin. Carolina Geological Society, 40th Anniversary Meeting, 83 pp. Billingsley, G. h. 1985. General stratigraphy of the Petrified Forest National Park, Arizona. In colbert, e. h. and JOHNSON, r. r. (eds.). The Petrified Forest through the ages. Bull. Mus. nth. Ariz. 54, 3-8. bock, w. 1953. American Triassic estherids. J . Paleont. 27, 62-76. breed, w. j. 1972. Invertebrates of the Chinle Formation. In breed, w. j. and breed, c. s. (eds.). Investigations in the Chinle Formation. Bull. Mus. nth. Ariz. 47, 19-22. brooks, H. K. 1969. Eocarida. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part R. Arthropoda4 {!), R 1 R399. Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas. MILLER AND ASH: TRIASSIC FRESHWATER DECAPOD 279 Daugherty, L. h. 1941. The Upper Triassic flora of Arizona. Pubis Carnegie Instn , 526, 1-108. felgenhauer, b. e. and abele, L. G. 1 983. Phylogenetic relationships among shrimp-like Decapoda. Crust. Issues , 1, 291-311. forster, r. 1967. Die reptanten Dekapoden der Trias. Neues Jb. Geol. Palaont. Abli. 128 (2), 136 194. glaessner, m. f. 1969. Decapoda. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part R. Arthropoda 4 (2), R399-R566. Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas. gore, p. j. w. 1986. Triassic notostracans in the Newark Supergroup: Culpeper Basin, northern Virginia, with a contribution on the palynology by Alfred Traverse. J. Paleont. 60, 1086-1096. litwin, r. j. 1985. Fertile organs and in situ spores of ferns from the Late Triassic Chinle Formation of Arizona and New Mexico, with discussion of the associated dispersed spores. Rev. Palaeobot. Palynol. 44, 101 146. olsen, p. e., remington, c. l., cornet, b. and Thomson, k. s. 1978. Cyclic change in Late Triassic lacustrine communities. Science , NY. 201, 729-733. ortman, a. e. 1905. The mutual affinities of the species of the genus Cambarus , and their dispersal over the United States. Proc. Am. phil. Soc. 44, 91 136. pennak, r. w. 1978. Fresh-water invertebrates of the United States, 2nd edn., 803 pp. Wiley Interscience, New schram, f. r. 1986. Crustacea, 606 pp. Oxford University Press, New York. — feldman, r. m. and copeland, m. t. 1978. The Late Devonian Palaeopalaemonidae and the earliest decapod crustaceans. J. Paleont. 52, 1375-1387. STEWART, J. H., POOLE, F. G., WILSON, R. F., CADIGAN, R. A., THORDARSON, W. and ALBEE, H. F. 1972. Stratigraphy and origin of the Chinle Formation and related Upper Triassic strata in the Colorado Plateau region. Prof. Pap. US geol. Surv. 690, 336 pp. tasch, p. 1978. Clam shrimps, 61-65. In ash, s. r. (ed.). Geology, paleontology, and paleoecology of a Late Triassic lake in western New Mexico. Brigham Young Univ. Geology Studies, 25, 1-95. van straelen, v. 1936. Sur des crustaces Decapoda Triasiques du Nevada. Bull. Mus. r. Hist. nat. Belg. 12 (29), York. I -7. GARY L MILLER Department of Zoology Weber State College Ogden, Utah 84408, USA SIDNEY R. ASH Typescript received 9 December 1986 Revised typescript received 21 May 1987 Department of Geology Weber State College Ogden, Utah 84408, USA CHANGES IN LIFE ORIENTATION DURING THE ONTOGENY OF SOME HETEROMORPH AMMONOIDS by TAKASHI OKAMOTO Abstract. To understand the mode oflife of Eubostrychoceras muramotoi Matsumoto, 1967, and some other heteromorph ammonoids from the Upper Cretaceous of Hokkaido, Japan, their life orientation and growth patterns were restored using a hydrostatic model and differential geometry. The adequacy of these restorations was tested using the obliquity of ribs and its change during ontogeny. Rib obliquity parallels the aperture at every growth stage and corresponds well to the inferred changes oflife orientation. The pattern of rib obliquity was also deduced from a computer simulation, programmed so that growth occurs only at the aperture which keeps a constant angle to the sea floor. In most heteromorphs, the computer-produced profiles approximate well to actual rib patterns. Although the rib obliquity of these ammonoids may appear to change somewhat capriciously, it must be functionally regulated. Except during the early orthoconic stage of growth, a floating or lightly touching benthonic mode of life is strongly suggested. T he ammonoid shell is composed of a phragmocone and a living chamber. During life, the average density of a living chamber filled with soft parts was greater than that of sea water, while the phragmocone was buoyant due to the gas it contained. If an ammonoid could float like extant Nautilus, the average density of the whole animal must have been approximately equal to the density of sea water. On this basis, Trueman (1941) estimated the buoyancy potential of some ammonoids; similar assumptions were made in the estimation of total average densities of various ammonoids by Reyment (1958, 1973), Heptonstall (1970), Tanabe (1975, 1977), and Ward and Westermann (1977). This type of hydrostatic model enables further inferences to be made about life orientation. When an ammonoid floated in sea water, buoyancy and gravity must have balanced and their respective centres must have lain on a vertical line. So if these centres are estimated in an actual ammonoid, its life orientation can be determined. Trueman (1941) first illustrated such living attitudes in both normally coiled and heteromorph ammonoids. Further developing this theory, Raup (1967) and Saunders and Shapiro (1986) restored the life orientation of some normally coiled ammonoids and discussed their functional morphology. However, the change of life orientation during the ontogeny of heteromorph ammonoids has been little studied (probably because their complicated coiling was difficult to model). Okamoto (1988) recently proposed a ‘growing tube model’ to reconstruct the three-dimensional coiling of heteromorph ammonoids. This model involves a modification of the Frenet frame (moving frame), an established method of differential geometry and useful for the analysis and description of any regular space curve. Computer simulations which combine Trueman’s concept and the growing tube model can aid our understanding of the life orientations of heteromorph ammonoids. Previous discussions of buoyancy and life orientation seem to have depended on speculative calculations, and the results have not been tested using independent evidence. The complicated but regular coiling patterns of most heteromorph ammonoids reveal more about life orientation and its change during ontogeny than planispiral cephalopods. (Palaeontology, Vol. 31, Part 2, pp. 281 -294.| © The Palaeontological Association 282 PALAEONTOLOGY, VOLUME 31 MATERIAL AND METHODS Abundant, well-preserved heteromorph ammonoids occur in the Upper Cretaceous of Hokkaido, Japan, with certain nostoceratids being particularly known for their curious mode of coiling. The coiling geometry and life orientation of several such species of Nostoceratidae are discussed below, especially that of the Coniacian Eubostrychoceras muramotoi Matsumoto, 1967. The repositories of the specimens examined are: UMUT, University Museum, University of Tokyo; GK, Department of Geology, Kyushu University; WEA, Institute of Earth Science, Waseda University; KPMG, Kanagawa Prefectural Museum. Eight specimens of E. muramotoi from the Obira area, Hokkaido, were available. The study of these and other previously described specimens reveals the following characters: 1. Coiling pattern. Matsumoto (1967) described the coiling pattern of the holotype (GK.H5589; text-fig. Id) as follows: The earliest shell is nearly straight, ascending and then followed by a subcircularly curved half whorl, which in turn is twisted obliquely down, passing to the main helical whorls, whose axis of coiling is along the earliest straight shaft’. The same mode of coiling is apparent in another nearly complete specimen (WEA 003T-1), where although the earliest stage is concealed by later helicoid whorls, the direction of the orthoconic shaft is inferred to be the coiling axis of the whorls (text-lig. 1a, b). In this sample, six specimens are dextrally coiled like the holotype (text-fig. Id) and two are sinistrally coiled (text-fig. 1c). text-fig. 1. Eubostrychoceras muramotoi Matsumoto. A, WEA 003T-1, from Obirashibe River (near mouth of Okufutamata-zawa tributary), Obira area; Coniacian; lateral view, x 1-5. b, apical view of the same specimen, x E5. c, WEA 004T, from the same locality; lateral view of sinistral specimen, x I.d, GK.H5589 from Pombetsu, Go-no-sawa, Ikushumbetsu area; Coniacian; upper view of holotype, x 2. e, WEA 003T-2, from the same locality as specimen in a; SEM photograph indicating conformable relation between ribs and growth lines, x 25. OKAMOTO: LIFE ORIENTATION OF HETEROMORPH AMMONOIDS 283 2. Change of rib obliquity. Oblique and gently flexuous ribs are regularly developed on the whorls. In the helicoid stage, there are about forty ribs per whorl. Matsumoto (1967) noted that a drastic change of rib obliquity occurs in the middle growth stage, one and a half or two whorls after the shell's turning-point. The change occurs in every specimen studied (text-fig. 1a, c). 3. Growth lines. Fine striae occur between the ribs (text-fig. 1 e). They run parallel to the ribs at every stage, and appear to be growth increments. The rib obliquity, therefore, seems to indicate the shape of the aperture at each growth stage. 4. Living chamber. The length of the living chamber in WEA 003T-1 appears to occupy about one and a quarter whorls of the helicoid stage. The length of the living chamber may be 40-50% of the total cone length, but because it is always more or less deformed or damaged, the precise estimation of this ratio is difficult. 5. Whorl cross-section. The shape of the whorl cross-section is almost circular. Shell thickness is about 2 % of whorl radius; this value is very small, but the shell seems to be reinforced by sharp ribs which occur regularly on the whorls. COILING GEOMETRY Expression of coiling form Raup (1966) proposed a generalized growth model to express the shell form of most gastropods, cephalopods, and bivalves. Heteromorph coiling forms, however, cannot be described completely by his model. Okamoto (1984) succeeded in modelling Nipponites (a Cretaceous heteromorph ammonoid) by regarding the shell as a meandering tube with circular cross-section. In this model, the locus of the tube’s centre R and the corresponding tube radius r define a tubular body: the tube’s surface is expressed by U = R+r and r R =0, where R is the differential of R and indicates the growth direction of the tube. Because normal cross-sections of the shell of E. muramotoi can also be treated as circles, it can be modelled similarly. In its early growth the shell of E. muramotoi approximates to an orthoconic tube. After a sharp bend, the shell coils helically around the earlier orthoconic shaft, and the whorls become almost isometrically helicoid. Consider now a cylindrical coordinate system in which the locus of the tube’s centre line is expressed by a parameter i. The direction of increase of i is the tube’s growth direction. Its projection on the X — Y plane, which is perpendicular to the coiling axis, is approximately an equiangular spiral (text-fig. 1b). If D is the distance between the coiling axis and the tube’s centre line, D increases simply with growth and may be expressed as an exponential function of i: D = I0a, + ao where a and a0 represent the whorl expansion rate and its initial value of D respectively. If Z represents the displacement of the tube’s centre line along the coiling axis, the tube's centre at every stage sits on one particular hyperbola (text-fig. 2a) and its trace follows this hyperbola’s revolution around an asymptotic line (text-fig. 2b). When the two asymptotic lines of the hyperbola are expressed as D = 0 (coiling axis) and Z = — 6D, Z is expressed as a function of D: cl Z = -bD-\ — D where the coefficient b is related to the apical angle of the specimen, and the coefficient d expresses the ‘pointedness’ near the whorl’s turning-point. If the revolution angle is nn , the angular velocity An/A/ is not constant: it is almost zero in the earliest orthoconic stage but maintains a constant value in the later helicoid stage. Therefore, the relationship between n and / can also be expressed as a hyperbola: i — n f (n > 0) n 284 PALAEONTOLOGY, VOLUME 31 where the coefficient /indicates the coordinates of i where the two asymptotic lines (i = 0, i — n—f ) intersect together. Finally, the whorl radius r is defined as a function of the length / of the tube centre line. If the radius increases isometrically relative to whorl length, then 2 r = cl, where coefficient c is the relative growth ratio of tube diameter to tube length. Therefore, four equations are required to express the shape of £. muramotoi , with six coefficients: a, a0, b , c, d , and / The values of a (whorl expansion rate), b (apical angle), and c (radius enlarging rate) were obtained directly (from WEA 003T-1), but the relationship between the other coefficients and shell shape is not so obvious. Approximate values of these latter coefficients were estimated empirically so as to fit the model most closely to the actual coiling pattern. The following values for the six coefficients of the hyperbolic model were determined: a — 0 075, a0 — 14, b = 2-97, c = 0 051, d = 4, and /= 2. This model can be applied to other heteromorph ammonoids, such as Muramotoceras yezoense Matsumoto and Ainoceras kamuy Matsumoto and Kanie; and, although not suitable for growth simulation, it is useful for the description of their shell form. text-fig. 2. Eubostrychoceras muramotoi Matsumoto; coiling pat- tern and its model, a, hypothetical whorl cross-section, b, revolving hyperbola model, indicating the whorl centre line. Analysis of growth Okamoto (1988) proposed a ‘growing tube model’ for analysing the growth pattern of any coiled shell. In this model, a coiled shell with circular whorl section can be generally described by three differential parameters (op. cit., text-fig. 7): £, radius enlarging rate; C, standardized curvature; and T, standardized torsion. Each parameter changes with a growth stage parameter s. Using this method, any heteromorph ammonoid can be uniquely described on a plane diagram involving the three functions of growth stage, £(s), C(s), and T{s). In practice, £, C, and T for £. muramotoi can be calculated by invoking the hyperbolic model described above, but modified in two ways to make it correspond more closely to real specimens (Okamoto 1988): 1, a higher £ value is estimated for the earlier growth stage; and 2, C is controlled so as not to exceed the theoretical limit (C = 1) throughout the growth. The growth pattern of this species can be divided into two stable stages (orthoconic and helicoid) with an intervening transitional interval (the turning-point). The sudden change of coiling mode at this transitional interval may indicate a change in mode of life. OKAMOTO: LIFE ORIENTATION OF HETEROMORPH AMMONOIDS 285 LIVING ATTITUDE Estimation of buoyancy potential If the particular growth pattern (£(s), C(s), T{s)) for a specimen is given, various physical characters (e.g. tube length, surface area, volume, centre of gravity, etc.) can be calculated from the growing tube model. Trueman (1941) argued for the potential buoyancy of some ammonoids by estimating the volume and density of phragmocone and living chamber. Raup and Chamberlain (1967) and Heptonstall (1970) also referred to the estimation of buoyancy potential. Tanabe (1975, 1977) and Ward and Westermann (1977) applied a similar method to the buoyancy potential of heteromorph ammonoids and discussed their functional morphology. Here I evaluate the adequacy of current methods for the estimation of ammonoid buoyancy. To estimate the total density of E. muramotoi, several density values are assumed (text-fig. 3a) for a simplified conical shell form. I adopt the densities of shell and soft parts estimated by Reyment (1958) and Denton and Gilpin-Brown (1966) using living Nautilus. Secondly, the following five values were measured or estimated from the actual specimen or its growing tube model: the ratio of phragmocone length to total cone length; diameter of aperture; whorl diameter at the phragmocone’s termination; shell thickness; and volume of septa. (Though shell thickness is taken to be c. 2 % of whorl radius, a somewhat higher value may be appropriate, because the volume of ribs should be taken into consideration.) The calculation gives a total density for this specimen of 1017, which is less than that of sea water, 1 026 (text-fig. 3a). However, the calculated total density is quite unreliable because it may be seriously influenced by Sea water density Air density Volume of septa Shell density Tissue density Total density ( 1 026 9/cm 3 I l 0 9 /cm I ( 4 '/. of phrag volume ) I 2. 62 g/cm3 ) ( 1 067 g/cm 5 I 1 017 9 /cm* B ~~ ___Tgtal density Measurements — _ 100 l 026 1 05 Phragmocone length b_ a 55”/. i 5 Diameter at aperture c 10.5 mm ± 1.1 Diameter at phragmo - cone terminus d 18 mm t 0.2 H h => Shell thickness e c 3 7. i 1 i 1 Volume of septa \ 4 7^ =+= Tissue density \ 1.067 g/cm3 Shell density \ 2.62 , 9/cm text-fig. 3. Calculation and error estimation of total density of Eubostrychoceras muramotoi Matsumoto, WEA 003T-1. A, estimated physical values, b, estimated error range for each value. Black rectangles show the fluctuation range of total average density when each value is 5% over- or underestimated. White rectangles show the fluctuation range as estimated from each measuring error. 286 PALAEONTOLOGY, VOLUME 31 slight errors in each estimated value. The confidence interval of the calculated total density is shown in text-fig. 3b. Black rectangles show the fluctuation range of total average density when each value is 5 % over- or understimated. Some error bars extend beyond the critical line of 1-026. While the value assumed for tissue density is unlikely to be as much as 5 % incorrect, that of the ratio of phragmocone length to total cone length and shell thickness cannot be estimated to within a 5 % margin of error. If a multiplication of these errors occurs, the confidence interval of calculated total average density must increase seriously. It is almost impossible to discuss sensibly the buoyancy regulatory ability of E. muramotoi from the result of this calculation. Life orientation Hydrostatic models for the life orientation of floating ammonoids were discussed by Trueman (1941), Raup (1967), Raup and Chamberlain (1967), and Saunders and Shapiro (1986). Trueman (1941) intuitively estimated the life orientation of some heteromorph ammonoids from the relationship between buoyancy and gravity. Klinger (1981) divided some heteromorph ammonoids into four groups, and discussed their mode of life from the standpoint of possible buoyancy control. Ebel (1985) argued that the hydrostatics of some ammonoids favoured a gastropod-like mode of life. However, little is known of the life orientation of heteromorph ammonoids from a hydrostatic standpoint. This is probably because no geometrical model had been devised which was capable of accurately describing heteromorph forms. Application of the growing tube model enables various physical properties to be calculated easily by microcomputer. Reconstructions of the life orientation of ammonoids based on this model require three assumptions to be made: 1, the floating position during life can be restored by assuming that the buoyancy of the phragmocone was just balanced by the weight of the living chamber in sea water. 2, for simplicity, the phragmocone and living chamber of ammonoids are regarded as being composed of homogeneous materials. 3, the ratio of phragmocone length to living chamber length was constant throughout growth; there is no certainty that the ratio was invariable in actual specimens, but the assumption is reasonable as a first approximation because the same total density was maintained at every growth stage if shell radius increased isometrically with shell length. If the centre of gravity of the phragmocone G{, and that of living chamber G2 are calculated at a given growth stage, the vector from Gj to G2 must be directed vertically. The centres of gravity can be determined in the following way. First, the cone is separated into many ‘slices’ by considering normal cross-sections parallel to the generating curve. The centre of gravity of each slice must be situated at the distance Cr/4 away from the slice centre in a direction towards the maximum growth point of the whorl surface, where C and r are the standardized curvature and the radius of the generating curve, respectively. The centres of gravity for the phragmocone and the living chamber are separately determined by integrations of the volume and the centre of gravity about each slice. If the centre of gravity of each slice and its volume are given by ( gx , gy, gz) and F, respectively, the centres of gravity for the two parts can be calculated as follows: In practice, both centres of gravity can be estimated by microcomputer through integration of numerous slices, each r/ 10 in length. Phragmocones and living chambers of real ammonoids were not, in fact, homogeneous: heavier shell formed the ‘outside’ while the ‘inside’ was filled with lighter materials, such as air, cameral liquid, and soft parts. Consider the extreme example of a hollow cone in which the mass is concentrated on its surface. In this case, the centre of gravity of the slice must be situated at a point Cr/2 away from the slice centre, towards the maximum growth point on the surface; the centre of gravity shifts Cr/4 further OKAMOTO: LIFE ORIENTATION OF HETEROMORPH AMMONOIDS 287 from its position when the shell is assumed to be homogeneous. Estimation of the vertical direction becomes somewhat more complicated, requiring knowledge of both the centre of gravity of the total mass and the centre of buoyancy. The former cannot be calculated without making assumptions about density. However, since the difference in results between the two methods is small enough not to seriously influence the vertical vector, the phragmocone and living chamber are both regarded here as effectively homogeneous. text-fig. 4. Stereographic projection on Wolff’s net (lower hemisphere) showing ontogenetic change of the vertical vector relative to the coiling axis for Eubostrycho- ceras muramotoi Matsumoto. Vertical vector is defined as direction from centre of buoyancy to centre of gravity; coiling axis is fixed at centre of diagram. The great merit of this method is that life orientation can be restored without making questionable assumptions concerning shell thickness and density of soft parts. In this hydrostatic model, only the growth pattern of the shell and one constant (ratio of phragmocone length to total cone length) are necessary. For E. muramotoi the former is estimated by moving frame analysis (text-fig. 5a; Okamoto 1988) and the latter measured at about 0 55 from a real specimen. From such data, the direction and magnitude of the vertical vector can be calculated at various growth stages. The change of direction of the vertical vector during ontogeny can be projected on a Wolff’s net (text-fig. 4). As the result of this simulation, the direction of the vertical vector is seen to turn over abruptly at stages 30-32. The magnitude of the vertical vector represents a kind of ’stability index1, because it measures distance between the two centres of gravity. Although this value generally increased with growth, it reached a minimum when the living attitude flipped over. RESULTS OF SIMULATIONS Simulation of Eubostrychoceras muramotoi In E. muramotoi , both the growth pattern obtained from moving frame analysis and the results of simulating life orientation are shown in text-fig. 5. There is a transitional interval in the growth pattern when the mode of coiling changes drastically, seemingly indicating a change in mode of life. The Z (coiling axis) component of gravitational force, standardized with a unit vector, is shown in text-fig. 5b. The rapid turnover of living attitude (stages 30-32) occurs after the initiation of helical coiling (stage 19). At turnover, the magnitude of the vertical vector is minimized (text-fig. 5c), and the living attitude may be unstable. Using the result of text-fig. 4, theoretical life orientations can be reconstructed for several growth stages by computer graphics (text-fig. 6). The turnover of life orientation occurs one and a half or two whorls after the turning-point in shell growth. Interestingly, though the mode of coiling does not change, an abrupt change in rib obliquity occurs just at this stage in real specimens (text-fig. I). 288 PALAEONTOLOGY, VOLUME 31 text-fig. 5. Diagrammatic figures showing change of some geometric and physical values during ontogeny of Eubostrychoceras muramotoi Matsumoto. a, moving frame analysis. B, Z component of unit vertical vector, c, length of vertical vector, a kind of stability index (see text). Simulation of other heteromorph ammonoids Okamoto (1988) also described the coiling patterns of several other heteromorph species by moving frame analysis, and the changes in their life orientation during ontogeny can be simulated by the same method. 1. Ainoceras kamuy Matsumoto and Kanie, 1967. The coiling pattern of A. kamuy in its early-middle growth stage resembles that of E. muramotoi (text-fig. 7a, b), and a similar change of life orientation is suggested (text-fig. 8a). Moreover, an analogous change of rib obliquity occurs at the corresponding stage (text-fig. 7b). After its helical coiling phase, this species goes through another significant change of coiling pattern, forming a loose and nearly planispiral whorl in later life. Life orientation also appears to change at the beginning of this last stage (text-fig. 8a), and it is interesting to note that rib obliquity changes from prorsiradiate to rectiradiate at this point. 2. M. yezoense Matsumoto, 1977. This species (text-fig. 7c) also has a similar coiling pattern to E. muramotoi. However, the loosely coiled helicoid shell, with its larger apical angle, suggests a more gradual and incomplete turnover of life orientation (text-fig. 8b). Rib obliquity also changes gradually at the corresponding growth stage (text-fig. 7c). 3. E. japonicum (Yabe, 1904). This species shows a different coiling pattern from the foregoing species. The shell exhibits crioceratoid coiling in early growth, then forms an open helicoid spiral OKAMOTO: LIFE ORIENTATION OF HETEROMORPH AMMONOIDS 289 text-fig. 6. Computer graphics showing inferred change of life orientation during ontogeny of Eubostrychoceras muramotoi Matsumoto. The turnover of life orientation occurs at one and a half or two whorls after the shell’s turning-point. (torticone) in its middle stages (text-fig. Id). Presumably some change of life orientation occurred between the two stages; rib orientation also changes from rectiradiate to prorsiradiate (text-fig. 8c). Rib obliquity seems to be influenced by life orientation. As stated above, the growth lines of E. muramotoi run parallel to the ribs, and rib obliquity indicates the shape of the aperture at every growth stage. This is also the case in other heteromorph species. It is supposed, as a working hypothesis, that these floating ammonoids grew oblique apertures to maintain a constant angle with the vertical vector. FURTHER OPERATIONS Simulation of rib obliquity On the basis of the above working hypothesis, a theoretical rib pattern can be simulated from the estimated life orientation. Rib shape and obliquity can be determined as follows (see also text-fig. 9). First, the ‘bottom margin’ and ‘top margin’ are defined on the generating curve of the growing tube model. When the growing tube model is oriented in life position, the bottom margin indicates the lowest point on the generating curve, and the top margin indicates the opposite point to it. The generating curve is then transformed into the ‘aperture curve’, which is generally elliptical in shape. In this transformation, each point on the generating curve moves in a progressive or regressive direction, and the point on the bottom margin or top margin shifts the furthest. By this procedure the apertural plane is determined at every growth stage, so as to keep a constant angle to the vertical vector. Text-fig. 10a shows a computer-produced profile of the rib pattern of E. muramotoi with the aperture angle set at 40° (determined from a real specimen, at the last whorl in this case). The theoretically produced profile of E. muramotoi is quite similar to the real specimen, in both its general outline and the abrupt change of rib obliquity in the early helicoid stage. The patterns of rib obliquity of the three other nostoceratids studied were simulated by the same method (text-fig. 290 PALAEONTOLOGY, VOLUME 31 text-fig. 7. Nostoceratids from central Hokkaido, Japan, showing characteristic change of rib obliquity, a, Ainoceras kamuy Matsumoto and Kanie, UMUT MM 17972, from the southern tributary of Rubeshibe-zawa, Saku area; Campanian; early growth stage, x 8. b, same species, GK.H5577, from the third tributary of the Nio-no-sawa, Saku area; Campanian, x 3. c, Muramotoceras yezoense Matsumoto, WEA 001Y, from Pankemoyuparozawa, Oyubari area; Turonian, x 1 -5. d, Eubostrychoceras japonicum( Yabe), KPMG 6373, from upper stream of Kotambetsu River, Kotambetsu area; Turonian, x 2. 10b, c, d). Each theoretical profile is sufficiently similar to the actual rib pattern for the hypothesis about the relationship between life orientation and rib obliquity to be generally accepted. So although the rib pattern of some nostoceratid ammonoids may appear to change somewhat capriciously, it is functionally regulated. In ordinary planispiral ammonoids, constant life orientation is supposed throughout growth because they maintained a geometrically similar shell form. Theoretical rib ob- liquity must also be constant; in fact, specimens of many normally coiled ammonoids exhibit an OKAMOTO: LIFE ORIENTATION OF HETF.ROMORPH AMMONOIDS 291 text-fig. 8. Stereographic projections on Wolff’s net showing the change of vertical vector relative to coiling axis during ontogeny; computer graphics show life orientation at each growth stage, a, Ainoceras kamuy Matsumoto and Kanie. B, Muramotoceras yezoense Matsumoto. c, Eubostrychoceras japonicum (Yabe). The phragmocone length/total cone length ratio in these three species is taken as 0 6, 0 5, and 0 7 respectively. 292 PALAEONTOLOGY, VOLUME 31 Top margin text-fig. 9. Transformation of generating curve of growing tube model to ‘aperture curve’. Each point on the generating curve moves in a progressive or regressive direction, so as to maintain a constant aperture angle. almost constant rib obliquity. This is compatible with the hypothesis of constant apertural angle, but the presumed life orientation is hardly testable in this case. However, it is uncertain whether or not such regularity is always maintained by heteromorph ammonoids. For example, rib obliquity in the orthoconic early stage of E. muramotoi cannot be explained by this model. The rib pattern of Polyptychoceras sp., which is prorsiradiate in the early orthoconic stage and nearly rectiradiate after the first U-turn, is also difficult to simulate. If actual rib obliquity does not conform to the theoretical rib pattern, one or more assumptions in the model may be wrong. Two possible interpretations are: 1, the estimation of life orientation is inadequate because the real mode of life was quite different (e.g. purely benthonic); or 2, the aperture does not always maintain a constant angle to the vertical vector. These exceptional cases are important for our understanding of the general mode of life of heteromorph ammonoids, and require further study. C text-fig. 10. Computer-graphics showing the theoretical rib obliquity with constant aperture angle, a, Eubostrychoceras mura- motoi Matsumoto. b, Ainoceras kamuy Matsumoto and Kanie. c, Muramotoceras yezoense Matsumoto. d, E.japonicum (Yabe). The aperture angles of these species are set at 40°, 60°, 40°, and 40° respectively. OKAMOTO: LIFE ORIENTATION OF HETEROMORPH AMMONOIDS 293 text-fig. 1 1. Possible life orientations of the simple orthoconic model. When buoyancy F1; gravity F 2, distance between the contact point and centre of buoyancy a, and distance between the contact point and centre of gravity b are given as in a, the three possible attitudes are determined by the interaction of two moments, as shown in b-d. e shows floating attitude. a Fi < bF2 a F,= b F2 a Fi> bF2 Fi~F2 Sinking attitude I have failed to obtain any direct evidence for a floating mode of life in nostoceratid ammonoids from the calculation of total density. However, reconstruction of a life orientation which cross-checks with observations of actual rib patterns is significant. But does this result really prove that heteromorph ammonoids floated during life? Let us examine whether a benthonic mode of life was possible. For simplicity, consider the sinking attitude of an orthoconic ammonoid (text-fig. 1 1a), where the centre of buoyancy, centre of gravity, and contact point on the bottom must lie along a straight line. When buoyancy F1, weight F2 , distance between contact point and centre of buoyancy n, and distance between contact point and centre of gravity b are given, the final attitude is determined by the interaction of two moments and bF2 . Three attitudes are possible: I, if the load acting on the contact point is sufficiently large (aFl < bF2 ), the shell lies down on the bottom; 2, if the load is below the limit (aF{ > bF2), the shell shows an upright posture equivalent to the floating state; or 3, if the two moments balance each other (uFj = £>F2), the shell assumes an oblique posture which is quite unstable (text-fig. 1 1). Generally, the sinking posture of an ammonoid shell is determined by the relationship between the centre of buoyancy, centre of gravity, contact point, and forces acting on these points. Actually an oblique posture may be possible when these three points are not situated on a straight line. But, if the load acting on the contact point were sufficiently small, the shell would take almost the same orientation as when floating. Therefore, even though the changing pattern of rib obliquity is satis- factorily explained by a floating mode of life, it is not necessarily proved ; text-fig. 1 1 o, E show the two possible life orientations for the main growth stages of those heteromorph ammonoids studied here. SUMMARY AND CONCLUSIONS This investigation of theoretically and hydrostatically possible life orientations for several nosto- ceratid ammonoids has shown that: 1, their shell form can be described by a revolving hyperbolic model. 2, by moving frame analysis, the growth pattern of F. muramotoi is separable into two stable stages, with a transitional interval indicating some change in mode of life. 3, the calculation of total density of ammonoids inevitably involves considerable error, and traditional methods involving speculative densities for various portions should not be used to estimate the buoyancy of heteromorph ammonoids. 4, changes of life orientation during ontogeny can be estimated by computer simulation, with a hydrostatic model; the turnover of life orientation corresponds well to the change of rib obliquity on real specimens of F. muramotoi and the three other heteromorph ammonoids studied. 5, these nostoceratids either floated freely or lightly touched the sea bottom. 6, nostoceratids with such a mode of life seem to have maintained their apertures at a constant angle to the vertical vector. 294 PALAEONTOLOGY, VOLUME 31 It is important to emphasize that the adequacy of this hydrostatic model was tested independently, i.e. by the change of rib obliquity during ontogeny. A constancy of apertural angle to the vertical was probably also maintained by ordinary planispiral ammonoids, but any resulting inference concerning their life position suffers from circular reasoning. The three-dimensional and complicated morpho- logy of heteromorph ammonoids promises to be a rich source of enquiry for the solution of various general problems of functional morphology in ectocochlian chambered cephalopods. Acknowledgements. I thank Itaru Hayami (University of Tokyo) for his critical reading of the manuscript, and Kiyotaka Chinzei (Kyoto University), David M. Raup (University of Chicago), Kazushige Tanabe (University of Tokyo), and Richard D. Norris (Harvard University) for their valuable suggestions. I am indebted to Tatsuro Matsunroto (Kyushu University), Hiromichi Hirano (Waseda University), and Yoshiaki Matsushima (Kana- gawa Prefectural Museum) who kindly lent me valuable specimens. I am grateful to Tatsuo Oji (University of Tokyo) and Haruyoshi Maeda (Kochi University) for their constructive discussion during my laboratory work, and Tomoko Yamashita for operating a SEM in the University Museum, University of Tokyo. REFERENCES denton, E. F. and gilpin-brown, j. B. 1966. On the buoyancy of the pearly Nautilus. J. mar. biol. Tss. U K. 46, 723-759. ebel, k. 1985. Gehausespirale und Septenform bei Ammoniten unter der Anname vagil benthischer Lebenweise. Paldont. Z. 59, 109-123. heptonstall, w. 1970. Buoyancy control in ammonoids. Lethaia , 3, 317-328. klinger, h. c. 1981. Speculations on buoyancy control and ecology in some heteromorph ammonites. Pp. 337-355. In house, m. r. and senior, j. r. (eds.). The ammonoidea. Spec. Vol. Syst. Tss. 18, 593 pp. matsumoto, t. 1967. Evolution of the Nostoceratidae (Cretaceous heteromorph ammonoids). Mem. Fac. Sci. Kyushu Univ. Ser. D, Geol. 18, 331-347, pis. 18 and 19. 1977. Some heteromorph ammonites from the Cretaceous of Hokkaido. Ibid. 23, 303-366, pis. 43-61. and kanie, Y. 1967. Ainoceras , a new heteromorph ammonoid genus from the Upper Cretaceous of Hokkaido. Ibid. 18, 349-359, pis. 20 and 21. okamoto, T. 1984. Theoretical morphology of Nipponites(a. heteromorph ammonoid). Kaseki (Fossils). Palaeont. Soc. Japan , 36, 37-51, pi. 1. [In Japanese.] 1988. Analysis of heteromorph ammonoids by differential geometry. Palaeontology, 31, 35-52, pi. 7. raup, d. m. 1966. Geometric analysis of shell coiling: general problems. J. Palaeont. 40, 1178-1190. — 1967. Geometric analysis of shell coiling: coiling in ammonoids. Ibid. 41, 43-65. and chamberlain, j. a. 1967. Equations for volume and center of gravity in ammonoid shells. Ibid. 566-574. reyment, R. A. 1958. Some factors in the distribution of fossil cephalopods. Stockh. Contr. Geol. 1, 97-184. 1973. Factors in the distribution of fossil cephalopods. Part 3: Experiments with exact models of certain shell types. Bull. geol. Instn Univ. Uppsala, NS, 4, 7-41. saunders, w. b. and shapiro, E. A. 1986. Calculation and simulation of ammonoid hydrostatics. Paleobiology, 12, 64-79. tanabe, k. 1975. Functional morphology of Otoscaphites puerculus (Jimbo), an Upper Cretaceous ammonite. Trans. Proc. palaeont. Soc. Japan, NS, 99, 109-132, pis. 10 and 1 1. 1977. Functional evolution of Otoscaphites puerculus (Jimbo) and Scaphites planus (Yabe), Upper Cretaceous Ammonites. Mem. Fac. Sci. Kyushu Univ. Ser. D. Geol. 23, 367-407, pis. 62-64. trueman, a. e. 1941 . The ammonite body chamber, with special reference to the buoyancy and mode of life of the living ammonite. Q. Jl geol. Soc. Loud. 96, 339-383. ward, p. d. and westermann, G. e. G. 1977. First occurrence, systematics, and functional morphology of Nipponites (Cretaceous Lytoceratina) from the Americas. J. Paleont. 51, 367-372, pi. 1. yabe, h. 1904. Cretaceous Cephalopoda from the Hokkaido. Part 2. J. Coll. Sci. imp. Univ. Tokyo, 20, 1-45, pis. 1-6. Typescript received 10 September 1986 Revised typescript received 7 August 1987 TAKASHI OKAMOTO Geological Institute University of Tokyo Tokyo 1 13, Japan MIDDLE JURASSIC AMMONITES OF TIBET AND THE AGE OF THE LOWER SPITI SHALES by G. E. G. WESTERMANN and WANG YI-GANG Abstract. Middle Jurassic ammonites of China are known only from Tibet. Eighteen localities are described, some new. Only Lower Bajocian and Lower Middle Callovian are established, whereas the other Middle Jurassic stages or substages are usually represented by hiati or non-marine facies. Two major ammonite faunas are distinguished: (1) Witchellia Fontannesia Association, including Fontannesia ki/iani [‘ Dorsetensia auct.], Laeviuscula Zone, known from the central part of Tethyan Himalaya and with strong affinity to northern and western Australasia of the extreme south-east Tethys; (2) ‘ Grayiceras ’ Association, including Grayicerasl gucuoi n. sp., late Calloviense Zone ( Subkossmatia opis Assemblage Zone of Kachchh), possibly widely distributed in the basal Spiti Shales (Belemnposis gerardi Beds) and equivalents of Tethyan Himalaya, to which it appears to be largely endemic, together with rare Indo-Madagascan (Ethiopian) and Mediterranean elements. In addition, the Early Bajocian Discites and ?Sauzei Humphriesianum Zones, and the Middle Callovian Coronatum Zone are indicated locally. The ‘ Grayiceras ’ fauna was dated as Oxfordian by most previous authors, but the stage is missing at least locally in South Tibet, where Kimmeridgian lies conformably on Lower Callovian. Biogeographic affinities support an origin of the Tethyan Himalaya from the Gondwana margin, not too distant from northern Australasia, whereas the limited North Tibetan faunas are consistent with a Eurasian position. Middle Jurassic marine strata and ammonites of China are more widespread than those of the Early or Late Jurassic. Their study began only recently (Zhao 1976), except for Arkell’s (1953) record of a small Middle Bajocian ammonite fauna collected by H. Hayden at Mekyigunru, Gamba county in the Tethyan Himalaya of South Tibet (our loc. 1; text-fig. 1). Subsequently, Wang and Chen (in Wang et a!. 1979) reported several Middle Jurassic ammonite species from North Tibet, and Yang (unpublished) recognized several Callovian specimens from South Tibet. The genera and species recorded in these papers are here revised, despite the poor and incomplete preservation of the fossils, and recent stratigraphic data about the ammonite-bearing localities are discussed. Some of the ammonite collections, however, lack strict stratigraphic control, while others which are stratigraphically controlled remain inadequate because of their small sample size and poor preservation. In 1985, we re-examined two important sections (Iocs. 2 and 13) and made additional fossil collections with good stratigraphic control. Since plate-tectonic theory holds that the Tethyan Himalaya is part of the Indian Plate of Gondwana, whereas North Tibet was part of Eurasia in the Jurassic, the present data, from both areas, are a significant contribution to the palaeogeography and ammonite biogeography of the Middle Jurassic in western China and central Asia. In the classic Spiti -Niti area in the western Himalaya (see text-fig. 1), the Lower Spiti Shales or Belemnop sis gerardi Beds, and the Ferruginous Oolite have been the subject of recent investigations. Based on new ammonite evidence, the basal oolites representing the transgressive phase belong to different parts of the Callovian with strong lateral diachroneity (Jadoul el al. 1985). Latex casts of this small fauna have been made available to Westermann by M. Gaetani while this paper was in press. The ‘ Pachyceras sp. ind.’, the only evidence for alleged Upper Callovian is a Middle Callovian Erymnoceras (confirmed by A. Zeiss); the ‘ Macrocephalites sp. ind.’ is a Grayicerasl waageni (Uhlig), of top Lower Callovian age. The B. gerardi Beds yielded no new ammonite fauna. They were dated as Oxfordian-Kimmeridgian mainly by stratigraphic interpolation (Jai Krishna and Singh 1982) and by correlation with Indonesia, based on specifically unidentified belemnites and inoceramids. This homeotaxis, however, has turned out to be controlled by biofacies rather (Palaeontology, Vol. 31, Part 2, 1988, pp. 295-339, pis. 20-25] ©The Palaeontological Association 296 PALAEONTOLOGY, VOLUME 31 text-fig. 1. Middle Jurassic ammonite localities and principal tectonic subdivision of Tibet. The Bangong Co-Nujiang (Tanggula) Deep Fracture Zone, with the Jurassic Ophiolite Flysh Belt (solid black), separates North from South Tibet; the Yarlung Zangbo Deep Fracture Zone (continuation of Indus Suture) with the Cretaceous Ophiolite Flysh Belt, divides South Tibet from the Tethyan Himalaya. See also Table 1 . than the biozones of these long-ranging taxa. The only known ammonites from the B. gerardi Beds appear to be the few specimens from the Spiti area described and discussed long ago by Uhlig ( 1 903- 1910). They have usually been identified and/or correlated with the Oxfordian mayaitid assemblage of Indonesia and the Indo-Madagascan (Ethiopian) Province (Uhlig 19106; Spath 1925, 1927-1933; Arkell 1956). The other fauna of the Ferruginous Oolite has not been described or illustrated so that even generic identifications, sometimes made by non-specialists, cannot be trusted. This includes the supposed macrocephalitids which could be misidentified ‘Grayiceras' . Similar faunas were recorded from the central Himalaya of Nepal (Bordet et al. 1971). Here the basal sandstones and/or ferruginous oolites are said to contain Callovian ammonites, and the Lower Spiti Shales may be developed in ammonite-bearing facies, yielding what appears to be the first good evidence of the Oxfordian, i.e. Perisphinctes spp. found loose. WESTERMANN AND WANG: JURASSIC AMMONITES FROM TIBET 297 The Menkatun Formation of the South Tibetan Tethyan Himalaya is clearly the continuation of the Spiti Shales (and thus a superfluous term in the senior author’s opinion) and, similarly, has at its base the Ferruginous Oolite which we would prefer to consider as a member of the ‘Menkatun’/Spiti Shales Formation. Here the basal part of the shales is developed in ammonite- bearing facies containing the ‘ Grayiceras ’ Assemblage here described, followed by poorly fossil- iferous beds yielding the rare Kimmeridgian bivalve Australobuchia spitiensis. Most of the Callovian and the entire Oxfordian are missing. THE MIDDLE JURASSIC GEOLOGY OF WESTERN CHINA AND PLATE TECTONICS In the conventional ‘fixist’ tectonic view, the Jurassic seas of China were evidently extensions of Tethys and Panthalassa, the ancient Pacific Ocean. The vast area of the Tibet-Qinghai Plateau (text-fig. 1), the most important Jurassic marine basin of China, was inundated from Tethys. Owing to the global regression at the end of the Triassic and the Indochina Orogeny, the Kunlun Mountains and their eastern extension into central Qinghai, the Hengduan Mountains, and the Yunnan-Guizhou Plateau became land. The coastline of the northern continent in China territory had shifted southward, approximately to a line from the Kunlun to the Hengduan Mountains, and most of the area south and west of this line received marine Jurassic deposits. We follow Wang and Sun (1983) in the divisions of the Jurassic sedimentary basins, as follows (from south to north, with the former designation by Wang and Sun in parentheses): Tethyan Himalaya District (District I1 + I2); Lhasa District (District I3); Qamdo District (District I7); Denquen-Shiquanhe District (District I4 + I5); Karakorum-Tanggula District (District I6). Our localities lie almost exclusively in the Qamdo and Tethyan Himalaya districts. The recent ‘mobilistic’ view based on plate-tectonic theory has greatly changed the reconstructions of Mesozoic southern Eurasia (Laurasia) as well as of Gondwana on the opposite side of Tethys (text-fig. 2). Much of southern and south-eastern Asia is a collage of continental blocks which assembled sequentially from the latest Palaeozoic to the Palaeogene. A variety of geologic histories for Tibet has been suggested in recent years (in Liu et al. 1981). One of the most lucid and most recent reconstructions is that of Sengor (1984, 1985). Text-figure 3 shows our modification of Sengor’s palinspastic map for the later part of the Jurassic, with north-eastern Gondwana according to Hamilton (1979, 1983) and the Somali Basin already open (Westermann 1975; Rabinowitz et al. 1983; Bosellini 1986). The more northerly position of the Indian Shield provides room for the terranes in the Indian Ocean, a few of them proven but most of them controversial, and improves the biogeographic ‘fit’ of Permian to Jurassic Himalayan faunas with those of the northern Australasian margin (eastern Indonesia-New Guinea, etc.). The age of the Jurassic magnetic ‘quiet zone’, when Madagascar is said to have separated from Africa, is now extended downward from Callovian-Oxfordian to Bajocian (Ogg and Steiner 1985, and pers. comm.), and abundant Bajocian oceanic ammonoids at Mombasa (Westermann 1975) indicate pre-Bajocian separation. Dominance of ‘leiostracan’ ammonoids implies proximity of a deep ocean, rather than a shallow embayment caused by crustal stretching. An alternative possible position for the Tethyan Himalaya is proposed in text-fig. 3. A rather similar Jurassic reconstruction of the northern margin of the Indian Shield to that of Hamilton’s is provided by Johnston and Veevers (1984, fig. 14). They double the size of the shield so that it extends along both Antarctica and western Australia, but the Himalayan basin remains in the conventional southern position. This vast added area is similar in size to the entire Tibet- Qinghai Plateau, but no significant underthrusting of southern continental blocks occurred beneath Eurasia; the thick ophiolite series of the Yarlung Zangbo Suture dips southward. According to Sengor (1984, 1985), the Cimmerian Continent, including the later North and South Tibet terranes, became separated from northern Gondwana at the end of the Palaeozoic. While moving northward across Tethys, this elongate palaeocontinent split lengthwise into two 298 PALAEONTOLOGY, VOLUME 31 parts (text-fig. 3). The northern part was assembled with Laurasia in the Early Mesozoic, thus closing the original Tethyan ocean (Palaeo-Tethys). Suturing occurred along the Lancan Jiang subduction zone (text-fig. 2). Northern Cimmeria is today the Cimmerides Orogenic Belt ranging from the Caucasus to Indo-china and includes the North Tibetan Qamdo terrane with our localities 5, 6, 12, 17, and 18 near or at its southern margin. text-fig. 2. General tectonic map of Tibet according to Sengor (modified from Sengor, 1984, 1985) showing the principal sutural zones with time of closing, the major tectonostratigraphic terrane (blocks), and the Middle Jurassic localities (X). CF (stippled) = Cimmerian flysh fill on oceanic crust: Palaeo-Tethys (closed); N = Niti area. Qamdo is also spelled Qantang. Same area as text-fig. 1. During the Middle and Late Jurassic the Tanggula Ocean separated the southern parts of Cimmeria (the later South Tibetan or Lhasa Block) from Eurasia (text-fig. 3). The Early Cretaceous subduction of the Tanggula Ocean along the margin of the Qamdo terrane closed this small ocean and welded the last part of Cimmeria to Laurasia/Eurasia. Since Sengor’s map is for the Late Jurassic, the Bajocian-Callovian Tanggula Ocean was probably somewhat larger than shown here. Loc. 7, the only locality on the north-eastern margin of the South Tibet or Lhasa terrane, yielded only a small fauna not seen by us. The early Cretaceous suture is along the Tanggula (Bongong Co-Nujiang) Suture (Fracture Zone) (text-fig. 2). The Indus-Yarlung Zangbo Fracture (Suture) defines the southern margin of the Lhasa terrane, marking the Eocene collision of the Indian Shield and the closure of Neo-Tethys. The Himalaya belongs to the Alpide Orogenic Belt stretching from the Mediterranean to Indonesia. Most of our localities, i.e. in the Tethyan Himalaya, are from near the northern shore (and ?slope) of the Indian Shield, now the western end of the Indo-Australian Plate (but see text-fig. 3). WESTERMANN AND WANG: JURASSIC AMMONITES FROM TIBET 299 text-fig. 3. The Middle to Late Jurassic position of the continents, a, according to Sengor (1984, 1985) for south-eastern Eurasia, and modified after Hamilton (1979 and see text), for the India Madagascar region, b, the alternative, conventional position of the India Madagascar region. Crosses indicate the Tibetan Middle Jurassic localities. A third alternative is here tentatively proposed: the Tethyan Himalaya was a terrane separated from the Indian Shield, lying alone in a more northern, west-east position. G, Geraldton district of Western Australia; M, Moluccas; NG, New Guinea; SN, Spiti Niti area; B, Broken Ridge; K, Kerguelen Plateau; N, Naturalista Plateau; S, Seychelles Plateau. The ammonoid provinces, or subprovinces of Indo-W. Pacific Province and oceanic ridge with extension, are indicated. table 1. Middle Jurassic ammonite localities of Tibet (see also text-fig. 1). Tethyan Himalaya, South Tibet Qamdo Block, North Tibet Locality 1 2 3 8 9 10 II 13 14 5 6 12 17 18 Unit/bcd 6 1 u 1-2 4-6 1-6 7-13 14-18 2 3 12 Sonninia s.l. sp. Euhoptoceras cf, subdccorala + + + IVitchellia cf. australica + + H'. libelica W. cf. sutneri + + Dorsetensia cf. romani Fontannesia kiliani + + + F. haydeni F. ? cf. arabica + + Oxycerites n. sp. A Jeanneiiceras cf. anomalum + + Pseudoioiiesl cf. sphaeroceroides Macrocephalites cf. macroceplialus M l cf. eilieridgei (l)Grayiceras nepaulense + + + G.l waageni + ? + + G.l gucuoi ? + + Subkossnialia cf. opis + Erymnoceras aff. coronation Choffatia cf. madani and balinensis C. cf. funata + + + + C. propinqua C. ( Indosphincles ) aff. urbana + + table 1. Middle Jurassic ammonite localities of Tibet (see also text-fig. 1). 300 PALAEONTOLOGY, VOLUME 31 X H £ o Z a + + + + + + + + + + e- + X H -c 3 O 00 6 X + + + + + + + + “a -is a, o a -! 2 ° Q .« oS •S 35,-1; -S <_ c tj c: o a .2 a a ? a c ■S ? 5 K Si a « ^ a « ° R -C C"' Q k, k, k, ^§1 v* Q d -S § § a or- I -S < &■ I ^ »: (J3 ^ ^ v i O' . ■o N S 60 £ *§! ~1 I I l I \ ~ Gucuo Ls. __ 9 — 9 S> Kim. &>- S 9 SO — -bys. bivalve AC.- 'U'>T[L belemnite \ o j ^ \ • — * »p. \ -i Oxf. mwA vMenkafun F. ~500m /Spiti Shales . ■ — -i — t- & ’ *e Lungmar T - 1 - , . \ ^ - Lst. — ^ Grayiceras; Ass. Fe -oolite Mb. Niehnieh Hsiungla F. Witch. -Font. Ass. * text-fig. 10. Tentative regional analysis of the Spiti Shales and equivalent Menkatun Formation, from north- eastern Pakistan to central South Tibet. Vertical chronostratigraphic scale approximating lime. Hiati, lithofacies, ammonite and belemnite/byssate bivalve biofacies, and ammonite associations indicated. Note the strongly diachronous Feruginous Oolite Member and lateral shaly equivalents at the base of the Spiti Shales/Mancatun Formation, followed at least in Nyalam County by a large hiatus (but see page 295). section at Ringdom Gomppa, whereas eastward toward Spiti, only Lower or Middle Callovian is indicated by the fauna, and the reduced easternmost sections at Tantak cannot be dated. Arkelfs review of the famous Niti area (1956, p. 408) is based on records in previous literature only, with the Upper Callovian based entirely on an unconfirmed record of ‘ Distichoceras cf. bicostatuni . We suggest that this genus could have been mistaken for an Early (or ?Middle) Callovian Jeanneticeras or Chanasia. Even if Upper Callovian should be present locally in the Ferruginous Oolite, this diachronous nearshore deposit in no way precludes the presence of any part of the Callovian in the offshore shaly facies within a few kilometres. It is noteworthy that no clearly Oxfordian ammonite fauna is known from the western Himalaya; that even Kimmeridgian ammonite records are very meagre and doubtful; and that Belemnopsis is long-ranging. The correlation of the B. gerardi Beds with the supposedly Late Oxfordian beds that bear Belemnopsis and inoceramids in the Moluccas, Indonesia, has two problems: (1) The belemnite/byssate bivalve-bearing sequence, without ammonites, is a biofacies (yet unexplained) which in the Moluccas is developed in the Kimmeridgian, above Oxfordian Epimayaites (although single inoceramids occur also in the ammonoids facies). (2) B. gerardi (Oppel) is poorly known (perhaps nomen dubium ), but the forms described by Uhlig (1910n) are related to latest Oxfordian 310 PALAEONTOLOGY, VOLUME 31 through Tithonian species (Challinor and Skwarko 1982). But superficially similar, large Belenmopsis occur in Indonesia from the Bathonian upward (‘Callovian’ of Challinor and Skwarko, 1982; associated ammonite faunas now re-dated, unpublished). (3) Similarly, the broad, coarsely rugose inoceramids mentioned by Uhlig (1910/?, 1911) could be latest Oxfordian-Kimmeridgian Retroceramus as in eastern Indonesia and New Zealand, but Retroceramus occurs with similar forms, i.e. R. eichwaldi (Koschelkina), already in the Bajocian-Callovian of the Northern Hemisphere and, together with the Indonesian 'Oxfordian' R. cf. galoi (Boehm), in the Callovian of the Andes (J. A. Crame, pers. comm.). Between the Niti valley and our locality 13, in the Thakkola area of central Nepal in the central Himalaya, Mouterde (in Bordet et al. 1971) has recorded ferruginous oolites and marls at the base of the Spiti Shales, yielding ‘ Indocephalites gr. urbanus , Macrocephalites sp., and large Oxycerites sp.' Only the last species was illustrated and it strongly resembles our Oxycerites n. sp. A. This is perhaps our ‘ Grayiceras ’ Association. Directly above are said to follow ‘ Epimayaites and Prograyiceras of the Upper Oxfordian’, but these could also be ' Grayiceras ’. From the shaly facies of this same area Mouterde gave the first reliable record of Oxfordian ammonites (found loose) in the Himalaya, i.e. Peltoceratoides sp. of the Lower Oxfordian and Perisphinctes ( Kranaosphinctes ) indogermanus Waagen, P. ( Arisphinctes ) gr. hellenae De Riaz, and P. ( Orthosphinctes ) sp. of the Upper Oxfordian, together with what appear to be true mayaitids. We suggest that the pre-Kimmeridgian hiatus documented for our locality 13 in the central Himalaya of Tibet may extend to other areas, with the higher Callovian and Oxfordian frequently missing (Bossoullet et al. 1977). Extensive stratigraphically controlled collecting will have to be done in the Lower Spiti Shales, however, before hiatuses can be distinguished from the regional absence of ammonoid biofacies (the Belenmopsis— byssate bivalve biofacies cannot be dated accurately), and before the age of the Lower Spiti Shales /Belenmites gerardi Beds can be clearly determined. BIOGEOGRAPHIC CONCLUSIONS According to the plate-tectonic subdivision of Tibet (text-figs. 2 and 3) the Middle Jurassic ammonite localities (Iocs. 5, 6, 12, 17, 18) belong to (a) the North Tibetan Qamdo block or terrane which since Early Jurassic was amalgamated with Eurasia; (b) the north-eastern margin of the South Tibetan/Lhasa block or terrane (loc. 7) which was still separated from Eurasia (including North Tibet) by the narrow Tanggula Ocean; and (c) the Tethyan Himalaya area (Iocs. 1-4, 8-1 1, 13-16) at or near the northern margin of the Indian Shield, then part of Gondwana at the south shore of Tethys. We will consider whether the fossil record supports the ‘mobilistic’ hypothesis. The samples from the first two areas are, unfortunately, small and poorly preserved, or small and not available to us for examination. The affinities of Himalayan ammonite species to other ammonite provinces have been listed in Table 2, including the northern and western Australasian area of extreme south-eastern Tethys (eastern Indonesia and Western Australia). In the table, definite species identifications are given the score of 1, uncertain identification 0-5. Bajocian The poorly known ammonite fauna from North Tibet includes Dorsetensia cf. romani, a species of mainly European, but also Andean and possibly East African distribution (Westermann 1975). The ammonite fauna from the Tethyan Himalaya of Tibet is mainly endemic (score = 2, see Table 2). Its closest affinity is to eastern Indonesia, i.e. the Sula Islands and Irian Jaya (1-5), whereas the pan-Tethyan to cosmopolitan, Indo-East African-Madagascan (Ethiopian) and West- Tethyan elements are minor (0-5 each). The scarcity of Ethiopian elements, however, is probably misleading, because coeval marine fauna is represented by only a few specimens from Madagascar (Collignon 1958, figs. 28-30) and Kenya (Westermann 1975, pi. 2, figs. 1 and 3). Current work by one of us (G. E. G. W.) in Kenya indicates indeed that Grayiceras- like ammonites are present WESTER MANN AND WANG: JURASSIC AMMONITES FROM TIBET 31 1 table 2. Biogeographic Analysis of Himalayan Fauna from Tibet. Species Pan- West- Ethiopian E. Indon. + Endemic Tethyan Tethyan W. Austr. Bajocian Euhoploceras cf. subdecorata Witchellia cf. australica W. tibetica W. cf. sutneri ( + ) ( + ) ( + ) ( + ) + Fontannesia haydeni F. kiliana + + Subtotal 0-5 0-5 0-5+ 1-5 2 Callovian Jeanneticeras cf. anomalum Grayiceras blanfordi ( + ) + + G. Iwaageni + G. Igucuoi Subkossmatia ? ( + ) cf. op is Erymnoceras aff. coronatum Choffatia + + cf. madani (?) ( + ) C. cf. balinensis ( + ) C. cf. funata C. ( Grossouvria ) ( + ) ( + ) propinqua + Subtotal 1 1 1-5 + 1 4-5 Total score 1-5 1-5 2-5 2-5 6-5 (Uncertain identifications placed in parenthesis and scored half unless genus restricted similarly.) there in the middle part of the Callovian. In addition, some Arabian affinity is suggested by FP. cf. and aff. arabica. Callovian The ammonite fauna of North Tibet is known only from a large incomplete Macrocephalites cf. macrocephalus of typically European appearance, and from the cosmopolitan Choffatia spp. The analysis of the ammonite fauna of the Tethyan Himalaya of South Tibet is also provisional. Endemism appears to be high (score = 4-5), with the affinities of the remaining elements being probably stronger to India-East Africa-Madagascar than to pan-Tethys, West-Tethys, and eastern Indonesia. Reports of single Boreal elements from South Tibet were presumably based on misidentifications. As in the case of the Bajocian fauna, however, resemblances to the Ethiopian and eastern Indonesian faunas may be underrated due to differences in biofacies and incomplete stratigraphic records. In conclusion, the affinities of the Middle Jurassic ammonite species from the Tethyan Himalaya 312 PALAEONTOLOGY, VOLUME 31 of Tibet support the ‘mobilistic’ (plate-tectonic) model, i.e. that their habitat was at the southern margin of Tethys, with connections to western India, East Africa, and Madagascar, as well as north-western Australasia. The limited evidence from North Tibet, on the other hand, is consistent with a Eurasian position. The latest Jurassic (Tithonian) ammonoid affinities of the Tethyan Himalaya become extremely close to north-western Australia which consequently Uhlig (1911) correctly included in the Himalayan Province. The Tethyan Himalaya, thus, lay at the junction (ecotone) of the north- south extending Ethiopian and the west-east trending Himalayan ammonoid provinces (or subprovinces of Indo-East Africa Province of Jai Krishna 1983). SYSTEMATIC PALAEONTOLOGY The specimens referred to below were deposited in the following institutions: NIGP, Nanjing Institute of Geology and Paleontology, Academia Sinica, Nanjing; A, Geological Bureau of Xizang, Lhasa; K, Geological Survey of India, Calcutta; and J, Department of Geology, McMaster University, Hamilton, Ontario. Family sonninidae Buckman, 1892 Our material from Tibet contains four genera; Sonninia, Euhop/oceras , Witchellia , and Dorsetensia. There are close similarities between Witchellia and Dorsetensia on the one hand, and Sonninia and Euhoploceras on the other (Westermann 1969; Eluf 1968; Westermann and Riccardi 1972; Morton 1972; Donovan et al. 1981). Euhoploceras has only recently been separated from Sonninia while Witchellia and Dorsetensia have long been recognized as genera. Intermediate forms between the genera exist, however, and are difficult to classify. Note that close affinities also exist between Dorsetensia and Fontannesia which is now placed in the Hildoceratidae (Westermann and Getty 1970). This suggests that Witchellia and Dorsetensia may also find a better place among the Hildoceratidae than among the Sonniniidae. Genus sonninia Bayle, 1879 Sonninia s.l. sp. v 1979 Cyclicoceras sp.; Wang and Chen in Wang et al., p. 58, pi. 17, fig. 9. Material. The poorly preserved NIGP 34074 from loc. 6. Description. The specimen is of medium size, somewhat involute, with a rather high keel. The flexed ribs generally bifurcate at mid-flank in small nodes, form a series of blunt nodes on the umbilical shoulder, and project on the ventral shoulder. Remarks. This specimen is undoubtedly a typical Sonninia , but the poor preservation prevents specific identification. Genus euhoploceras Buckman, 1913 The ‘genotype’ of ‘ Sherhornites' (Buckman 1923, pi. 411) is intermediate between Euhoploceras and Papilliceras , with a spinous stage on the inner whorls as in the former and a series of lateral tubercules on the body-chamber as in the latter. Its geographic distribution is cosmopolitan. Euhoploceras cf. suhdecoratum Buckman, 1893 ($?) 1953 Sonninia aff. dominans Buckman; Arkell, p. 322, pi. 14, fig. 8 a, b. Material. Arkell’s (1953) description of the small and entirely septate specimen (K9/247) from our loc. 1. Description. The evolute shell has an oval whorl-section and a well-developed spinous stage to approximately 40 mm diameter. One flank of the last half-whorl has approximately twenty ribs. WESTER MANN AND WANG: JURASSIC AMMONITES FROM TIBET 313 Discussion. The specimen is very close to Sonninia subdecorata Buckman (1893, pi. 84, figs. 9-11) except for the somewhat more distant ribbing. Westermann tentatively considers E. subdecoratum to be the microconch of Euhoploceras adicra. Age and distribution. E. subdecorata and related macroconchs mark the Aalenian-Bajocian boundary beds of Europe. Genus witchellia Buckman, 1889 Although the generic names Witchellia and Dorsetensia were established in the last century, problems of matching the macroconchs with the microconchs remain, i.e. the correspondence between Witchellia $ and Pelekodites <$, and whether ‘ Zugophorites ’ is synonymous with Witchellia , and ‘ Hyalinites ’ with Witchellia or Dorsetensia. Witchellia cf. australica Arkell, 1954 1953 Witchellia aff. platymorpha Buckman; Arkell, p. 333, pi. 14, fig. 6a, b. * ? 1954 Witchellia australica Arkell in Arkell and Playford, pp. 561-563, 584-585, figs. 4 and 9. Material. Arkell’s (1953) description of specimens K9/231, 248, and possibly also 235 from our loc. 1; one fragmentary large septate specimen J2186 from loc. 2. Remarks. This form was originally compared with W. platymorpha Buckman on the basis of ‘wide, distant, feeble ribbing lost early and a high flat-sided whorl with tabulate, strong tricarinate venter’ (Arkell 1953). Whereas both forms undoubtedly belong to the involute W. laeviscula group (Westermann 1969), the Tibetan form is more weakly ribbed becoming smooth on the mature flanks. Arkell (in Arkell and Playford 1954) compared one Tibetan specimen (K9/231) with W. australica Arkell. Although the Tibetan specimens resemble the holotype in the gradual umbilical slope and rounded shoulder, they differ in the ‘broad tabulate venter bearing a large but blunt medial keel flanked by grooves’ and ‘faint ribbing of the nucleus, and thereafter the whorls are smooth or carry only faint indefinite falcoid fold’. Specimen K9/248 is also similar to the holotype of W. australica in venter and ornamentation, but has a rather steep umbilical wall and a subangular umbilical shoulder. Age and distribution. W. australica is known from the Laeviuscula Zone of Western Australia. Witchellia tibetica Arkell, 1953 1953 Witchellia tibetica Arkell, p. 332, pi. 14, fig. la. b. [holotype]. Material. Arkell’s (1953) descriptions of specimens K.9/245 and 240 from our loc. 1. Remarks. This species was established on the basis of the holotype (K9/245) and one questionable specimen (240) only. Arkell considered W. glauca Buckman as the nearest European species, with W. tibetica being more evolute and more finely ribbed. Westermann (1969) suggested that W. ‘ glauca', W. " actinophora' , and W. ‘ falcata ’ Buckman are synonymous with W. sutneri (Branco); but all have falcate ribs with bullae-like primaries, which W. tibetica lacks. W. tibetica appears to be closest to the European morphospecies ''Zugophorites'' zugophorus, ‘ Gelasinites ’ gelasinus Buckman from the lower Laeviuscula Zone, and W. sayni (Haug), as is evident in the rather evolute coiling, the strong, dense ribs, the tricarinate-bisulcate venter, the steep and low umbilical wall, and the subangular umbilical margin. However, we could not re-examine the Tibetan specimens which are deposited in the Geological Survey of India. Distribution. Possibly endemic to Tibet. 314 PALAEONTOLOGY, VOLUME 31 Witchellia cf. sutneri (Branco, 1879) Plate 20, figs. 8-9 v 1976 Witchellia tibetica Arkell; Wang in Zhao, p. 517, pi. 3, figs. 9 12 [non Arkell, 1953]. v 1985 Witchellia tibetica Arkell; Wang, pi. 1, fig. 7. Material. Septate NIGP 30420 and 30421 from loc. 2 (coll. Jsb79). Remarks. Both specimens undoubtedly belong to Witchellia on the basis of the falcate ribs and bisulcate venter. They are morphologically between W. sutneri , with evolute coiling, and W. tibetica Arkell, with dense ribs and lacking bullae-like primaries, but more similar to the former. The septal suture on NIGP 30421 has a wide E lobe with low median saddle, a wide E/L saddle divided by a small incision, and smaller and narrow L and U lobes. W. glauka Buckman and W. falcata Buckman from the English Laeviuscula Zone are closely affiliated or conspecific. Age and distribution. W. sutneri occurs in the Laeviuscula Zone of north-west Europe. Family sonniniidae or hildoceratidae Genus dorsetensia Buckman, 1922 Dorsetensia cf. romani (Oppel, 1857) $ Plate 20, figs. 10-11 v ? 1976 Dorsetensia xizangensis Wang in Zhao, p. 519, pi. 4, figs. 17 and 18. Material. NIGP 84768 and possibly also 84765 from loc. 5 (coll. 76Z9) in North Tibet. Remarks. The fragmentary macroconch NIGP 84768 (d ~ 113 mm) closely resembles the holotype of D. romani which was refigured by Huf (1968, pi. 13, fig. 6 a-e). The flanks of the body-chamber are flat and converge toward the fastigate, sharp venter, and some irregular, feeble, slightly flexed folds can be observed on its medial and inner flanks. The umbilical wall is steep with subangular margin. D. xizangensis Wang is also similar, but it is based on an incomplete, undiagnostic specimen and is here considered a nomen dubium. Age and distribution. D. romani occurs in Europe and the Andes of Chile (Westermann and Riccardi 1972). It is the index of the Romani Subzone, Humphriesianum Zone. Family hildoceratidae Hyatt, 1867 Subfamily grammoceratinae Buckman, 1905 Genus fontannesia Buckman, 1902 We follow Westermann and Getty (1970, p. 240) in transferring the genus from the Sonniniidae to the Grammoceratinae. EXPLANATION OF PLATE 20 Figs. 1-4, 6, 7. Fontannesia kiliani (Kruizinga), from loc. 2 (coll. Jsb 79), x 1. 1 and 2, ($), NIGP 30426. 3 and 4, (?), NIGP 30427. 6 and 7, (39, NIGP 30429. Fig. 5. Oxycerites n. sp. A, (d), NIGP 84763 (coll. FdIV-1 1), loc. 3, x 1. Figs. 8 and 9. Witchellia cf. sutneri (Branco), (d), NIGP 30421 (coll. Jsb 79), loc. 2, x 1. Figs. 10 and 11. Dorsetensia cf. romani (Oppel), ($), NIGP 84768 (coll. 7639), loc. 5, x0-5. Figs. 12-14. F.l cf. arabica Arkell, (?), NIGP 30423 (coll. FdIV-19), loc. 3, x 0-6. PLATE 20 WESTERMANN and WANG, Fontannesia , Oxycerites , Witchellia , Dorsetensia 316 PALAEONTOLOGY, VOLUME 31 Fontannesia hay deni (Arkell, 1953) * 1953 Dorsetensia haydeni Arkell, p. 334, pi. 13, fig. 5 a-c, pi. 14, fig. 9a, b. non v 1976 Dorsetensia haydeni Arkell; Wang in Zhao, p. 518, pi. 4, figs. 12-16. non v 1985 Dorsetensia haydeni Arkell; Wang, pi. 1, figs. 9 and 10. Material. Arkell’s (1953) description of specimens K9/246, 239, and 244 from loc. 1. Description. This is an evolute, rather compressed, medium-sized form with a subova te whorl- section and a blunt keel on the narrow-rounded venter. The slightly flexed and dense ribs are restricted to the inner whorls and gradually weaken outward to become irregular and obsolete. The umbilicus is large with gentle slope. Discussion. Arkell (1953 and 1954 in Arkell and Playford) pointed out that this form comes close to the Australian ‘T>.’ clarkei Crick and the very similar 'D.' whitehousei Arkell which he correctly believed should probably be transferred to Fontannesia. The Tibetan specimens resemble F. clarkei var. whitehousei but are even weaker in ornamentation, probably beyond the range of variability of F. clarkei. We thus retain the species F. haydeni for the time being. Age and distribution. Whereas F. haydeni appears to be endemic to Tibet, the closest relative, F. clarkei, occurs in the Laeviuscula Zone of Australia. Fontannesia kiliani (Kruizinga, 1926) $ (and e?) Plate 20, figs. 1 -4, 6 7 v * 1926 Grammoceras Kiliani Kruizinga, p. 38, pi. 7, fig. 2 [$]. ? 1953 Dorsetensia cf. regrediens (Haug); Arkell, p. 13, pi. 13, figs. 2a, b, 3 [$\. v 1970 Fontannesia aff. F. clarkei (Crick) ?subsp. kiliani, Westermann and Getty, p. 238, pi. 48, figs. I 4; pi. 49, figs. 1-4, text-figs. 4 and 5 part [dj. v 1976 Dorsetensia cf. edouardiana (d’Orbigny); Wang in Zhao, p. 518, pi. 4, figs. 10 and 1 1 [d1]. v 1976 Witchellia sayni Haug; Wang in Zhao, p. 518, pi. 3, figs. 4-6. 1985 Dorsetensia haydeni Arkell; Wang, pi. 1, figs. 9 and 10. Material. NIGP 30425-20427, 30428, 30429, 730422, and J2187 from loc. 2; Arkell’s (1953) microconchs K9/232, 243 from our loc. 1. Description. The macroconch is of medium size (60-90 mm), moderately evolute, and strongly compressed with tabulate, keeled venter of medium height. It bears slightly flexed and moderately projected simple ribs on middle and outer flanks which become blunt, rarely obscure, on the outermost whorl. The umbilical slope of the inner whorls is shallow or absent, but becomes steep and subangular on the outermost whorl. The microconch is only 30 to 40 mm large, with evolute and slightly compressed whorls, narrowly tabulate venter with blunt rounded keel, and prominent simple ribs over the entire flanks. The ribs are slightly sinuous and projected on the ventral shoulder, as in the macroconch. Discussion. There is good resemblance to the European D. deltafalcata (Quenstedt) in macroconchs and microconchs (cf. Huf 1968, pi. 9, fig. 6 a-c for microconch, pi. 10, figs. 2 a-c, 3a-c for macroconchs). This Tibetan form was, in fact, identified by us with that species of the Humphriesianum Zone before being found in association with Witchellia indicating the Laeviuscula Zone. The perfect match of the macroconch is, however, with F. kiliani from the Moluccas and Irian Jaya, Indonesia, as illustrated by the holotype and better, new material by Westermann and Getty (1970). The smoother inner flank with gentle slope of the inner whorls distinguish it from Dorsetensia. The West Australian F. clarkei (Crick) and its several, probably synonymous, close allies from the Laeviuscula Zone (Arkell and Playford 1954) differ in the more rounded whorls, with less developed umbilical shoulder and ventral tabulation particularly in the adult, and in the WESTERMANN AND WANG: JURASSIC AMMONITES FROM TIBET 317 widely spaced costae becoming obsolete. Significantly, the other probable Fontannesia from the Moluccas, F. baumbergeri (Kruizinga), closely resembles F. clarkei. Age and distribution. Moluccas and Irian Jaya, Indonesia, undated (found loose). Fontannesia! (n. subgen.?) cf. arabica Arkell, 1952 $ Plate 20, figs. 12 14 v 1976 Witchellia laeviusculus (Sowerby); Wang in Zhao, p. 517, pi. 3, figs. 15 and 16. Material. NIGP 30423 and 84779 from loc. 3. Description. The better preserved fragment NIGP 30423 is about 122 mm in diameter and entirely septate. The inner whorls of both specimens have eighteen to twenty simple ribs per half-whorl. The outer whorl has only very weak folds on the inner and medial Hanks and a rounded, floored keel on the subtabulate venter. The septal suture is very simple and with wide E/L saddle. Discussion. Whereas our specimens resemble somewhat the European and Andean macroconchs of Dorsetensia , e.g. D. liostraca forma subtecta Buckman (1892), the closest resemblance is to the earlier central Arabian 1 D .’ arabica (Arkell 1952) in their simple immature ribs. Although there is close superficial resemblance to W. laeviuscula (Sow.), the simplified septal suture, the probably floored keel without ventral sulci, and the regular simple ribs militate against assigning the specimens to Witchellia. Additional material and associated fauna is needed to solve the problems of generic taxonomy. We suggest that this could be a Dorsetensia homoeomorph affiliated to Fontannesia with which it is contemporaneous, and it may be classified as a new subgenus of Fontannesia. Age and distribution. D.' arabica has recently been dated as basal Bajocian in central Arabia (Enay and Mangold 1984). Fontannesia ? n. sp. aff. arabica 1953 Dorsetensia cf. romanoides (Douville); Arkell, p. 333, pi. 13, fig. la, b. Material. Arkell’s (1953) description of specimen K9/237 from our loc. I. Remarks. Arkell (1953) noted this macroconch resembles D. pulchra Buckman, except for the more inflated whorl section (EI/W = 1-65 from illustration) which resembles that of D. romanoides. However, both D. pulchra and D. romanoides are strongly compressed and more bluntly ribbed than the Tibetan form and believed by Huf (1968, pp. 86-87) to be conspecific with D. romani (Oppel). Westermann and Riccardi (1972, p. 98) noted that the Tibetan macroconch closely resembles the European D. edouardiana and probably also ‘ D .’ arabica Arkell. The possibility exists, however, that this form is an unusual Fontannesia macroconch, particularly since the (?associated) microconch described by Arkell (1953) under D. cf. regrediens belongs indeed to F. gr. kiliana and " D .' arabica is earliest Bajocian (see below). Arkelfs (1953) D. cf. romanoides came from the Lungmar Limestone and was found together with species of the Discites-Laeviuscula zones. This unnamed species differs from F.‘i arabica , probably its closest ally, in the larger diameter and more involute and compressed-lanceolate whorls. Family oppeliidae Douville, 1890 Subfamily oppeliinae Douville, 1890 Genus oxycerites Rollier, 1909 Oxycerites n. sp. A. $ Plate 20, fig. 5 Material. A small septate specimen NIGP 84763 from loc. 3 (FdIV-l I ). PALAEONTOLOGY, VOLUME 31 text-fig. 1 1 Lateral and apcrtural views of Jean- neticeras cf. anomalum Elmi, f, J2189. Complete microconcli from ‘ Grayiceras ’ Association, bed A5 of loc. 13, x I . Description. The phragmocone (20 mm diameter) appears to be adult. It is moderately involute with sharp ventrolateral ribs on the last half-whorl and probably a microconch. The flanks are a little convex and converge gradually towards the acutely fastigate venter (h ~ 10 mm, b ~ 5-5 mm). The inner whorls are entirely smooth; broad, short ribs start to appear on the outer flank and ventral shoulder only on the last septate whorl (Wh > 3-9 mm) whereas the inner half of the flank remains smooth. The ribs widen towards the venter where they project sharply and die out close to the fastigation. The umbilicus slopes gently and is rather small (3-7 mm). Discussion. The specimen is very close to ‘ Oppelia fused of Boehm (1912, p. 143, pi. 33, fig. 3o-3; pi. 34, figs, la, b , 2, 3) from the upper Lower Callovian of Indonesia. That form is currently being classified as a new species of Oxycerites (Westermann and Callomon, unpubl.) based mainly on the ribbing which changes gradually from dense to widely spaced, and the acutely fastigate venter. Age and distribution. The closest ally occurs in the Calloviense/Gracilis Zone of the Moluccas, Indonesia. Subfamily hecticoceratinae Hyatt, 1900 Genus jeanneticeras Zeiss, 1956 Elmi (1967) raised the original subgenus to the genus level and discussed the taxonomy and distribution of Jeanneticeras in detail. The genus has been known only from the Lower Callovian of the Mediterranean and Submediterranean provinces. This ‘microconch genus’ appears to correspond to several hecticoceratid genera in their restricted sense (Elmi 1967). Jeanneticeras cf. anomalum Elmi, 1967 <$ Text-fig. I la, b cf. 1967 Jeanneticeras anomalum Elmi, p. 763, pi. 16, figs. 1 3, 5; text-figs. 184, 195, 196. Material. One complete microconch J2189 from bed A5, loc. 13. Description. The complete microconch is 45 mm in diameter. The septate whorls are involute and compressed trapezoidal, with broadly rounded venter. Their ornamentation consists of subradial primaries and dense, projected secondaries which terminate on the ventrolateral shoulder in small tubercles. The septal suture is moderately incised; approximation at the end of the phragmocone indicates maturity. The body-chamber is slightly less than a half-whorl long, agresses strongly at the umbilical seam, and terminates in partially preserved lateral lappets. The costae become increasingly falcate and blunt on the lower and middle flanks; the secondaries become much more widely spaced, and the terminal tubercles elongate into well-developed clavi. Discussion. Resemblance to ‘ Oppelia ( Oecotraustes )’ adela Uhlig from the basal Spiti Shales is WESTERMANN AND WANG: JURASSIC AMMONITES FROM TIBET 319 apparently superficial. According to Uhlig (1910a, p. 72, pi. 43, fig. 3 a-d), the holotype (only specimen) bears fine mid-ventral tubercles, i.e. a serrated keel, not seen by the illustrator. We have received photographs of the supposed holotype (no. 7837) from the Indian Geological Survey which show septation on the last half-whorl and no egression of the large umbilicus, indicating that this is an inner whorl. The minute protuberances at mid-venter appear to sit on the partially exposed siphuncle, whereas the venter is generally rounded. It is possible that the protuberances are conellae, i.e. diagenetic products of a keel floor. This species therefore differs significantly from our Jeanne tic eras. S. Elmi has seen the photographs and commented that "O.' adela appears to be very close to the middle Late Callovian Putealiceras. Remarks and occurrence. The specimen was identified for us from photographs by S. Elmi and dated as latest Early Callovian (late Calloviense/Gracilis Zone). J. anomalum has been known only from the Chanasia ardescica Subzone of south-east France and is said to be an excellent guide. Family otoitidae Mascke, 1907 Genus pseudotoites Spath, 1939 Pseudotoites! cf. sphaeroceroides (Tornquist, 1898) 1953 Emileia (Frogdenites) sp.; Arkell, pi. 13, fig. 4 a-c. Material. Arkell’s (1953) description of a single immature specimen K9/249 from our loc. 1. Remarks. The specimen is a very small (d = 18 mm), rather evolute cadicone with spinous nodes at the furcation point of dense ribs, close to the umbilical margin. This specimen does not permit even generic identification and could be the inner whorls of either micro- or macroconch. It is similar to the extremely rare north-west European Frogdenites , but also resembles juvenile Emileia and, especially, the common Pacific Pseudotoites. Age and distribution. The genus Pseudotoites is known from the Laeviuscula Zone of south Alaska, the Southern Andes, Western Australia, and Indonesia. Inflated species resembling P. sphaeroceroides from the Andes occur also in Western Australia, i.e. P. corona (Arkell in Arkell and Playford 1954). Family sphaeroceratidae Spath, 1920 Subfamily macrocephalitinae Salfeld, 1921 Genus macrocephalites Zittel, 1884 Macrocephalites cf. macrocephalus (Schlotheim, 1813) $ Plate 21, figs. 1-2 Material. The poorly preserved specimen NIGP 84775 from unit 12 of loc. 18, North Tibet. Remarks. The entirely septate shell is large, involute and sub-globular. The outer flanks and venter are covered by dense, fine ribs, approximately forty per quarter-whorl; the inner flanks (corroded) were either smooth or had only extremely blunt primaries. These features are typical for the group M. macrocephalus (for definition of the species see Callomon 1971, 1980). The complicated suture has a subradial saddle envelope and the laterodorsal parts of the septum are not oblique to the shell radius as in Eucycloceratinae and Mayaitinae. Age and distribution. M. macrocephalus is well known from (he (?)Macrocephalus Zone of Europe. Macrocephalites ? cf. etheridgei (Spath, 1928) Plate 21, figs. 3 4 cf. * 1928 Kamptocephalites etheridgei Spath, p. 200, pi. 69, fig. 3 [holotype refigured]. Material. One fragment with incomplete body-chamber NIGP 84774 from levels Cl 2 of loc. 8. Remarks. This relatively evolute sphaeroceratid has the septal suture of Macrocephalites s.l. (Thierry 1978), i.e. the saddle envelope is subradial and not protracted as in the otherwise similar "Grayiceras' . 320 PALAEONTOLOGY, VOLUME 31 There is good resemblance to M. etheridgei also in the inflated whorls with steep umbilical slope and well-developed umbilical shoulder. Age and distribution. M. etheridgei is known from the Upper Bathonian- Lower Callovian of eastern Indonesia and New Guinea (Sato 1975; Westermann and Callomon, unpubl.). Subfamily eucycloceratinae Spath, 1928 The family-group taxon Grayiceratidae was introduced by Spath (1925) in a paper on Somali fossils, based on a dubious nominate genus— and as obscurely ‘deleted" by him 3 years later (Spath 1928, p. 224) by ‘replacing’ it with the (Tithonian) ‘Paraboliceratidae’ (the newly examined septal suture of the genotype was considered to be ‘perisphinctoid’). The name Grayiceratidae was therefore suppressed in favour of the Mayaitidae Spath, 1928 by the International Commission on Zoological Nomenclature, opinion 471 (Henning 1957), based on Arkell’s (1955) submission that the nominate genus is a nomen dubium (see below). The family Eucycloceratidae Spath, 1928 was founded for the Callovian homoeomorphs of the Oxfordian Mayaitidae Spath, 1928; included in the Macrocephalitidae without distinction in the ''Treatise' (Arkell in Arkell et al. 1957); retained at least at subfamily level by Westermann (1968) as the postulated phyletic intermediate between Macrocephalitidae and Mayaitidae; and questionably retained as a subfamily of the Sphaeroceratidae in the most recent classification by Donovan et al. (1981). Four genera, all from the latest Early-early Middle Callovian of the Indo- East African-Madagascan area (Ethiopian Province), were originally included in the family: Eucycloceras Spath, 1924 (type species Stephanoceras eucyclum Waagen), Subkossmatia Spath, 1924 ( Ammonites opis J. de C. Sowerby), Idiocy clocer as Spath, 1928 (I. perisphinctoides Spath), and Nothocephalites Spath, 1928 (TV. asaphus Spath). They differ from macrocephalitins in the protracted, sometimes arched and simplified septal suture, and in the egressing body-chamber which is particularly coarsely ribbed, even in the macroconch of Idiocy clocer as. The Mayaitinae Spath, 1928, also restricted to south-eastern Tethys, have usually been regarded as being confined to the Upper Oxfordian (when divided into two substages). Although many mayaitins perfectly resemble the eucycloceratins, the supposed Late Callovian-Early Oxfordian gap swayed most researchers to derive the mayaitins from the Callovian-Oxfordian Pachyceratidae (e.g. Arkell et al. 1957). Westermann (1957), however, disregarded the supposed gap in the record and, based on a comparison of septal morphology, proposed that the eucycloceratins are the phyletic intermediates between macrocephalitins and mayaitins. Recently, Thierry (1975) has confirmed the identity in sutural ontogeny between macrocephalitins and mayaitins, and Wester- mann (in Sato et al. 1978) documented Early Oxfordian (Cordatum Zone) Epimayaites with typically protracted sutures in the Sula Islands, Indonesia. The recent arguments of Donovan et al. (1981) against a direct relationship between eucycloceratins and mayaitins is based on the old records from Kachchh, India (Spath 1927-1933) where Epimayaites occur mainly above Mayaites. The latter genus is characterized by more inflated whorls and a suture with radial saddle envelope. We also point out that the Upper Callovian-Lower Oxfordian tends to be represented by a hiatus due to world-wide eustatic regression, and often is poorly documented by ammonites in the Tethyan Realm. The macroconch Epimayaites Spath, 1928 and the microconchs Paryphoceras Spath, 1928, Dhosaites Spath, 1924, and Prograyiceras Spath, 1928 are all characterized by protracted sutures EXPLANATION OF PLATE 21 Figs. 1 and 2. Macrocephalites cf. macrocephalus (Scliloth.), ($), NIGP 84775, unit 12 of loc. 18 in North Tibet, x 1 . Figs. 3 and 4. MP. cf. etheridgei (Spath), (<39, NIGP 84774 (coll. FdIV-11), damaged phragmocone with incomplete body-chamber, levels Cl-2 of loc. 3, x 1 . PLATE 21 WESTERMANN and WANG, Macrocephalites 322 PALAEONTOLOGY, VOLUME 31 (opposite of retracted), coarsening body-chamber ribbing often even on the flanks of the macroconchs, variable ventral rib projection, and egression of the terminal body-chamber; these are all features shared with the eucycloceratins. Only the macroconch of Mayaites Spath, 1924 is said to differ by the straight ribs and subradial suture. But it remains unknown whether transitions exist among contemporaneous assemblages (populations?), i.e. whether Mayaites is a separate clade from Epimayaites. A study of Spath’s monograph reveals that the sutures were often drawn with little attention to radial orientation, and that the umbilical elements of Mayaites nearest to the seam are consistently vertical to inclined away from the seam. This contrasts with Macrocephalites s.l. where these elements (U3 + ?) incline towards the seam, i.e. they are vestiges of retracted umbilical (sutural) lobes. Spath divided the mayaitid macroconchs according to straightness of the secondaries and/or sutural inclination. But he also separated them stratigraphically by placing the slightly earlier forms from the Dhosa Oolite of Kachchh in Mayaites and those from the higher Kontkote Sandstone into Epimayaites, even when contradicted by morphology and suture, e.g. ‘Ed axioides Spath. The suture is unknown from several ‘species’, and similar variations in sutural inclination and ventral ornamentation exist in the Callovian eucycloceratins, especially Idiocycloceras. We are therefore inclined to believe that the Oxfordian mayaitins form a single clade, and may even belong to a single dimorphic genus with interspecific and intraspecific morphologic variation resembling that of Macrocephalites s.l. (Westermann and Callomon, unpubl.). The only marked morphologic distinction of the Callovian eucycloceratins from the Oxfordian mayaitins as illustrated in the Kachchh monographs (Waagen 1875; Spath 1927-33) is the apparent absence of Subkossmatia macroconchs. A cursory inspection of a eucycloceratins collection from Jaiselmer (north of Kachchh), made by Jai Krishna, however, indicates the scarce presence not only of incomplete macroconchs of S. cf. opis closely resembling Epimayaites 9 (together with its much more abundant microconch), but even the possible presence of ‘G.’ nepaulense (Gray). We conclude that the root of the mayaitin clade is in the eucycloceratins of the latest Early Callovian S. opis zone (Jai Krishna and Westermann 1985), branching off from late, but not the latest macrocephalitins. Eucycloceratins hardly differ at the family-group level from mayaitins and may therefore be combined in the Mayaitinae Spath, 1928 (a much more frequently used name than Eucycloceratinae, erected in the same paper). Additional work on eucycloceratins of Western India and Madagascar (also ?Kenya) is required to decide on the classification. Genus grayiceras Spath, 1923 History of genus name. The complicated history of the genus was detailed by Arkell (1955) who concluded that the name is a nomen duhiunr, it should not be used unless new field evidence becomes available. Together with Mr D. Phillips (Keeper of Ammonites), Westermann has examined the British Museum (Natural History) type collections. Spath (1923) based the genus-group name Grayiceras on 'the group of Simbirskites nepalensis [ rede nepaulensis] and S. mexicanus Burckhardt’ (our brackets); in the following year Spath (1924) again named 'the new genus Grayiceras ’, with the ‘genotype G. blanfordi n. sp. = S. nepaulensis Blanford 1865: non Gray, in Uhlig [1910u]’. Thus Spath created two homonymous genera based on different type species, and the first designation stands. Type species. We support the opinion of Crick (1903), Uhlig (1910«), and Arkell’s proposal (1955) that Blanford’s type specimen of S. nepaulensis , i.e. the holotype of G. blanfordi (refigured by Uhlig 1910 a, pi. 45a, fig. \a-c) is very probably the same as A. nepaulensis Gray (1830 1832). G. blanfordi therefore appears to be a junior synonym of A. nepaulensis which then becomes the type species of Grayiceras. The probable lectotype (Blanford in Salter and Blanford 1865) of G. nepaulensis , however, is not Gray’s syntype of figure 1 as assumed by Crick, but that of Ins figure 2 which was mislaid or unrecognized at Crick’s time. Both syntypes are here reproduced photographically for the first time (PI. 22; PI. 23, fig. 1), noting that indubitable identification of the specimens with the brush illustrations of Gray will remain impossible. The probable lectotype bears the label ‘Spiti’ and the remaining syntype (paralectotype) 'Niti Pass’, whereas Gray recorded ‘Sulgranees’, westernmost Nepal. The labels appear to be of a later date. The age of the specimens even as WESTERMANN AND WANG: JURASSIC AMMONITES FROM TIBET 323 to stage is also unknown since the specimens were probably collected loose if not purchased, and the type locality has not been revisited. Although our fragmentary specimens from the Early Callovian of Tibet closely resemble G. nepaulense, the homoeomorphism with Oxfordian microconchs of Paryphoceras (cJ) is so close that we cannot be sure of specific or even generic identity within the present classification. The marked egression of the terminal body- chamber in the lectotype cannot be compared with our specimens because of their incomplete preservation, and the generic affinity of the Early Callovian GP. waageni and gucuoi with normally coiled body-chambers, remains uncertain. The age of the lectotype G. nepaulense could therefore possibly be Oxfordian as assumed by Uhlig (191 Oc/, b, 1911), Spath (1927-1933), and Arkell (1956), rather than Callovian. Consequently, the generic name Grayiceras remains something of an enigma. We therefore use queries or inverted commas, rather than creating a new genus based on our poor material. Affinities. Besides the type species, Spath (1927-1933, p. 224) also placed ‘5. koeneni ' Uhlig in Grayiceras and noted that "M." waageni and kitcheni , Uhlig spp., are intermediate between Grayiceras and Epimayaites. The ‘5. koeneni ' holotype (the only specimen), however, is a large body-chamber resembling the Callovian \Subkossmatia ramosd Spath $ and I. singulare Spath $ from Kachchh; but the specimen is too incomplete for stricter comparison and therefore a nomen dubium. The small 'M." waageni and the probably conspccific 'M.' kitcheni are relatively abundant in South Tibet and almost perfect homoeomorphs of the Oxfordian Epimayaites. The body-chamber is coiled more or less normally, however, and the ribbing is not modified on the mature ultimate whorl as in all microconchs of other Callovian Oxfordian Eucycloceratinae and Mayaitinae. The septal suture is strongly protracted, distinguishing GP waageni (including ‘A/.’ kitcheni) from compressed Macrocephalites microconchs, i.e. ' Dolikephalites'. GP gucuoi n. sp., similarly, closely resembles inflated, evolute Macrocephalites microconchs (f), i.e. ‘ Kamptokephalites' , and also Oxfordian Dhosaites f. The Late Oxfordian Prograyiceras , type species P. grayi Spath, is distinguished by the much coarser and strongly projected secondary ribs. The Callovian mayaitin closest to Grayiceras is probably Eucycloceras eucyclum (Waagen) which has septate whorls similar to G. nepaulense and GP waageni , but differs in the ornamentation of the body-chamber. Subkossmatia is more compressed with denser, prosocline ornamentation. GP. gucuoi resembled Idiocycloceras from Kachchh but differs in the broader whorls bearing shorter primaries. All eucycloceratins, however, differ in the trapezoidal, not rounded, inner whorls with vertical umbilical slope and sharp margin. (?) Grayiceras nepaulense (Gray, 1830-1832) (f ?) Plate 22; Plate 23, fig. 1; text-fig. 12a b cf. v * 1830 1832 cf. v 1865 non v 1875 cf. 1 9 1 Or/ cf. v 1924 cf. v 1928 Ammonites nepaulensis Gray, pi. 10, figs. 1 and 2 [lectotype]. Ammonites nepalensis Gray; Blanford in Salter and Blanford, p. 77, pi. 14, fig. la, b. Stephanoceras nepaulense Gray; Waagen, p. 136, pi. 35, figs. 2 and 3 [= Prograyiceras grayi Spath], Simbirskites nepaulensis Gray; Uhlig, p. 271, pi. 45a, fig. 1 a-c [Blanford, 1865, refigured]. Grayiceras blanf or di Spath, p. 11 [for Uhlig, 1910], Grayiceras blanfordi Spath, p. 224, pi. 27, fig. 3 [septal suture of holotype]. Lectotype. The almost complete, fully grown lectotype of G. nepaulense (PI. 22) has a slightly compressed elliptical whorl-section with rounded umbilical shoulders and a rather narrow, evenly curved venter. The body-chamber, preserved with two to three whorls, was about three-quarters of a whorl long. It egresses markedly only with the ultimate quarter-whorl. The ribs consist of gently curved primaries which increase in prominence and length on the body-chamber, and subfasciculate to ataxiocerid secondaries which form a moderately convex arc on the venter. The primaries reach maximum height at mid-flank on the phragmocone, but extend to two-fifths whorl height on the adult body-chamber. The suture is moderately complicated with deep E and L lobes followed by three much smaller U lobes. The saddle envelope rises strongly toward the umbilical seam, i.e. the suture is protracted. Measurements. In mm of lectotype (C25182). d b h b/h u Near aperture 120 48 55 0-87 29 (0-24) End phragmocone 102 ~29 33 ~0-95 2 1 -5(0-20) 324 PALAEONTOLOGY, VOLUME 31 text-fig. 12. ‘ Grayiceras' Association, a, b, probable G. nepaulense (Gray) body-chamber fragments, from loc. 13, x 1. a, J21827>, from bed Al; b, J21886, from bed Bl. c-e, G.7 gucuoi n. sp., from bed A5, loc. 13, x 1. c, d, J2177a, complete specimen with latex cast of phragmocone; E, J21786, incomplete body-chamber with latex mould of phragmocone. Material. Two fragments of large body-chambers, J2182a, b , one with last septum, from bed Al, and several fragments, J2188, from bed Bl, loc. 13. Descriptions. The whorl section is subcircular-ovate, slightly depressed (b = 43-5 mm, h = 41-5 mm), with maximum width at two-fifths whorl height and steep umbilical slope. The section resembles that of the lectotype at the equivalent growth stage, i.e. the beginning of the body- chamber at about 90 mm diameter. The long, prominent primaries divide at three-fifths whorl height into slightly projected secondaries and some ribs are intercalated, just as in the lectotype; the primaries on our specimens vary in degree of flexure, some being as flexuous as in the lectotype. The septal suture is strongly protracted, with the entire septal surface being inclined to the whorl section by c. 35°. Discussion. Our fragments closely resemble the lectotype. The strongly oblique umbilical seam in Spath’s (1928, pi. 27, fig. 3) illustration of the lectotype indicates that the suture is much more protracted than shown by him. For comparison with other species, see under genus. Grayiceras? waageni (Uhlig, 1910) (rf ?) Plate 23, figs. 2-5; Plate 24, figs. 16; text-fig. 13 * 1910a Macrocephalites waageni Uhlig, p. 270, pi. 77, figs. 1, 2a, b , 3 a c. ? 1910« Macrocephalites kitcheni Uhlig, p. 271, pi. 77, fig. 6 a d. v 1976 Macrocephalites compressus (Quenstedt); Zhao, pi. 5, figs. 1 3. EXPLANATION OF PLATE 22 Figs. 1-3. Grayiceras nepaulense (Gray), lectotype [holotype of G. blanfordi Spath], British Museum Nat. Hist. C25182, from ‘Spiti’ (label) or ‘Sulgraness’ (text), x 1. PLATE 22 WESTERMANN and WANG, Grayiceras 326 PALAEONTOLOGY, VOLUME 31 TEXT-FIG. 13. Grayiceras ? waageni (Uhlig), reproduction of Uhlig’s illustration (1910a, pi. 77, fig. 3 a-c, Geol. Surv. India no. 10030), x 1, from ‘Gieumal’, Spiti. This specimen was used to ‘characterise the new species’ and clearly shows the oblique (pro- tracted) septem; but the specimen is now badly deteriorated by pyrite oxidation (see also PI. 23, figs. 4-6). v 1976 Dolikephalites cf. typicus (Blake); Zhao, pi. 7, fig. 79; pi. 7, fig. 10; pi. 12, figs. 1 and 2. v ? 1976 Macroceplialites sp. 1; Zhao, pi. 6, figs. 7 and 8. v ? 1979 Dolicephalites sp.; Wang and Cheng in Wang et ah, pi. 17, figs. 1 and 2. 1985 Macroceplialites compressus Quenstedt; Wang, pi. 1, fig. la, b. Lectotype. Here designated: Uhlig, 1910a, pi. 77, fig. 2a, b (Geol. Sur. India no. 10026), a somewhat crushed but probably almost complete shell of 65 mm diameter, from 'Gieumal' (Giumal) in Spiti area. We have received new photographs from the Geological Survey of India (Calcutta) of the three syntypes from ‘Gieumal’. The small (d = 35 mm) and entirely septate specimen (Geol. Sur. India no. 10027; Uhlig 1910a, pi. 77, fig. 3a-c) was said to be the best-preserved specimen ‘which has been chiefly made use of in establishing the character of the species’. Unfortunately, this specimen has meanwhile deteriorated so badly (pyrite oxidation) that it cannot be designated as the lectotype. It also did not show the typical ornamentation of the outer whorl. The two larger syntypes, of which the best is here designated the lectotype, are both somewhat crushed, but otherwise satisfactorily preserved. The photo of the lectotype shows the septal suture (probably protracted) at about three-fifths of the whorl before the broken end of what appears to be the body-chamber. The primary costae become very prominent and markedly curved; the whorls are involute and were originally more or less strongly compressed ovate; the venter is narrowly convex and crossed by prominent sharp ribs. Material. Specimens NIGP 30435-30436, 30470 from loc. 8; specimens NIGP 30439 and 730446 from beds Kp3, 4 of loc. 10; fragmentary body-chambers J2180a-e, one of which has mould of incomplete phragmocone, from bed A5, and (?) two juvenile specimens, J2 1 8 1 a, b from bed Al, loc. 13; specimens A058 and 7A057 from loc. 14. Description. An involute, small species (or microconch) with subtrapezoidal whorls, somewhat compressed to as broad as high. The narrow size range of our body-chamber fragments indicates a full diameter of only 60-70 mm and our largest specimen has approximated sutures at 46 mm diameter, in good agreement with the type series. Ribbing is dense, sharp where the shell is preserved, and highly prominent and more or less strongly flexed on the inner flank. There are about fifteen to twenty primaries and forty to fifty secondaries per half-whorl. The primaries are concave and divide irregularly near mid-flank into the less prominent secondaries; some ribs are EXPLANATION OF PLATE 23 Fig. I. Grayiceras nepaulense (Gray), paralectotype, British Museum Nat. Hist. C5052 [original to Gray, pi. 100, fig. I], from ‘Niti Pass’ (label) or ‘Sulgraness’ (text). Figs. 2-5. GP. waageni (Uhlig). 2, 3, lectotype, probably almost complete but somewhat crushed, lateral and ventral views (Uhlig 1910a, pi. 77, fig. 2 a b) from ‘Gieumal, Spiti’. 4, 5, specimen in the British Museum Nat. Hist, collection, closely similar to paralectotype illustrated in text-fig. 13, unlabelled but stored together with syntypes of B. nepaulense and same preservation. Figs. 6, 7. GP. aff. waageni (Uhlig), NIGP 30445, from units Kpl 4 of loc. 10. Figs. 8, 9. Subkossmatia cf. op is (J. de C. Sow.), J21786, from bed A5 of loc. 13. All figures x 1 . PLATE 23 WESTERMANN and WANG, Grayiceras , Subkossmatia 328 PALAEONTOLOGY, VOLUME 31 intercalated. The ratio of secondaries to primaries is approximately 2:5. The secondaries are rectiradiate to slightly prorsiradiate and cross the venter straight to weakly convex, becoming markedly convex on the second half of the body-chamber. The body-chamber shows no significant change or only slight reduction in coiling, at least with the first half-whorl. The aperture is unknown. The septum is typically bullate, with two complete (paired) saddle and lobe axes, and markedly inclined to the radial plane (about 20°); the suture is therefore strongly protracted. Sutural complication is moderate with a graded sequence of lobes and saddles along the protracted saddle envelope. Measurements. In mm (J2180a). d b h b/h u Body-chamber 65 31-5 33 0-95 1 1 2(0-17) End phragmocone 46 22-4 ~21-5 — ■ 1 -05 9-0(0-20) Discussion. Our specimen closely resembles the illustrations of the syntypes (type series) collected by Stoliczka at ‘GieumaF in Spiti valley, and the only specimen (holotype) of ‘M. kitcheni ’ Uhlig from the same collection. The holotype of "M. kitcheni', however, is now badly disintegrated due to pyrite oxidation (new photo received) and no longer serves as a type specimen. The original illustration (Uhlig 1910a, pi. 77, fig. 6 a-c) indicates that the holotype was close to, and probably a conspecific variant of, G.l waageni (Uhlig’s illustration does not show the ‘strongly deflected forward’ ornamentation mentioned in his text). Sutural complication is moderate with a graded sequence of lobes and saddles along the protracted saddle envelope. There is rather close similarity to the microconch of M. keeuwensis Boehm (1912) from the Calloviensis/Gracilis Zone of the Sula Islands, except for the much more strongly protracted septal suture and the more prominent ribbing of G.l waageni. Significantly, that species has long been suggested as the hypothetical ancestor of the ‘Euclycloceratinae’ (Spath 1928). Almost perfect homoeomorphy exists, however, to the juvenile or inner whorls of the Oxfordian Epimayaites, e.g. the type species E. transiens (Waagen), or of its microconch Paryphoceras, which without stratigraphic control may be indistinguishable. Mayaitinae, however, tend to have variocostate ribbing in both dimorphs, with the nricroconchs getting extremely coarse ribs on the ultimate one or two whorls whereas macroconchs tend to become smooth, and with an egressing body- chamber. G.l waageni could be a microconch, possibly to G. nepaulense whose inner whorls appear to be closely similar but are poorly known. But since both presumed dimorphs of ‘ Grayiceras ' appear to have some coarsening of ornamentation on the adult body-chamber, both forms could alternatively be microconchs of different species. G.l waageni could even be a macroconch of an even smaller form, but these latter may be juveniles and/or incomplete. Grayiceras ? aff. waageni (Uhlig) Plate 23, figs. 6 and 7 ? v 1976 Indocephalites aff. transitorius Spath; Zhao, pi. 8, figs. 10 and 11. ? v 1976 Indocephalites cf. indicus Spath; Zhao, pi. 6, figs. 3 and 4. EXPLANATION OF PLATE 24 Figs. 1-6. Grayiceras ? waageni (Uhlig), from ‘ Grayiceras ’ Association of Loc. 13. 1-4, body-chamber with latex cast of phragmocone, J2180a, bed A5. 5, body-chamber fragment, J2 1 807?, bed A5. 6, almost complete juvenile(?), J218G, from bed Al. Figs. 7-14. G.l gucuoi n. sp., from bed A5 of loc. 13. 7-10, holotype, J2176, complete but phragmocone as latex cast. 1 1, J21772, complete juvenile)?). 12 14, J21776, (?)juvenile body-chamber. All figures x 1 . PLATE 24 WESTERMANN and WANG, Gravicerasl 330 PALAEONTOLOGY, VOLUME 31 Material. Three incomplete and/or distorted body-chambers NIGP 30445, 30449, 730437, 784764, 730440 from beds Kpl 4(6), loc. 10; possibly also NIGP 30438, 30441, and 30444 from loc. 9; fragments from beds A and Bl, loc. 13. Description. The body-chamber is approximately 80-90 mm in diameter and three-quarter whorls long. Coiling is moderately involute (u/d = 0-20 to 0-25) with slight egression at the end. The whorl section is rounded-ovate, about as high as wide, to markedly depressed. The ribbing is sharp and dense, with about twenty concave primaries and forty-five to fifty secondaries per half-whorl. The primaries divide irregularly (subfasciculate) at about mid-flank into two or three secondaries which, together with some intercalated ribs, cross straight over the venter. The septal suture is strongly protracted. Discussion. This form has the ornamentation of Gd waageni but differs in the more inflated, depressed whorls and the somewhat larger size. G. nepaulense has a similar whorl section, but differs in the larger size and the much coarser costation on the body-chamber. Almost perfect homoeomorphs are Epimayaites subtumidus (Waagen) and the dubiously distinct E. axioides Spath from the Upper Oxfordian of Kachchh, but their phragnrocone has only biplicate ribs while the body-chamber is unknown from the type specimen. Grayiceras ? gucuoi n. sp. (,y ?) Plate 24, figs. 7 14; text-fig. I2c-e 7 1910 Macrocephalites sp., Uhlig, 1910, pi. 77, fig. 5 a-c. 7 1958 Idiocycloceras rebillyi Collignon, pi. 21, fig. 86. v 7 1976 Indocephalites diadematus (Waagen); Zhao, pi. 6, figs. 5 and 6; pi. 5, fig. 45; pi. 7, figs. 3 and 4. 7 1985 Indocephalites diadematus (Waagen); Wang, pi. 1, figs. 5 and 6. Holotype. The almost complete specimen J2176 with phragmocone, mould (latex cast), Plate 24, figs. 7-10, from bed A5 of loc. 13, 4 km upstream in valley from bridge at km 613/4 of Nyalam Lhasa highway, Tethyan Himalaya, in South Tibet. Name. Derived from village of Gucuo, north of type locality. Other material. Five topotypes, J2177«-e, one of which is almost complete with phragmocone mould and the others incomplete and, in part, juvenile; two incomplete moulds, J2177a, 6, from bed A5, loc. 13; probably also NIGP 84776-84778 from upper part and NIGP 30438 from bed Kp8, loc. 10. Diagnosis. Small (5(06)07 /un, width O6(0-7)O8 /an; upper surfaces sculped with distinct spines which are up to 0-6 /mi tall and randomly placed, not organized in opposite pairs and not united to form transverse ridges; lower surfaces smooth, unsculped, columellae absent. Nexine smooth, separated from sexine by a gap which may be very wide; accurate nexine measurement obstructed by sexine, range 14 16 /mi. Nexine may be rotated inside the sexine. The details of the aperture were not observed, there being no suitably orientated grains. Variation record N Recorded under M. Number of specimens L 4. Locality C Mersa Matruh borehole, N.-W. Desert, Egypt. Grid ref. 31° 19' 43.00" N„ 27° 16' 07.00" E. Rock formation D Kharita. Sample position E MMX-1 5400*, at depth 5400 ft. Sample lithology F Dark shale. Preservation P Good. Repository R Dept. Earth Sciences, Cambridge University, UK. Preparation JP 041. Stubs JPS 73, 74. Earlier records S None. Conclusion T Ends. PENNY: CRETACEOUS ACOLUMELLATE SEMITECTATE POLLEN 411 table 13. Biorecord(cand): retimono-pimple. REFERENCE TAXON DESCRIPTION Group of organisms J Monocolpate pollen. Sequence age G Mesozoic/Cretaceous/Aptian. Originator A Penny, J. H. J. Cambridge University, UK. Origination date B 1987, 9 July, seventeen 45. Taxon name K Biorecord (cand): retimono-pimple. Description M (All observations made with SEM.) Monocolpate pollen, rounded, maximum diameter 14-8(1 5-5)1 61 pi n. Exine semitectate, reticulate, lumina small, rounded or irregularly subcircular, maximum internal diameter 1-4(1 -8)2-4 pm. Microlumina absent. Muri rounded in cross-section, height 0-5(0-54)0-6 pm, width 0-56(0-58)0-8 pm; upper surfaces with sculpture of truncated spines; sides and lower surfaces unsculped; columellae absent. Aperture long, extending half-way round the circumference of the grain and bordered by wide tectate margins that have heavily concentrated sculpture. Lumina adjacent to aperture, smaller than those on the main body of the grain. Nexine smooth, rounded, separated from sexine by a narrow gap. Precise measurement of nexine obstructed by sexine. Variation record N Recorded under M. Number of specimens L 2. Locality C Mersa Matruh borehole, N.-W. Desert, Egypt. Grid ref. 31° 19' 43.00" N„ 27° 16' 07.00" E. Rock formation D Kharita. Sample position E MMX-1 7890, at depth 7890 ft. Sample lithology F Medium pale yellow sandstone. Preservation P Good. Repository R Dept. Earth Sciences, Cambridge University, UK. Preparations. JP 066, 180. Stubs JPS 228, 229, 230, 249. Earlier records S None. Conclusion T Ends. 412 PALAEONTOLOGY, VOLUME 31 Occurrence of retimono-bighole ( cami and CfA records). The first reliable occurrence of grains of this morphology is in sample MMX-1 8183 (mid-Aptian). They then range up through the sequence, the youngest occurrence being in sample MMX-1 5203 (early Albian). There is one occurrence outside this range in cuttings sample MMX-1 10240* (early Aptian). This form is rare, becoming slightly more frequent towards the top part of its range. Biorecord(cand): retimono-pimple Plate 38, figs. 7, 8, 10, 1 1; Table 13 Comparison and remarks: retimono-pimple. This grain type is rare, being represented by only three specimens. Nevertheless it is very easily distinguished by the broad tectate aperture margins and by the markedly smaller size of the lumina adjacent to the aperture. There are no clearly comparable published examples. Occurrence of retimono-pimple. Grains with this morphology occur in only two samples, MMX-1 7890 and MMX-1 8818, both of which are mid-Aptian. Deposited records of retimono-pimple (sample number MMX-1 . ./number of specimens). Biorecord(cand): 7890/2; CfA: 8818/1. DISCUSSION Diversity and taxonomic status This study has shown that detailed SEM examination of large numbers of specimens makes it possible to distinguish many different forms within the large Retimonocolpites peroreticulatus-reticulatus group. There were several other varieties in addition to the thirteen described above, but these are not currently known in sufficient detail to be described. In the past, assignments to several genera have been suggested for grains of the R. peroreticulatus- reticulatus group, two of these being Liliacidites Couper, 1953 and Retimonocolpites Pierce, 1961. The form genus Liliacidites was described by Couper (1953) from the Upper Cretaceous to Eocene in New Zealand, the distinctive features being monosulcate (occasionally trichotomosulcate) apertures, elongated outlines and reticulate, columellate sexine with lumina in the reticulum being of variable size. The precise circumscription of this genus is still disputed, but Doyle et al. (1975) restrict it to grains with 'differentiation of the reticulum into coarsely and finely reticulate areas’, this being their reason for using Retimonocolpites for R. peroreticulatus , R. reticulatus, and related forms. Un- fortunately Retimonocolpites is also unsuitable. It was erected for 'reticulate, monocolpate pollen’, but the type species, R. dividuus Pierce, 1961, is described as columellate although this feature was not specified in the generic description because at the time acolumellate pollen was not known making the explanation of plate 38 Figs. 1, 2, 4, 5. retimono- hairy. 1 and 4, grain number JPR 770/34 (spot), sample MMX-1 7310, prep. JP 195, stub JPS 245, coordinates 785 x 352. 1, x 1600, neg. 240/15; 4, x 7000, neg. 240/16. 2 and 5, grain number JPR 769/2 (CfA), sample MMX-1 5430*, prep. JP 042, stub JPS 75, coordinates 815 x 315. 2, x 1600, neg. 60/5; 5, x 7000, neg. 60/6. Figs. 3, 6, 9, 12. retimono-bighole. 3 and 6, grain number JPR 243B/13 (cand), sample MMX-1 5400*, prep. JP 041, stub JPS 73, coordinates 768 x 248. 3, x 1 600, neg. 39/29; 6, x 7000, neg. 39/30. 9 and 12, grain number JPR I34B/24 (CfA), sample MMX-1 5203, prep. JP 063, stub JPS 103, coordinates 876x290. 9, x 1600 neg. 70/32; 12, x 7000, neg. 70/33. Figs. 7, 8, 10, 1 1. retimono-pimple. 7 and 10, grain number JPR 763/26 (cand), sample MMX-1 7890, prep. JP 066, stub JPS 228, coordinates 754x365. 7, x 1600, neg. 193/22; 10, x 7000, neg. 193/23. 8 and 11, grain number JPR 763/35 (cand), sample MMX-1 7890, prep. JP 066, stub JPS 228, coordinates 742x 338. 8, x 1600, neg. 193/29; 11, x 7000, neg. 193/30. PLATE 38 PENNY, Retimono-hairy, bighole, pimple 414 PALAEONTOLOGY, VOLUME 31 distinction unnecessary. There has since been much discussion about what can be included in this form genus. Walker and Walker (1984) made combined LM/SEM/TEM studies of single grain preparations and in their discussion they noted that R. peroreticulatus should be placed in a separate genus, although they reserved that separation for the future when more specimens had been examined. Juhasz and Goczan (1985) attempted to clarify the classification of early Cretaceous angiosperm pollen, describing several new genera and species from their Albian Hungarian material. One of these new genera, Brenneripollis , was partly distinguished by the tendency for sexine and nexine to be loosely connected, and R. peroreticulatus and R. reticulatus were transferred to it. However, these species are still not clearly comparable with Brenneripollis because they are acolumellate while the genus description specifically refers to columellate varieties. Juhasz and Goczan used only light microscopy, thus limiting their scope for clear recognition of taxa, but they nevertheless recognized the probability that there were many more species with similar morphologies than R. peroreticulatus and R. reticulatus , suggesting that in the past limited observation had led to misidentifications. Thus, R. peroreticulatus and R. reticulatus still remain as problematic taxa; neither can be properly compared with the forms described in this study because SEM detail is unavailable for the type material and even SEM examination of new specimens from the type locality leaves the problem unresolved (Doyle et al. 1975). Both Doyle et al. (1975) and Walker and Walker (1984) provided SEM illustrations of grains referred to as R. peroreticulatus. These grains are different in size, although they share the feature of randomly placed spines which distinguishes retimono-spotspines and retimono-hedgehog. The specimen illustrated by Doyle et al. (1975) compares quite well with retimono-hedgehog while that of Walker and Walker (1984) is more similar to retimono-spotspines, these similarities also extending to their respective observed stratigraphic ranges. Clearly it is possible that these two published specimens might indeed belong to completely separate forms, although this could only be confirmed by SEM examination of more specimens from North America. This being the case the problem arises as to which, if either, should be called R. peroreticulatus. In view of the wide range of forms which can now be distinguished I feel there is justification for the complete separation of grains of this morphology from any affinity with the established generic groups, one of the main criteria for this separation being their complete lack of columellae. Furthermore, in the light of the obvious difficulties associated with attributions to the two existing species R. peroreticulatus and R. reticulatus , I feel that these two species should be retained only for the storage of the lower resolution LM data, while new forms which are more precisely distinguished with SEM detail must be accommodated in entirely new taxa, thus avoiding the problems associated with misattribution and consequent ballooning of taxa for which the whole ranges of morphology are not available. Stratigraphic information The earliest recorded grains of this group are of possible late Barremian or early Aptian age from the Potomac group of North America (Brenner 1963; Doyle 1969; Doyle and Robbins 1977; Hickey and Doyle 1977), Alberta (Singh 1971, 1975), and Israel (Brenner 1974). Unfortunately the dating control of these observations is uncertain, either because of the lack of independent stratigraphic evidence or because of the possibility of sample contamination (e.g. through caving (Brenner 1974)). However, comparison with better dated sequences such as those from southern England (e.g. Hughes et al. 1 979) indicates that these first occurrences are certainly not younger than early Aptian. Definite Aptian and Albian occurrences are confirmed by many observations, sometimes backed by finer stratigraphic control through megafossils (e.g. Singh 1975) or lithostratigraphic correlation (Schrank 1983). The topmost occurrences are of probable Cenomanian age (Laing 1975), although Morgan (1976) figures some similar grains as L. textus from sediments which are possibly as recent as Turonian. The stratigraphic ranges of the forms described in this study are illustrated in the range chart (text-fig. 7). The earliest appearance is in sample MMX-1 10825, from which only three grains were recovered. One of these is CfB to retimono-hedgehog and one is CfB to retimono-smallhole. The PENNY: CRETACEOUS ACOLUMELLATE SEMITECTATE POLLEN 415 — OOOOP0 00 OOOOvj^^J SS3$8Sg3s$gS§?g rogRSg 3 W?r?vv^ YvT? — JS §8. 8. -W- L, k £>0 cn oM K -SPINEROW <■ •NECKLACE <1 -SMALLHOLE Late Barremian Early Aptian Aptian Late Aptian Early Albian Cl text-fig. 7. Stratigraphic distribution of the retimono forms in Mersa Matruh 1 and their abundance expressed as a percentage of the total angiosperm pollen. Solid circles, core sample occurrences; open circles, cuttings sample occurrences; open triangles, no older core sample occurrences; solid triangles, no younger core sample occurrences; solid line, stratigraphic range in core samples; dashed line, distribution in cuttings above older core occurrences; dotted line, probable distribution in cuttings below younger core occurrences. 416 PALAEONTOLOGY, VOLUME 31 age of this sample is ?Barremian to early Aptian, making this first appearance roughly equivalent to the earliest records of similar grains elsewhere. Considerable diversification then takes place in the mid Aptian, with a peak diversity in the mid to late Aptian followed by a slight decline in the late Aptian to early Albian part of the sequence. Numerical abundance, expressed as percentage of the total angiosperm pollen grains recovered, follows the same trend (text-fig. 7). It can be seen from text-fig. 7 that there is a good potential for the use of these grain types in biostratigraphy. Obviously the usefulness of this information will be greatly influenced not only by the morphological detail of any new records but also on the numbers of specimens which are available for comparison, particularly as some of the forms described above are easily confused. The forms with the greatest stratigraphic potential are those that can be recognized most easily. In this respect retimono-ridged would be an ideal marker for the early to mid Aptian because it is easily picked out by its distinctive transverse ribbing, retimono-pimple has similar advantages but its rarity will limit its usefulness. The later part of the sequence is characterized by the presence of RETiMONO-BiGHOLE, which is fairly easily distinguished by its large lumina, and retimono-spotspines, which is more difficult to identify without examination of many specimens because it can easily be confused with the slightly larger retimono-hedgehog. Single or few grains are therefore not very useful in this separation, because there is an overlap in size range between the two types. However, it seems that most of the published examples are of the spinate forms, being similar to retimono- spotspines or retimono-hedgehog (e.g. Doyle et al. 1975; Walker and Walker 1984) or to the forms with spines in pairs such as retimono-spinerow (e.g. Schrank 1983). The apparently exaggerated rarity of the non-spinous varieties might be due to the extensive use of LM only, with which it is impossible to distinguish them. Nevertheless, the much greater numerical abundance of the spinous forms gives them the main potential for future stratigraphic correlation. It is, therefore, most important that future comparisons are made only when SEM detail and large numbers of specimens are available, otherwise this potential will be lost and the distinction between the various forms confused by misattribution and consequent taxonomic ballooning. Speculation on the evolutionary relationships of the group One of the most interesting and distinctive features of the pollen types in this group is the complete absence of columellae. Doyle et al. (1975) regarded this as a feature which distinguished the group as an ‘extinct experimental line’ that had become secondarily acolumellate by reduction, hinting that the tendency towards the acolumellate condition might also account for the relatively sparse columellae of forms such as R. dividuus Pierce, 1961. The earlier appearance of columellate varieties has been well documented, the earliest being recorded from southern England (Hughes et al. 1979; Hughes and McDougall 1987). In addition to the columellate varieties Hughes et al. (1979) recovered pollen which was very similar to retimono-spinerow. These were recorded as Biorecord(cand): retisulc-dubdent and exhibited the interesting feature of short basal remnants of columellae. Independent dating reveals that these are the oldest accurately dated examples of acolumellate pollen, being of earliest Aptian age. The similarity of retisulc-dubdent to the retimono forms described above, together with the possession of remnants of columellae, places this form in an intermediate position between the early columellates and the typical acolumellate varieties. In my view this provides an important clue to the origin of the acolumellate forms supporting the suggestion of Doyle et al. (1975) that the group is indeed secondarily acolumellate (text-fig. 8). Acknowlegements. This work had the financial support of a NERC case award (no. GT4/82/GS/19) with Robertson Research International Limited of North Wales, to whom I am also grateful for providing the samples and obtaining permission from the original drilling company for publication of my results. I am greatly indebted to my supervisor Dr N. F. Hughes who provided much advice and useful discussion. Drs M. C. Boulter and J. A. Doyle made very helpful criticisms of the text. I thank Mr D. Newling and Mr R. Lee for their assistance with the photography and G. W. J. Penny for assistance with the preparation of the plates. PENNY: CRETACEOUS ACOLUMELLATE SEMITECTATE POLLEN 417 0 d EXTINCTION? dQ Id El » ■ $ SEMITECTATE ACOLUMELLATE TECTATE PERFORATE COLUMELLATE RECENT CENOMANI AN/TURONIAN APTIAN BARREMIAN TECTATE IMPEREOF i MTE COLUMELLATE j 'd >d’d| GRANULATE INCIPIE INT COLUMELLATE ‘ ANGIOSPERM GYMNOSPERM GRANULATE EARLY CRETACEOUS text-fig. 8. A chronological scheme for the early evolution of tectate angiosperm exine structure (modified after Walker 1976). 418 PALAEONTOLOGY, VOLUME 31 REFERENCES brenner, G. J. 1963. The spores and pollen grains of the Potomac Group of Maryland. Bull. Mil. Dep. Geol. Mines , 27, 215 pp. 1 974. Palynostratigraphy of the Lower Cretaceous Gevar'am and Talme Yafe formations in the Gevar'am 2 well. Bull. Geol. Surv. Israel, 59, 1-27. chapman, J. L. 1982. Morphology, classification and interpretation of Aptian and Albian angiosperm pollen from Portugal. Ph.D. thesis (unpublished), University of Cambridge. couper, r. a. 1953. Upper Mesozoic and Cainozoic spores and pollen from New Zealand. Palaeont. Bull. Wellington, 22, 1 77. doyle, j. a. 1969. Cretaceous angiosperm pollen from the Atlantic coastal plain and its evolutionary significance. J. Arnold Arbor. 50, 1-35. — jardine, s. and doerenkamp, a. 1982. Afropollis, a new genus of early angiosperm pollen, with notes on the Cretaceous palynostratigraphy and paleoenvironments of northern Gondwana. Bull. Centres Reck. Explor- Prod. Elf- Aquitaine, 6(1), 39-117. — and bobbins, e. i. 1977. Angiosperm pollen zonation of the continental Cretaceous of the Atlantic coastal plain and its application to deep wells in the Salisbury Embayment. Palynology , 1, Proc. VUIth Ann. Mtg. AASP, Houston, 43-78. — van campo, M. and lugardon, b. 1975. Observations on exine structure of Eucommiidites and Lower Cretaceous angiosperm pollen. Pollen Spores, 17 (3). 429-486. hickey, l. j. and doyle, j. a. 1977. Early Cretaceous fossil evidence for angiosperm evolution. Bot. Rev. 43 (1), 3-104. hughes, N. F. 1976. Palaeobiology of angiosperm origins, 242 pp. Cambridge University Press, Cambridge. 1986. The problems of data-handling for early angiosperm-like pollen. In spicer, r. a. and thomas, b. a. (eds.). Systematic and taxonomic approaches to palaeobotany. Systematic Association series no. 31, 235-253. — drewry, g. e. and laing, j. f. 1979. Barremian earliest angiosperm pollen. Palaeontology, 22 (3), 515-535. and mcdougall, a. b. 1987. Records of angiospermid pollen entry into the English Early Cretaceous succession. Rev. Palaeobot. Palynol. 50, 255-272. juhasz, m. and goczan, f. 1985. Comparative study of Albian monosulcate angiosperm pollen grains. Acta bioi, Szeged. 31, 147-172. laing, j. f. 1973. Angiosperm and gymnosperm pollen from the upper Albian to middle Cenomanian of southern England. Ph.D. thesis (unpublished). University of Cambridge. — 1975. Mid-Cretaceous angiosperm pollen from southern England and northern France. Palaeontology, 18 (4), 775-808. MORGAN, R. p. 1976. Albian-Senonian palynology of site 364, Angola Basin. I nit. Rep. DSDP, 40, 915-951. penny, j. h. J. 1986a. Early Cretaceous angiosperm pollen from Egypt. Ph.D. thesis (unpublished). University of Cambridge. 19866. An early Cretaceous angiosperm pollen assemblage from Egypt. Spec. Papers in Palaeontology, 35, 119-132. pierce, R. L. 1961. Lower Upper Cretaceous plant microfossils from Minnesota. Bull. Minn. geol. Surv. 42, 1-86. schrank, e. 1982. Kretazische Pollen und Sporen aus dent Nubischen Sandstein des Dakhla-Beckens (Agypten). Berliner Geowiss. Abb. A 40, 87-109. 1983. Scanning electron and light microscopic investigations of angiosperm pollen from the Lower Cretaceous of Egypt. Pollen Spores, 25 (2), 213-242. singh, c. 1971. Lower Cretaceous microfloras of the Peace River area, northwestern Alberta (2 vols. and appendix). Bull. Res. Coun. Alberta , 28, 1-299 and 301-542. 1975. Stratigraphic significance of early angiosperm pollen in the mid Cretaceous strata of Alberta. Spec. Pap. geol. Ass. Can. 13, 365-389. walker, j. w. and walker, a. g. 1984. Ultrastructure of lower Cretaceous angiosperm pollen and the origin and early evolution of flowering plants. Ann. Mo. Bot. Gdn. 71, 464-521. j. h. j. penny Department of Earth Sciences Cambridge University Typescript received 12 January 1987 Downing Street Revised typescript received 17 July 1987 Cambridge CB2 3EQ FORT1PECTEN TAKAHASHIE A RECLINING PECTINID FROM THE PLIOCENE OF NORTH JAPAN by ITARU HAYAMI Cllld ICHIRO HOSODA Abstract. The adaptive morphology and mode of life of a large bizarre pectinid, Fortipecten takahashii (Yokoyama) from the Pliocene of north Japan and Sakhalin, were examined mainly from the standpoint of relative growth. In spite of its similarity to some extant species of Patinopecten in the early growth stage, Gryphaea- like and unusually heavy valves are formed after the middle stage by a drastic change of growth pattern. In the later stage, unlike many swimming pectinids, the weight of the valves becomes positively allometric to the cube of shell size, indicating remarkable relative thickening of the shell. The decrease of umbonal angle, abrupt inward bending of the shell surface, and disappearance of anterodorsal and postero- dorsal gapes also suggest rapid loss of swimming ability in the middle stage. After it escaped from predators in the young stage, this pectinid probably abandoned a swimming strategy and became an immobile recliner on soft substrates. F. takahashii is thus regarded as an exceptional Cainozoic bivalve which succeeded, though only temporarily, in resurrecting Mesozoic-type reclining life habits in some inland seas of the north-western Pacific region. The diversity of immobile suspension feeders living freely on soft sea-bottoms appears to have declined significantly with geologic time. In the Mesozoic Gryphaea , Exogyra, and some other coiled oysters were undoubtedly full-time recliners, but after the "Late Mesozoic marine revolution (Vermeij 1977)’ gryphaeid oysters only rarely occupied this niche. Gryphaea- like gross morphology is known also in various Mesozoic non-ostreacean bivalves (Carter 1972; Jablonski and Bottjer 1983; Seilacher 1984). In modern seas only a few species of Placuna are regarded as recliners (snowshoe strategists), and all other free-living bivalves on soft substrates have escape strategies such as burrowing and swimming. As was interpreted by Stanley (1970), Thayer (1975, 1979), and LaBarbera (1981), the rarity of full-time recliners in the Cainozoic-Recent seas is most certainly due to increased predation pressure as well as increased bioturbation by deposit feeders. Fortipecten takahashii (Yokoyama, 1930) and its related species from the Pliocene (partly Upper Miocene) of the north-west Pacific region (north Japan, Sakhalin, Kamchatka, and Alaska) reveal Gryphaea- like morphology in the adult stage which is characterized by an unusually heavy shell for pectinids and a bowl-like inflated right (lower) valve. From this and some other reasons one of us (I. H.) has long supposed that the adult individuals of these species may have been almost immobile recliners on soft substrates. If this is true, Fortipecten can be regarded as a unique bivalve which succeeded in developing such a reclining strategy in the post-Cretaceous seas. Since Fortipecten is a conspicuous and relatively short-ranging taxon (commonly treated as a subgenus of Patinopecten or a distinct genus), many authors have investigated its geographic distribution and biostratigraphic significance in addition to making systematic descriptions (see Masuda 1962a, 1978; Kafanov 1986a, b). On the other hand, little has been discussed about the mode of life and the functional significance of its peculiar morphology. In this paper we intend to analyse the morphology of Fortipecten , especially its type-species F. takahashii , to elucidate the mode of life. Because the young individuals of this species are biconvex and thin-shelled like many swimming species of Patinopecten , the analysis of its allometric growth as well as comparative studies with extant free-living pectinids and Mesozoic Gryphaea- like homeomorphs will be important for this purpose. | Palaeontology, Vol. 31, Part 2, 1988, pp. 419-444, pis. 39-40.| © The Palaeontological Association 420 PALAEONTOLOGY, VOLUME 31 PREVIOUS STUDIES AND GENERAL MORPHOLOGY OF FORT1PECTEN Yokoyama (1930) first described Pecten takaliashii from the Pliocene beds at Isousi of the Motodomari area, south Sakhalin. P. agnatus, which was simultaneously erected by him from the same area, is undoubtedly synonymous, because it is, as was interpreted by Masuda (1962u) and some others, merely an immature specimen of P. takahashii. The shell morphology of this pectinid, in fact, changes so greatly with growth that the young specimens look as if they were specifically different from the adult ones. Subsequently the occurrence of P. takahashii was recorded from the Pliocene of other areas of Sakhalin (Khomenko 1931, described under the name of P. pilutunensis; Kryshtofovich 1964), Kamchatka (Ilyina 1963; Gladenkov 1984), various areas of Hokkaido (Yokoyama 1932; Nomura 1935; Takeuchi and Sanbonsugi 1938; Fujie 1958; Akamatsu el al. 1979; Oka and Akamatsu 1979; Uozumi et al. 1986), as well as the Pacific side of north Honshu (Nomura 1938; Masuda 19626; Hayasaka and Hangai 1966; Noda and Masuda 1968). Though considerably wide intrapopulational and geographic variation have been known (Yabe and Hatai 1940; Suzuki 1979), this species is generally characterized by the heavy test, strong convexity of the adult right valve, unusually large auricles, and eleven to seventeen radial costae which are often of two orders of prominence in the adult left valve. Since Kuroda (1932) assigned P. takahashii to Patinopecten , many authors have considered that this pectinid should be taxonomically placed in or near that genus. In fact, the young shell of text-fig. 1. Geographic distribution of Forti- pecten in north Japan and south Sakhalin. Local- ities of studied samples of Fortipecten takahashii and Patinopecten yessoensis are indicated by symbols. HAYAMI AND HOSODA: PLIOCENE RECLINING PECTINID 421 Pecten takahashii, as shown later in detail, is weakly convex, thin-shelled, and considerably similar to Patinopecten ( Mizuhopecten ) yessoensis (Jay, 1857), a commercial scallop in the north-western Pacific. Since Yabe and Hatai (1940) proposed the subgenus Fortipecten for Pecten takahashii , most palaeontologists have regarded it as a valid genus-group name, but some different taxonomic evaluations have been made about the peculiar morphology of this species. For example, Akiyama (1962) regarded the heavy and strongly inflated shell as due to adaptation to some lagoonal environment and denied its subgeneric distinction from Patinopecten. In contrast, Masuda (1962u) treated Fortipecten as a distinct genus, and furthermore a new subfamily Fortipectininae was proposed (Masuda 1963). In addition to Pecten takahashii , the following species seem to belong to Fortipecten , though their diagnostic characters are not necessarily clear: Pecten ( Plagioctenium ) hallae Dali. 1921, from the Pliocene of east Alaska. [? = P. { Patinopecten ) rhytidus Dali, 1921 1. P. mironovi Khomenko, 1934, from the Pliocene of Sakhalin. P. (Pd) sachalinensis Ilyina, 1957, from the Pliocene of Sakhalin. P. ( Fortipecten ) kenyoshiensis Chinzei, 1960, from the Pliocene of north Honshu. P. (F.) makarovi Kryshtofovich, 1964, from the Pliocene of Sakhalin. Fortipecten mollerensis MacNeil, 1967, from the Pliocene or Upper Miocene of south Alaska. F. kuroishiensis Kotaka and Noda, 1967, from the Upper Miocene of north Honshu. Because the intrapopulational and geographic variation of each species may be wide, its taxonomic discrimination and phylogenetic relation to other species should be studied at the population level In this paper we do not discuss this problem, because the examined material is still restricted to several samples from north Japan and south Sakhalin. A REVIEW OF SUZUKI’S BIOMETRIC STUDY An elaborate biometric study of F. takahashii was carried out by Suzuki (1979). He investigated various characters of the shell (mainly right valve) such as shell height, convexity, surface curvature, weight, and development of each shell layer on the basis of four samples from different areas of Hokkaido, and discussed the mode of shell growth and geographic variation. His data and conclusions seem to be reliable. Because his paper appeared in a special Japanese publication of limited distribution, we review here, with his permission, his important conclusions in the following section. 1. The samples of F. takahashii are morphologically classifiable into two forms. Though the variation in each sample is wide. Form A, occurring in the Japan Sea side of Hokkaido (i.e. Takikawa and Ishikari-numata areas) as well as south Sakhalin is characterized by a heavier test and stronger convexity of the right valve than Form B from the Pacific side of Hokkaido (i.e. Tokachi and Akan areas) and north Honshu. 2. The surface of the right valve of F. takahashii often bends abruptly inward at the middle growth stage when shell height attains about 70 mm. After this stage the shell becomes much thicker and convexity becomes stronger especially in the samples belonging to Form A. The crook- backed appearance and strong convexity of right valves are mainly attributable to this bending rather than the change of surface curvature. 3. The subvertical section of an adult right valve shows almost uniform robustness of the whole shell which is formed by the much thickened outer layer in the ventral area and the thickly accumulated inner layer in the umbonal area. Suzuki (1979, text-fig. 4) examined the relation between shell height (H) and weight (IF) of right valves in his sample from the Ishikari-numata area which is composed of many well-preserved specimens of various growth stages. His data were plotted again by us on a double logarithmic scatter diagram in order to recognize their allometric relation more clearly. From this diagram 422 PALAEONTOLOGY, VOLUME 31 the relation between H and W was recognized as diphasic allometry; that is, in the early growth stages ( H < 70 mm) W increases nearly in proportion to H 3, whereas the growth ratio (or the slope of the best-fit in the diagram) in the later growth stages ( H > 70 mm) is as large as 1-4. Though it may be partly due to the crook-backed shape of right valves, it is strongly suggested that the shell is at first thickened nearly isometrically, and that the heaviness of adult valves is formed by the significant positive allometry of shell thickness to shell size after the middle growth stage. Similar change of the growth pattern was also ascertained in the samples collected by us (see text-fig. 7). MATERIAL AND EXAMINED CHARACTERS In order to elucidate the adaptive significance of the peculiar morphology of F. takahashii, the relative growth of the shell was analysed on the basis of the following samples (articulated, right and left valves are indicated by CV, RV, and LV, respectively, and observed ranges of length ( L ), height (//), and weight (IV) are shown in parentheses). Sample In [UMUT CM 181 16], 25CV (L = 34-3 147 0mm, H = 35-4- 1 54-9 mm, W= 3-37 600- 1 4 g), 152RV (L = 13 0 137 0 mm, H= 13 0 146-5 mm, W = 0-12-284 08 g), and 151LV (L = 12-2- 1 33-9 mm, H = 12-3 136-6 mm, W = 0-07-227-25 g). Locality: Lower Pliocene Horokaoshirarika Formation at the river floor of the Horoshintachibetsu, about 3-5 km north-west of Ishikari-numata JR Station, Numata Town, Uryu County, central Hokkaido. This sample contains individuals of various growth stages and suitable for the study of relative growth. Sample Tk [UMUT CM 18117]. 22CV (L = 63-2 170 6 mm, H = 62-9 169-5 mm, W= 19-23-707-88 g), 38 RV (L= 22-3-159-5 mm, H = 23-0 164-Omm, W = I -22-422-99 g), and 49LV (L = 44-5-159-5 mm, H = 48-0 152-0 mm, W = 3-43 364-25 g). Locality: Lower Pliocene Takikawa Lormation at the river floor of the Sorachi, East 2-chome of Takikawa City, about 4 km east of Takikawa JR Station, central Hokkaido (partly collected by Y. Iwasaki). Most speciments are well-preserved adult individuals. Sample Sd [UMUT CM18118], 5CV (L = 102-0 161-2 mm, H = 98-9 164-Omm), 4RV (L = 126-1 160-2 mm, H = 1214 154-5 mm), and 7LV (L = 84-5 1 50-5 mm, H — 80-8 144-1 mm). Locality: Lower Pliocene Tatsunokuchi Lormation near Yodomibashi Bridge, Sendai City, Miyagi Prefecture. Though the sample size is small, this is the best-preserved sample from the Pacific side. The relative growth of a living species, Patinopecten yessoensis, was also examined and compared with that of F. takahashii on the basis of the following samples. Sample Ak [UMUT RM18U9], 49CV (L = 18-2-165-8 mm, H = 17-6 160-9 mm, W = 0-39-265-18 g). Locality: Akkeshi Bay, off Akkeshi Town, Kushiro County, eastern Hokkaido. Sample Nt [UMUT RM18120], 131CV (L = 1 5-4- 146-5 mm, H = 15-9 144-4 mm, W = 014-1801 1 g). Locality: Notoro Estuarine Lake, near Abashiri City, northern Hokkaido. In spite of the large sample size, this sample is not necessarily ideal, because most of the specimens are probably cultivated individuals. The shell is often abnormally thin and sometimes distorted. Sample Wk [UMUT RM 18121], 14CV (L = 122-5 192- 1 mm, H = 118-7 173-5 mm, W= 1 13-58-331 00 g). Locality: Soya Strait, off Wakkanai City, northern Hokkaido. This sample represents only the later growth stages. Various morphological terms relating to pectinid valves in this study are adapted mainly from those defined by Waller (1969). We examined various linear measurements, angles, and weight of these specimens. Among others, the following measurements are useful because their interrelations indicate allometric change of shell form. L: Maximum length of disc, measured in the direction parallel to outer ligament (or hinge axis). H: Distance between the origin of growth and the most ventral point of the valve, measured in the direction perpendicular to outer ligament. D: Length of outer ligament, which also represents the total length of the two auricles. T: Maximum thickness of articulated valves (in closed condition), measured in the direction perpendicular to commissure plane. HAYAMI AND HOSODA: PLIOCENE RECLINING PECTINID 423 U\ Umbonal angle (in degrees) between two lines from the origin of growth to the anterodorsal and posterodorsal shoulders of disc, as was defined by Stanley (1970, p. 20). Only left valves were used. W: Weight of a valve (or articulated valves). Computations for statistics were made separately on each valve excluding some broken specimens which were inadequate for measurement. Besides, the extent of anterodorsal and posterodorsal gapes, marginal discrepancy of discs, symmetricity of disc, surface curvature, spiral angle in vertical section, and size, position, and obliquity of striated (quick) and smooth (slow) adductor muscles may be intimately related to the difference of life habit. These characters were, though mostly qualitatively, examined and evaluated at need. All the specimens used in this study are preserved in the University Museum, University of Tokyo (UMUT). ALLOMETRIC CHANGE OF SHELL FORM As vigorously discussed by Gould (1971 ), isometric (proportional) growth of pectinid valves would result in the decline of swimming ability, because the gravitational force scales at the cube of the length (L3) while the lifting force scales at smaller powers of the length (generally supposed as L2, provided that the swimming velocity is independent of shell size). In order to maintain swimming ability, many free-living pectinids appear to undergo various allometric changes especially in the later growth stages. In the early growth stages F. takahashii and P. yessoensis share various characteristics, e.g. thin and subequiconvex valves, simply rounded radial costae, shagreen microsculpture on the left valve, and distinct anterodorsal and posterodorsal gapes between discs. As commonly recognized in many extant free-living scallops, the early dissoconchs of F. takahashii and P. yessoensis possess several denticles of active ctenolium along the anterodorsal margin of the right disc, which indicate the presence of a byssate stage before the free-living stage. The general morphological similarity seems to suggest that the mode of life in the early growth stages is nearly identical between the two species. The peculiar Gryphaea- like outline and heavy test formed in the later stages of F. takahashii , as demonstrated by Suzuki (1979), are largely attributable to the highly allometric growth after the middle stage when the shell height (or length) attains about 70 mm. The change of growth pattern at this size is drastic enough to assume some significant change in the mode of life. In P. yessoensis the outline of the shell also changes from subequiconvex to right-convex, but no critical change of growth pattern is perceptible. The shell appears to grow nearly isometrically or slightly allometrically until the latest growth stage. In the following section we describe the relative growth of these two pectinids by means of current bivariate methods (the results are collectively shown in Table 1) and discuss the adaptive significance of the differences. Shell elongation and umbonal angle The ventral elongation of pectinid discs (or the ratio of height/length) seems to be inversely correlated to the umbonal angle. Stanley (1970) pointed out that free-living pectinids are generally characterized by a broader (lower) shell and a larger umbonal angle than byssate pectinids. He interpreted that a large umbonal angle is advantageous for swimming motion, because the currents expelled from the gapes of anterodorsal and posterodorsal margins pass more directly in a dorsal direction, so that the resultant propelling force is increased. Moreover, by analogy with a hydrofoil, the broader disc (or smaller ratio of height/length) must produce a greater ratio of lift/drag (Karman and Burgers 1963). We examined the relationship between the umbonal angle ( U ) and shell height (H) through the growth of F. takahashii and P. yessoensis. As shown in text-fig. 2, the mean umbonal angle is nearly the same in the early growth stages of the two species, but ontogenetically it becomes smaller in F. takahashii and larger in P. yessoensis. Owing to the wide intrapopulational variation 424 PALAEONTOLOGY, VOLUME 31 table 1. Bivariate analyses for the recognition of allometric growth, a, aa, and b: slope, its standard error and E-intercept of the best-fit (reduced major axis, Y = aX+b ), respectively. Species Sample Valve Size X Y N r a Oa b F. takahashii In RV All Log L Log H 159 0-9979 1 -003 0-005 1-002 In RV All Log L Log D 81 0-9945 1-150 0-013 0-390 In RV H > 70 Log H Log W 57 0-9739 4-148 0-125 3-878 x 10-7 In RV H < 70 Log H Log W 81 0-9881 3-025 0-052 4-429 x 10-5 In LV All Log L Log H 152 0-9988 0-978 0-004 1-114 In LV All Log L Log D 65 0-9933 1189 0-017 0-349 In LV H > 70 Log H Log W 55 0-9678 3-749 0-127 2-384 x 10^6 In LV H <70 Log H Log W 80 0-9888 3-081 0-051 3-514 x 10~5 In cv All Log H Log T 23 0-9825 1-380 0-054 6-950 x 10^2 Tk RV All Log L Log H 56 0-9906 1-025 0-019 0-908 Tk RV All Log L Log D 35 0-9526 1-156 0-059 0-354 Tk RV H > 70 Log H Log W 35 0-9672 4-100 0-176 4 519 x 10-7 Tk LV All Log L Log H 68 0-9928 1-010 0-015 0-976 Tk LV All Log L Log D 32 0-9703 1-172 0-050 0-353 Tk LV H > 70 Log H Log W 46 0-9612 3-749 0-152 2-512 x 10^6 Tk CV All Log H Log T 22 0-9093 1-917 0-170 4-887 x 10-3 P. vessoensis Ak RV All Log L Log H 49 0-9989 0-999 0-007 0-986 Ak RV All Log L Log D 41 0-9960 0-996 0-014 0-542 Ak RV H > 90 Log H Log W 19 0 9656 2-937 0-175 4-743 x 10-5 Ak RV H <90 Log H Log W 30 0-9953 3-188 0-056 1-691 x 10“5 Ak LV All Log L Log H 49 0-9989 0-992 0-007 1-029 Ak LV All Log L Log D 43 0-9955 0-991 0-014 0-564 Ak LV H > 90 Log H Log W 19 0-9777 2-778 0-134 9-861 x 10-5 Ak LV H <90 Log H Log W 30 0-9952 3-218 0-057 1-538 x 10^5 Ak CV All Log H Log T 48 0-9903 1-009 0-020 0-256 Wk RV All Log L Log H 13 0-9841 0-846 0-042 2-077 Wk RV All Log L Log D 11 0-9756 1 088 0-072 0-332 Wk RV All Log H Log W 13 0-9682 2-774 0-192 1 016 x 10~4 Wk LV All Log L Log H 14 0-9879 0-880 0-036 1-777 Wk LV All Log L Log D 13 0-9103 1-210 0-139 0-186 Wk LV All Log H Log W 14 0-9389 3-012 0-277 2-825 x 10“5 Wk CV All Log H Log T 13 0-9357 0-828 0-081 0-596 the absolute values of the correlation coefficient (r) between H and U are not high (0-4028 in F. takahashii and 0-7004 in P. yessoensis ), but the correlations are certainly significant with 99-9 % confidence. The ontogenetic enlargement of the umbonal angle is certainly effective in reducing difficulty of swimming. In other words the growth pattern of P. yessoensis is normal for swimming pectinids, but that of F. takahashii is quite abnormal. The ontogenetic change of shell elongation can be quantified by the relative growth of the height (//) to the length (L). The reduced major axes on double logarithmic scatter diagrams (not illustrated here) indicate that in both species H increases nearly isoinetrically to L. We could not detect any statistically significant difference between the growth ratios of the most suitable samples In and Ak. It must be noted, however, that the growth patterns of the two species become considerably different toward the latest stage, because the growth ratios in the adult-rich samples of the two species differ significantly from each other (1 -025 + 0-019 in Sample Tk, and 0-846 + 0-042 in Sample Wk). We often encountered very tall adult individuals of F. takahashii in which H is much larger than L. Such a large form ratio is never found in the adult specimens of P. yessoensis and other extant pectinids with adept swimming ability. Development of auricles The anterior and posterior auricles of F. takahashii are unusually large. This is generally regarded as a diagnostic character of the genus Fortipecten (Masuda 1962#; Hertlein in Cox el al. 1969; HAYAMI AND HOSODA: PLIOCENE RECLINING PECT1NID 425 Shell height (H mm) text-fig. 2. Relation between shell size and umbonal angle. The umbonal angle becomes larger with growth in Patinopecten yessoensis (as in many other swimming pectinids) but smaller in Fortipecten takahashii. The left valves of samples Ak and In were analysed. The absolute values of r are not high, but the correlations are very significant (P < 0 001). Kafanov 1986 b). The ontogenetic change of the auricular development can be roughly shown by the relative growth of the length of outer ligament ( D ) to the overall length of disc (L). As shown in text-fig. 3, the reduced major axes signify that D is nearly isometrically increased to L in P. yessoensis, but that the relative growth is decidedly positively allometric in F. takahashii. The calculated growth ratios (slope of reduced major axis) in the latter species (Sample In) is IT 50 ±00 13 for the right valve and IT 89 + 0 0 17 for the left valve. F. takahashii possesses somewhat larger auricles than P. yessoensis even in early growth stages, and the difference becomes more striking with growth. Because nearly isometric or negatively allometric growth of auricles is commonly seen in ordinary pectinids (both byssate and free-living), the growth pattern of F. takahashii must be said to be unusual. The functional meaning of auricles in non-byssate pectinids is poorly known, but the symmetricity of auricles may be required for swimming straight forward. The large auricles of F. takahashii may be useful for a reclining life, because they probably stabilize the horizontal living attitude on soft substrates. Outer ligament length D mm 426 PALAEONTOLOGY, VOLUME 31 text-fig. 3. Allometric relation between shell length (L) and outer ligament length (D). The unusually large auricles of Fortipecten takahashii are the product of a positively allometric growth of D against L. The samples In and Ak were analysed. Marginal discrepancy of discs Marginal discrepancy of discs is generally insignificant in byssate pectinids, but the right (lower) valve is often somewhat larger and overlaps the ventral margin of the left (upper) valve in right- convex free-living genera (e.g. Pecten , Euvola, and Patinopecten). The degree of discrepancy varies among species but generally seems to relate to the convexity of the right valve. Both in P. yessoensis and F. takahashii the discrepancy is small in the early growth stages and becomes gradually larger with growth. Though the marginal part of the right valve is apt to be broken in articulated specimens of F. takahashii , the discrepancy sometimes exceeds 10 mm in the latest stage. Considering the life habits of plano-convex free-living pectinids (Stanley 1970; our observations in aquaria), the overlapping ventral margin of the lower valve appears to be effective in permitting a horizontal and half-buried living attitude without clogging so that only the tentacled mantle margin is exposed above the soft substrate. In plano-convex immobile recliners this feature may be similarly efficient in permitting cleansing of the mantle cavity. Anterodorsal and posterodorsal gapes It has been noticed that many free-living pectinids with an adept swimming habit are characterized by the development of gapes along the anterodorsal and posterodorsal margins of discs (Waller 1969; Stanley 1970; and others). Such gapes are generally undeveloped in byssate pectinids, though in a few species (e.g. Aequipecten opercularis) this feature coexists with an active ctenolium whicl indicates the presence of a byssus. The gapes seem to be effective in swimming because the animal can expel jet currents through them. The young individuals of P. yessoensis and F. takahashii are similarly characterized by narrow but distinct anterodorsal and posterodorsal gapes. The gapes are maintained throughout growth and are somewhat widened toward the later growth stages in P. yessoensis , whereas the anterodorsal HAYAMI AND HOSODA: PLIOCENE RECLINING PECTINID 427 and posterodorsal margins are almost closed after the middle growth stage in F. takahashii (PI. 39, fig. lc, d). The disappearance of gapes seems to occur immediately after the change of growth pattern when the shell size attains about 70 mm in length. The size of gapes does not seem to be proportionate to swimming ability, but their disappearance in the middle stage of F. takahashii indicates a loss of swimming ability. For reclining life such gapes must be unnecessary and rather disadvantageous. Symmetricity of disc The disc of F. takahashii , unlike P. yessoensis and many other swimming pectinids, is commonly asymmetric with a more or less prosoclinal demarcation line. The obliquity is as large as 15 in some specimens (e.g. UMUT CM 18117b; PI. 40, fig. 3 a). Such an asymmetric disc is probably disadvantageous for swimming motion and has never been observed in P. yessoensis and other adept swimming pectinids. Shell convexity The relation of shell convexity to swimming ability in pectinids has been discussed by some authors (see Waller 1969). Judging from the great variability of shell form of swimming species involving left-convex, equiconvex, right-convex, and concavo-convex, the mode and extent of shell convexity are not necessarily decisive factors for swimming, even though they are intimately related to swimming orientations (Waller 1969; Stanley 1970). The convexity of a pectinid valve, especially that of F. takahashii , is somewhat difficult to quantify with accuracy, because the commissure of both valves does not he within one plane. Therefore, we analysed the allometric relation between shell height ( H ) and maximum convexity ( T ) in articulated individuals (in closed condition). As shown in text-fig. 4, the relative growth Shell height H mm text-fig. 4. Allometric relation between shell height (H) and shell convexity (T) in articulated specimens (in closed condition). The relative growth is nearly isometric in Patinopecten yessoensis (Sample Ak), but T is highly positively allometric against H in Fortipecten takahashii (Sample In). 428 PALAEONTOLOGY, VOLUME 31 Fortipecten takahashii Patinopecten yessoensis 30mm text-fig. 5. Subvertical sections of full-grown valves of Fortipecten takahashii (UMUT CM18117c) and Patinopecten yessoensis (UMUT RM18119a). Note the thin outer layer in the earlier part and the abrupt thickening of each shell layer in F. takahashii. Solid part: myostracum (prismatic aragonite), dotted part: inner shell layer (crossed lamellar aragonite), open part: outer shell layer (foliated calcite), broken lines: projection of the margins of the adductor muscle on the subvertical section. is nearly isometric in P. yessoensis (Sample Ak) but positively allometric in F. takahashii (Sample In). If only large specimens are considered, the growth ratio of the latter species, as shown by the Sample Tk, is as large as 1-9. Because the convexity of the left valve is generally weak throughout growth, the highly allometric nature is largely attributable to ontogenetic change of the right valve. EXPLANATION OF PLATE 39 Fig. I. Fortipecten takahashii (Yokoyama, 1930). Full-grown individual, UMUT CM18117a, x 0 65. 1 a: right view, 1 b: left view, lc: posterior view, \d : anterior view. Loc.: Pliocene Takikawa Formation at the river floor of the Sorachi, East 2-chome of Takikawa City, central Hokkaido (collected by Y. Iwasaki). PLATE 39 HAYAMI and HOSODA, Fortipecten 430 PALAEONTOLOGY, VOLUME 31 Fortipecten takahashii Patinopecten yessoensis M- +, * ibii lift db. text-fig. 6. Computer graphics showing the method to obtain the best-tit of an equiangular spiral and its spiral angle. The dots were obtained from an actual specimen along the top of a radial costa near the demarcation line. The spiral angle is about 58° in Fortipecten takahashii (UMUT CM18117a) and about 33° in Patinopecten yessoensis (UMUT RM181 19b). The outer surface of a pectinid valve in the vertical section was approximated by an equiangular spiral lacking the initial part near the origin instead of a spiral with enlarging spiral angle. In vertical sections of F. takahashii , as described by Suzuki (1979) and reconfirmed here (text-fig. 5), there is often a critical bending point at about 70 mm distance from the beak, where the relative convexity begins to increase at an unusually high rate. Such a remarkable ontogenetic change of shell convexity has not been observed in swimming pectinids. Spiral angle and surface curvature As pointed out by Fukutomi ( 1953) and Raup (1966) a valve of any bivalve species is fundamentally a rapidly expanding spiral tube, the section of which along the demarcation line can be approximated by an equiangular (logarithmic) spiral. The early part of the spiral near the origin, however, is commonly more or less skipped and not represented by the dissoconch. The geometric nature of an equiangular spiral is, in theory, entirely determined by the spiral angle, which is the constant angle between the radius and the tangential line. The right valve of F. takahashii is bowl-like and intuitively shows a much larger spiral angle (or smaller expansion rate) in comparison with that of P. yessoensis. It is, however, not necessarily easy to measure the spiral angle of a valve, because the origin of the supposed equiangular spiral is not shown in the vertical section. Therefore, we estimated its approximate value by the following procedure (see also text-fig. 6). 1 . Using a three-dimensional digitizer we obtained a line drawing which shows a perpendicular projection of the top of a radial costa nearest the demarcation line. 2. Two-dimensional co-ordinates with arbitrary axes were determined for more than twenty points (including the beak and a point on the ventral margin) on the line drawing. 3. Using a microcomputer, the best-fit equiangular spiral for these points was determined by the least- square method, and its origin, expansion rate, and spiral angle were calculated. EXPLANATION OF PLATE 40 Figs. 1 3. Fortipecten takahashii (Yokoyama, 1930). 1 a-c, young individual (swimming stage), UMUT CM18116a, x 1. la: right view, 16: left view, lc: posterior view. Loc.: Pliocene Horokaoshirarika Formation at the river floor of the Horoshintachibetsu, Numata Town, central Hokkaido. 2 a-c, middle-aged individual showing relatively strong convexity of right valve, UMUT CM 181 16b, x 0-65. 2a: right view, 2b: left view, 2c: posterior view. Loc.: the same as above. 3 a, b , middle-aged individual showing a prosoclinal disc and abrupt inward bending of right valve, UMUT CM 18117b, xO-65. 3a: right view, 36: posterior view, Loc.: the same as Plate 39, fig. 1 a-d. PLATE 40 HAYAMI and HOSODA, Fortipecten 432 PALAEONTOLOGY, VOLUME 31 As shown in the computer graphics (text-fig. 6), the actual curve of the outer surface in the right valve of F. takahashii (and also P. yessoensis) can be well approximated by some part of an equiangular spiral, though considerable deviation may occur near the inner bending point in F. takahashii and in the latest stage of P. yessoensis. The spiral angle (or expansion rate) is remarkably different between the two species. The result of our calculations on a few right valves of each species indicates that the average spiral angle is about 56° in F. takahashii but only about 33° in P. yessoensis. The profile of an adult right valve of F. takahashii looks semicircular not only in vertical section but also in transverse section. In other words, the valve looks hemispherical, because the early part of the expanding tube is largely skipped. As observed in the samples In and Tk of F. takahashii , the extent of surface curvature seems to vary greatly among individuals. Some specimens are of considerable size and relatively weakly inflated, while there are some small specimens with strong convexity (e.g. PI. 40, fig. 2 a-c). The remarkable variation of convexity may be partly due to the variability of the spiral angle but, we suppose, largely attributable to the difference in the extent of skipped portion near the origin of an equiangular spiral. It is a general tendency that an individual with a large skipped portion can grow to a large ultimate size. Nevertheless, the surface curvature of the young right valve is not much different between F. takahashii and P. yessoensis. The curvature becomes weaker much more slowly with growth in F. takahashii. Such a hemispherical shape of the lower valve is probably advantageous for a stable reclining attitude against water currents and biological disturbance from any direction. Shell thickness and weight Because isometric growth of pectinid valves has to result in a steady decrease of the ratio of lift/gravity (Gould 1971), relative thinning of the shell will be particularly effective in maintaining swimming ability until the later growth stages. Jefferies and Minton (1965, text-fig. 10) suggested such a negatively allometric growth of shell thickness to shell size in Bositra huchi from the Jurassic, which was regarded by them as a nektoplanktonic bivalve. Tanabe (1973) also documented a significantly negative allometry of the thickness of the outer prismatic layer to the shell height in Sphenoceramus naumanni and the early growth stages of S. schmidtii, for both of which a pseudoplanktonic mode of life was suggested. Though we are not aware of any well-documented biometric study treating this problem with extant pectinids, it would not be surprising if there were a marked difference in the growth patterns between byssate and swimming species. The thickness of a pectinid shell, however, is somewhat difficult to measure with accuracy, because, even in a vertical section, it is strongly influenced by the radial costation, and because comparison of different individuals at a homologous point is not easy. In the circumstances we examined the allometric relation between shell size (height in this case) and shell weight in each valve. Because weight generally scales at the cube of size, isometric growth will be shown by a growth ratio of 3. The results of our bivariate analyses are collectively shown in text-fig. 7, which indicates considerably different modes of allometric growth between F. takahashii and P. yessoensis. In F. takahashii the relative growth can be approximated by diphasic allometry; that is, in the early stages it is nearly isometric, whereas in the later stages it evidently turned to positive allometry. This change is observable in both valves. As shown in the vertical section (text-fig. 5), the outer layer and myostracum become suddenly thicker and the accumulation rate of the inner layer must become much larger after the middle stage. Though some artificial division of growth stages is inevitable to analyse the diphasic nature, the calculated growth ratios in Sample In are significantly different between the smaller ( H < 70 mm) and larger (// > 70 m) individuals (3 025 + 0 052 and 4-148 ±0-1 25 for the right valve, and 3-081 +0-051 and 3-749 + 0-127 for the left valve, respectively). We presume that the more drastic change of growth pattern in the right valve is related to the surface bending in the middle stage, but the relative thickening of the shell with growth is obvious in both valves of F. takahashii. In P. yessoensis the relative growth can also be regarded as diphasic, but the direction of change Shell weight W g Shel1 weight HAYAMI AND HOSODA: PLIOCENE RECLINING PECTINID 433 -I I I L- 0) 100 ■ so- io - 5 - 1 . 0.5 - 0.1 4 F . takahashii larger specimens Rv JT 'A- T v LV W = (3.878 * 10’7]H4'148 • y* • w larger specimens & = (2.384 *10"‘)H3'7" r af JC smaller specimens W =(4.429 x 10‘5)H3'025 r / */ •y * T 20 smaller specimens W = (3.514 x 10"5]H3018 -i 1 1 1 1 1 — r— r 50 100 200 Shell height H mm Shell height H mm text-fig. 7. Allometric relation between shell height (H) and shell weight (IV), which indirectly indicates the ontogenetic change of relative shell thickness. In the later stages the shell becomes relatively much thicker in Fortipecten takahashii but slightly thinner in Patinopecten yessoensis. The scale of abscissae is three times larger than that of ordinates. 434 PALAEONTOLOGY, VOLUME 31 is just the reverse. Though the calculated growth ratios are slightly larger than 3 in the early stages of Sample Ak, relative thinning of the shell seems to occur in the later stages. If shell length (L) is applied as a variable for shell size, more obvious negative allometry will be concluded. In conclusion, the growth pattern of P. yessoensis seems to reduce the swimming difficulty in the later stages, whereas in F. takahashii the relative shell thickening indicates rapid loss of swimming potential after the middle stage. Obliquity , size, and position of adductor muscles Thayer (1972) pointed out that the oblique orientation of adductor muscles (especially the longitudinal axis of the striated portion) to commissure plane is a functionally meaningful structure in swimming pectinids, because it enables quicker clapping motion of valves. He studied the obliquity in various free-living and attached species of monomyarian bivalves, and ascertained an intimate relation between this feature and swimming ability; that is, the obliquity is generally large in adept swimmers such as Amusium japonicum and Placopecten magellanicus but very small in byssate and cemented species. Moreover, in Hinnites multirugosus it was said that the striated muscle is notably oblique in the early (free-living) stages but almost perpendicular in the later (cemented) stages. In all the growth stages of Patinopecten yessoensis and early stages of F. takahashii, as partly shown in text-figs. 5 and 8, the striated muscle is certainly oblique (more than 15° as measured in a plane normal to the hinge) like many other swimming pectinids. In the later stages of F. takahashii, however, the obliquity becomes almost zero. The ontogenetic change of the size and position of striated muscle must be also important in relation to swimming mechanics. As interpreted by Gould (1971), an allometric enlargement of striated muscle and its dislocation from a posterodorsal to a more central portion of the disc, which are observable in Placopecten magellanicus and some other free-living pectinids, seem to result in a more powerful clapping motion and reduce the difficulty of swimming in the later stages. Such an improvement of musculature actually occurs in Patinopecten yessoensis and probably in the early stages of F. takahashii. However, relative enlargement of the smooth portion (instead of the striated portion) seems to occur in the later stages of F. takahashii (text-fig. 8). The massive smooth muscle in adult individuals of this species must have produced strong tension in holding the valves tightly closed, which is probably more necessary for a reclining life. Incidentally, remarkable relative enlargement of smooth muscle occurs in Chesapecten jeffersonius from the Pliocene beds around Chesapeake Bay (Gould 1971). Because the adult valves of such large species of Chesapecten, described by Ward and Blackwelder (1975), are often as heavy as those of F. takahashii, swimming ability may have been completely lost in the later stages, as assumed by Gould (1971) and Miyazaki and Mickevich (1982). The shell form of Chesapecten, however, is nearly equiconvex throughout growth and much different from that of F. takahashii. Allometric change of some other characters In pectinids, as studied by Trueman (1953), the central non-calcified portion of the resilium (inner ligament) serves to open the valves by its compressive elasticity, when the contractional force of the adductor muscles is released. Therefore, it may be natural that thick-shelled pectinids generally require a larger resilium than thin-shelled pectinids of similar size. The relative size of resilial insertion appears to be nearly equal between P. yessoensis and F. takahashii in their early stages. In adult specimens of F. takahashii, however, it becomes unusually large in response to the relative thickening of the valves. The apical angle of the resilial pit is also much widened, and the ratio of its area/valve surface becomes larger in the later stages (text-fig. 8). In P. yessoensis, like many free-living and byssate pectinids, the shape and relative size of resilial pit are almost invariant throughout growth. The ontogenetic change of surface ornamentation is also remarkable in F. takahashii. In the early stages the radial costae are commonly simple and subequal in prominence like many species of Patinopecten. In the later stages, however, the costae becomes relatively narrower (almost HAYAMI AND HOSODA: PLIOCENE RECLINING PECTINID 435 text-fig. 8. Ontogenetic change of muscle insertions and resilium pit in Fortipecten takahashii. The smooth portion of the adductor muscle and resilium pit become relatively large with growth, re: resilium pit, st: striated portion, sm: smooth portion, il: distribution of inner shell layer. invariant in absolute width) in the right valve and show two orders of prominence in the left valve. In every sample of F. takahashii the adult left valve has commonly five strong radial costae, each interval of which has two (rarely one) weaker costae. This arrangement is somewhat similar to that of the Cretaceous genus Neithea. In some middle-sized specimens of F. takahashii there are small openings at the ventral ends of these five stronger costae, though their function is unknown. Range of morphological variation As stated above, it is noticed that the intrapopulational variation of various morphometric characters (e.g. umbonal angle, disc obliquity, shell convexity, and surface curvature) is much broader in F. takahashii than in P. yessoensis. The reason is not exactly clear, but one possible 436 PALAEONTOLOGY, VOLUME 31 explanation is that a reclining life habit does not require such strict design of shell form as a swimming habit. HABITAT AND GEOGRAPHIC VARIATION OF F. TAKAHASHII F. takahashii has been regarded as a characteristic element of the Pliocene Takikawa-Tatsunokuchi Fauna which was extensively distributed in Hokkaido and the Pacific side of north Honshu (text- fig. 1). This fauna contrasts with the Omma-Manganji Fauna on the Japan Sea side of Honshu (and south Hokkaido), and also with the Kakegawa Fauna on the Pacific side of south-west Honshu (Chinzei 1978, 1986). The three faunas are at least in part coeval. Constituent species of the Pectinidae as well as other molluscs are largely different between them. The Takikawa- Tatsunokuchi Fauna was distributed in south Sakhalin and further north and east, and is regarded as having lived under the influence of cold water currents, though embayment conditions are generally suggested by the sedimentary facies and constituent molluscs. At various localities of Hokkaido and north Honshu F. takahashii occurs predominantly in muddy sand. Associated molluscs are generally of low diversity but considerably different between localities. In the Takikawa and Ishikari-numata areas of central Hokkaido, for example, this pectinid is commonly accompanied by Acila , Macoma , Mya , and Turritella in one assemblage, and by Anadara , Spisula , and Macoma in another assemblage (Matsui, pers. comm.). In the Sendai Basin (the type Tatsunokuchi Formation) of north-east Honshu, dominant associated molluscs are Anadara , Dosinia, Pseudamiantis, Peronidia, and Neverita (see Chinzei 1978 for the species names and other details). Burrowing bivalves are generally common, while open sea and rocky shore elements are rarely represented in this fauna. F. takahashii seems to have adapted to the muddy sand bottom of shallow bays or inland seas, though it is not clear whether the substrates were soupy or not. Uozumi et al. (1986) studied the stratigraphic range of F. takahashii on the basis of various evidence from palaeomagnetism, K-Ar datings and fission-track datings. According to them, this pectinid lived during the greater part of the Pliocene from 6 0 to 2 0 Ma in the Tokachi area on the Pacific side of Hokkaido, but its fossil records disappear much early in the Takikawa and other areas on the Japan Sea side of central-northern Hokkaido, where the Upper Pliocene is commonly represented by non-marine sediments. It is implausible to explain the morphological difference between the Pacific and Japan Sea populations by phylogenetic change. According to Gladenkov (1984) and some other Russian biostratigraphers, this species is also restricted to the Pliocene (mainly lower) in Sakhalin and Kamchatka. Fortipecten is likely to have been derived from some Miocene stock of Patinopecten ( Mizuho - pecten). Though there are a few uncommon and debatable species of Fortipecten in the Upper Miocene of Japan and Alaska, the first appearance of F. takahashii seems to have been sudden in every examined stratigraphic section. Since the Gryphaea- like morphology of this species can be regarded as largely functional, the evolutionary change may have been rapid. It would be futile, therefore, to try to verify the ancestry by finding a gradual morphological change in the fossil record. As clarified by Suzuki (1979), F. takahashii shows a remarkably wide range of geographic variation; the samples from the Pacific sides of Hokkaido and north Honshu (Suzuki’s Form B) are characterized by a less inflated right valve and not so heavy shell in comparison with the samples from the Japan Sea side of Hokkaido and Sakhalin (Suzuki’s Form A or the typical form). The analyses of relative growth in the present paper are based on the samples In and Tk which belong to the Form A, and only the small sample Sd represents the Form B. F. kenyoshiensis from the Late Pliocene Togawa Formation in north Honshu is characterized by the much weaker convexity of the right valve, so that Chinzei (1960) regarded it as intermediate between Patinopecten and typical Fortipecten. The relative growth of this species and its taxonomic relation to the weakly inflated form of F. takahashii has not been sufficiently studied, but they were probably also almost immobile recliners, because their adult valves are still much heavier HAYAMI AND HOSODA: PLIOCENE RECLINING PECTINID 437 text-fig. 9. Full-grown individual (Suzuki’s Form B) of Fortipecten taka- hashii, showing relatively weak convexity of right valve and large critical size, UMUT CM18118, xO-65. a: right view, b: posterior view. Loc.: Pliocene Tatsunokuchi Formation at the river cliff near Yodomibashi Bridge, Sendai City, north Honshu. than those of various species of Patinopecten and other swimming genera. Because abrupt inward bending of the shell commonly occurs at a larger size (about 100 mm in height) (Chinzei 1960; text-fig. 9), it is highly probable that the change of life habit from a swimmer to a full-time recliner was more or less retarded in these forms. The retardation may be natural, if we assume that relatively thin-shelled forms require larger size before the beginning of reclining life, because immobile recliners must resist predators only by the strength of their shells. These relatively thin- shelled forms are not necessarily ancestral to the typical form of F. takahashii , because their first appearance seems to be slightly later than that of the typical form. Yabe and Hatai (1940) and Akiyama (1962) regarded the peculiar morphology of F. takahashii and its morphologic difference between sedimentary basins as largely due to some adaptive response to different environments (or ecophenotypic effect in modern views). Ecophenotypic effect is possible but difficult to verify in this case, and we tentatively assume that the dichotomous geographic variation of this species is due to some geographic isolation of populations between the Japan Sea side and the Pacific side. DISCUSSION OF MODE OF LIFE Inferred life mode of F. takahashii Free-living pectinids, as emphasized by Gould (1971), reveal various allometric changes in shell form and musculature, so that swimming difficulty in later growth stages is reduced. Such allometric changes certainly occur in P. yessoensis\ e.g. the umbonal angle becomes larger and the thickness of the shell is negatively allometric to shell size. F. takahashii, on the contrary, seems to be an unusual pectinid, because of (1) the ontogenetic decrease of the umbonal angle, (2) the disappearance of disc gapes along the anterodorsal and posterodorsal margins, (3) the significant relative thickening of the shell, and (4) the relative enlargement of smooth adductor muscle (see also 438 PALAEONTOLOGY, VOLUME 31 table 2. Allometric changes of various characters in the later growth stages of Fortipecten takahashii and Patinopecten yessoensis. Character F. takahashii P. yessoensis Umbonal angle Form ratio (H/L) Size of auricles Marginal discrepancy Marginal gapes Disc symmetricity Shell convexity (RV) Shell convexity (LV) Shell convexity (CV) Shell thickness Shell weight Size of striated muscle Size of smooth muscle Obliquity of muscles Size of resilium pit Radial costae (RV) Radial costae (LV) Individual variation Becoming smaller Becoming larger Positively allometric Positively allometric Positively allometric Nearly isometric Positively allometric Decreased toward zero Positively allometric Almost invariable in width Of two orders of prominence Comparatively broad Positively allometric Positively allometric Almost disappeared Becoming prosoclinal Positively allometric Slightly positively allometric Becoming larger Becoming smaller Nearly isometric Positively allometric Becoming wider Aclinal throughout growth Slightly positively allometric Slightly negatively allometric Nearly isometric Slightly negatively allometric Slightly negatively allometric Positively allometric Negatively allometric About 15° throughout growth Nearly isometric Isometrically widened Of a single order of prominence Comparatively narrow Table 2). Judging from these allometric features, it is concluded that F. takahashii was a swimmer only in the early growth stages but became a full-time recliner after the middle stage. Such a Gryphaea- like heavy shell with a bowl-like lower valve seems to be advantageous for a reclining life in the following ways: 1 . The centre of gravity of the whole organism lies inside the cavity of the lower valve, so that the reclining life position, even if physically or biologically disturbed, is readily restored to the original state only by gravity. 2. The strongly inflated lower valve is effective in raising the commissure above the sediment surface, when the organism ‘floats’ on a muddy substrate. 3. Comparatively large and heavy valves are advantageous for safety, because full-time recliners need to resist various predatory and boring organisms without any escape strategy. The ontogenetic change of F. takahashii seems to satisfy these requirements. The adult individual of this species probably attained its stable life position by burying the greater part of the lower valve in a soft (if not soupy) substrate. Epibionts and boring organisms are very common in the left valves (and the marginal part of the right valves) but much rarer in the main part of the right valve of the same articulated individual. According to our rough calculation, even a full-grown individual (e.g. UMUT CM18117a) could ‘float’ on soft mud, if the specific gravity of the mud exceeded 1 -4. Comparison with some Mesozoic Gryphaea -like homeomorphs In gross morphology F. takahashii resembles, though of course only superficially, some Mesozoic pectinaceans, e.g. Neithea from the Cretaceous and Weyla from the Lower Jurassic. The plano- convex shells of these two genera, as discussed before (Hayami and Noda 1977), indicate that they were free-lying on soft substrates but probably not so excellent swimmers as living Pecten (s.s.). Carter (1972) and Jablonski and Bottjer (1983) also regarded N. quinquecostata from the European Chalk as an iceberg strategist. As the result of our observations of some well-preserved specimens of N. quinquecostata from the Chalk, N. texana from the Comanche Group of the Gulf Coast, HAYAMI AND HOSODA: PLIOCENE RECLINING PECTINID 439 and W. alata from the Andean Liassic limestones, the following features also suggest the diffi- culty of their swimming movements: 1. The outline of Neithea and Weyla is generally taller than that of extant swimming pectinids. Their umbonal angles are generally small and almost invariable throughout growth. 2. Disc gapes are generally undeveloped in Neithea and Weyla. 3. The valves of Neithea and Weyla look generally heavier than those of extant swimming pectinids of similar size. The aragonitic inner layer is often completely lost during the course of fossilization, but it is noticed that the thickness of the calcitic outer layer is usually positively allometric to shell size. 4. The mode of life of these Mesozoic pectinaceans in early growth stages has not been clarified. Yet, it would not be surprising if many immobile recliners smaller and more thin-shelled than Fortipecten could have lived on Mesozoic level bottoms, because predation pressure and bioturbation were probably much weaker in comparison with Cenozoic and modern seas. Among other Mesozoic pectinaceans some large species of Boreionectes [= Maclearnia ], e.g. B. imperialis asiaticus from the lower Neocomian of arctic Siberia (Zakharov 1965, 1966) are noteworthy, because they reveal thick plano-convex shells. Their gross morphology is somewhat similar to that of Fortipecten , but they are left-convex instead of right-convex. So far as we observed Zakharov’s illustrations of some articulated valves, epibionts and boring organisms are common in the right valves but rare in the left valves. The adult individuals of these species were probably also full-time recliners with the left valve directed down. In the Mesozoic there are a large number of Gryphaea- like homeomorphs in various unrelated families. In addition to above-mentioned pectinaceans, for example, Gervilleioperna and Lithioperna of the Isognomonidae, Gervillaria of the Bakevelliidae, Volviceramus and Inoceramus (only /. lamarcki group) of the Inoceramidae, and Ctenostreon of the Limidae can be regarded as full-time recliners. These pteriomorphs are generally thick-shelled and share strong convexity of the lower valve and a well-developed posterior wing without any strong ornamentation. The large posterior wing possibly acted as a stabilizer for maintaining a reclining attitude much as the large auricles of F. takahashii. The morphology and evolutionary change of Liassic Gryphaea in western Europe have been studied by a number of authors. As recognized and experimented upon by Zeuner (1933), the narrow and tightly-coiled shell of G. arcuata from the Lower Lias is most stable on soft substrates when the left valve is directed down with a nearly horizontal commissure plane. Hallam (1968), having opposed Trueman’s (1922) classical orthogenesis theory, considered that the Middle and Upper Liassic bowl-like and relatively thin-shelled species, G. gigantea , was derived from the Lower Liassic species through an intermediate species. He interpreted that the evolution was probably promoted by natural selection, because the bowl-like form is more stable against a strong water current of any direction than a narrow and tightly coiled form. The gross morphology of G. gigantea , as well as that of G. dilatata from the Upper Jurassic, is interestingly similar to the discs of F. takahashii. They are commonly characterized by the nearly uniform thickness of the shell with thickly accumulated inner layer and broadly rounded surface of the lower valve. Their hemispherical appearance is due to the growth pattern approximated by a large-angled logarithmic spiral, the early part of which is largely omitted. This can be regarded as approximate to the Rudwickian paradigm which is the most advantageous for an immobile reclining life habit. Rarity of reclining bivalves in Cenozoic seas The increase of predation pressure with geologic time (especially after the Mesozoic) and its impact on the course of marine organic evolution are an interesting subject in palaeobiology (Vermeij 1977, 1983; and others). As was discussed by him, armoured strategy is common in gastropods but seems to be rare in bivalves (except for some cemented and byssate genera in warm seas). Almost all the extant bivalves on level bottoms are very large, cryptic, endobyssate, or mobile (burrowing, jumping, or swimming). It is, however, plausible that F. takahashii alone succeeded 440 PALAEONTOLOGY, VOLUME 31 in a reclining life by forming robust valves after the middle growth stage. How rare are reclining bivalves in the Tertiary? Though we have not carried out an exhaustive study of this problem, bowl-shaped bivalves appear to have become much rarer after the Cretaceous. Only Hinnites crispus from the Pliocene of Italy, as was interpreted by Seilacher (1984), may have been a secondary recliner after the cemented stage. In the Tertiary there was a remarkable radiation of a plano-convex group of the Pectinidae, which comprises Pecten , Flabellipecten , Oppenheimopecten , and Euvola (Fleming 1957; and others). The escape behaviour of some extant species of this group was observed by several authors (Dakin 1909; Baird and Gibson 1956; Rees 1957; Baird 1958; Stanley 1970; Thomas and Gruffydd 1971) and also by us in aquaria. Their swimming ability probably declines with growth as a result of lift/gravity relation, and the gerontic individuals of large species may be almost immobile. They may no more need to swim if the smooth adductor muscle grows powerful enough to resist the shell-opening force of starfish which are generally regarded as natural enemies. Lethal boring by naticids and other organisms seems to be rare in the shells of the extant and fossil pectinids treated in this study. The reclining mode of life of F. takahashii may be basically similar to that of the gerontic stage of large species of Pecten. Yet the most distinguished peculiarity of F. takahashii is that the change of life mode from swimming to reclining seems to have occurred at an unusually young growth stage and, what is more, is well represented by the adaptive morphology. So far as we observed a number of specimens of P. albicans from Japan and P. maximus from England, the disc gapes are persistent throughout growth, and the relative shell thickening does not occur until the latest stage. Provided that the swimming velocity of pectinids is independent of shell size (Gould 1971), the swimming potential seems to be indicated by the ratio of surface area/weight, because the lift is considered to be dependent upon the surface area. For simplicity, the surface area is substituted by Lx H (in cm2), and the soft part is assumed to be equal to the sea water in specific gravity. According to our rough calculation of L x HI IV, the ratio rarely falls below 10 even in the gerontic individuals of large species of Patinopecten and Pecten. In F. takahashii, however, the ratio falls below 10 in the middle stage (about 70-80 mm in shell height) and becomes 0-4 or still smaller in the full-grown individuals. There is no positive evidence to assume that the habitat of F. takahashii was under a special condition with low predation pressure. It is, however, noteworthy that the associated molluscan fauna is always of unusually low specific diversity and that this exceptional pectinid appeared in Shell height Omm 50mm 100mm 150mm 1 1 l Fortipecten takahashii (strongly inflated form) !■ 1 Bys.j Swimming J Reclining 1 1 Fortipecten takahashii (weakly inflated form) 1 1 Bys.j Swimming | Reclining 1 Patinopecten yessoensis l Bys. 1 Swimming l Pecten albicans 1 Bys. j Swimming 1 text-fig. 10. Diagram showing the relation between shell size (height) and life habits in Fortipecten takahashii and two extant swimming species. Bys: byssate stage, the limit of which is estimated by the disappearance of active ctenolium. HAYAMI AND HOSODA: PLIOCENE RECLINING PECTINID 441 some cold-water embayments of the north-western Pacific region. From various evidence it is generally supposed that shell-crushing predators are abundant in tropical shallow seas but less common in cold seas. CONCLUSION: A RESURRECTION OF MESOZOIC-TYPE RECLINING LIFE Except for some sedentary species of Hinnites , few extant pectinids seem to show such a remarkable allometric change of the dissoconch as F. takahashii. Judging from the result of our biometric analysis and various other evidence, it is strongly suggested that this bizarre pectinid abandoned a swimming strategy in the middle growth stage and took an iceberg strategy on soft substrates. Remarkable relative thickening of the shell and common inward bending of each valve, which indicate the abrupt change of life mode, occur at this stage, when the shell height reaches about 70 mm in the strongly inflated typical form from the Japan Sea coast (about 100 mm in the less inflated form from the Pacific coast). The critical size indicating the strategic change is inversely correlated to the convexity of the right valve not only geographically but also within one and the same sample. In Sample Tk, for instance, adult individuals with large critical size generally show weak surface curvature and relatively thin shells. On the contrary, individuals with strong curvature seem to be characterized by thick shells and relatively small critical and ultimate size. Therefore, the heaviest individual is not necessarily the largest individual. The relationship between the heaviness and convexity may have been controlled by some ecological factor, presumably by the buoyancy of the right valve on soft mud. Although the ancestry of F. takahashii is unknown, its geographic and intrapopulational variation seem to suggest that the iceberg strategy was developed at first in the gerontic stage and then phyletically accelerated to the middle growth stage. The abundance of F. takahashii and its related species in the Pliocene (partly also Late Miocene) seas of the north Pacific region can be regarded as an exceptional resurrection of a Mesozoic-type reclining life. Such bowl-like heavy valves with a low centre of gravity seem to guarantee a stable living attitude against physical and biological disturbances, as already tested by Gryphaea and many other cup-shaped genera in the Mesozoic. In most Mesozoic homeomorphs it is believed that the reclining stage followed directly the byssate or cemented stage. Their reclining life appears to have initiated at a much smaller size in comparison with F. takahashii which overcame the preying danger in the young stages by swimming. Cainozoic level bottom, in general, must have been a much severer environment for recliners than in the Mesozoic, because of the stronger predation pressure and increased bioturbation by deposit feeders. Fortipecten succeeded once in developing the iceberg strategy in some cold-water embayments of the north Pacific region, but the success did not continue long, as it became extinct by the beginning of the Pleistocene. Acknowledgements. We express our sincere thanks to the following persons, without whose generous co- operation this work could not have been accomplished: Dr David J. Bottjer (University of Southern California) for his kind review of the manuscript. Dr Thomas R. Waller (Smithonian Institution) for various stimulation about the functional morphology of pectinids. Professor Kiyotaka Chinzei (Kyoto University) for his discussion, Mr Takashi Okamoto (University of Tokyo) for his assistance in computer graphics. Professor Yasuhide Iwasaki (Kumamoto University) for our free access to his collection, and Dr Tomoki Kase (Nat. Sci. Mus. Tokyo) and Dr Hisao Ando (Waseda University) for their assistance in field-work. REFERENCES akamatsu, m., Suzuki, s., kagawa, y. and nakata, m. 1979. A new occurrence of Patinopecten takahashii from the Pliocene deposits in ‘green tuff regions’, southwest Hokkaido, Japan. Ann. Rep. Hist. Mas. Hokkaido , 7, 115. akiyama, m. 1962. Studies on the phylogeny of Patinopecten in Japan. Sci. Rep. Tokyo Kyoiku Daigaku , sec. C, 74, 63 122, pis. 1 8. 442 PALAEONTOLOGY, VOLUME 31 baird, R. H. 1958. On the swimming behaviour of scallops (Pecten maximus L.). Proc. Malac. Soc. London , 33, 67-71. — and gibson, F. a. 1956. Underwater observations on escallop ( Pecten maximus L) beds. J. Mar. Biol. Ass. UK, 35, 555-562. carter, r. m. 1972. Adaptations of British Chalk Bivalvia. J. Paleontol. 46, 325-340, pis. 1-3. chinzei, k. 1960. A new Fortipecten from the Pliocene Sannohe Group in Aomori Prefecture, northeast Japan. Japan. J. Geol. Geogr. 31, 63-69, pi. 7. — 1978. Neogene molluscan faunas in the Japanese Islands: an ecologic and zoogeographic synthesis. Veliger, 21, 155-170. — 1986. Faunal succession and geographic distribution of Neogene molluscan faunas in Japan. In kotaka, t. (ed.). Japanese Cenozoic molluscs — their origin and migration. Palaeont. Soc. Japan , Spec. Papers , 29, 17-32. cox, L. R. et al. 1969. Bivalvia. In moore, r. c. (ed.). Treatise on Invertebrate Paleontology. Part N. Mol/usca 6, NL-N952. Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas. dakin, w. j. 1909. Pecten. Memoirs on typical British marine plants and animals, XVII, 136 pp., 9 pis. Liverpool Marine Biology Committee. Fleming, c. A. 1957. The genus Pecten in New Zealand. New Zealand Geol. Surv. Palaeont. Bull. 26, 1-69, pis. 1-15. fujie, t. 1958. Illustrated Cenozoic fossils. Takikawa-Honbetsu fossil fauna and distribution of Patinopecten ( F .) takahashii (Yokoyama). Cenozoic Research , 26, 34-38, pi. 25. [In Japanese.] fukutomi, T. 1953. A general equation indicating the regular form of Mollusca shells, and its application to geology, especailly in paleontology ( 1 ). Hokkaido Univ. Geophys. Bull. 3, 63-82. gladenkov, y. b. 1984. Neogene stratigraphy of northeast Asia. In ikebe, n. and tsuchi, r. (eds.). Pacific Neogene Datum Planes , 235-243. University of Tokyo Press. GOULD, s. J. 1971. Muscular mechanics and the ontogeny of swimming in scallops. Palaeontology, 14, 61-94. hallam, a. 1968. Morphology, palaeoecology and evolution of the genus Gryphaea in the British Lias. Philos. Trans. Roy. Soc. London, ser. B, 254 (792), 91 128, pis. 10-12. hayami, i. and noda, m. 1977. Notes on the morphology of Neithea (Cretaceous pectinids) with taxonomic revision of Japanese species. Trans. Proc. Palaeont. Soc. Japan , ns 105, 27 54, pis. 5 and 6. hayasaka, s. and hangai, t. 1966. A new locality of Fortipecten takahashii (Yokoyama) in Fukushima Prefecture, with reference to its associated fauna. Saito Ho-on Kcii Mus. Res. Bull. 34, 29-37, pi. I. ilyina, a. p. 1963. Neogene molluscs from Kamchatka. Trans. Geol. Prosp. Serv. USSR, 202, 1-242, pis. 1 54. [In Russian.] jablonski, d. and bottjer, d. j. 1983. Soft-bottom epifaunal suspension-feeding assemblages in the Late Cretaceous. Implications for the evolution of benthic paleocommunities. In tevesz, m. j. s. and mccall, p. L. (eds.). Biotic Interactions in Recent and Fossil Benthic Communities, 749-812. Plenum Publications. jefferies, R. p. s. and minton, p. 1965. The mode of life of two Jurassic species of ‘ Posidonia ’ (Bivalvia). Palaeontology, 8, 156 185, pi. 19. kafanov, a. l 1986n. Systematics and geological history of the subfamily Fortipectinmae Masuda, 1963 (Bivalvia, Pectinidae). In kafanov, a. i. (ed.). Paleogene Neogene bivalve molluscs of the Far East and Eastern Paratethys, 18-46. Far East Sci. Center, USSR Acad. Sci. Vladivostok. [In Russian.] 19866. Comparison of the geographical and stratigraphical ranges of Fortipectininae and Patinopectini- nae (Bivalvia: Pectinidae). Monogr. Mizunami Fossil Mus. 6, 23 40. karman, t. von and burgers, j. m. 1963. General aerodynamic theory— perfect fluids. In durand, w. f. (ed.). Aeordynamic Theory, vol. 2, 367 pp. Dover Publications. khomenko, j. 1931. Materials on the stratigraphy of the Tertiary beds of the eastern Sakhalin oilfield. Trans. Geol. Prospect Serv. USSR, 79, I 126, pis. 1 12. kotaka, t. and noda, h. 1967. Miocene Mollusca from the Minami-tsugaru district, Aomori Prefecture, northeast Japan. Saito Ho-on Kai Mus. Res. Bull. 36, 33 47, pis. 1 and 2. kryshtofovich, L. v. 1964. Molluscs from the Tertiary sediments of Sakhalin. Trans. Soviet Petrol. Sci. Geol. Exped. Inst. 232, I 344, pis. 1-55. kuroda, t. 1932. An illustrated catalogue of the Japanese shells (10). Venus, 3, appendix 87 102. [In Japanese.] labarbera, m. 1981. The ecology of Mesozoic Gryphaea, Exogyra and Ilymatogyra (Bivalvia: Mollusca) in a modern ocean. Paleobiology, 7, 510-526. HAYAMI AND HOSODA: PLIOCENE RECLINING PECTINID 443 macneil, f. s. 1967. Cenozoic pectinids of Alaska, Iceland, and other northern regions. Geol. Surv. Prof. Paper , 553, 1-57, pis. 1-25. masuda, k. 1962«. Tertiary Pectinidae of Japan. Sci. Rep. Tohoku Univ. ser. 2, 33, 117 238, pis. 18 27. — 1962 b. Pectinid fauna from the Pliocene Sannohe Group, Aomori Prefecture, northeast Honshu, Japan. Sailo Ho-on Kai Mus. Res. Bull. 31, 20 26, pis. 1 and 2. — 1963. The so-called Patinopecten of Japan. Trans. Proc. Palaeont. Soc. Japan , NS 52, 145 153, pis. 22 and 23. miyazaki, j. M. and mickevich, M. f. 1982. Evolution of Chesapecten (Mollusca: Bivalvia, Miocene Pliocene) and the biogenetic law. Evol. Biol. 15, 369 409. noda, h. and masuda, k. 1968. On the early Miyagian marine fauna from the Ojika Peninsula, Miyagi Prefecture, Japan. Saito Ho-on Kai Mus. Res. Bull. 37, I 9, pi. I. nomura, s. 1935. A note on some fossil Mollusca from the Takikawa Beds of northeastern part of Hokkaido. Sci. Rep. Tohoku Imp. Univ. ser. 2, 18, 31 39. — 1938. Molluscan fossils from the Tatunokuti shell bed exposed at Goroku cliff in the western border of Sendai. Ibid 19, 235-275, pis. 33-36. oka, t. and akamatsu, m. 1979. On the new occurrence of Patinopecten (Fortipecten) takahashii (Yokoyama) in the Tokachi Province, Hokkaido. ./. Geol. Soc. Japan , 85, 691 693. [In Japanese ] raup, d. m. 1966. Geometric analysis of shell coiling: general problems. J. Paleontol. 40, 1178 1190. rees, w. j. 1957. The living scallop. In cox, i. (ed. ). The Scallop: Studies of a Shell and its Influences on Humankind , 15 32. Shell Trans. Trad. Co. seilacher, a. 1984. Constructional morphology of bivalves: evolutionary pathways in primary versus secondary soft-bottom dwellers. Palaeontology. 27, 207-237. Stanley, s. m. 1970. Relation of shell form to life habits in the Bivalvia (Mollusca). Mem. Geol. Soc. America , 125, 1-296. SUZUKI, s. 1979. Variation and growth of Patinopecten ( Fortipecten ) takahashii. Taira Chigaku Dokokai. Special Vol. 48 51. [In Japanese.] takeuchi, k. and sanbonsugi, m. 1938. Explanatory text of the geological map of Urakawa. Hokkaido Tech. Fab. Rep. 1, 1-23, pis. 1 and 2. [In Japanese.] tanabe, k. 1973. Evolution and mode of life of Inoceramus ( Sphenoceramus ) naumanni Yokoyama emend., an Upper Cretaceous bivalve. Trans. Proc. Palaeont. Soc. Japan, ns 93, 163-184, pis. 27 and 28. thayer, c. w. 1972. Adaptive features of swimming monomyarian bivalve (Mollusca). Forma et Functio. 5, 1-32. — 1975. Morphologic adaptations of benthic invertebrates to soft substrata. J. Mar. Res. 33, 177 189. — 1979. Biological bulldozers and the evolution of marine benthic communities. Science, 203, 458- 461. thomas, G. E. and gruffydd, L. d. 1971 . The types of escape reactions elicited in the scallop Pecten maximus by selected sea-star species. Mar. Biol. 10, 87 -93. trueman, a. E. 1922. The use of Gryphaea in the correlation of the Lower Lias. Geol. Mag. 59, 256-268. trueman, E. r. 1953. The ligament of Pecten. Quart. J. Micro. Sci. 94, 193-202. uozumi, s., akamatsu, m. and takagi, t. 1986. Takikawa-Honbetsu and Tatsunokuchi faunas. In kotaka, t. (ed.). Japanese Cenozoic molluscs- their origin and migration. Palaeont. Soc. Japan. Spec. Papers. 29, 21 1 226, pis. 19 and 20. vermeij, G. j. 1977. The Mesozoic marine revolution: evidence from snails, predators and grazers. Paleobiology, 3, 245-258. 1983. Traces and trends of predation, with special reference to bivalved animals. Palae ontology. 26, 455-465. waller, T. r. 1969. The evolution of the Argopecten gibbus stock (Mollusca: Bivalvia), with emphasis on the Tertiary and Quaternary species of eastern North America. Paleont. Soc. Mem. 3, 1 125. ward, l. w. and blackwelder, b. w. 1975. Chesapecten , a new genus of Pectinidae (Mollusca: Bivalvia) from the Miocene and Pliocene of eastern North America. Geol. Surv. Prof. Paper, 861, 1 24, pis. 1-7. yabe, h. and hatai, k. 1940. A note on Pecten (Fortipecten, subgen. nov.) takahashii Yokoyama and its bearing on the Neogene deposits of Japan. Sci. Rep. Tohoku Imp. Univ. ser. 2, 21, 147 160, pis. 34 and 35. yokoyama, m. 1930. Tertiary Mollusca from south Karafuto. J. Fac. Sci. Imp. Univ. Tokyo, sec. 2, 2, 407 418, pis. 77-80. — 1932. Tertiary Mollusca from the coalfield of Uryu, Ishikari. Ibid. 3, 221-247, pis. I 4. 444 PALAEONTOLOGY, VOLUME 31 zakharov, v. a. 1965. New Upper Jurassic and Lower Cretaceous Camptonectes (Pectinidae, Bivalvia) from arctic Siberia. In SAKS, n. (ed.). Stratigraphy and Paleontology of the Mesozoic Deposits of Northern Siberia , 72 80, pis. I 6. Acad. Nauka USSR. [In Russian.] — 1966. Late Jurassic and Early Cretaceous Bivalve Molluscs of Northern Siberia , 189 pp., 46 pis. Acad. Nauka USSR. [In Russian.] zeuner, f. 1933. Die Lebensweise der Gryphaen. Palaeobiologica , 5, 307-320. itaru hayami Geological Institute University of Tokyo Hongo 7-3-1, Bunkyo-ku Tokyo 1 1 3, Japan ICHIRO HOSODA Yotsuya Commercial High School Typescript received 1 February 1987 Kami-saginomiya 5-1 l-l Revised typescript received I May 1987 Nakano-ku, Tokyo 165, Japan A LATE TRIASSIC CYNODONT FROM THE AMERICAN SOUTH-WEST by SPENCER G. LUCAS and WAYNE OAKES Abstract. A right dentary fragment and two postcanine teeth from the upper shale member of the Chinle Formation (Late Triassic: Norian) at Bull Canyon, Guadalupe County, New Mexico represent a new species of cynodont, Pseudotriconodon chatterjeei. P. chatter jeei differs from P. wildi from the Norian of Luxembourg, type and only other known species of Pseudotriconodon , by its smaller size and striated enamel. The term Dromatheriidae has been used to embrace small cynodonts with multicuspate, laterally compressed postcanine teeth, but too little is known of the dromatheriids to confirm the unity of the family or assess unambiguously their phylogenetic relationships to other cynodonts. Although Late Triassic vertebrates have been known from the American South-west for more than a century, and recent collecting has increased their diversity substantially, quarrying and screenwashing have revealed hitherto unknown taxa, particularly of smaller vertebrates. New finds include dinosaurs, ictidosaurs, pterosaurs, and sphenodontids (e.g. Murry 1982; Chatterjee 1983, 1984; Lucas et al. 1985/?; Murry 1986). These new discoveries have enhanced our knowledge of vertebrate faunal composition near the end of the Triassic and, among other things, provide important data for the more precise correlation of the Upper Triassic strata of the South-west with those of eastern North America and western Europe. Our efforts to collect Late Triassic microvertebrates have focused on the upper shale member of the Chinle Formation at Bull Canyon, east-central New Mexico. This stratigraphic unit has produced fossil vertebrates for nearly a century as well as megafossil plants and nonmarine invertebrates (Ash 1972; Kues 1985; Lucas et al. 1985u). The vertebrate fauna indicates a Late Triassic (Norian) age and is dominated by fossils of parasuchian reptiles ( Rutiodon spp. and Nicrosaurus gregorii ) and metoposaurid labyrinthodonts. Among the many fish scales and reptile teeth recovered from our screenwashing operation are a miniscule jaw fragment and teeth of a cynodont. These fossils appear to represent a new species of the family Dromatheriidae, a poorly known group of Triassic cynodonts not previously reported from the American South-west. SYSTEMATIC PALAEONTOLOGY Order therapsida Broom, 1905 Infraorder cynodontia Owen, 1861 Family ‘dromatheriidae’ Gill, 1872 Genus pseudotriconodon Hahn, Lepage and Wouters, 1984 Pseudotriconodon chatterjeei sp. nov. Text-fig. 1a-d 1985 Cynodontia: Lucas, Oakes and Froehlich, p. 205, fig. 6. 1986 Dromatheriidae: Oakes and Lucas, p. 22. Etymology. The specific name is for Dr Sankar Chatterjee of Texas Tech. University in recognition of his many contributions to the vertebrate palaeontology of the Triassic. Holotype. UNM MV-518, a right dentary fragment and two teeth (text-fig. 1) in the collection of the University of New Mexico, Albuquerque. | Palaeontology, Vol. 31, Part 2, 1988, pp. 445-449.| ©The Palaeontological Association 446 PALAEONTOLOGY, VOLUME 31 0.1 m m A — B “ C,D text-fig. 1. UNM MV-518, liolotype of Pseudotriconodon chatter jeei sp. nov. A, right lateral view of postcanine tooth in dentary. b, right lateral view of isolated postcanine tooth, c, d, right lateral (c) and occlusal (d) views of right dentary fragment. Horizon and locality. Upper shale member of the Chinle Formation, SW1/4 SW1/4 SW1/4 section 28, T9N, R26E, Bull Canyon, Guadalupe County, New Mexico, USA. See Lucas et al. ( 1 985A) for more detailed stratigraphic information. Holotype dimensions. Postcanine tooth in dentary: length 0-71 mm, width 0-25 mm, height of main cuspid 0-41 mm, height of anterior cuspid 0-23 mm, height of posterior cuspid 0-21 mm; depth of dentary below postcanine tooth 1-15 mm; length of isolated postcanine tooth 0-63 mm. Diagnosis. Postcanine teeth about 30-40% smaller than those of Pseudotriconodon wildi; also differs from P. wildi in having many distinct striations on the labial and lingual aspects of the postcanine teeth that extend towards, but do not reach, the base of the crown. Description. UNM MV-518 is a right dentary fragment bearing three root-filled alveoli followed by a postcanine tooth (text-fig. lc, d), and an isolated postcanine tooth (text-fig. 1b). Because of the close association of the dentary fragment and the isolated tooth, we believe the isolated tooth is from the dentary fragment, although its eroded base cannot be attached to the dentary fragment. In the event that the isolated tooth is later shown not to be from the dentary fragment, the holotype of P. chatterjeei should be restricted to the dentary fragment. The postcanine teeth are tricuspid, with two cuspids arranged symmetrically on the anterior and posterior edges of the main (central) cuspid. The main cuspid is about twice as tall as the other two cuspids, which are of subequal size and, in lateral view, the main cuspid is a low broad triangle. Striations on the labial and lingual aspects of the tooth run dorsoventrally but do not reach the base of the crown. The wear on the cuspids is apical. Tooth cross-section is semi-oval so there is essentially no mediolateral expansion of the crown in occlusal view. The crown has no basal cingulids, and no basal groove separates the crown from the root. A single large root supports the crown and what is visible of this root, and its base within the dentary, show no evidence of bifurcation. The single, large bone (dentary) which supports the three alveoli and the postcanine tooth is slightly concave lingually and convex labially. LUCAS AND OAKES: AMERICAN TRIASS1C CYNODONT 447 Discussion. The postcanine teeth of P. chatter jeei superficially resemble those of the Triassic pterosaur Eudimorphodon and the prolacertiform Tanystropheus in being tricuspid. However, the central cusps of the teeth of Eudimorphodon are much taller relative to the other two cusps than is the case in P. chatterjeei (Wild 1978, fig. 8), the cross-section of the teeth of Eudimorphodon is rectangular at the base of the crown and widely and irregularly biconvex in a plane through the apices of the anterior and posterior cusps (R. Wild, pers. comm. 1985), whereas the cross-section of the teeth of P. chatterjeei is narrowly and regularly biconvex, and the juvenile teeth of Eudimorphodon , although more similar to P. chatterjeei in cross-section and cusp proportions than the adult teeth, have smooth enamel (lack striations) (R. Wild, pers. comm. 1985). Similar differences in cusp proportions, crown cross-section, and enamel texture distinguish the teeth of P. chatterjeei from those of Tanystropheus (e.g. Wild 1980, fig. 4) and other prolacertiforms with tricuspid teeth. The teeth of prosauropod dinosaurs are multicuspate, but bear many cuspules along their occlusal edges (Galton 1985) and do not closely resemble those of P. chatterjeei. Similarly, the postcanine teeth of triconodont mammals, with their basal cingula, smooth enamel, complex wear patterns, and bifurcated roots, do not closely resemble those of P. chatterjeei. Indeed, the closest similarity of the teeth of P. chatterjeei is with small, multicuspid teeth from North Carolina (e.g. Simpson, 1926 6), Luxembourg (Hahn et al. 1984), and Switzerland (Peyer 1956; Clemens 1980) that have been assigned to the Cynodontia. Among these cynodonts, cusp proportions, root shape, and size of the postcanine teeth of P. chatterjeei are closest to the ‘group I’ type of tooth of P. midi , a taxon known only from isolated teeth from the Norian of Medernach, Luxembourg (Hary and Muller 1967; Hahn et al. 1984). Particularly striking is the similarity of the postcanine teeth of P. chatterjeei to a tooth of P. wildi illustrated by Hahn et al. (1984, pi. 1, fig. 1). Smaller postcanine tooth size and the numerous striations on the tooth crown, however, distinguish P. chatterjeei from P. wildi. Of course, it is possible that P. chatterjeei belongs to a genus distinct from P. wildi , but there is no morphological evidence now available to support such a conclusion. Moreover, there are no replacement teeth in the dentary of the holotype of P. chatterjeei , indicating that it is not the juvenile morphology of P. wildi or another cynodont. The morphological evidence thus best supports assignment of UNM MV-518 to the cynodont genus Pseudotriconodon and its recognition as a species distinct from P. wildi , the type and only known species of that genus. THE DROMATHERIIDAE Gill (1872) coined the term Dromatheriidae (considered by him to be Marsupialia incertae sedis) for Emmons' (1857) Dromatherium , based on a lower jaw fragment from the Upper Triassic (Carnian: Olsen et al. 1982) Cumnock Formation of the Sanford Basin, North Carolina. Osborn (1886) subsequently named Microconodon for a second lower jaw fragment from the same locality. Early debate as to the mammalian affinities of these two genera was extensive (see Simpson 19266, pp. 88 9 1 for a review). However, Simpson (19266) argued convincingly that these genera are not mammals, but instead belong to the Cynodontia incertae sedis. However, Simpson ( 1926a, p. 549) also stated that Dromatherium , as well as Microconodon , ‘must probably be considered the type of a distinct family in view of the great differences between them in tooth pattern and jaw form’. Parrington (1947, p. 726) later characterized Dromatherium and Microconodon , together with the South African taxa Lycorhinus and, perhaps, Pachygenelus as ‘unspecialized survivors of the Cynognathidae’. This may have been the impetus for Haughton and Brink’s (1954) inclusion of Pachygenelus , Lycorhinus, and Karroomys in the Dromatheriidae. Hahn etal. (1984) recently redefined the Dromatheriidae to include Dromatherium , Microconodon , European Pseudotriconodon, and South American Therioherpeton. The incomplete skull of Therioherpeton is known (Bonaparte and Barberena 1975) and justifies its inclusion in the Cynodontia. Indeed, the skull of Therioherpeton is rather similar to that of the trithelodontid Diarthr ognathus in lacking prefrontals and a postorbital bar and having slender zygomatic arches and a broad brain case, among other features (Bonaparte and Barberena 1975; Kemp 1982). On 448 PALAEONTOLOGY, VOLUME 31 the basis of these similarities, Kemp (1982) placed Therioherpeton in the Trithelodontidae, although, as stressed by Bonaparte and Barberena (1975), the postcanine teeth of Therioherpeton differ significantly from those of trithelodontids (see Gow 1980 for the teeth of trithelodontids). Indeed, the postcanine teeth of Therioherpeton are multicuspate, lack any mediolateral expansion of the crowns, and show an incipient stage of root bifurcation. They share these features with the postcanine teeth of Dromatherium , Microconodon , and (with the exception of root bifurcation) Pseudotriconodon. If the dental resemblances between Dromatherium , Microconodon , Pseudotricon- odon , and Therioherpeton are indicative of close relationship, then the affinities of the Dromatheriidae arguably lie with the Trithelodontidae. The problems with strongly supporting this conclusion are largely the same now as they were 60 years ago when Simpson studied the dromatheriids: the crania of Dromatherium and Microconodon are still unknown, and the two genera are, except for some similarities in their postcanine dentition, very different. The same problems attend definite assessment of the relationships of Pseudotriconodon. Convergence on a multicuspate, laterally compressed postcanine tooth by Hahn et al.'s (1984) four dromatheriid genera also must be considered seriously. We believe that current knowledge of the Dromatheriidae is an inadequate basis with which to determine unambiguously their phylogenetic relationships. The similarity of P. chatterjeei to P. wildi from the Norian of Europe lends some support to recent assignment of a Norian age by Lucas et al. (1985a), Chatterjee (1986), and Murry (1986) to the uppermost Dockum Group of western Texas and its equivalent in east-central New Mexico (upper shale member of Chinle Formation). The discovery of Pseudotriconodon in New Mexico broadly extends the geographic distribution of small cynodonts with multicuspid, laterally compressed postcanine teeth. The New Mexican dromatheriid thus increases the faunal similarity of the Upper Triassic Chinle-Dockum strata of the American South-west and age-equivalent strata in eastern North America and western Europe. Acknowledgements . We thank S. Chatterjee, P. Murry, K. Padian, and R. Wild for comments on the identity of UNM MV-518; P. Houlihan for permission to collect in Bull Canyon; the University of New Mexico for financial support; J. W. Froehlich, A. Hunt, and P. Reser for assistance in the laboratory and the field; two anonymous reviewers for comments on an earlier version of this manuscript; and R. Pence for drawing text-fig. 1 . REFERENCES ash, s. r. 1972. Upper Triassic Dockum flora of eastern New Mexico and Texas. In kelley, v. c. and trauger, F. D. (eds.). Guidebook of east-central New Mexico , 124 -128. New Mexico Geological Society, Albuquerque. bonaparte, J. and barberena, m. c. 1975. A possible mammalian ancestor from the Middle Triassic of Brazil (Therapsida-Cynodontia). J. Paleont. 49, 931-936. chatterjee, s. 1983. An ictidosaur fossil from North America. Science, NY, 220, 1151 1153. 1984. A new ornithischian dinosaur from the Triassic of North America. Naturwissenschaften , 71, 630- 631. - 1986. The Late Triassic Dockum vertebrates: their stratigraphic and paleobiogeographic significance. In padian, k. (ed.). The beginning of the age of dinosaurs, 139-150. Cambridge University Press, Cambridge. clemens, w. a. 1980. Rhaeto-Liassic mammals from Switzerland and West Germany. Zitteliana, 5, 51-92. EMMONS, E. 1857. American geology, part VI, 152 pp. Albany, NY. galton, p. 1985. Diet of prosauropod dinosaurs from the Late Triassic and Early Jurassic. Lethaia, 18, 105 123. gill, t. 1872. On the characteristics of the primary groups of the class of mammals. Proc. Am. Ass. Advmt. Sci. 20, 284 306. gow, c. e. 1980. The dentitions of the Trithelodontidae (Therapsida: Cynodonlia). Proc. R. Soc. B208, 461 481. hahn, G., lepage, J. c. and wouters, G. 1984. Cynodontier-Zahne aus der Ober-Trias von Medernach. Grossherzogtum, Luxemburg. Bull. Soc. beige Geol. Paleont. Hydrol. 93, 357-373. LUCAS AND OAKES: AMERICAN TRIASSIC CYNODONT 449 hary, a. and muller, a. 1967. Zur stratigraphischen Stellung des Bonebeds von Medernach (Luxemburg). Neues Jb. Geol. Paldont. Mb. 6, 333 341 haughton, s. h. and brink, a. s. 1954. A bibliographic list of Reptilia from the Karroo beds of Africa. Palaeont. afr. 2, 1-187. kemp, T. s. 1982. Mammal-like reptiles and the origin of mammals, 363 pp. Academic Press, London. kues, b. s. 1985. Nonmarine molluscs from the Chinle Formation, Dockum Group (Upper Triassic), of Bull Canyon, Guadalupe County, New Mexico. In lucas, s. g. and zidek, j. (eds.). Santa Rosa-Tucumcari region, 185 196. New Mexico Geological Society, Albuquerque. lucas, s. G., hunt, a. p. and morales, m. 1985a. Stratigraphic nomenclature and correlation of Triassic rocks of east-central New Mexico: a preliminary report. Ibid., 171-184. oakes, w. and froehlich, j. w. 19856. Triassic microvertebrate locality, Chinle Formation, east-central New Mexico. Ibid., 205-212. murry, p. A. 1982. Biostratigraphy and paleoecology of the Dockum Group (Triassic) of Texas, 459 pp. Ph.D. thesis (unpublished). Southern Methodist University, Dallas. — 1986. Vertebrate paleontology of the Dockum Group, western Texas and eastern New Mexico. In padian, k. (ed.). The beginning of the age of dinosaurs, 109 137. Cambridge University Press, Cambridge. oakes, w. and lucas, s. g. 1986. Triassic cynodont (Reptilia) from New Mexico. New Mex. Geol. 8, 22. olsen, p. e., mccune, a. r. and Thomson, k. s. 1982. Correlation of the early Mesozoic Newark Supergroup by vertebrates, principally fishes. Am. J. Sci. 282, 1 44. osborn, h. f. 1886. A new mammal from the American Triassic. Science, NY, 8, 540. parrington, f. r. 1947. On a collection of Rhaetic mammalian teeth. Proc. zool. Soc. Lond. 116, 707-728. peyer, b. 1956. Uber Zahne von Haramyiden, von Triconodonten und von wahrscheinlich synapsiden Reptilien aus den Rhat von Hallau, Kt. Schafifhausen, Schweizerische. Schweiz, palaeont. Abli. 72, 1-72. simpson, G. G. 1926a. Are Dromatherium and Microconodon mammals? Science, NY, 63, 548-549. — 19266. Mesozoic Mammalia. V. Dromatherium and Microconodon. Am. J. Sci. 12, 87-108. wild, r. 1978. Die Flugsaurier (Reptilia, Pterosauria) aus der Oberen Trias von Cene bei Bergamo, Italien. Boll. Soc. paleont. ital. 17, 176-256. — 1980. Neue Funde von Tanystropheus (Reptilia, Squamata). Schweiz, palaeont. Abh. 102, 1 43. SPENCER G. LUCAS and WAYNE OAKES Department of Geology University of New Mexico Albuquerque, New Mexico 87131 USA Typescript received 29 February 1987 Revised typescript received 9 July 1987 FAUNAL AND FACIES DYNAMICS IN THE UPPER SILURIAN OF THE ANGLO-WELSH BASIN by LESLEY CHERNS Abstract. Faunal data are matched against major sedimentary facies to interpret benthic palaeoecology and palaeogeographical evolution through the Lower Leintwardine Formation (Ludlow Series, upper Silurian) of the Anglo-Welsh basin. In shelf areas, original patchiness of level-bottom benthic epifauna, opportunistic species, and wide faunal belts characterize the storm-influenced subtidal environments. Monospecific assemblages suggest high environmental stress. Breaks in shelf deposition, erosion, and hardground formation introduced the sequence, particularly in inshore areas, as carbonate shelf environments were replaced by clastic silt sedimentation. The most marked lateral faunal change, from skeletal benthos to graptolite assemblages, takes place across the offshore shelf margin, where tectonic controls are reflected in downslope slumping and submarine channelling. In low energy trough environments, scarcity of benthos other than infaunal lingulides suggests unfavourable bottom conditions. Leintwardinian faunas are subdivided broadly into three epifaunal brachiopod-dominated shelf associations, a brachiopod-ostracode offshore shelf assem- blage, and a thanatocoenotic graptolite-lingulide trough association. Offshore spread of shelf environments in late Leintwardinian times was accompanied by immigration of new arthropod faunas, and shelf-wide domination by the distinctive Upper Leintwardine Formation association. Wider circulation patterns and regression heralded the final silting-up of the basin in response to late Caledonian tectonism. The Silurian depositional basin across the Welsh Borderland and Wales comprised an onshore platform area with a well-differentiated offshore trough to the west; the distribution of shelf and trough facies is well-established (e.g. Holland and Lawson 1963; Ziegler 1970). The narrow trough was aligned NE-SW through central Wales, although its present apparently linear axis and elongate configuration are somewhat accentuated by NW-SE Caledonian crustal shortening (min. 43 km: Coward and Siddans 1980). The closely defined stratigraphical framework for the Anglo-Welsh basin has allowed it to be a test area for palaeoecological research, following Ziegler’s (1965) pioneer study. However, most surveys have covered broad stratigraphical ranges, and in relating faunal distributions to regional palaeogeography little attention has been given to substrate- organism relationships. This study focuses on a limited part of the sequence, the Lower Leintwardine Formation, and analyses in detail the faunas in relation to sedimentary facies. Stratigraphical background. The Lower Leintwardine Formation is of late Silurian age (Ludlow Series, Ludfordian Stage) (Holland et al. 1959, 1963, 1980; Holland 1980). The boundary stratotypes for this formation and the overlying Upper Leintwardine Formation are in the Ludlow anticline, in shelly facies. The shelly faunal criteria thus strictly facilitate correlation only across the shelf region. In the graptolitic facies of trough areas, the Lower and Upper Leintwardine formations correlate with the Saetograptus leintwardinensis Biozone. Correlation of trough and shelf sequences relies on a limited graptolite distribution among shelf faunas and on reference to the transitional, mixed shelly and graptolitic faunas of shelf edge areas. The main field transect runs SE-NW across the outcrop area, from shelf to basin (text-fig. 1). The Lower Leintwardine succession is considered in detail from the following sections: in shelf facies, from inliers at May Hill, N. Woolhope (Perton), the Malverns (Chances Pitch) and Abberlcys (Woodbury), and in the main outcrop area from the type area of the Ludlow anticline; in the shelf edge region, from Aymestrey and Leintwardine; and in the trough, at Kerry. Complementary studies of parts of the sucession were made from the West Midlands inlier at Lye, in the Wenlock Edge district, Knighton, Lyepole Bridge, the Brecon anticlinal area, Builth, Usk, Llandovery— Llandeilo, and the Cennen Valley; sequences were also examined in the Dean and Brookend boreholes of the British Geological Survey. | Palaeontology, Vol. 31, Part 2, 1988, pp. 451-502, pi. 41. | © The Palaeontological Association 452 PALAEONTOLOGY, VOLUME 31 text-fig. 1 . The major outcrop areas of Ludlow rocks, showing the main distribution of Leintwardinian rocks and key localities. The primary study sections (capitals) compose a shelf to basin transect (marked). MAJOR FACIES TYPES Three principal facies types are distinguished as a background to discussion of faunal distribution data: 1, Aymestry Limestone; 2, calcareous siltstone; and 3, laminated siltstone facies. The calcareous siltstone facies includes sequences dominated by sheet-laminated units and a more thickly flaggy variety. Locally, the laminated siltstone facies passes up through ‘intermediate’ sequences into this more thickly flaggy facies. The facies are described below, and are used on the stratigraphical range charts (text-figs. 5-12). Beyond the main transect, a fourth facies type, of CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 453 poorly calcareous elastics, is distinguished for the SW of the outcrop area, and is used on palaeogeographical maps and sections (text-figs. 13 and 14). Conglomerates and phosphatized horizons within the major facies types are discussed separately. 1. Aymestry Limestone facies The basal beds of the Lower Leintwardine Formation at the stratotype (SO 4953 7255) and across much of the shelf region are developed as carbonates within the top part of the Aymestry Limestone facies and its equivalents in shelf inliers. The major portion of the carbonate facies, however, belongs to the underlying Upper Bringewood Formation, in which facies variation was related by Watkins and Aithie (1980) to a west-east geographical gradient from high energy, shelf edge barrier environments (e.g. at View Edge, SO 4261 8071), to low energy, mud deposition in the back-barrier, inner shelf (e.g. at Woodbury, Frith Wood, and Perton; see text-fig. 1 for localities). Further south-east, such as at May Hill (Lawson 1955) and S. Woolhope (Squirrell and Tucker 1960), there is no carbonate sequence below the Lower Leintwardine Formation. From the Ludlow anticline eastwards across the shelf inliers to Perton, the Aymestry Limestone of the Lower Leintwardine Formation comprises nodular, silty, and argillaceous limestones and calcareous siltstones. Terrigenous clay contents are high, and angular quartz silt forms a variable and commonly important component. Intense burrowing of the carbonate mud sediment is evident from the mottled, pelletal texture; this accounts for the destruction of much bedding fabric and for erratic mixing of skeletal grains through the strata. There are numerous fine-grained limestone (microsparitic) nodules, and local enrichment in carbonate is common. Nodule formation appears to have been early diagenetic, prior to compaction (Whitaker 1962; Watkins 1979). The nodular carbonates are irregular, in medium to thick units where bedding is picked out mainly by silty and shaly intercalations. The latter include bentonites, both in the basal Lower Leintwardine Formation and high in the Upper Bringewood Formation (text-figs. 5, 8, 9, 10; White and Lawson 1978). The lithologies of these sequences represent sediment dominated by mixed carbonate and terrigenous mud and silt, including limited amounts (mostly < 10%) of skeletal material (= calcilutites of Jaanusson in Jaanusson et al. 1979; mudstones and wackestones of Dunham 1962), and deposited in conditions of low current energy. Beds of skeletal sand (calcarenites) among the muddy sediments are mostly mud-supported (wackestones), although there are also beds or lenses of variably sparitic deposits which are at least in part grain-supported (packstones). The latter, which have sharp lower contacts, represent intervals of higher current energy. The base of local units equivalent to the Lower Leintwardine Formation in the shelf inliers is placed at the appearance of conglomeratic beds among the skeletal calcarenites (e.g. Squirrell and Tucker 1960; Phipps and Reeve 1967); these conglomerates are discussed separately (p. 460). There is a transition up into calcareous siltstone sequences, through nodular facies (e.g. PI. 41; text-fig. 9) where calcareous siltstones replace limestones as the dominant lithology. This indicates increased silt influx in relation to carbonate deposition. Through this interval, thin biodetrital layers of fine skeletal sand together with some larger skeletal debris become common, as well as thicker, sparitic skeletal calcarenites, including conglomerates. The increased frequency of skeletal sands is accompanied by higher numerical abundance of the skeletal fauna. West of Ludlow, in the shelf edge region at Shelderton, north of Leintwardine, and in the south- west part of Wenlock Edge, the Aymestry Limestone facies of the Lower Leintwardine Formation comprises higher energy carbonates. This trend corresponds to that shown by the underlying Upper Bringewood Formation. At Shelderton (SO 4157 7778), the upper part of the Upper Bringewood Formation is composed of thick, tabular, and cross-bedded sparitic calcarenites and crinoidal calcirudites (= grainstones), with thick interbeds of closely packed Kirkidium knightii valves. The sequence represents well-winnowed, grain-supported skeletal sands and gravels of high energy carbonate environments (‘facies 6’ of Watkins and Aithie 1980, as at View Edge). In this same area, at Lawnwell Dingle (SO 4163 7677), the basal Lower Leintwardine carbonates are dominantly thinner, nodular, silty skeletal calcarenites ( = wackestones and packstones), pass- ing up through transitional, highly nodular facies before muddy calcareous siltstones become 454 PALAEONTOLOGY, VOLUME 31 the main lithology (text-fig. 11b). Beds throughout this sequence are crowded with the small articulate brachiopod Dayia navicula. Many basal skeletal layers are lenticular, although the thicker skeletal beds appear more laterally extensive and some pass up directly into fiat-laminated, tabular sheets of hard calcareous siltstones. The sequence indicates fluctuating energy levels, with the winnowed skeletal sands representing numerous periods of strong current disturbance (such as storm waves) in an environment where carbonate and terrigenous mud and silt accumulated during intervals of quieter water conditions. By comparison with the coarse-grained, sparitic carbonates of the Upper Bringewood Formation, these basal Lower Leintwardine sediments represent a lowering of depositional energy levels. Beyond the main transect, in the SW part of Wenlock Edge, the Lower Leintwardine Formation caps a pronounced limestone ridge (e.g. Shergold and Shirley 1968). Throughout the area the change within the Lower Leintwardine Formation up into calcareous siltstone facies is transitional ( = ‘basal — high Lower Leintwardine Beds’ of Shergold and Shirley 1968). Lateral variation in the carbonate facies suggests higher depositional energy in the southern area, which is again comparable to the trend shown by the Upper Bringewood Formation. In the south (e.g. Norton Camp, SO 4451 8186), the Upper Bringewood carbonates are medium to thick, sparitic calcarenites which, by comparison with the Shelderton sequence, are somewhat finer grained and silty, and commonly contain a significant component of terrigenous clay. This suggests less turbulent environments for deposition of the skeletal sands, among which K. knightii, Strophonella euglypha, and corals are prominent (Greig et al. 1968). Watkins and Aithie (1980) interpreted sequences of this type as flanking, overwash deposits from the shelf edge bank environments ( = ‘facies 5’). The overlying basal Lower Leintwardine carbonates are mostly thinner, tabular calcarenites (wackestones and packstones), representing variably winnowed, mud- to grain-supported skeletal sands of moderate to high current energy. Towards the north-east, the Upper Bringewood carbonate sequences pass laterally into thinner bedded, siltier facies, a change which is reflected faunally in the absence of K. knightii and its replacement by a strophomenid-rich shelly fauna (e.g. ‘strophomenid siltstone facies’ of Holland and Lawson 1963, fig. 6). The basal Lower Leintwardine silty and muddy carbonates (e.g. north-west of Diddlebury, SO 4992 8688) include some tabular skeletal calcarenites, but there is a dominance of carbonate mud and silt deposition. One further area where the basal Lower Leintwardine Formation is developed in carbonate facies is the West Midlands inlier at Lye (SO 930 845), where these beds belong to the top of the Sedgley Limestone (e.g. Holland et al. 1963). Thick, fine-grained and nodular Upper Bringewood limestones with shaly partings pass up into thinly bedded, skeletal limestones with siltstone interbeds, which have a Lower Leintwardine fauna (c. 2-4 m sequence; King and Lewis 1912). The limestones in this upper part are sparitic, commonly conglomeratic calcarenites (mostly packstones) representing winnowed skeletal sands and indicating conditions of high energy. They are interbedded with fine, muddy siltstones laid down in low energy environments. The sequence compares most closely with the conglomeratic intervals of SE shelf areas (p. 460). Through most of the outcrop area of the Aymestry Limestone, the top part of the carbonate EXPLANATION OF PLATE 41 Figs. I 6. Lower Leinlwardine Formation facies. 1, nodular carbonates at top of Aymestry Limestone, Perton (SO 5952 3995). 2, sheet laminated siltstone unit in calcareous siltstone facies, Ludlow (SO 4893 7245), x 1 -25; note erosional base to internal skeletal lens, and escape burrow trace (arrowed) from upper internal skeletal layer. 3, skeletal calcarenite in calcareous siltstone facies, Perton (SO 5952 3995), x 7-5. 4, isolated coarse siltstone sheet in transition from laminated siltstone facies to more thickly flaggy calcareous siltstone facies, Crickadarn (SO 0850 4258), x F25; note low angle lamination and erosional contact with underlying fine muddy siltstone; narrow Chondrites burrow traces are evident through the laminated bed. 5, Lingula lata valves in bedding surface assemblage, laminated siltstone facies, Lyepole Bridge (SO 4014 6530), x3. 6, graded siltstone in poorly calcareous clastic facies, Cennen Valley (SN 6102 1906), x 15. PLATE 41 CHERNS, Lower Leintwardine Formation facies 456 PALAEONTOLOGY, VOLUME 31 formation, which corresponds closely with the Upper Bringewood Formation-Lower Leintwardine Formation faunal boundary, is marked by increasing silt deposition and a general lowering of depositional energy levels, although there were interruptions of higher current energy when winnowed skeletal sands were deposited. The turbulent, barrier edge environments of the Upper Bringewood Formation were replaced by lower energy conditions, and some eastward migration of carbonate swell environments is evident in early Leintwardinian times before siliciclastic sedimentation became dominant across the whole shelf area. 2. Calcareous siltstone facies The shelf sequences of the Lower Leintwardine Formation are predominantly thinly bedded clastic sediments, comprising carbonate-rich quartz siltstones, interbedded with thin carbonates that mostly represent layers or laterally impersistent lenticular concentrations of skeletal material. The sequences are richly fossiliferous; sediments are generally decalcified, so that shells are preserved as moulds. The siltstones are composed of medium to coarse, sub-rounded to angular silt, together with variable, but commonly high amounts of fine silt and clay and a low content of fine sand. Fresh samples have microspar carbonate cement, and the porous texture of decalcified areas suggests that most of these shelf sediments were previously calcareous. Some of the siltstones are finely laminated (individual layers < 1 mm), from carbonate-rich silts alternating with clay-rich and micaceous silts. In the thin micaceous laminae there are abundant, finely disseminated anhedral flakes of mica; in thicker bedded siltstones the content of mica, occurring as small flakes dispersed through the beds, is generally < 8%. Feldspars (potash) and other detrital minerals form only minor components (< 4%). Bioturbation traces are widespread throughout the shelf succession, although a fairly high proportion of beds retain at least some primary bedding fabric. Among laminated sediments, branching networks of narrow, Chondrites traces are commonly evident, traversing bedding surfaces and passing shallowly and obliquely through the beds. More extensive deformative reworking by deposit feeders is indicated by beds with badly disrupted, remnant fabric or a patchy and mottled texture. Skeletal material has an irregular distribution, mainly concentrated into layers of skeletal sand as micritic and sparitic calcarenites (PI. 41). These deposits range from layers comprising a single blanket of shells to medium beds (> 15 cm), but typically they are thin (5-10 cm) and lenticular, representing bedding surface accumulations of shells which appear to extend up to only a few metres laterally. The thicker beds have sharply defined lower contacts but more transitional, silty, and bioturbated upper boundaries, grading up into calcareous siltstones. In some cases, it is evident that the skeletal bed forms the coarse basal layer of a sheet of laminated siltstone, with a rapid transition through finer skeletal debris up into the silt bed; these units are discussed separately below. Among the siltstones many beds include relatively few skeletal grains, and irregular mixing of small patches of skeletal sand appears to relate to burrowing activity. However, there are also beds in which skeletal material is common and widely distributed throughout, without the carbonate-rich matrix associated with shells that is typical of the skeletal calcarenites; these siltstones commonly have a rather homogeneous texture, indicating intense bioturbation. The calcareous siltstone sequences represent sediment of carbonate-rich, muddy silts interbedded with mud- and grain-supported skeletal sands. Even in the micritic skeletal layers, mud left in intragranular cavities of skeletal grains indicates that the deposits are winnowed. The skeletal beds represent lag deposits resulting from episodes of higher current energy in an environment where silt-grade clastic material accumulated during lower energy intervals. 2a. Sheet-laminated units. This facies designation has been used to distinguish those parts of sequences in calcareous siltstone facies which include prominent beds of hard, flat-laminated calcareous siltstones, generally associated with basal skeletal concentrations. There is a marked regional variation in the distribution and nature of this facies (e.g. text-figs. 5-13). In the shelf inliers, beds of this type are thin (< 5 cm), and commonly the fabric of the siltstone portion is CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 457 only patchily preserved because of bioturbation; it is reasonable to assume that many other, more extensively reworked beds in these sequences represent comparable deposits. (However, only where these units are well-developed through an interval of sediments has the facies been indicated on the range charts.) The main occurrence is in the Ludlow anticline, where medium units (mostly 10-15 cm) dominate the sequence in the middle Lower Leintwardine Formation (text-fig. 6b). Higher in the sequence, laminated units occur as isolated sheets and are mostly thinner (< 8 cm). Text-fig. 6b shows a section at Ludlow (SO 4893 7245) where the lower part is dominated by thin to medium sheets of carbonate-rich, laminated siltstones, interbedded with thin, strongly bioturbated calcareous siltstones. The beds are richly fossiliferous. Individual units form flat- laminated, lenticular sheets (up to 20 cm, thickest centrally), extending up to a few metres across. The lamination, produced by alternation of silty and muddy layers, sometimes shows a rhythmite grading up through units, from coarser, silt-dominated into muddier, more finely laminated sediment. Individual laminae are generally < 1 mm although they locally reach 2-3 mm. The basal part of each unit is formed by closely packed, lenticular skeletal concentrations in carbonate-rich, mud-supported (more rarely grain-supported) calcarenites. The geometry of the sheets shows planar to gently undulating basal contacts that are sharp and commonly slightly erosional on beds beneath. Many of the thicker laminated sheets comprise multiple sets (up to a few cm thick), forming low-angle cross-strata infilling shallow troughs which each have coarse, skeletal basal layers (hummocky cross-stratification of Harms 1975). Some sheets display current ripple cross- lamination. Bioturbation of the top, muddy part of the units results in irregular silty patches which have a mottled, bioturbate texture, and the boundary with overlying thin siltstones is transitional. Within multiple units, the tops of individual sets also commonly show some burrow-disrupted fabric. However, in the main part of the laminated sheets, bioturbation traces are limited to a few, discrete escape burrows, steeply inclined to the bedding (e.g. PI. 41; Watkins 1979, pi. 2). In overall lithology, the laminated units do not differ significantly from interbedded siltstones, although the grading of the former means that the coarser basal layers contain a rather higher proportion of fine sand (< 10%) than is normal in the calcareous siltstones (< 5%). The composition of faunas in the laminated units and interbedded flags is discussed on pp. 469-473. The nature of the laminated units, with their slightly erosional basal contacts and coarse skeletal layers, suggests that they represent settling of suspension load by a waning current after a period of high current energy, comparable to Recent sheet deposits resulting from offshore transport of sediment by storm tidal and ebb currents (as documented from the North Sea, e.g. by Reineck and Singh 1980). The general lithological similarity to interbedded sediments suggests that these are fairly distal, open shelf deposits. In the shelf edge region (e.g. at Lawnwell Dingle and south-west Wenlock Edge), the basal calcareous siltstone facies of the Lower Leintwardine Formation includes some thin (< 5 cm), finely laminated graded silt sheets. More prominent, isolated, laterally extensive tabular sheets of carbonate-rich siltstones with basal skeletal layers occur in the underlying Aymestry Limestone and transitional nodular carbonate facies in these sequences. 2b. More thickly flaggy, calcareous siltstone facies. A lithological change to more thickly flaggy sediments takes place in late Leintwardinian times at Ludlow and in some other areas further west. At the stratotype for the Upper Leintwardine Formation this transition occurs only in the uppermost metre of the higher division, and in other sections at Ludlow is above the important Aegiria grayi-Neobeyrichia lauensis faunal horizon (e.g. Lawson 1960); but at Leintwardine, Aymestrey, Brecon, and Builth it occurs below or close to this latter level. This facies continues in the overlying Lower Whitcliffe Formation in areas from Ludlow westwards (e.g. Holland and Lawson 1963, fig. 9: ‘massively bedded’). The sediments are thin to mainly medium bedded, coarse- and medium-grained calcareous siltstones which display blocky or irregular fracture. Many beds have a homogeneous or burrow mottled texture, though others retain a laminated fabric. Isolated sheet-laminated deposits include 458 PALAEONTOLOGY, VOLUME 31 ripple cross-laminated coarse siltstone/fine sandstone units, particularly among higher beds (e.g. Crickadarn Mill, SO 0850 4259). Skeletal material is commonly concentrated in thin layers of limited lateral extent among otherwise poorly fossiliferous beds, although sheet deposits may have prominent basal skeletal layers. The typical thin, decalcified seams with grayi-lauensis assemblages may occur separately from other, brachiopod dominated accumulations. At localities in the Leintwardine to Aymestrey region, some winnowed skeletal layers are associated with phosphatized pebble beds (e.g. text-fig. 11c). The thickly flaggy calcareous siltstones represent mainly well-bioturbated silts of low to moderate energy environments, with intercalations of rapidly deposited, higher energy, laminated and sometimes micaceous silt/fine sand sheets that lie with flat, slightly erosional contact on underlying beds. Skeletal layers are winnowed current accumulations, and some thick and thinner skeletal layers associated with phosphatized material indicate periods of low net sedimentation and erosional reworking. However, the generally thicker beds suggest increased overall sedimentation rate compared with lower beds, and the silt sediment is somewhat coarser. 3. Laminated silt stone facies The thick clastic sequences of the trough areas, which have limited, graptolitic faunas, comprise very thinly bedded, closely laminated siltstones interbedded with shaly, dark mudstones and fine siltstones ( = ‘laminated siltstone facies’ of Holland and Lawson 1963). The proportions of muddy and silty lithologies vary both vertically and laterally; in general, shaly sequences are typical of western areas in the lower part of the succession, passing up transitionally into finely laminated siltstone sequences (e.g. Kerry, Clun Forest: Earp 1938, 1940). The laminated siltstones are fine to medium grained, slightly calcareous and muddy, with a narrow, parallel layering being produced by small differences in mean silt grade, carbonate, mica, and clay contents. Quartz grains are sub- rounded to angular, and comprise 30-40 % in a matrix of terrigenous clay with some microspar cement. Muscovite is a widespread minor detrital component, but occurs also in high concentrations of well-aligned, small anhedral flakes in some thin laminae. Opaque iron minerals occur in accessory amounts ( < 2 % ). Occasional thin beds of argillaceous calcilutite indicate that clastic sedimentation rate fluctuated. Minor current structures are evident in many beds. Thin laminae and shallow lenses of bioclastic debris and skeletal material have an irregular distribution among the siltstones, with inconspicuous layers of finely disseminated shell meal being rather more common than deposits of larger skeletal grains. The latter are mostly restricted to slump bands, both in western districts and in the intermediate, marginal shelf to trough areas of Builth, Cwm Graig Ddu (Straw 1937, 1953), and the Brecon anticlinal disturbance (Kirk 1951). The sparse, mainly planktic and lingulide faunas (e.g. PI. 41; Cherns 1979) of the laminated siltstones contrast strongly with the shelly benthos of slumped beds. Among the laminated sediments, bioturbation traces are few or lacking. Apart from minor local facies developments of silty, thinly bedded carbonates with Bringewoodian ‘Aymestry Limestone’ faunas in the Brecon region (e.g. Kirk 1951), fine clastic sediments dominate the trough succession through the Ludlow. In the shelf edge areas of Aymestrey and Leintwardine (text-fig. 11), the very thinly bedded siltstones are rather more calcareous and fossiliferous, and with their mixed shelly and graptolitic faunas these sequences represent a transition between the calcareous siltstone facies of the shelf and the laminated siltstones of the trough. At Aymestrey, the laminated siltstones include some thin (< 3 cm) graded silt units which have skeletal basal layers, equivalent to the sheet-laminated deposits of the shelf sequences. There are also many thin, mud-supported skeletal calcarenites which represent winnowed shell accumulations. Patchy spreading of carbonate cement out from skeletal layers into adjacent sediments indicates diagenetic movement of carbonate. The fine terrigenous muds and silts represented by the laminated siltstone facies were laid down in low energy environments. The restricted indigenous shelly benthos in trough areas, together with the reduced amount, or absence, of bioturbation suggests conditions inimical to most benthic organisms. CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 459 3 a. ‘ Intermediate ’, laminated siltstone — more thickly flaggy, calcareous siltstone facies. The ‘inter- mediate’ facies occurs in the shelf edge region, through a transition from laminated siltstones up into more thickly flaggy calcareous siltstone facies in the higher Lower Leintwardine Formation (text-fig. 11a, c). The ‘intermediate’ sequence is mainly thinly bedded, medium-coarse, more calcareous siltstones, only some of which are laminated, and with harder very calcareous nodular beds and isolated laminated silt sheets (PI. 41). In more fossiliferous sequences (e.g. at Shelderton; text-fig. 1 lc), as well as bedding surface skeletal layers there are beds with more dispersed skeletal material and homogeneous or mottled, bioturbate texture. However, in poorly fossiliferous sequences of similar lithology (e.g. Aymestrey; text-fig. 1 1a), few beds show traces of bioturbation and the sparse skeletal material is largely confined to occasional thin, impersistent, winnowed skeletal seams. The scarcity of fossils in intervening beds here suggests that epibenthic fauna was largely absent. The ‘intermediate’ facies represents deposition of coarser and less muddy silts, which in some areas were associated with an active endofauna, by contrast to the underlying laminated siltstone facies. 4. Poorly calcareous clastic facies This facies refers to the thinly bedded and shaly mudstones, siltstones, and fine sandstones, with thin skeletal horizons, that comprise the Leintwardinian sequences of the SE flank of the Towy anticline in the south-west, Llandovery-Llandeilo region. It corresponds to the Cennen Beds of Squirrell and White (1978) and the Upper Cwm Clyd Beds through Lower Roman Camp Beds of Potter and Price (1965) (e.g. at Sawdde Gorge). Towards the north-east this facies passes laterally into more argillaceous and very thinly bedded, commonly laminated and slightly calcareous sediments which make up the whole of the Cwm Clyd beds (e.g. Clawdd British) and which are transitional into the laminated siltstone facies of the basin areas. Characteristic of this facies type is the poorly or non-calcareous nature of the sediments (excluding skeletal layers). Coarser beds in thin section display tightly packed mosaics mainly of equigranular fine quartz sand or coarse silt. In the Cennen Valley the hard, coarser beds are dominantly of fine-medium sand grade, with angular to sub-rounded quartz grains that are separated by chloritic clay matrix, and with some layered fabric being picked out by more clayey laminae. Opaque iron minerals are a common accessory, and the decalcified skeletal layers appear ferruginous. Detrital mica occurs in minor amounts. In beds that have more tightly packed quartz mosaics the grains show sutured contacts. Small rounded shale or siltstone clasts and quartz gravel or coarse sand occur in some beds. At Sawdde Gorge, the sequence fines up overall into mainly siltstones (e.g. PI. 41) and mudstones, but the basal, more sandy beds overlying the red, Bringewoodian Trichrug Beds (Potter and Price 1965) also include some sharply based units of coarse quartz/small pebble conglomerates and tabular sandstones. The conglomerates have grains of rounded, sometimes polycrystalline quartz of coarse sand and gravel grade, small rounded rock particles (shale, siltstone, quartzitic sandstone), and rare pellets among finer, angular to sub- rounded quartz sand and silt in a reddish stained clayey matrix. In the muddy layers and some laminated sandstones and siltstones, bioturbation traces are evident. Skeletal material is largely confined to thin and lenticular beds. By contrast to the Trichrug Beds, where almost the only fauna is rare Lingula and Orbiculoidea found in thin grey beds thought to represent minor marine or estuarine incursions (Potter and Price 1965), the brachiopod-molluscan faunal assemblage of the overlying greenish grey beds is clearly marine. These poorly calcareous clastic sequences appear to indicate deposition in low to moderate energy, marginal subtidal environments where mud and silt deposition alternated with influxes of coarser silt and sand, and where winnowed skeletal layers accumulated among the bioturbated muds as a result of minor current disturbances. Transgression over the deltaic Trichrug deposits of Sawdde Gorge was followed by markedly lower energy sedimentation, albeit with storm episodes of stronger offshore currents which produced the conglomeratic beds among the basal cover. 460 PALAEONTOLOGY, VOLUME 31 Conglomerates The stratigraphical range charts for localities along the main transect show that for shelf sequences (text-figs. 5-10), except in the type area of Ludlow, conglomeratic horizons occur widely around the Upper Bringewood-Lower Leintwardine boundary and in the lower part of the younger division, both in Aymestry Limestone facies and in calcareous siltstone facies. The base of local divisions equivalent to the Lower Leintwardine Formation in the shelf inliers is marked by conglomeratic beds (e.g. Lawson 1955; Squirrell and Tucker 1960; Phipps and Reeve 1967); the faunal data for this correlation are discussed below (pp. 474, 475). The interpretation of the conglomeratic horizons has been described elsewhere (Cherns 1980); widespread evidence for early lithification of carbonates and in situ hardground formation suggests that many of these beds represent prolonged breaks in sedimentation with episodic erosion, rather than the products of minor intraformational storm disturbances (cf. Watkins 1979) which dominate sedimentation higher in the Lower Leintwardine Formation. In the south-east of the shelf, where the Aymestry Limestone is not developed, the lower part of the Lower Leintwardine Formation comprises conspicuous skeletal limestone conglomerates interbedded with fine, muddy calcareous siltstones and mudstones. The conglomeratic lower portion of the sequence thickens southwards from May Hill (text-fig. 7a; = Lower Blaisdon Beds of Lawson 1955) to the BGS Brookend borehole at Tites Point (e.g. text-figs. 1 and 14; Cave and White 1971). The underlying beds are muddy calcareous siltstones (May Hill) or mudstones (Brookend) with thin argillaceous limestones, which contain Bringewoodian shelly faunas ( = Upper Flaxley Beds of Lawson 1955; Cave and White 1978). The thickest and coarsest beds of conglomerate occur in the basal Lower Leintwardine Formation (e.g. text-fig. 7a). These are sparitic skeletal calcarenites, containing well-rounded to elongated and flattened clasts of limestone and calcareous siltstone. Many clasts have bored and encrusted, corroded rims, and complex depositional histories (Cherns 1980, figs. 5-7). Where the distribution of bored clasts is closely associated with internal hardground surfaces, it appears that these lithified clasts subsequently became reworked only locally. In other beds, the thin, platy form of heavily bored clasts, and lack of abrasion of attached epizoan skeletons, also suggest only limited reworking. The high energy disturbances associated with the winnowed skeletal conglomerates alternated with low energy intervals during which interbedded fine, laminated clastic sediments were deposited. There is a strong contrast between the rich skeletal assemblages of these limestones and the otherwise generally sparse faunas (e.g. text-fig. 7a). In the Aymestry Limestone and overlying calcareous siltstones of the shelf inliers (text-figs. 5- 10, 13, 14), skeletal limestone conglomerates are common through the lower part of the Lower Leintwardine Formation. These beds represent winnowed, high energy concentrations of skeletal material, bioclastic debris, and commonly ill-sorted and variably rounded clasts of limestone and calcareous siltstone. In conglomerates from Perton, Woodbury, and Chances Pitch, borings are less numerous than in sequences further south. However, hardgrounds formed in situ in all these localities, mainly on carbonate sand substrates (e.g. Cherns 1980, fig. 7b, e). At Lye, the conglomerates include many bored calcarenite clasts but there is no evidence here for in situ hardening. It is apparent that comparable depositional conditions, involving intermittent sedimentation and episodic erosion, prevailed widely over the shelf areas in early Lower Leintwardine Formation times. In the BGS Dean borehole (SO 678 000; Coppack and White 1974) on Wenlock Edge, in the northern shelf, there is a single conglomeratic horizon at the base of the Lower Leintwardine Formation, within well-bioturbated nodular argillaceous calcilutites and calcareous siltstones of the Aymestry Group. Here, in situ hardground formation affected carbonate mud sediment (Cherns 1980). At Aymestrey, near the shelf edge, Lawson (1973) reported a conglomeratic bed locally in the top of the Aymestry Limestone. At Usk, Walmsley (1959) recorded a quartz conglomerate (25-35 cm thick) in temporary sections near Llandegveth in the extreme south of the inlier at around the base of the Lower Leintwardine Formation, i.e. in the lower part of the Upper Llanbadoc Beds. At one other locality, c. 2-4 km CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 461 further north-east, he found a thin seam (c. 1-3 cm) with rounded quartz pebbles at this level, which he interpreted as the feather edge of a pebbly deposit derived from and thickening towards the south (also at Darran Plantation, ST 3276 9805: Squirrell and Downing 1969). The Llandegveth conglomerate (ST 3339 9603; ST 3332 9600), which cuts down erosively into the underlying calcareous siltstone, is grain supported, with rounded clasts of coarse sand to pebble grade, of quartz, quartzite, and volcanic rocks in a calcareous (sparitic) and shelly matrix that includes thelodont denticles (Walmsley 1959; Squirrell and Downing 1969). The quartz-rich composition of the Usk conglomerate contrasts with deposits in more easterly shelf inliers, and apparently relates to an unstable, southern source area. However, it indicates a depositional break also in the Usk region around the Upper Bringewood-Lower Leintwardine transition. Phosphatized horizons There are thin phosphatic horizons and matrix supported pebble beds at several localities high in the Lower Leintwardine Formation and close to the Upper Leintwardine Formation boundary, occurring in the thinly bedded calcareous siltstone facies of shelf sequences and in the more thickly flaggy facies of shelf edge sequences. In shelf areas, such beds occur at Longhope (text-fig. 7b), above the ?Lower Leintwardine Formation at the base of Whitcliffian strata at Gorsley (Lawson 1954), and at Perton (text-fig. 9; also Turner 1973). These beds are richly shelly or bioclastic with small rounded phosphatized clasts, thelodonts and acanthodian scales, and phosphatic (apatitic) skeletal fragments. Similar phosphatic pebble horizons or bone beds are also present at the top of the Upper Leintwardine Formation and in the Whitcliffian strata in these areas (Lawson 1955; Squirrell and Tucker 1960), and more widely, e.g. in the Brookend borehole, Tites Point, and Newnham inliers (Cave and White 1971), at the top of the Upper Leintwardine Formation at Usk (Squirrell and Downing 1969), and at the top of the Sedgley Limestone at Gornal (Ball 1951). At localities in the shelf edge region of Leintwardine there are horizons with worn and encrusted skeletal material (commonly large orthocones) high in the Lower Leintwardine Formation, and phosphatized pebble beds below the grayi-lauensis level (e.g. Shelderton; text-fig. 1 lc). These horizons are richly shelly siltstones, which include variably sized phosphatized clasts of muddy limestone. At Aymestrey there is a phosphatized conglomerate lower in the Lower Leintwardine Formation (text-fig. 11a), and then further thin phosphatized beds at the grayi-lauensis level (Lawson 1973). On Wenlock Edge Shergold and Shirley (1968) reported a ‘detrital limestone’ horizon at this higher level in the central/NE district. The significance of the phosphatized pebble-bearing horizons lies in their indication of further periods of non-deposition and erosion, the derived clasts becoming phosphatized prior to burial. The most favourable conditions for deposition of phosphorites and phosphatization of sediment in platform settings are shallow warm seas, in areas with a low net sedimentation rate, and low energy environments where winnowing out of fine sediment results in enrichment of phosphatic material (e.g. review in Bromley 1967). At Shelderton (text-fig. 11c), closely spaced thin skeletal horizons which merge locally into prominent skeletal bands associated with worn and encrusted, derived material, indicate condensed sequences from episodic winnowing and reworking during periods of low net deposition. Both in the shelf inliers and shelf edge areas, the shelly faunas of the phosphatized beds are rich and diverse, clearly marine, and the occurrence of such beds within bioturbated and laminated silt sediments suggests that these are subtidal lag deposits. SAMPLING AND RECORDING OF DATA In Lower Leintwardine Formation shelly facies, skeletal material is characteristically concentrated into bands and lenses, with relatively little in intervening sediments. Also, there are considerable compositional differences between successive, and commonly narrowly spaced, assemblages, particularly in inner shelf sequences. Therefore, ‘spaced collection’ methods do not provide a representative record of the faunal pattern through a profile (cf. Calef and Hancock 1974; Watkins 1979), and bed-by-bed, continuous collecting was adopted for analysis of faunal composition and variation. Faunal records are based largely on field data, modified by 462 PALAEONTOLOGY, VOLUME 3 1 subsequent laboratory examination of selected material. Full generic names of taxa discussed below are given in text-fig. 5. Faunas were recorded qualitatively and semi-quantitatively, i.e. up to a maximum of eighteen individuals per species, for each unit, as follows: For bivalved macrofossils (brachiopods, bivalves), articulated specimens plus the greater number of either single valve and half the indeterminate valves. In shelf sequences, the majority of skeletal assemblages are dominated numerically by articulate brachiopods. All individuals and fragments of graptolites, gastropods, nautiloid cephalopods, solitary rugose and tabulate corals, tube-dwelling annelids (e.g. Serpulites , Keilorites), tentaculitids, and cornulitids. Since these groups form only minor components of shelly assemblages, the exaggeration of abundances inherent in including fragments is not important. For trilobites, complete individuals plus one-third of the total of pygidia and glabellae; smaller elements recorded as present (i.e. 1 ) only if these others were missing. The scoring represents a compromise between allowance for growth by ecdysis (e.g. Harrington 1959: individuals cf. total remains = perhaps 10 %) and the relative infrequency of the group. For bryozoans, encrusting and erect colonies plus half the fragments of ramose and bifoliate forms. Overestimation is chiefly important where fragmentation is high. For crinoids, articulated stem sections plus one-tenth of isolated ossicles were broadly estimated. This arbitrary fraction was taken with regard to the normal disintegration of the plated skeleton prior to fossilization being countered by the inconspicuous appearance of individual ossicles among skeletal assemblages. The records obtained are thus of limited application (Jaanusson 1984); thin section examination indicates a wider distribution than shown on the range charts, but echinoderms are generally no more than minor components of faunas above the Aymestry Limestone. For ostracodes, which belong with the meiofauna to small macrofauna (< 5 mm), a visual estimate was made of one-tenth of individuals and single valves. Counts were restricted to superficial examination of bedding surfaces and freshly cut faces of samples. The methodological problems in combining widely different size groups (e.g. meio- and macrofauna) in quantitative faunal analyses are discussed by Jaanusson (1984). Difficulties of inter alia effective sampling throw into question the reliability of results for small organisms such as ostracodes in an integrated study. The reduction factor applied makes some allowance for ecdysis (Benson 1961: 7-9 moults per individual) and also the trend of ‘frequency dominance of the smallest’ (Jaanusson 1979). The records shown are undoubtedly incomplete and a least approximation of abundance. Examination of skeletal sand in thin sections of carbonates shows a more ubiquitous distribution than the range charts indicate. However, in a series of limestone samples taken along the transect, although ostracodes are locally common (particularly non-palaeocopes) in shelf sequences they are generally only a minor component (< 5 %) of the skeletal sand, whereas among beds from higher Lower Leintwardine and Upper Leintwardine strata of the shelf edge region they form a quantitatively much more important fraction (up to 52 %). The field data also reflect this distributional feature (e.g. text-fig. 11c), since in the latter area ostracodes appear conspicuous among the limited skeletal faunas. Collection size varied according to density of skeletal material, bed thickness and accessibility, and faunal composition. Representative samples of lithology and fauna were collected. There is sedimentological evidence for widely fluctuating rates of deposition locally within sequences, and regional thickness variations indicate differences between localities along the transect (e.g. text-fig. 14). Together with the heterogeneous distribution of fossils this means that standard volume does not provide an objective basis for data collection. Within localities the sample sizes were relatively consistent, so that individual range charts reflect absolute variations in local faunal abundance. However, particularly in the trough succession, where fossils are scarce, each sample took considerably longer to collect and thus abundances of taxa are exaggerated relative to the shelf data. As a control on the sampling method, at Aymestrey and Woodbury Quarry parallel records of the first fifty macrofossils (except bryozoans, crinoid ossicles) in each sample were taken for comparison with the standard sampling method. Calef and Hancock’s (1974) rarefaction graphs for May Hill Ludlow brachiopods show decreased rate of addition of new species by this sample size, and Watkins (1979) used fifty specimens for consideration of diversity and relative abundance within samples from Ludlow shelf sequences. The CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 463 text-fig. 2. Faunal records for the first fifty macrofossils counted (stippled area), superimposed on records from the standard sampling procedure (black), at a, Aymestrey and b, Woodbury (cf. text-figs. 11a and 10, respectively). The maximum, ‘cut-off level (arrowed) for the standard method is indicated (dashed line) where exceeded by the ‘first fifty’ record. 464 PALAEONTOLOGY, VOLUME 31 results obtained by the two methods (text-fig. 2) mostly show comparable quantitative variations in species abundance from bed to bed for the more common taxa. However, the ‘fifty specimen’ chart shows lower diversity in the fauna at a number of levels, indicating that this sample size was not always adequate with regard to more minor elements of faunas (also Watkins 1979, fig. 3 1 E from the Lower Leintwardine Beds at Chances Pitch). More importantly, the limiting of data to macrofossils, which is implicit in taking a standard number of individuals, becomes increasingly inadequate to represent the fauna towards the shelf edge, where meiofauna— especially beyrichiacean ostracodes— form a quantitatively conspicuous component of many assemblages (e.g. text-fig. I lc). At Aymestrey, the striking difference between methods for the distributions of Schizocrania striatal is explained by the very inconspicuous appearance of the small, thin-shelled specimens, since the faunal records for this inarticulate brachiopod derive mainly from subsequent laboratory examination of material. Limestone samples taken from the measured profiles were used for comparison of composition of skeletal sand with that of skeletal assemblages. Only limited results were obtained for the former because in recrystallized carbonates high proportions of grains are unidentifiable (commonly 25-50 %). Echinoderm abundances are probably least affected, since these particles retain relatively well-defined optical properties. Among the brachiopod-dominated sequences of shelf areas, articulate brachiopods also generally form the major component (commonly 40-50 %) of identifiable grains in the skeletal sand, with lesser amounts (10 20 %) of bryozoans and echinoderms, while molluscs, arthropods, and pellets form only minor components (mostly < 5 %). In the basal Lower Leintwardine Formation limestone conglomerates of the shelf inliers, bryozoans are conspicuous and locally become the dominant component of skeletal sand. In the shelf edge region, skeletal sand of richly shelly carbonates of the basal Lower Leintwardine Formation in Aymestry Limestone facies is dominated by brachiopods and echinoderms. There is a marked overall increase in brachiopod abundance by comparison with the underlying Upper Bringewood Formation in which, although brachiopod grains are commonly an important component (20-35 %), echinoderms (crinoids) are quantitatively dominant (40-70 %) in most beds (also Watkins and Aithie 1980). In the higher Lower (and Upper) Leintwardine of the shelf edge region, where arthropods become increasingly important in limited skeletal assemblages, the general decalcification of sediments limits compositional description to data from peels. In the thin skeletal layers, there is apparently relatively little fragmentation of many smaller skeletal grains, such as individual ostracode and Aegiria grayi valves, but brachiopods form the major component among associated shell debris. Raw faunal data for the stratigraphical range charts (text-figs. 5-12) are deposited with the British Library, Boston Spa, Yorkshire, UK as Supplementary Publication No. SUP 14033 (48 pages). They may be purchased from the British Library, Lending Division, Boston Spa, Wetherby, Yorks. LS23 7BQ. Prepaid coupons for such purposes are held by many technical and university libraries throughout the world. SKELETAL DISTRIBUTION AND THE EFFECTS OF TRANSPORT In the top part of the Aymestry Limestone facies and in the calcareous siltstone facies of the shelf, the main occurrence of skeletal material is in thin, laterally impersistent carbonate-rich skeletal layers. These occur at irregular but commonly narrow (< 2 cm) intervals. Intervening siltstones include finely laminated beds among others with patchy bedding fabric or a mottled texture, indicating bioturbational reworking. Beds with dispersed skeletal material are mostly well bioturbated. Burrowing activity appears to account for locally high fragmentation of shells and for some mixing of skeletal grains out from bedding surface assemblages, although many layers are little disturbed. Thinner skeletal beds are mostly mud-supported; they range from winnowed accumulations to assemblages of scattered shells. In thicker skeletal lenses the main concentration of shells is grain-supported, but peripheral parts of these same deposits are mud-supported. Sheet- laminated siltstones commonly have a marked basal skeletal bed sitting with sharp or erosional contact on beds below, and there are also thinner internal skeletal layers. Calcareous siltstone facies , bedding surface assemblages The nature of thinner skeletal layers was examined from sequences at Ludlow, Wood Green, and Frith Wood. Text-fig. 3 and Table 1 show the quantitative species composition of bedding surface assemblages within a 20 cm square quadrat. 100-i 50- B SAMPLE-* 10 14 5 1 13 3 4 6 SIZE — 100 284 349 169 103 202 79 128 43 15 16 17 18 19 20 21 22 65 54 37 80 60 76 94 24 50-1 23 24 25 26 27 28 29 30 31 32 112 44 92 36 103 43 44 78 22 ___| M z-y.‘ 21 A B C 94 65 73 SAMPLES MEAN SIZE 1-14 : Ludlow 133 15-22: Wood Green 61 23-32: Frith Wood 66 29 A B C 43 66 63 text-fig. 3. Bedding surface assemblages (quadrat sites 400 sq. cm) from the Lower Leintwardine Formation at a, Ludlow (I 14; SO 4895 7240); b. Wood Green (15-22; SO 6945 1665); and c, Frith Wood (23-32; SO 723 404). d shows laterally adjacent sites for four bedding surfaces (20, 21, 26, 29). Sample sizes and means for the localities are given. 466 PALAEONTOLOGY, VOLUME 31 TABLE 1. Composition of bedding surface skeletal assemblages (20 cm square quadrat) at Ludlow (1-14; SO 4895 7240), Wood Green (15-22; SO 6945 1665), and Frith Wood (23-32; SO 723 404). The proportions of convex-up values (%), pedicle valves of Isorthis orbicularis , Protochonetes ludloviensis , and Salopina lunata (P%: populations ^ 20), and right valves of Fuchsella amygdalina (R%: populations ^ 15) are shown, convex-up; concave-up; t, min. total; T, max. total; P, pedicle valve; B, brachial valve; R, right valve; A > qI B Xl c D 3 c| I E Qd left F w| /alve G W| Other I brachiopods F. amygdalina — Other macrofossils TOTAL % convex- up r~ I. orbicularis %P . S P. ludloviensis %P z S. lunata %P O P c >. § u>\ Sample P B P B P B P B P B P B R L 1 . convex-up -~ concave- up max. T 35 56 91 91 6 2 1 6 9 2 2 1 1 1 100 103 43.6 2. T 22 2 24 24 15 9 12 21 46 1 1 1 46 61 77.0 32.6 3. - T 32 3 35 35 6 3 8 7 14 24 6 1 1 1 2 6 2 6 8 4 4 66 79 72.5 58.3 4. 26 26 52 52 12 14 10 24 36 2 2 25 12 25 37 1 1 1 104 128 71.4 33.3 5. T 103 50 153 153 3 2 4 6 9 5 5 2 2 2 166 169 67.1 6. - T 2 16 18 18 10 2 1 5 11 18 1 1 2 2 1 4 4 5 35 43 46.5 7. - 1 1 1 24 20 2 9 29 55 1 1 1 31 57 80.1 47.3 8. T 7 7 7 70 38 4 1 74 113 1 1 82 121 95.8 65.5 9. ~ T 3 3 3 1 4 5 5 2 2 2 2 10 5 10 15 22 27 82.6 66.7 10. t T 238 46 284 284 284 284 83.8 11. " T 10 lo 10 92 48 7 99 147 2 2 111 159 95.5 67.3 12. ~ T 11 4 1 2 12 18 3 3 62 62 1 1 1 78 84 13. " 136 16 152 152 1 1 1 12 12 1 1 1 36 36 202 202 89.0 14. T 246 96 342 342 3 1 3 3 3 349 349 72.2 15. 5 5 5 3 3 3 3 13 4 1 14 21 11 11 7 4 7 11 7 7 7 7 54 65 87.5 33.3 CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 467 TABLE I ( COllt .) A B c D E F G H 1 J K L M N 0 P 16. T 19 19 19 1 2 3 9 5 6 4 15 24 8 8 42 54 73.9 1- 62.5 17. T 6 2 1 6 9 13 13 3 3 3 2 12 IO IO 32 37 95.8 18. T 8 8 20 6 26 45 12 7 3 6 21 8 4 12 3 3 3 3 3 52 80 66.7 , 57.8 42.9 19. T 2 1 3 3 14 14 11 3 14 20 5 1 13 3 16 23 6 1 47 60 90.4 69.6 70.0 20. T 2 1 2 4 5 12 12 2 2 2 57 57 75 76 21. T 14 14 44 1 45 80 26 9 59 94 87.5 56.3 22. t T 1 1 1 2 3 8 8 6 3 9 1 1 11 1 1 1 1 1 20 24 23- T 3 3 3 26 15 14 12 40 67 1 2 8 5 5 1 2 2 3 50 81 61.7 59.7 24. t T 47 15 62 67 3 2 IO 11 3 2 13 26 9 9 13 4 ' 6 6 90 112 79.2 50.0 25. T 18 18 12 1 13 26 13 31 44 96.2 50.0 26. t T 2 2 4 5 1 17 6 13 1 30 37 8 1 27 12 6 18 16 16 23 5 2 68 92 72.8 81.1 69.6 27. T 1 2 1 2 5 5 5 7 7 9 9 17 6 2 3 3 26 36 80.8 33.3 52.9 28. t T 36 10 46 50 4 11 6 17 10 27 12 8 20 2 2 2 4 81 103 83.8 60.0 29. T 7 1 8 4 12 1 5 1 6 7 6 1 2 9 15 15 36 43 82.1 30. t T 7 3 io 2 12 6 4 1 7 11 7 1 8 15 7 4 4 2 2 2 31 44 87.5 31. T 2 2 2 5 5 13 13 20 7 30 8 38 51 9 58 78 83.6 65.0 74.5 32. l 1 1 1 12 12 2 2 3 1 2 2 4 4 4 21 22 Most assemblages are numerically dominated by epifaunal articulate brachiopods, and apart from rhynchonellides the majority of shells are single valves. Bivalves are also locally important (e.g. samples 9, 25, 27: Fuchsella amygdalina), and typically are disarticulated. The resistance of shells to post-mortem disarticulation relates not only to energy of the environment and speed of burial, but also to hinge dentition and articulation. In brachiopods, the cyrtomatodont hinge (e.g. rhynchonellides) is more resistant to disarticulation than a deltidiodont one (e.g. orthides and strophomenides) (Jaanusson 1971). The length of exposure after death is a contributory factor, since retaining soft tissues decay rapidly; it follows that populations of conjoined deltidiodont 468 PALAEONTOLOGY, VOLUME 31 brachiopods, or bivalves (where closure of the shell is controlled solely by soft tissues), were locally derived and quickly buried. Overall disarticulation ratios were not calculated because this would primarily reflect qualitative composition of assemblages. In general, rhynchonellides are more common among faunas at Wood Green (samples 15-22) and Frith Wood (samples 23-32), whereas Ludlow assemblages (samples 1-14) are dominated by Dayia navicula and Isorthis orbicularis. The orientation of single valves (where common) among assemblages was used to calculate the proportions of convex-up to concave-up valves (Table 1 ). The convex-up orientation is dominant (> 67% in twenty-two of twenty-five assemblages). Preponderance of the hydrodynamically stable attitude (e.g. Brenchley and Newall 1970) indicates the effects of current transport on disarticulated shells, and possibly lack of much subsequent bioturbational reworking (e.g. Emery 1968). The ratio of opposing valves was calculated for I. orbicularis and Salopina lunata , Protochonetes ludloviensis, and F. amygdalina (Table 1). This ratio has been suggested as a measure of current sorting, relating to commonly appreciable differences in size, shape, and weight between the two valves, and comparable to the right/left separation of bivalve shells along beaches (e.g. Boucot et al. 1958; Lever 1958; Craig 1967). The larger, pedicle valves are more common in the majority of the brachiopod assemblages. For D. navicula , where the two valves differ considerably, there are extremely few brachial valves (Table 1). Current sorting seems unlikely to explain the differential since no assemblages have the reverse composition. Differences in mechanical strength between dissimilar valves may be important, because small and thin shells become damaged and destroyed more quickly by turbulent conditions (Trewin and Walsh 1972). By contrast, current sorting may explain the markedly higher number of one valve in two out of four bivalve assemblages (samples 9, 17), although there is no apparent current alignment of shells. There is rapid fluctuation in individual species abundance. Each assemblage characteristically has one or two numerically dominant species, and there is marked compositional variation from bed to bed. The total number of specimens per sample also varies widely, with the highest concentrations in virtually monospecific assemblages (e.g. samples 14, 10). The overall density of fossils is markedly higher at Ludlow (av. 120, cf. 61 and 65 per quadrat; samples 1-14; Table 1), where D. navicula and I. orbicularis show a generally inverse relationship in abundance although other species may dominate (e.g. sample 12, Sphaerirhynchia wilsoni; sample 9, F. amygdalina). These Ludlow assemblages illustrate the heterogeneity of faunas but are not representative of the relative proportions of assemblages in which the various individual species are prevalent. Successive faunas from Wood Green (samples 30-32) illustrate rapid faunal changes, the first having several common species, the next dominated by Salopina lunata , and then by Micro sphaeridiorhynchus nucula. Composition and density of assemblages laterally show comparable variation, with considerable local differences (text-fig. 3; Table 2). For example, in sample 26 the numerically table 2. Lateral variation in quadrat assemblages. Wood Green (samples 20, 21; SO 6945 1665) and Frith Wood (samples 26, 29; SO 723 404). Sample 20 21 26 29 A B C A B C A B C A B C Dayia navicula 4 Howellella elegans 2 7 2 12 12 1 Isorthis orbicularis 5 12 12 37 15 5 7 4 Microsphearidiorhynchus nucula 12 21 6 14 12 38 3 14 8 16 Protochonetes ludloviensis 2 3 6 80 27 24 28 38 20 9 19 2 Salopina lunata 5 10 2 Sphaerirhynchia wilsoni 56 32 23 1 15 20 43 Fuchsella amygdalina 3 10 1 23 3 1 Other macrofossils 1 6 3 3 1 Total 76 71 54 94 65 73 92 61 43 43 66 63 CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 469 dominant I. orbicularis is replaced laterally by P. ludloviensis and is only a minor element in the third site, while the number of specimens per site ranges from ninety-two to forty-three. The bedding surface layers represent concentrations of skeletal material, chiefly of benthic epifaunal organisms, and mostly occurring as disarticulated, convex-up shells. This suggests formation by periodic reworking of surface sediment and transport of skeletal grains. Since many shell beds are mud-supported and do not cut into beds beneath, current energy was only moderate and relatively little mixing and reworking have taken place. Assemblages retain an original heterogeneity in faunal distribution and are probably fairly locally derived. Resistance of shells to breakage decreases with exposure so that the generally low proportion of damaged shells (< 5%), together with the dominance of convex-up orientation and lack of evidence for much bioturbational disturbance, suggest rapid burial of skeletal assemblages. Monospecific faunas The wide fluctuations in species abundance from bed to bed result in many virtually monospecific assemblages (e.g. samples 14, 10: D. navicula). In more extensive, higher energy accumulations, including sheet-laminated units, the closely packed faunas are commonly of low diversity. Two major categories of skeletal distribution can be distinguished for these monospecific assemblages: type 1, skeletal concentrations in shell layers of variable thickness and extent; and type 2, a dispersed distribution of shells through beds. Populations of D. navicula , the two enteletaceans I. orbicularis and S. hmata, Sphaerirhynchia wi/soni , and Shaleria ornatella were analysed. As well as shell orientation and valve ratios, size frequency diagrams were constructed for populations. Unfortunately, there is no general agreement on interpretation of size frequency data (e.g. see discussion in Williams et al. 1981). Many shell assemblages contain a relatively low proportion of small members. Few small specimens are also characteristic of many living brachiopod and bivalve populations (e.g. Rudwick 1965, 1970; Hallam 1967; Thayer 1975). However, the wide variation known from Recent forms prevents categorization of an original population structure; growth and mortality rates are highly variable (e.g. Hallam 1967), and recruitment can be patchy, irregular, and periodic (e.g. Jackson et al. 1971; Thayer 1975). In articulate brachiopods, where the free- living larval phase is short (e.g. Rudwick 1970), the occurrence of juveniles attached to adults (e.g. Thayer 1975; Curry 1981), prolonged brooding (e.g. McCammon and Buchsbaum 1968; Rickwood 1968), and polymodal size distributions (e.g. Craig and Hallam 1963; Jackson et al. 1971) suggest that dispersal of larvae is commonly limited. Reduction in the smaller size fractions is inherent both through selective removal by current action and because of relatively low fossilization potential, e.g. by solution and mechanical destruction (e.g. Hallam 1967; Curry 1981). Many common Lower Leintwardine brachiopods, such as D. navicula and S. ornatella , are pedically attached only in juvenile stages (p. 480), so that young individuals are particularly susceptible to removal; a comparable life strategy in bivalves involves the change from byssal attachment to free- living (e.g. Yonge 1962; Stanley 1970). For the size frequency data here, the absence of small forms is not taken alone as an indication of size sorting, but winnowing is extrapolated from variance relative to other populations, mean size, and from sedimentological and petrographical evidence. D. navicula (text-fig. 4, samples 1-15). This species has its greatest abundance from Ludlow to the outer shelf. In the latter region these are the Dayia or Mocktree Shales of Elies and Slater (1906), and many assemblages through several metres of sequence are virtually monospecific. Populations were taken from calcareous siltstone facies at Ludlow (text-fig. 6b, both from sheet-laminated units and from the interbedded siltstones: samples 1-8), and from Aymestry Limestone facies in the Leintwardine area, from Shelderton (SO 416 778: sample 9) and Lawnwell Dingle (text-fig. 11b: samples 10-15). Type 1 skeletal concentrations include: a , samples from sheet-laminated units from convex-up shell sheets which are evidently transported (samples 3a-d, adjacent sites from a single horizon; sample 4); b, closely packed current accumulations from interbedded siltstones (convex-up, sample 5; concave-up, samples 6, 7); c, grain-supported beds of convex-up shells from carbonate sequences 470 PALAEONTOLOGY, VOLUME 31 (sample 12; samples 9a-c, sites from one bed); and d, mud-supported carbonates with variable shell orientation and many conjoined individuals (samples 13, 14; 65%, 50%, respectively). Where shells are in close contact and convex-up, disarticulation cannot be quantified since the hinge region of most shells is obscured. However, sections through shell layers suggest a high proportion of single pedicle valves, and concave-up assemblages (samples 6, 7) are almost entirely such. The distribution of convex-up shells in assemblages 3D-7 is indicated separately. The proportion of damaged shells is greater in highly concentrated assemblages (e.g. samples 9a-c, >65%), and density of shells shows wide local variation (e.g. samples 3a-d, 65-91 individuals/quadrat, av. 75; samples 9a-c, 43-79 individuals/quadrat, av. 59). Two type 1 assemblages are e , single shell thickness layers of limited lateral extent which have mainly unbroken, conjoined individuals (samples 8, 15; 75%, 70%, respectively) and large mean size compared with other populations (8-2 cf. 6-5 mm av. at Ludlow; 81 cf. 6-7 mm av. at Leintwardine); one (sample 15) has individuals predominantly with the umbo embedded in the sediment, apparently in situ (Tucker 1964). Type 2 assemblages of widely dispersed shells in siltstones (samples 1,2, 10, 11) show a random distribution and orientation of skeletal grains. Poorly defined bedding fabric indicates biogenic reworking, but since there is little mixing outwards from intercalated skeletal layers the dispersed shells are probably locally derived. Damage from reworking is indicated by locally high fragmen- tation in areas of mottled, bioturbate texture. The size frequency graphs show considerable variation, with as much variability within populations from a single current-derived bed as between beds (e.g. samples 9a-c, the y2 probabilities reject at 95% confidence level any pair having similar distribution). Small individuals are scarce, and there are very few brachial valves. Especially in high energy accumulations, relative fragility of these shells by comparison with the secondarily thickened, adult pedicle valves (e.g. Tucker 1968), and post-mortem removal of juveniles, seem likely explanations. In both areas D. navicula is prevalent in type 1 transported, current accumulations of moderate to high energy environments but also occurs as locally derived populations in type 2 and some type 1 assemblages in interbedded sediments. This suggests that, either the higher energy deposits result from relatively local reworking of the surface sediment, or this faunal belt covered a wide area. At Ludlow, D. navicula is one of several prevalent species, and its abundance fluctuates rapidly from bed to bed. It occurs similarly in sheet-laminated and in interbedded siltstones. At Leintwardine this species has sole prevalence through the lower part of the succession. Disarticu- lation is high in both transported and locally derived, reworked assemblages. A large proportion of conjoined shells in some transported assemblages (samples 13, 14) suggests rapid burial. I. orbicularis and Salopina lunata (text-fig. 4, samples 16-30). Valve ratios and orientations, as well as size frequency diagrams, were analyzed from enteletacean populations at Ludlow (samples 16- 27), Wood Green (sample 28), and Frith Wood (samples 29, 30). More commonly, assemblages are dominated by one species (but cf. samples 16, 17); I. orbicularis occurs in abundance through the Lower Leintwardine shelf sequence, whereas S. lunata becomes increasingly common in the middle to upper part. Assemblages of either in type 1 transported accumulations are typically text-fig. 4. Size frequency diagrams for articulate brachiopod populations of Dayia navicula , Isorthis orbicularis and Salopina lunata , Sphaerirhynchia wilsoni , and Shaleria ornatella. Means and standard deviations are shown, as well as sample size and number of individuals per quadrat, and type of assemblage (dominantly articulated shells, convex-up or concave-up valves). Samples: D. navicula , 1-8 = Ludlow (SO 4895 7240), 9 = Shelderton (SO 416 778), 10-15 = Lawnwell Dingle (SO 4163 7677); /. orbicularis and Salopina lunata , 16-27 = Ludlow (SO 4895 7240), 28 = Wood Green (SO 6945 1665), 29 and 30 = Frith Wood (SO 723 404); Sphaerirhynchia wilsoni , 31-34 = Perton (SO 5952 3995), 35 and 36 = Chances Pitch (SO 7475 4019), 37 41 = Woodbury (SO 7430 6368); Shaleria ornatella , 42-46 = Ludlow (SO 5071 7428), 47 and 48 = Woodbury (SO 7430 6368), 49 = Chances Pitch (SO 7486 4020), 50 = Perton (SO 5952 3995). ' S A j p U I ON * S A { p u { -ON 4/5 " ® A ! p U I ON CO -SAipui ON D. navicula ■ 1 60 0 75 i 2 0105 3 A O □ 67 A. B - 0 75 1 c n 1 066 iJ O ] 4 □ 91 | trilfr-i , , jr f ^ ^ 5 r\ 60 □ 120 JiL 5 ’ 7 ,9 l'l ' Length (mm) 6 □ 66 ' Jrou 'll 8 O 70 045 , i ' 9/ J m O I B j 0 79 Wk i , -M ■ □ 56 c ^ □ 43 10 11 90 □ 45 12 13 90 70 90 ‘ 80 60 80 □ 45 □ 58 □ 75 0 73 □ 75 0 64 A JL jL 1 A.la \JL 14 15 O 60 1 80 □ 75 0 64 lililill 3D-7 : convex-up valves orbicularis & S. lunata . wilsoni ornatella 6 10 14 18 22 Hinge width (mm) 49 _ I 50 n sample size □ n no. indivs. per 400 cm2 quadrat Assemblage type: O articulated shells s~\ convex-up valves w concave-up valves 472 PALAEONTOLOGY, VOLUME 31 single valves (> 90%), associated with variable amounts of fragmental debris. However, in some higher energy, carbonate-rich beds, a greater proportion of conjoined individuals (10-1 5 %) suggests rapid deposition. Samples come from type 1 assemblages in a, prominent skeletal limestones (samples 19, 20, 29) and b, thin, bedding surface accumulations. At Ludlow, type 2 assemblages from beds of dispersed valves are interbedded with type 1 skeletal layers where convex-up orientations are dominant. There is a clear dominance of one valve in most assemblages (e.g. samples 21, 26, 27), which suggests current sorting. Difference in mean size reflects different hydrodynamic properties for the dissimilar valves. Discrete size frequency diagrams for mixed populations (e.g. for the two species in samples 16, 17, and for two laterally spaced populations of I. orbicularis in sample 28) indicate limited post-depositional winnowing, whereas low variance (e.g. samples 21, 23) indicates stronger reworking of some beds. At Ludlow, transported I. orbicularis assemblages form type 1 skeletal layers both in sheet-laminated units and interbedded siltstones. Thus, as with D. navicula there are no faunal grounds for differentiating these. Sphaerirhynchia wilsoni (text-fig. 4, samples 31-41 ). The greatest abundance of the rhynchonellacean S. wilsoni is in the inshore shelf region. Populations were taken from Perton (samples 31-34), Chances Pitch (samples 35-36), and Woodbury (samples 37-41); at the first two, S. wilsoni composes many monospecific assemblages, while at Woodbury many skeletal bands dominated by S. wilsoni also include M. nucula (and vice versa). The type 1 skeletal concentrations are mostly lenticular, mud- to grain-supported carbonates, representing closely packed accumulations of shells up to several metres across. The mud-supported silty edges of these beds pass into more carbonate- rich, grain-supported sediment encompassing the main concentration of skeletal material. Thicker deposits have sparitic cement and much fragmental debris, indicating conditions of high current energy. Thinner, less conspicuous skeletal layers represent lower depositional energy levels. Even in high energy, well-winnowed skeletal beds there are many conjoined individuals, which indicates the strong articulation of the shell (p. 467). All assemblages analysed are from type 1 skeletal beds. Samples from Perton and Chances Pitch are grain-supported concentrations: a, monospecific (samples 31-33, each with three lateral sites; samples 35, 36); and b , more diverse (sample 34: text-fig. 9, 77). From Woodbury, samples 37-41 are c, relatively thin, rhynchonellide (see above) shell layers (< 5 cm), from variably mud- to grain- supported sediment. Size frequency graphs (max. length of the shell) show marked differences in population profile. Samples with low variance (samples 31-33, 35, 36, 39) appear to have undergone size sorting as a result of current action. For lateral sites of two beds (samples 31, 32) the low standard deviation for each implies that, despite variation in means of <0-5 mm, there is a significant x2 difference between sites. This suggests sorting analogous to Recent shell banks (e.g. Lever and Thijsson 1968). Virtually monospecific assemblages could also result from extreme current sorting, or be an original distributional feature. The latter appears more probable, since rapid fluctuations in abundance of several common species, commonly without evidence of strong current energy or sorting, suggest an originally mosaic distribution of species. The assemblages from Woodbury, which have larger mean size (12-6 mm, cf. 10-4 mm at Perton), differ widely in variance (e.g. samples 37, 40, cf. 39). Beds are mud-supported, which suggests lower depositional energy environments than at the other localities. Although somewhat equivocal, unless the lower contacts of beds are clearly erosional there is no evidence that assemblages of conjoined shells have been transported far; type 1 skeletal layers which comprise small 'clusters’ of shells and fragmental material among the calcareous siltstones may represent very locally derived populations. Shaleria ornatella (text-fig. 4, samples 42-50). This stropheodontide has its main abundance in shelf facies of the Upper Leintwardine Formation (e.g. Lawson 1960), although it is also locally common at scattered horizons lower in the sequence in these areas and becomes a major faunal element of the lower division in the east of the Ludlow anticline (text-fig. 5). Many assemblages comprise closely packed, single valves in bedding surface accumulations. CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 473 Populations were taken frpm Lower Leintwardine shelf strata (Ludlow, samples 42-46; Chances Pitch, sample 49; Perton, sample 50), and across the Lower-Upper Leintwardine boundary (Woodbury, samples 47, 48; = text-fig. 10: 49, 50 respectively). Samples from type 1 assemblages include a, fairly diverse, brachiopod-dominated limestone faunas (samples 43, 50), and b, monospecific assemblages from thick beds of tightly packed and imbricated valves (samples 44, 47, 49). These imbricated assemblages and mixed faunas represent winnowed, higher energy deposits, at least in part grain-supported and associated with much fragmental debris. There are also c, monospecific assemblages from thin layers of bedding-parallel valves (samples 45, 46, 48). Convex-up/concave-up ratios (samples 45, 46, 48) indicate no major preferred orientation, and little difference in the numbers of pedicle and brachial valves. The greater concentrations in higher energy deposits include sheet-laminated units (samples 43, 44) which were apparently rapidly buried. A type 2 dispersed population (sample 42), which represents a low energy, local deposit, shows normal distribution and wide variance. At Ludlow (text-fig. 5), S. ornatella forms many type 1 transported current accumulations, but its occurrence in interbedded sediments as more locally derived type 2 assemblages indicates that it was also indigenous here. This distribution pattern is characteristic right across the shelf area in the Upper Leintwardine sequences. In the lower part of the Lower Leintwardine Formation, the species is abundant only in local horizons, in transported type 1 accumulations of skeletal limestones (e.g. text-fig. 6, samples 30-40; text-fig. 5, samples 19-23). Then, both S. ornatella and the other stropheodontide Leptostrophia filosa are essentially restricted to higher energy deposits and are infrequent in interbedded siltstones. A similar pattern is seen also at Perton (text-fig. 9, samples 30-38). This suggests that the assemblages may have become introduced into these areas periodically; stropheodontides are unattached, and the wide, flat and light, easily disarticulated shells appear susceptible to current scour. Analysis of monospecific faunas of five brachiopod species among calcareous siltstone facies indicates variable reworking and levels of current energy. For D. navicula , /. orbicularis , and Salopina lunata, monospecific populations of largely single valves in higher energy, type 1 deposits are interbedded with similar assemblages of types 1 and 2 in less reworked beds, and for D. navicula there are even some local, articulated populations. Lack of faunal distinction between prominent sheet laminated siltstone units and interbedded siltstones is a possible indication of wide faunal belts. For Sphaerirhynchia wilsoni , where shells typically remain conjoined, rigorous reworking of some accumulations is deduced from size frequency distributions. Again, other assemblages in the same sequences are current derived deposits of lower energy, and there is no indication of extensive faunal mixing as a result of transport. For Shaleria ornatella , which occurs as disarticulated valves, a wide faunal belt is apparent in the acme for the species in high Lower Leintwardine-Upper Leintwardine strata. The more limited and local occurrences of this stropheodontide and also L. filosa lower in the Lower Leintwardine Formation, restricted to well-winnowed accumulations, may represent non-indigenous faunas introduced by more extensive transport. They could, however, also be interpreted as temporary faunal shifts. The occurrence of monospecific brachiopod faunas is important in its indication of environmental stress. Such assemblages are uncommon in Recent level-bottom communities, where articulate brachiopods show slow growth and low recruitment, and hence poor competitive ability (e.g. Thayer 1981). By contrast, the Silurian faunas suggest "opportunistic’ species distribution; articulate brachiopods dominated the majority of assemblages and the monospecific assemblages indicate rapid, temporary expansions in population size. Laminated siltstone facies , bedding surface assemblages In the laminated siltstone facies there is a pronounced difference in skeletal distribution between the offshore shelf, Aymestrey to Leintwardine region and those areas further to the west. However, in all sequences the finely laminated fabric of many sediments indicates a lack of bioturbational reworking. 474 PALAEONTOLOGY, VOLUME 31 At Aymestrey the skeletal assemblages of the lower part of the Lower Leintwardine Formation are characteristically thin, impersistent bedding surface accumulations of disarticulated shells, ranging from closely packed, imbricated deposits, through single layer shell blankets to widely dispersed skeletal grains. Convex-up orientation is greatly dominant, and a number of the low diversity assemblages also show preferred alignment of skeletal grains. (For example, all one assemblage of twenty-two F. amygdalina valves are convex-up, right valves dominate 17:5, and seventeen lie with the long axis aligned within one 30° interval ENE-WSW; in this same bed scour grooves have the same alignment.) Degree of breakage is variable, but preservation of the delicate hinge spines of Shagamella ludloviensis pedicle valves, and unbroken valves of the thin shelled Lingula lata and small Schizocrania striatal , indicate only moderate water energy and reworking. Low diversity and local species prevalence are typical of the brachiopod and bivalve dominated assemblages, and appear to represent an original distribution feature. Graptolites occur commonly in some beds, including assemblages of aligned rhabdosomes, and occasionally together with L. lata valves. At Lyepole Bridge quarry (SO 4014 6530), c. 2-2 km west of Aymestrey, and west of the outcrop area of the Aymestry Limestone, very thinly bedded and shaly, dark laminated siltstones are interbedded with thin units (< 1 m) of slumped strata which appear to be derived from the south- east. In the interbedded undisturbed siltstones, where skeletal material is mainly confined to very thin bedding surface layers, the indigenous’ fauna is mostly L. lata valves and graptolites. Both these form monospecific bedding surface assemblages but they also occur together. Graptolite assemblages include some with aligned rhabdosomes; L. lata assemblages range from widely scattered valves to densely packed layers of single valves several shells thick, and much of the breakage appears due to post-depositional compaction (e.g. Cherns 1979). These assemblages, though current derived, are of low to moderate energy environments. By contrast, the slumped beds have closely packed, thick shelly deposits dominated by articulate brachiopods (especially D. navicula) and associated with much bioclastic debris. These faunas presumably derived, along with the slumped beds, from areas higher up the depositional slope. Further west in trough areas, where faunas are considerably sparser, the faunal distribution is similar and shelly assemblages are largely restricted to slumped sequences. BIOSTRATIGRAPHY OF THE LOWER LEINTWARDINE FORMATION Base of Lower Leintwardine Formation Holland et at. (1963) placed the base of the Lower Leintwardine Formation within the top of the Aymestry Limestone facies at the stratotype (text-fig. 5). The boundary is essentially coincident with a level characterized by the loss of diagnostic Upper Bringewood fossils; many of the common Lower Leintwardine forms are long-ranging species and continue from the older formation. The Upper Bringewood indices include Kirkidium knightii, Amphistrophia funiculata, Gypidula lata , Strophonella euglypha, tabulate corals, Dalmanites myops, and Poleumita g/obosa (e.g. Lawson 1960). At Ludlow, faunas are commonly sparse through the boundary interval and dominated numerically by the brachiopods Atrypa reticularis and I. orbicularis. Through to the top of the carbonate unit, species diversity and abundance of macrofossils increase. The lithological transition up into calcareous siltstone facies is more distinct at the stratotype than in more westerly sections (text-fig. 6a). In the shelf inliers, the base of the Lower Leintwardine Formation is placed well below ( c . 5 m) the top of the Aymestry Limestone at Woodbury (text-fig. 10), within transitional facies at the top of the carbonate development at Chances Pitch (text-fig. 8; = Aymestry Limestone Member of Phipps and Reeve 1967 at these two localities), and below a transition ( c . 4-5 m) through nodular carbonate facies at Perton (text-fig. 9; = Upper Sleaves Oak Beds of Squirrell and Tucker 1960). (In the stratigraphical range charts, dominance of calcareous siltstones within the transitional CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 475 facies is taken to denote calcareous siltstone facies.) The faunal changes span several metres through poorly fossiliferous beds, and correspond closely to the incoming of shelly conglomeratic limestones. At Chances Pitch and Perton the boundary can conveniently be taken at the lowermost conglomerate, but since at Woodbury the equivalent horizon contained a typically Upper Bringewood fauna the boundary here was placed at a conglomerate c. 2 m higher where Upper Bringewood fossils are no more than rare. The sparse fauna of the topmost Upper Bringewood beds is dominated by A. reticularis , with some diagnostic Upper Bringewood forms. In south-eastern shelf areas where there is no carbonate formation in the Ludlow Series, faunal changes at the base of the Lower Leintwardine Formation also correspond to the incoming of prominent limestone conglomerates, among muddy calcareous siltstone sequences (e.g. Lawson 1955; Cave and White 1971). The diverse brachiopod-bryozoan faunas of these limestones replace Bringewoodian strophomenid-coral faunas, and only rare Bringewoodian indices are found above this level (e.g. text-fig. 7). To the west from Ludlow, in the Leintwardine and Wenlock Edge districts, the base of the Lower Leintwardine Formation locally occurs well within the Limestone. In the classic area for the ’Dayia shales’ of Elies and Slater (1906) the base of the Lower Leintwardine Formation is placed at the faunal change from the characteristic high-energy Upper Bringewood assemblage of the massive ‘Conchidium Limestone’ (p. 453) into the Dayia rich beds. The section from Lawnwell Dingle (text-fig. 1 1b) starts in very fossiliferous basal Lower Leintwardine beds c. 1-2 m above the boundary. In the south-west part of Wenlock Edge, the boundary is a similar marked faunal change up into rich and diverse small brachiopod associations. As well as D. navicula , A. reticularis , I. orbicularis, Sphaerirhynchia wilsoni, M. nucula, ShagameUa ludloviensis, Howel/e/la eiegans , and Hyattidina canalis are also very common. North-eastwards, in the siltier facies of the Aymestry Limestone, strophomenid-rich Upper Bringewood assemblages are replaced by brachiopod faunas dominated by A. reticularis, Sphaerirhynchia wilsoni, I. orbicularis, HoweUeUa eiegans , and M. nucula. In shelf edge districts, the base of the Lower Leintwardine Formation corresponds to a sharp lithological change at the top of the Aymestry Limestone (e.g. text-fig. 11a). In south Shelderton (e.g. Bow Bridge, SO 4304 7314) laminated siltstone facies lies with sharp but apparently conformable contact on thick, nodular Upper Bringewood limestones. The lowermost siltstones include rich, monospecific concentrations of D. navicula and /. orbicularis valves. At Aymestrey (text-fig. 1 1a), the top of the Limestone is locally an erosional planed surface (cf. Lawson 1973). There are truncated fossils, and surficial shallow, silt and clay filled, weathered pits. Winnowed skeletal layers with abraded limestone pebbles and Upper Bringewood fossils form lag deposits. In a nearby track section (SO 4190 6558), Lawson (1973) noted a conglomeratic bed (20 cm) in the top of the Limestone. The basal Lower Leintwardine siltstones are richly fossiliferous. At Usk, sequences correlatable faunally with the Upper Bringewood Aymestry Limestone at Ludlow, i.e. the Lower Llanbadoc Beds (Walmsley 1959), are calcareous siltstones and nodular limestones. Despite carbonate facies influence on the fauna, Walmsley noted marked faunal differences from Ludlow, such as the absence from Usk of K. knightii and Hyattidina canalis, yet presence of Protochonetes ludloviensis. At the base of the overlying Upper Llanbadoc Beds, which correlate with the Lower Leintwardine Formation, a local conglomerate is followed by calcareous siltstone facies with an abundance of D. navicula (Walmsley 1959). Base of Upper Leintwardine Formation The Upper Leintwardine fauna has a number of diagnostic fossils common only in these beds, e.g. Neobeyrichia lauensis, Calymene puellaris, and Encrinurus stubblefieldi. Also notable are Shaleria ornatella and Aegiria grayi (Holland et cd. 1963). The stratotype at Ludlow (SO 5071 7428) represents an intermediate sequence between two distinctive faunal units, the S. ornatella horizon of shelf inliers and the A. grayi-N. lauensis fauna of offshore shelf to trough areas. At the stratotype, the lower part of the S. ornatella acme is within the Lower Leintwardine Formation, and neobeyrichiid ostracodes have not been found. In other Ludlow sections these ostracodes appear 476 PALAEONTOLOGY, VOLUME 31 ' S erpulltes" sp. CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 477 MAJOR FACIES Aymestry Limestone calcareous siltstones » - sheet laminated units ii - more thickly flaggy laminated siltstones ■I - slumps ^ intermediate, laminated siltstones - calcareous siltstones poorly calcareous elastics bentonite FAUNAL ABUNDANCE 18 + 13-17 8- 1 2 - 3-7 - 1,2 ULF Upper Leintwardine Formation LLF Lower Leintwardine Formation UBF-, Lower and Upper Bringewood LE3F-J formations conglomerates, • • ® phosphatised shallow erosional scours Trichrug sandstones submarine channels text-fig. 5. The stratigraphical range charts on text-figs. 5-12 show the major facies types and faunal abundance (for the main fauna); records for ostracodes (strictly, meiofauna) are included despite the less reliable quality of these data (see text); sample levels and numbers are shown, and positions of formation boundaries are indicated; this key refers to all range charts and palaeogeographical maps (text-figs. 5-14). Ludlow (eastern part), Sunnyhill (SO 4953 7255-4971 7246), including stratotype for the base of the Lower Leintwardine Formation. some way above the base of the S. ornatella acme; e.g. in text-fig. 5 this level corresponds to the A. grayi- trilobite horizon, but in text-fig. 6c it is well above where A. grayi comes in and also above the first occurrence of C. puellaris and E. stubblefieldi. Comparable divergence of the A. grayi, trilobite, and N. lauensis-N . scissa horizons is more evident in westerly, shelf edge sequences, which lack a S. ornatella fauna and where a lower, N. nutans horizon (Siveter 1978) is also found (text-fig. 11c). Hence, since the characteristic horizon on a wider scale is the A. grayi-N. lauensis association, the addition of the ostracode assemblage has been taken, where possible, to indicate the base of the Upper Leintwardine Formation (cf. Shergold and Shirley 1968: = ‘higher Upper Leintwardine Beds’). In the shelf inliers A. grayi and N. lauensis are rare, and known only from a few specimens at one locality at May Hill (Lawson and Whitaker 1969), A. grayi recorded questionably at Woolhope (Squirrell and Tucker 1960), N. lauensis from the Malverns (Phipps and Reeve 1967). N. nutans was found at Chances Pitch. Apart from this last record which is from Lower Leintwardine beds, all the others come from the thin divisions which encompass the S. ornatella acme. Where diagnostic trilobites are found it is mainly at this same level, except for an earlier appearance of C. puellaris at May Hill (text-fig. 7; Lawson (1955) took this index to mark the base of the Lower Longhope Beds) and at Chances Pitch (text-fig. 8). As at Ludlow there is no lithological change 478 PALAEONTOLOGY, VOLUME 31 text-fig. 6. Ludlow (western part), a, lower part of succession (SO 4561 7354, 4618 7537, and 4620 7363); b, middle Lower Leintwardine Formation (SO 4893 7245); c, Haye Park, upper part of succession (SO 4876 7117 4883 7115). See text-fig. 5 description and key. CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 479 text-fig. 7. May Hill, a, Wood Green (SO 6945 1665 and 6943 1672); b, Longhope (SO 6935 1854 6939 1850). See text-fig. 5 description and key. at the base of the Upper Leintwardine Formation in the shelf inliers. However, thin, phosphatic and conglomeratic horizons occur at several localities at around this level (p. 461). The S. ornatella horizon, in its major development across the shelf region, has a rich and distinctive Upper Leintwardine Formation fauna (Holland et al. 1963). By contrast to Ludlow, at Woodbury and Perton the trilobite and S. ornatella horizons correspond to the rapid decline of Sphaerirhynchia wilsoni, while at Chances Pitch and Longhope the appearance of C. puellaris is below these changes in brachiopod emphasis (text-figs. 7-10). Also, the gap at Ludlow between the loss of S. wilsoni and first appearance of N. lauensis contrasts with a record of the two together in the Malverns (Lawson and Whitaker 1969). The base of the Upper Leintwardine Formation in shelf inliers has been taken at the level where Shaleria ornatella becomes abundant but also Sphaerirhynchia wilsoni declines rapidly: where these two overlap (e.g. at May Hill), the boundary was placed at the latter, marked faunal change. § 8 £ 5 5 S i “ CHERNS. SILURIAN FAUNAL AND FACIES DYNAMICS 480 PALAEONTOLOGY, VOLUME 31 FAUNAL COMPOSITION The numerous Lower Leintwardine skeletal assemblages in the calcareous siltstone facies and the top part of the Aymestry Limestone are characteristically dominated by benthic epifauna, usually articulate brachiopods (commonly > 80% of macrofauna). By contrast, in the laminated siltstone facies, benthic skeletal faunas are sparse except in intermediate, shelf edge areas and in slumped beds, and in trough sequences they are widely represented only by endofaunal inarticulate brachiopods. The morphological adaptations of Silurian brachiopods, all of which were sessile suspension feeders, appear to relate chiefly to life attitude on or within the sediment, and to the nature of the substrate, or in broad terms to hard (rocky) or soft (level) bottom environments. All brachiopods are anchored in the juvenile, post-larval stages, the majority by pedicle attachment, and they then require hard or firm substrates. Unlike Recent brachiopods, of which the most common forms remain attached to hard bottoms throughout life, many Silurian groups changed life habit through ontogeny and became able to live on soft substrates, i.e. they were ‘ambitopic’ (Jaanusson 1979; ‘liberosessile’ of Bassett 1984). In composition early Palaeozoic benthic faunas do not demonstrate the clear distinction found in Recent seas between hard- and soft-bottom environments, where there is little ecological interaction. More varied modes of life, by comparison with Recent stocks, account for the great range of shell morphologies and sculptures in Silurian brachiopods. Most epifaunal species which retain an open pedicle foramen throughout life are assumed to have been pedunculate as adults (‘fixosessile’ of Bassett 1984). Of the common Lower Leintwardine Formation articulate species this includes the enteletaceans /. orbicularis and Salopian lunata, the rhynchonellacean M. nucula , and some spiriferaceans, e.g. Howellella elegans and Hyattidina canalis. These brachiopods are usually considered as hard-bottom fauna, although hard substrates for pedicle anchorage need not necessarily be more extensive than local hard patches within otherwise soft sediment, e.g. pebbles, skeletal grains. Also, the pedicle in some attached groups may have functioned largely as a tether for recumbent shells on either hard or soft substrates (e.g. Westbroek et al. 1975). Among populations of pedunculate brachiopods, the post-mortem detachment of shells from the substrate might provide adequate numbers of local attachment surfaces for continuing settlement of spat, although clearly such loose grains will be susceptible to current transport. Among inarticulates, O. rugata and Schizocrania striata , which have a marginal pedicle notch on flat or slightly concave pedicle valves, represent attached, hard-bottom forms. Records of these discinaceans from upper Ludlow rocks indicate that some individuals adopted an encrusting habit on other shells, some possibly during the life of the host (Holland 1971; Lockley and Antia 1980). The shells are moulded closely to the substrate and were presumably held in place by the pedicle (cf. Recent Discinisca : Rudwick 1965). A few Lower Leintwardine Formation individuals of S. striata were found attached to larger skeletal grains, but at the only locality where this species occurs in abundance (at Aymestrey; text-fig. 11a), the numerous small, thin valves occurred separately in bedding plane assemblages, detached from their original substrates. Many common Silurian brachiopods were ambitopic. The pedicle opening present in younger stages closed during ontogeny and the adults rested free (recumbent) on the sediment, as soft- bottom fauna. The free-living adults show various morphological adaptations for maintaining stability on, or partially embedded in (‘quasi-infaunal’ of Bassett 1984), the sediment; the latter habit can grade towards an endofaunal existence. In concavo-convex chonetaceans such as P. ludloviensis and Shagamella ludloviensis, the pedicle was lost early in ontogeny. Oblique hinge spines on the convex pedicle valve may have functioned as stabilizers (perhaps also attached distally by mucal secretion from the spine tips: Bassett 1984) in an orientation resting on that valve, possibly partially buried but with the commissure and growing mantle edges raised off the sediment. They may have been able to adjust the orientation by vigorous ‘snapping’ of the valves (e.g. Rudwick 1970). Strophomenaceans commonly became detached and recumbent in later stages; stropheodontides such as Shaleria ornatella and Leptostrophia filosa had relatively thin, broad, CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 481 m- text-fig. 8. Chances Pitch (SO 7473 4018-7498 4020). See text-fig. 5 description and key. Pitch (SO 7473 4018 7498 4020). See text-fig 5 2 PALAEONTOLOGY, VOLUME 31 text-fig. 9. Perton (SO 5952 3995 and 5963 4012). Sec text-tig. 5 description and key. CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 483 text-fig. 10. Woodbury (SO 7430 6368). See text-fig. 5 description and key. 8 CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS PALAEONTOLOGY, VOLUME 31 484 Formation (SO 4163 7677); c, Shelderton (SO 4172 7790 4178 7792). See text-fig. 5 description and key. CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 485 text-fig. 12. Kerry, a. Ring Hole gorge, upper part of slump sequence (SO 1242 8365); b, Pant-y-llidiart lane to Cwm, upper part of M. leintwardinensis Shales (SO 1615 8747-1630 8743); c, Pant-y-llidiart dingle, upper part of M. leintwardinensis Shales (SO 1650 8770-1647 8735). Approx, gaps (m) between sample points shown in parentheses. See text-fig. 5 description and key. I CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 486 PALAEONTOLOGY, VOLUME 31 gently concavo-convex shells where the large surface area would provide support on the sediment surface. In Leptaena depressa , the wide geniculate adult shell with a pronounced fold and trail may represent adaptation to a quasi-infaunal habit where the commissure remained free of the substrate even though the shell was largely buried (e.g. Rudwick 1961; Bassett 1984). Among the common Upper Bringewood Formation forms, Strophonella euglypha and Amphistrophia funiculata show similar morphological adaptation, with geniculate adult shells. Adult forms of the biconvex athyridacean Dayia navicula and the atrypacean Atrypa reticularis show marked secondary thickening of the umbo region in the pedicle valve, while incurvature of the beak area and sealing by plate structures during growth resulted in reduction or loss of the pedicle opening (e.g. Tucker 1964). The distribution of these species suggests that a clustered, gregarious life habit was a common, though not invariable, feature of populations. Stability was maintained partly by embedding of the posteriorly weighted shell in the sediment, but additionally by close juxtaposition with neighbouring individuals; this strategy was termed ‘co-supportive’ by Bassett (1984). Worsley and Broadhurst (1975) described clustered populations of A. reticularis in the Llandovery of the Oslo region, for which loss of the pedicle varied among individuals but where most larger, adult shells were detached, resting on the flatter pedicle valve with its broad fringe or supported by adjacent shells. Many Silurian pentameraceans display this type of occurrence in fairly high energy environments, e.g. Pentamerus oblongus and Stricklandia lens in the Llandovery (Ziegler et al. 1966), and K. knightii in the high energy carbonate facies of the Upper Bringewood Formation. A comparable strategy may have applied also to close packed populations of some rhynchonellace- ans (e.g. Sphaerirhynchia wilsoni) to supplement the anchoring function of the pedicle. Although the shell is not thickened posteriorly, there was umbonal incurvature resulting in reduction or loss of the pedicle. In S. wilsoni an umbo-down posture was apparently maintained by increased globosity and geniculation, the adults adopting a quasi-infaunal habit partially buried in the sediment (Westbroek et al. 1975; Fiirsich and Hurst 1981). The small ambitopic inarticulate lingulacean Craniops implicatus has a cicatrix for direct attachment to a hard grain in early ontogeny. It became detached still at a relatively young stage but at a size that allowed independent stability. The adult has a limbus which would tend to stabilize the recumbent shell (e.g. Bassett 1984). In general, the distribution of both hard- and soft-bottom brachiopods in the Leintwardine formations shows a marked concentration to shelf areas (text-figs. 5-12). However, the small, thin- shelled plectambonitacean Aegiria grayi, which has a narrow pedicle foramen and was probably pedically attached (Cocks 1970), has a distribution notable for its concentration in western, outer shelf areas. Its occurrence there is commonly among restricted assemblages, in association with ostracodes but largely separate from assemblages of other epifaunal brachiopods. The small, light concavo-convex shell, which is similar morphologically to the recumbent to quasi-infaunal chonetaceans and strophomenaceans, has been discussed variously as an adaptation to low-energy offshore, soft-bottom conditions or to an epiplanktic mode of life attached to floating algae (cf. Sargassum , e.g. Fiirsich and Hurst 1974; Watkins and Berry 1977). An epiplanktic habit has been proposed for several Ordovician and Silurian articulate brachiopod genera with similar general morphology, mostly to explain their occurrence in offshore, graptolitic shales, e.g. Sericoidea (Bergstrom 1968), Clionetoidea (Havlicek and Vanek 1966; Havlicek 1967), Aegiria and Shagamella (Watkins and Berry 1977). Except apparently for Aegiria, these brachiopods lost a functional pedicle during ontogeny and thus lacked the means of attachment to an algal or other floating substrate (unless the short hinge spines of Shagamella could achieve this: Bergstrom 1968). For unattached forms an epiplanktic habit would seem unlikely, and it is notable that some Recent benthic brachiopods inhabit zones down to abyssal depths (e.g. Pelagodiscus: Rudwick 1970). Cocks (1970) noted the main Llandovery distribution for Aegiria as being in the ‘deepest' benthic, Clorinda community, but suggested that occurrences in trough sequences where benthic fauna was otherwise scarce might be explained by offshore transport along with an uprooted algal substrate. Sheehan (1977) proposed alternatively that fallen seaweed might provide patches of relatively firm CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 487 substrate for benthic organisms in offshore areas of soft sediment. For A. grayi in the Lower and Upper Leintwardine formations, a pedunculate, possibly recumbent benthic habit in offshore environments explains its distribution, yet an epiplanktic life-style cannot be discounted. However, for the heavier shelled, ambitopic S. ludloviensis , which is concentrated among benthic shelf assemblages, a benthic life habit seems probable (cf. Watkins 1979, for Lower Bringewood Formation). Of the benthic fauna associated with brachiopods across the shelf, bivalves form minor components of many skeletal assemblages and are even prevalent locally, although not generally in comparable numbers with brachiopod dominated assemblages. Except for pterineids, bivalves also are concentrated mainly on the shelf. All the more common Lower Leintwardine Formation bivalves represent suspension-feeding benthos. Most were byssally attached, either endobyssate, i.e. partly buried in soft sediment, e.g. Cypricardinia , Goniophora , Pteronitella , or epibyssate on hard substrates, e.g. MytUarca , Pterinea (e.g. Stanley 1972). The most common bivalve, Fuchsella, may have been a slow, shallow burrower since it lacks a byssal sinus or gape; otherwise it was probably endobyssate. Bryozoans occur widely in shelf sequences, but their usual fragmentation among skeletal assemblages means that diversity may be underestimated. They are hard-bottom, suspension feeding epifauna, and most require firm though not necessarily large attachment sites (e.g. Brood 1979, 1984). In the Lower Leintwardine Formation ramose forms appear best represented, although the bifoliate cryptostome Ptilodictya lanceolata is conspicuous in shelly conglomerates. Most bryozoans are dissociated from their original substrates. Encrusting and fenestellate forms are however found on some skeletal grains, most prominently on abraded fossils in phosphatized horizons of high Lower Leintwardine-Upper Leintwardine Formation beds. Crinoids and solitary corals are further attached, suspension feeding organisms that form minor parts of hard-bottom epifaunas. Like bryozoans, they may have required only local hard sites for attachment; a few corals are attached to a skeletal fragment, but crinoids are dissociated from their original substrates and largely disaggregated. Both these groups are considerably less abundant than in the underlying Upper Bringewood Formation. The hard-bottom endofauna is generally poorly represented in early Palaeozoic benthic faunas by comparison with Mesozoic and Recent examples (e.g. Palmer 1 982). The only evidence of this group in the Lower Leintwardine Formation is from small borings ( Trypanites ) which affect clasts in con- glomerates, and from a few small circular bored cavities on skeletal grains. The low density of these organisms suggests that their activity was unimportant as a destructive factor on hard substrates. Apart from quasi-infaunal brachiopods and endobyssate bivalves, the endofauna of soft substrates is represented in skeletal assemblages chiefly by lingulide brachiopods. Both the common Lower Leintwardine Formation species Lingula lewisii and L. lata appear, from the not uncommon occurrence of individuals perpendicular to the bedding and thus apparently in situ (Cherns 1979), to have been burrowers like Recent Glottidia and Lingula (e.g. Thayer and Steele-Petrovic 1975; Emig 1982; Hammond 1983). The very different distribution patterns for the species appear to result from different palaeoecological tolerances, comparable to the range known from Recent environments (e.g. Emig et al. 1978; Plaziat et al. 1978), and not as Watkins and Berry (1977) suggested to adoption of an aberrant, epiplanktic existence by L. lata (Cherns 1979). L. lata typically represents the only benthic skeletal fauna through much of the laminated siltstone facies in trough areas. Other endofaunal or semi-endofaunal skeletal organisms which occur in very minor numbers are tentaculitids and cornulitids (e.g. Larsson 1979), and tube-dwelling worms such as ‘ Serpulites' longissimus and Keilorites (e.g. Brood 1979). However, endofaunal activity is indicated throughout the shelf by commonly intense bioturbation of sediments, caused by vagile benthos of which the major proportion was probably soft-bodied organisms, as in Recent environments (e.g. Craig and Jones 1966). By contrast to the sessile benthos, this sediment-reworking vagile benthos represents mainly deposit feeders. In the trough sequences of the laminated siltstone facies, the scarcity or absence of bioturbation traces indicates that conditions were inimical to endofaunal deposit feeders. 488 PALAEONTOLOGY, VOLUME 31 The vagile benthos, which is less restricted by substrate or habitat than sessile organisms, comprised mainly arthropods and molluscs. Quantitatively, this faunal component is poorly represented through much of the Lower Leintwardine Formation. Of the trilobites, some calymenids were apparently capable of burrowing (e.g. Osgood 1970) and encrinurids may, like some cybelines (Ingham 1968), have been able to conceal themselves in the sediment. However, it is probable that all also spent part of the time actively searching for food, whether as deposit feeders, hunters, or scavengers. Eurypterids, as active predators, were probably similarly epifaunal, crawling or swimming but also able to burrow or hide temporarily within the sediment. Ostracodes, which in size belong to the small macrofauna or meiofauna, include both palaeocope and non-palaeocope groups; beyrichiid ostracodes were apparently benthic, but it is uncertain to what extent they lived in or on the sediment or associated with algal vegetation (e.g. Siveter 1984). Of the vagile molluscs, early Palaeozoic archaeogastropods, which occur only in minor numbers, are thought chiefly to represent epifaunal deposit-feeding microherbivores or algal browsers. Some forms, e.g. bellerophontids, were apparently adapted to soft substrata, by contrast to most Recent representa- tives of this order (e.g. Peel and Wangberg-Eriksson 1979). Orthocone and cyrtocone nautiloids are generally poorly preserved, occurring widely but in low numbers. For Silurian nautiloids, Mutvei (1979) noted the commonly small muscle attachment scars and he interpreted the functional morphology of the siphonal tube as regulatory. He suggested that many nautiloids were possibly pseudoplanktic rather than benthic, as scavengers or rather inactive predators. The chambered, buoyant shell might drift extensively after death of the animal, and nautiloids are most numerous among laminated siltstone sequences of trough areas, both dispersed and in current-aligned assemblages (Hewitt and Watkins 1980). Graptolites, as planktic organisms, were independent of the substrate type and benthic communities. Their dominantly troughward distribution presumably reflects prevailing currents or water mass distribution, modified by preservational factors. Particularly in shelf areas the more turbulent environments were likely to destroy the fragile skeletons. Faunal associations — palaeocommunities? Much has been written on the composition and palaeoecological significance of recurrent benthic fossil assemblages from Lower Palaeozoic sequences. Approaches encompass various taxonomic levels, commonly with an emphasis upon articulate brachiopods as representing the major elements of faunas, although some studies have also included associations dominated by other groups, and with assemblages ranging from local to global in extent (e.g. Ziegler et al. 1968; Boucot 1975, 1981; Jaanusson 1979; Williams et al. 1981). Various interpretations have been made with regard to controls by factors such as depth, substrate, and water energy (e.g. Hancock et al. 1974; Hurst 1975; Lawson 1975; Noble 1979; Jaanusson 1979); a broader shoreline-related environmental classification was proposed for brachiopods by Boucot (1975). More recently the concept of depth regulation of Silurian brachiopod communities has been applied to models for regional sea-level changes (e.g. Johnson et al. 1981). In referring to recurrent fossil associations as ‘communities’, or using ecological terms to describe them, it should be stressed that the composition of fossil assemblages bears only limited relation to any original community structure, and that the evidence from sediments for controls on species distribution is inadequate (e.g. Craig and Jones 1966). However, as the best means available for comparison of fossil with Recent faunas and ecosystems the use of such terminology has become widespread (e.g. Williams et al. 1981; Boucot 1981). Here, the Lower Leintwardine shelf faunas have been subdivided into several facies-related benthic associations, each subject to fluctuating composition in terms of diversity, density, frequency of occurrence, extent, and in the nature of the embedding sediments. Whether or not the composition of an association represents an ecological distribution depends largely upon the degree of transport and mixing of faunas involved during sediment accumulation. In addition, a thanatocoenotic association characteristic of basin sequences is described. The faunal data derived from the main transect from shelf to basin are displayed in text-figs. CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 489 5-12. It is evident that the complexity of faunal distribution relates closely to the sampling frequency (e.g. the detailed patterns of text-fig. 7, cf. more spaced collections in the upper part of text-fig. 5). It is characteristic of individual assemblages that a single species is locally dominant, yet the prevalent species changes from bed to bed and even laterally along bedding surface layers (text-fig. 3). Analysis of skeletal distribution in shelf sequences suggests that this highly variable composition of assemblages represents an original pattern of rapid and local fluctuations in relative species abundance (p. 469). Some monospecific assemblages appear to have undergone more rigorous current reworking, but there is no indication that this has caused significant mixing of faunal ‘belts’ (p. 473). Comparison of statistically defined brachiopod communities for the Ludlow (Calef and Hancock 1974) with the Lower Leintwardine faunas would suggest change in ‘depth- related’ community from bed to bed, but no single, low diversity assemblage can be taken alone as a broader environmental index (e.g. Lawson 1975; Cherns 1979; Watkins 1979). A wider approach encompassing faunal distribution related to overall facies definition, and hence to environmental factors, presents a more representative picture of the faunas (also Watkins 1979). Thus, while there is clearly an overlap in individual species distribution, several associations can be recognized which relate broadly to position and sedimentation patterns along a shelf to trough transect, and also to stratigraphical level. Lower Leintwardine Formation faunal associations Within the areas of calcareous siltstone facies the different ranges occupied by individual brachiopod species were apparently a response primarily to physical stresses along environmental gradients. Within a species range, large local fluctuations in relative abundance appear to represent more temporary controlling factors, e.g. larval recruitment. In Recent environments the most marked faunal boundaries relate to changing sedimentation zones along shelf to trough profiles (e.g. Dorjes 1971; Reineck and Singh 1980). In the Lower Leintwardine Formation the greatest lateral faunal change, from dominantly shelly to graptolitic sequences, corresponds through much of the unit to the offshore edge of the calcareous siltstone facies belt. Only in the basal part of the laminated siltstone sequences of shelf edge areas (e.g. Aymestrey) were shelly faunas common, and these disappeared rapidly towards the areas of greater subsidence not far further west (e.g. Lyepole Bridge). Subdivision may be made primarily within the calcareous siltstone facies of the shelf region into three broad, subtidal benthic faunal associations dominated by epifaunal, suspension- feeding articulate brachiopods, which characterized the Leintwardine formations. A lack of distinct facies boundaries explains the transitional and intergrading composition of these associations, which describe overall changes in faunal emphasis. In general, changing faunal associations represent faunal shifts; immigrations (sensu Jaanusson 1979) are few, the most notable being the incoming in late Leintwardinian times of neobeyrichiid ostracode and trilobite faunas of Baltic aspect. In relation to Boucot’s (1975) classification of benthic faunas these associations correspond to Benthic Assemblages 2 to 4. The first two associations represent lateral equivalents which characterized more inshore and offshore shelf areas respectively through the major part of the Lower Leintwardine Formation. They replaced the Upper Bringewood lower energy carbonate faunas which Watkins (1979) termed the ‘A try pa reticularis- coral Association’, the high energy carbonate faunas with Kirkidium knight ii and tabulate corals (‘A. knightii Association’: Watkins and Aithie 1 980), and also the strophomenid- rich faunas of adjacent areas of silt deposition. The Lower Leintwardine sedimentary facies, which dominantly represent shelf silt environments, indicate frequent higher energy disturbances leading to accumulation mainly of epifaunal but also shallow endofaunal skeletal fossils. Marked and rapid, local faunal shifts and low diversity assemblages are concluded to relate to original patch distributions and to domination by opportunistic species. It is apparent that the carpets of shells and skeletal debris produced by episodes of scouring of surficial sediment might provide areas of ‘hard’ substrate, yet the relatively infrequent destruction and disturbance of skeletal beds as a result of endofaunal activity suggest that many such layers rapidly became buried by sediment, perhaps as a result of the same erosional events. 490 PALAEONTOLOGY, VOLUME 31 Association A. In inshore shelf sequences, and also through the top of the carbonate formation into calcareous siltstone facies in the Wenlock Edge region, the fauna concentrated into winnowed skeletal layers is dominated particularly by an abundance of the rhynchonellides Sphaerirhynchia wilsoni and Microsphaeridiorhynchus nucula , and Isorthis orbicularis. A number of other brachiopod species also occur in abundance and prevalently, although less widely or through more limited parts of sequences, e.g. D. navicula , Howellella elegans , Hyattidina canalis , Protochonetes ludloviensis, Salopian lunata, and Shagamella ludloviensis. These brachiopods include attached, hard substrate species, and cosupportive forms, e.g. D. navicula , Sphaerirhynchia wilsoni , but also recumbent, soft-bottom forms, such as the chonetaceans. Bryozoans and crinoids are associated hard-bottom faunas, although these smaller and fragmented skeletal grains may represent more extensively transported sand and gravel grade material. The soft-bottom, semi-endofaunal or burrowing bivalve F. amygdalina occurs widely and is prevalent in some bedding surface assemblages, but the main indication of local endofauna, apart from in situ L. lewisii, is from the commonly intense bioturbation of sediments. The same faunal association also characterizes the high energy limestone conglomerates of the lower part of the Lower Leintwardine Formation, where in situ hardground formation occurred widely (even within individual deposits). The bryozoan Ptilodictya lanceolata , which anchored by a cemented holdfast, is common in these beds, and attached, strongly ribbed, and robust shelled brachiopods are widely, though not invariably, prevalent. Association B. In areas further out on the shelf, where sheet-laminated deposits are prominent among sequences (e.g. Ludlow, text-fig. 6b), and through the lateral transition into the basal laminated siltstone facies of the Leintwardine to Aymestrey region, a mid-shelf to offshore association is dominated by D. navicula , I. orbicularis , and Shagamella ludloviensis. Of these, only I. orbicularis was a hard-bottom form, the other two species being ambitopic and adapted as adults to a recumbent mode of life. Assemblages are commonly virtually monospecific. Other common species of the first association also occur in some abundance in the mid-shelf part of this zone (e.g. at Ludlow). However, by comparison, Protochonetes ludloviensis does not become common here until higher beds (cf. Lower and Upper Whitcliffe formations), and Salopina lunata has a more limited distribution (again more common in younger beds). Sphaerirhynchia wilsoni and Howellella elegans are chiefly common only in the lower part of the Lower Leintwardine Formation, through the carbonate-silt transition, and the former is much less prominent overall. Hyattidina canalis is not common. Through its main development, this association has fewer common brachiopod species. The infaunal L. lewisii is widely distributed and becomes locally numerous in situ in some beds, in what presumably represent rapidly buried assemblages among the sheet-laminated facies. The soft-bottom bivalve F. amygdalina dominates some assemblages. As in the first association intense bioturbation indicates the activity of soft-bottom endofauna. Currents which produced the sheet-laminated beds scoured the surface sediment, and these beds include accumulations of skeletal epifauna and shallow infauna. The similarities of sediment grade and faunal composition to interbedded sediments argue against extensive transport and faunal mixing having taken place. Towards the shelf edge, I. orbicularis is found in comparable abundance only in the lowermost Lower Leintwardine beds (also, Alexander 1936). D. navicula occurs in enormous, monospecific abundance at the top of the higher energy carbonate facies through into the siltstones (e.g. text- fig. 1 1b), and Shagamella ludloviensis is prevalent with it in the very thinly bedded, basal laminated siltstone facies (e.g. text-fig. 1 1a). In the latter facies (e.g. Aymestrey), valves of the smaller infaunal lingulide L. lata are very common, and also of the epifaunal attached inarticulate Schizocrania striatal However, the two small, soft substrate species D. navicula and Shagamella ludloviensis are greatly dominant. The absence or limited traces of bioturbation in the sediments suggest a greatly reduced soft-bottom, deposit-feeding endofauna. In relation to Ludlow, the main brachiopod fauna is less diverse, and it is notable that small, smooth or finely ribbed, and thin shelled forms are more prominent. The action of offshore, shelf to trough currents is evident from, for example, the linear scour grooves at Aymestrey, and there are some thin, laminated silt sheets. Small and light valves might be relatively easily transported, yet the limited damage among derived, convex-up CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 491 assemblages indicates only moderate current energy and reworking, and generally lower energy conditions prevailed in these outer shelf areas of muddy, finer grained sediment. Above the lower part of the laminated siltstone facies in these areas, skeletal faunas become relatively sparse. Association C. The third association of the calcareous siltstone facies is the Shaleria omatella fauna of the high Lower Leintwardine-Upper Leintwardine beds, which has a very wide but stratigraphically restricted distribution across the shelf region. It is characterized by marked, even monospecific abundance of this recumbent stropheodontide, widely corresponding closely to the disappearance from sequences of Sphaerirhynchia wilsoni and Shagamella ludloviensis, and to minor acmes of Leptaena depressa and Atrypa reticularis. Salopian lunata, P. ludloviensis , and M. nucula are also notably common at this level. Arthropods, molluscs, and bryozoans form minor faunal elements. The association is thus dominated by large, soft-bottom epifaunal to quasi-infaunal brachiopods, together with some hard substrate species. There are both epi- and endobyssate or shallow burrowing bivalves, and diverse vagile skeletal benthos. The sedimentary facies indicate similar physical conditions to beds below, and the presence of an active soft-bottom endofauna. Comparable stropheodontide assemblages lower in the Lower Leintwardine Formation, which have limited and local distribution, are restricted essentially to high energy, winnowed skeletal layers. These may represent introduced faunas or a fairly transient distribution in various shelf areas earlier in Lower Leintwardine Formation times. The major Shaleria omatella fauna evidently occupied a wide tract of the shelf in late Leintwardinian times. The presence of Calymene puellaris, Encrinurus stubblefieldi , and limited occurrence in this association of the ostracode assemblage Neobeyrichia lauensis - scissa-confluens are important since they represent faunal immigrations (e.g. Siveter 1978; Siveter 1983). Association D. Two further faunal associations characterized sequences of the Lower Leintwardine- lower Upper Leintwardine formations. The first of these is the Lingula Iata^~ graptolite fauna of western areas of laminated siltstone facies, including thick trough sequences. The concentration of the small, thin-shelled endofaunal lingulide species in shelf edge and trough sequences (Cherns 1979, cf. epiplanktic mode of life proposed by Watkins and Berry 1977) is notable with regard to the general absence of other skeletal benthos and of endofaunal reworking, which suggests conditions inimical to most benthos. Planktic graptolites form the only other common fauna in these sequences, except in slump units where assemblages of shelly fauna were apparently derived with the sediment from outer shelf environments. Small pterineids, nautiloids, rare cardiolids, and D. navicula valves form minor faunal components. The thin-shelled pterineids represent epibyssate organisms; although they mostly occur as isolated valves, rare articulated specimens at Kerry and Aymestrey indicate local derivation. At this level in the sequence these bivalves are infrequent by comparison with the crowded, transported assemblages of single valves, predominantly of one type, described from the older, Wilsonia Shales formation at Builth (Straw 1937). Orthocones are mostly small, infrequent, and dispersed, although larger and aligned specimens occur in some bedding assemblages. A trough distribution is not necessarily original for these Ludlow cephalopods, though relation chiefly to an offshore ‘graptolitic water mass’ has been suggested (Watkins and Berry 1977; Hewitt and Watkins 1980). It is evident that the association of lingulides and graptolites represents a thanatocoenosis. The sparse and dispersed distribution of both in many sediments contrasts strongly with the rich bedding surface assemblages— if only occasional— which represent current accumulations of each or of the two together (e.g. Lyepole Bridge; Aberedw, Wood 1900). Association E. The remaining association is the Aegiria grayi — ostracode faunas which characterize the high Lower-Upper Leintwardine beds in mid-shelf to trough areas. The association has a wide facies distribution; it occurs in calcareous siltstone facies, particularly the more thickly flaggy type, in the ‘intermediate’ facies to the latter from the laminated siltstone facies, in laminated siltstone facies, and in the poorly calcareous facies of south-western areas. To some extent this association overlaps with the S. omatella shelf fauna (e.g. at Ludlow). The assemblages are typically in thin, winnowed skeletal sand layers where the small size of the grains and limited extent of deposits 492 PALAEONTOLOGY, VOLUME 31 suggest only moderate current energy. In most areas the sediments at this level are well bioturbated, indicating an active soft-bottom endofauna. In the Builth area it is notable that the lithological change up into thickly flaggy calcareous siltstone facies follows closely upon rapid increase in bioturbation and the incoming of this fauna at the top of the laminated siltstone facies. The sequence at Shelderton shows a similar pattern, though spread over a wider interval (text-fig. 11c). However, at Kerry the association first occurs as thin skeletal sand layers among laminated siltstone facies which lack traces of bioturbation (text-fig. 12). The assemblages comprise the small brachiopod A. grayi, beyrichiacean and non-palaeocope ostracodes, associated with variable amounts of fine skeletal sand debris, small crinoid ossicles, trilobites, and D. navicula. It is arguable whether A. grayi represents a benthic or epiplanktic form (p. 486), but beyrichiaceans were apparently vagile benthos. With regard to the ecological significance of this association it is important to note that there is some succession in the beyrichiacean ostracode assemblages; the first occurrences at Ludlow and in shelf edge areas are of A. grayi with small beyrichiaceans, followed in higher beds by the incoming of the neobeyrichiid species N. nutans , and then by the main assemblage of N. lauensis-scissa-confluens. Around this same level in sequences there is also an increase in vagile benthos represented by trilobites. The distribution and composition of the association suggest a benthic epifaunal brachiopod assemblage of outer shelf environments, associated with a significant vagile arthropod skeletal benthos and with an active, largely soft- bodied soft-bottom endofauna. Its spread westwards at the scissa-lauensis level, mainly closely linked to lithological change to coarser sediments, suggests regional environmental shifts. Whether the distribution of algal material was an important factor either in the benthic environment or as a transporting medium, particularly in areas where benthos was otherwise scarce, cannot be assessed. The incoming of the association as skeletal sand layers into areas of fine silts which previously lacked, or had limited benthic organisms, might represent an early stage during colonization by benthic faunas (e.g. Jaanusson 1984). PALAEOGEOGRAPHY The faunal shifts which define the base of the Lower Leintwardine Formation in the type area correlate broadly with the transition from low to moderate energy, carbonate mud and silt depositional environments into dominantly clastic silt sedimentation (text-figs. 13a, b and 14). Similar carbonates, which Watkins and Aithie (1980) interpreted as back-barrier shelf facies, extended inshore across the shelf, and in these areas the faunal boundary is again near the top of the carbonate development. At the boundary, fairly sparse though diverse ‘residual' assemblages of beds high in the Upper Bringewood Formation were replaced, with loss of a number of diagnostic fossils, mainly by articulate brachiopod assemblages of long-ranging forms. Further offshore, to the west, where the Upper Bringewood Aymestry Limestone is in high and moderate energy, carbonate sand and gravel facies that represent an outer shelf barrier belt (Watkins and Aithie 1980), the corresponding faunal changes lie well within the carbonate formation. Here, richer Bringewoodian assemblages of large recumbent brachiopods (pentameraceans, strophomen- aceans) and tabulate corals in sand grade, mud- to grain-supported carbonates gave way to prolific, hard- and soft-bottom ‘small’ brachiopod faunas, particularly D. navicula , in facies of fluctuating but generally lower depositional energy levels. Clastic silt deposition soon became dominant also in these areas, although carbonate environments persisted, with some eastward shift in depositional focus, through lower energy mud- and silt-grade facies across part of the Wenlock Edge region. The Upper Bringewood-Lower Leintwardine faunal boundary across the inshore shelf areas (i.e. as seen in the infers) is associated widely with breaks in deposition, episodic erosion, and hardground formation, indicated by skeletal conglomerates that occur both within the Aymestry Limestone muddy carbonate belt and in laterally adjacent carbonate-rich clastic depositional zones. The latter, which extend southwards from south Woolhope, are thinner sequences in which beds of winnowed, coarse conglomeratic skeletal sand indicate episodes of raised current energy among CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 493 text-fig. 13. a, isopach map (in m) for Leintwardinian strata, also showing major tectonic features. CSF = Church Stretton Fault; PLF = Pontesford-Linley Fault; TL = Towy lineament, b-d, palaeogeo- graphical maps for the Leintwardinian; b, ‘basal Lower Leintwardine Formation’ times; c, ‘middle Lower Leintwardine Formation’ times; d, ‘basal Upper Leintwardine Formation’ times. See text-fig. 5 key. generally fine, low energy muddy silts. Early lithification of conglomerates affects even some mud- grade carbonates, which suggests that net deposition in these areas was low. Possible stromatolitic drapes across one hardground would indicate that very shallow conditions prevailed locally (Cherns 1980, fig. 5). The replacement of Bringewoodian strophomenid-coral assemblages by Lower Leintwardine brachiopod-bryozoan faunas corresponds closely to the onset ol reduced and interrupted sedimentation in these areas. The single non-skeletal conglomerate which represents an in situ carbonate mud hardground horizon in the north-east Wenlock Edge Dean borehole Cennen Valley 494 PALAEONTOLOGY, VOLUME 31 text-fig. 14. Facies and thickness variations for Leintwardinian strata along transects a, NW-SE (= primary study transect from shelf to trough) and b, SW-NE across the main outcrop areas. See text- fig. 5 key. CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 495 provides evidence that depositional breaks affected a very wide tract of the inner shelf region. The same time interval also saw increasing clastic silt influx across the entire shelf area. Along the offshore, well-defined western edge of the Upper Bringewood carbonate belt (text-fig. 14, e.g. at Aymestrey and south Shelderton), a sharp and locally unconformable junction at the top of the Limestone, followed by laminated siltstone facies, corresponds to the faunal boundary. The abrupt facies change brought in fine clastic sedimentation and markedly lower energy environments, where there was apparently little or no endofaunal reworking. Instability and interrupted sedimentation also in this area in late Bringewoodian -early Leintwardinian times are indicated locally by an intraformational conglomerate in the Limestone, and by an interformational discontinuity as represented by an eroded, planed, and pitted top surface to the Limestone (text- fig. 11a). Overlying thin, laminated siltstones include numerous small scour grooves which have ENE-WSW current alignment indicative of offshore current flow. Shelf to trough slumping at this level, seen not far to the west (Lyepole Bridge) in markedly thicker fine clastic sequences, suggests marked instability across an important structural lineament which controlled the ‘hinge’ region between shelf and trough sequences (notably, associated with the Church Stretton strike-slip tectonic zone). The correspondingly sharp faunal shift at the Limestone-laminated siltstone boundary introduced a mixed assemblage of small, largely soft-bottom epifaunal brachiopods, together with the ‘trough association’ of endofaunal lingulides and planktic graptolites. The broad facies relation of faunal shifts at the base of the Lower Leintwardine Formation, and long stratigraphical range of the characteristic Leintwardinian shelly benthos make uncertain the precise biostratigraphical significance of this boundary across the shelf region in relation to the leintwardinensis Biozone. In trough areas, change in the graptolite fauna occurs within thick sequences of laminated siltstones. Wherever graptolites occur in the Lower Leintwardine beds of shelf sequences they belong to the leintwardinensis Biozone, and they are sufficiently widespread to provide reasonable faunal control (text-figs. 5-12); records of Saetograptus n. sub-sp. from Ludlow, Perton, and May Hill, all from the lower part of the formation, are interesting. Brachiopod distributions suggest diachroneity in detail with regard to the top of the carbonate facies (e.g. text- fig. 14). The reduced and interrupted sedimentation which accompanied the faunal changes, and which also heralded the end of carbonate deposition across much of the shelf, indicates prolonged periods of sediment starvation affecting areas right across the level bottom shelf although having most influence in the more marginal, inshore areas. Also in south-western areas the faunal shifts are associated with uplift and instability. At Usk the single quartz conglomerate, of probable southern provenance, corresponds closely to a change from silty carbonate to silt deposition, and to the faunal changes. In the Llandovery-Llandeilo region, marine transgression, at least in the eastern part, brought in poorly to non-calcareous sandstones and mudstones that have modified Leintwardinian faunas, overlying apparently Bringewoodian, southerly derived deltaic and fluviatile Trichrug Beds. The environmental changes which led to clastic silt deposition across the shelf region resulted in a wide, level bottom subtidal belt that lay within reach of storm currents (text-fig. 13c). The high energy, offshore barrier environments of the Upper Bringewood Formation carbonates lay within wave base, so that facies changes here apparently involved transgression. Loss of an elevated barrier belt was accompanied by increased circulation in previously sheltered back barrier shelf areas, where despite relative increase in clastic sediment supply the net deposition was very slow. Intermittent breaks in deposition are most apparent around possible submarine swells of the inner shelf, where sequences are much condensed (text-fig. 14). Ebb currents, perhaps from storm disturbances, swept the shelf region, across the remnant carbonate zone. Offshore flow possibly triggered slumping beyond the unstable shelf margins down into the basin. On the western margin of the depositional trough the major downslope slumping from the NW which began in Eltonian times continued into the leintwardinensis Biozone only in the northern Kerry-Clun Forest area. Trough deposition of laminated silts continued uninterrupted from Bringewoodian, twnescens/ incipiens to leintwardinensis biozones without facies change. Along the trough axis, indigenous 496 PALAEONTOLOGY, VOLUME 31 benthos was largely restricted to the endofaunal L. lata , and conditions across wide areas were apparently unsuitable for most epifaunal and endofaunal benthos. The calcareous siltstone facies through the major part of the Lower Leintwardine Formation in shelf areas, above the basal conglomeratic sequences, was characterized by more continuous deposition, although the closely spaced, winnowed skeletal layers that occur among well-bioturbated sediments indicate numerous higher energy, relatively minor storm disturbances. Sheet laminated silt units, which represent offshore transport of clastic material by storm-generated traction or turbidity currents, dominated deposition across the mid-shelf region of Ludlow. Their basal skeletal layers are accumulations mainly of benthic epifauna local to (except possibly for certain stropheodontide horizons) the depositional area. In inshore shelf areas the sheet-laminated units are fewer and thinner, or perhaps are not evident because of the intense reworking of surface sediment which is typical of these areas of lower net sedimentation. Across the level bottom shelf, skeletal faunas were very common and mainly dominated by suspension-feeding articulate brachiopods. The two major faunal associations described corresponded to more onshore and offshore parts of the shelf. In the shelf edge region, in laminated siltstone facies, shelly faunas become notably sparser above the basal sequences. Traces of endofaunal reworking are also few, indicating that benthic faunas became much reduced or largely absent from these areas as they were from trough areas further to the west. Whether this change represents further transgression and deepening, or perhaps restricted circulation, remains equivocal. In the higher part of shelf sequences, sheet-laminated silt units diminish in frequency. In the eastern Ludlow area, a further faunal shift brought in Shaleria ornatella in abundance among the Lower Leintwardine associations, without facies change. Towards the west at Ludlow, there was an increasing component of deposit-feeding arthropods among skeletal faunas, which occur rather more dispersed in the sediments. Among both trilobites and ostracodes this level saw the first immigration of typically Baltoscandian species. The small epifaunal, or possibly epiplanktic, brachiopod A. grayi returned among Ludlow faunas, in what represents a faunal shift and its first appearance in the Leintwardine beds. Planktic graptolites also became relatively common in the mid-shelf region of Ludlow. These faunal changes might suggest that rather wider current circulation replaced the prevalently offshore flow which had influenced earlier shelf deposition, and on a regional scale immigration from the Baltic area was now taking place. However, continuing offshore-downslope currents and instability at the shelf margin are evident from troughward channelling along a series of submarine canyons, which were active and involved considerable downcutting in late Leintwardinian times (Whitaker 1962). The Lower-Upper Leintwardine boundary interval (text-fig. 13d) in shelf areas of calcareous siltstone facies shows renewed evidence of interrupted and reduced sedimentation. Thin conglomer- atic and phosphatized horizons occur widely, from inshore shelf areas (e.g. May Hill, Perton) to Ludlow, and across to Aymestrey and Leintwardine. Across the inshore shelf, the S. ornatella faunal association rapidly became prevalent, corresponding to loss of several common Lower Leintwardine brachiopods, e.g. Sphaerirhynchia wilsoni and Shagamella ludloviensis , and introducing the index Upper Leintwardine trilobites. The Upper Leintwardine shelly unit is thin, spanning an interval of condensed sedimentation which culminated in yet more widespread breaks in shelf deposition, commonly with formation of bone-beds (e.g. Holland et al. 1963). Towards the offshore shelf region, where the Shaleria ornatella fauna was absent or greatly reduced, the perhaps slightly younger though overlapping A. grayz'-neobeyrichiacean association characterized late Leintwardinian sequences. Shortly below this, reintroduction of benthic endo- fauna and skeletal epifaunas accompanied a facies transition which involved a fairly rapid upward coarsening of sediment, from laminated siltstones through ‘intermediate’ into more thickly flaggy, calcareous siltstone facies. Deposit-feeding vagile arthropods became conspicuous among the fairly rich, offshore skeletal fauna of well-bioturbated sediments, in which the ‘basin association’ of graptolites and L. lata also occurred commonly. A. grayi appeared here in abundance some way below the diagnostic ostracode faunas. Similar, though poorly fossiliferous and virtually CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 497 unbioturbated siltstones filled the heads of nearby active submarine canyons. The offshore spread of diverse benthic faunas indicated in the shelf edge region is also notable for the markedly higher proportion of deposit-feeding skeletal benthos than in earlier and inshore associations. Periods of reduced and interrupted sedimentation also affected offshore shelf areas, as evidenced by phosphatized and conglomeratic horizons. The loss of graptolites from Leintwardine corresponds closely to the coarser silt influx into this area (e.g. text-fig. 1 lc). Across a wide marginal tract of the south-west trough area (e.g. Builth: Straw 1937, 1953) there was similar facies development in late Leintwardinian times. Upward coarsening of sediment and increasing current influence were accompanied by offshore spread of shelly benthos and endofaunal reworking. The late Leintwardinian palaeogeographical changes suggest regression in response to late Caledonian regional development, leading to coarser silt influx into the trough, wider current circulation, and offshore migration of shelf environments. Only in the north-western part of the trough, further down the axial slope, was laminated silt facies being deposited at this time, though slumping from the trough margins had ceased also in these areas. Thin beds with the A. grayi- neobeyrichiacean association which occur among the laminated siltstones (e.g. Kerry) suggest that perhaps even the low energy trough environments were occasionally colonized by offshore skeletal epibenthos. Some areas remained, however, inimical to benthic faunas, as at Knighton where thick barren sequences continued through into Whitcliflfian times (Holland 1959). The late Leintwardinian regression continued through the Whitcliflfian, culminating in transition to non-marine environments. As in the Leintwardinian, there was considerable differential subsidence from shelf to trough areas. Shelf sequences are thin and interrupted, in calcareous siltstone facies, while ‘trough’ sequences are far thicker, in the more thickly flaggy, calcareous siltstone facies which extended from the Ludlow area westwards. The latter facies includes an increasing proportion of sheet-laminated units (e.g. Watkins 1979). The trough filled rapidly with silt and fine sand, and shelly shelf benthos spread throughout the outcrop area. The fauna are mostly long-ranging forms continuing from earlier beds, though with loss of a number of diagnostic Leintwardine forms (Holland et al. 1963). Watkins (1979) interpreted the Ludlow Series sediments of the shelf areas as an overall regressive sequence of mainly terrigenous facies, interrupted by a carbonate development in late Bringewoodian times. A ‘S', ornatella Association’, equivalent to that described here, was considered an ecological anomaly, interrupting a continuous faunal gradation between two associations which characterized Lower Leintwardine and Whitcliffe formations respectively. It was concluded that the S. omcitella fauna indicated a temporary shelf-wide change in current system which transported a distal shelf community inshore across the shelf. However, the wider distribution of S. ornatella and A. grayi- neobeyrichiacean associations as described here indicates that the former was strongly concentrated in inshore to mid-shelf regions, and virtually absent from offshore shelf regions where the latter became prevalent. Wider circulation patterns and some regression appear probable, with breakdown of certain regional barriers to migration. If S. ornatella were conspecific with the shaleriid which became prevalent on Gotland with the lauensis-scissa fauna (Bassett and Cocks 1974), this might also represent a late Leintwardinian immigration to the Anglo-Welsh area. The distribution pattern suggests opportunistic spread, and S. ornatella dominated inshore shelf environments through a period of limited sedimentation. An important consideration for Leintwardinian palaeogeography relates to trough environments. The absence of most benthos from laminated siltstone facies through much of the Biozone has generally been ascribed to depth, although the wide distribution of L. lata suggests that perhaps other factors, such as low oxygen concentrations, may have been limiting. Most Ludlow trough facies have been considered as deep water, with intervals of distal turbiditic facies representing axial current flow, and major synsedimentary slide sheets indicating significant marginal slopes (Bailey 1969; Woodcock 1976a, b). Whitaker (1962) estimated Leintwardinian submarine canyons as cutting relatively steeply through marginal shelf areas. However, Woodcock (1984) noted that most Ludlow turbidites might be distal storm-generated shelf turbidites laterally equivalent to shelf storm sand and silt units, which would imply that trough areas were relatively shallow. It is notable 498 PALAEONTOLOGY, VOLUME 31 that the leintxvardinensis laminated siltstones of the basin areas are not in graded, apparently turbiditic units despite the evidence that storm current ebb flows influenced shelf deposition through this period. By late Leintwardinian times there was offshore migration of shelf facies and faunas, probably as a result of regression rather than tectonic controls on sediment supply. There remained marked differential subsidence from shelf to trough (e.g. text-fig. 14). The correspondence of facies shifts with the evidence of regional faunal immigrations suggests that wider circulation patterns were important in the rapid offshore spread of shelf benthos, and not purely depth controls, since in shelf areas there was no significant change in facies. Acknowledgements. I am grateful to R. B. Rickards (Cambridge) and D. J. Siveter (Leicester) for identification of specimens, to D. E. White (British Geological Survey) for access to borehole data, specimens, and field notes, and to M. G. Bassett (Cardiff), V. Jaanusson (Stockholm), and J. D. Lawson (Glasgow) for their help and for constructive criticism of the manuscript. The study was carried out during the tenure of a NERC studentship at Glasgow. The collections upon which this study is based are housed at the British Geological Survey, Keyworth (for Ludlow area: BGS Zs3874 4503) and the National Museum of Wales, Cardiff (all other areas: Accession No. 78.35G). REFERENCES Alexander, F. e. s 1936. The Aymestry Limestone of the main outcrop. Q. J / geol. Soc. Loud. 92, 103-1 15, pi. 8. bailey, r. j. 1969. Ludlovian sedimentation in south central Wales. Pp. 283-304. In wood, a. (ed. ). The Precambrian and Lower Palaeozoic rocks of Wales. University of Wales Press, Cardiff. ball, h. w. 1951. The Silurian and Devonian rocks of Turner’s Hill and Gornal, south Staffordshire. Proc. Geol. Ass. 62, 225-236. bassett, m. G. 1984. Life strategies of Silurian brachiopods. In bassett, m. g. and lawson, j. d. (eds.). Autecology of Silurian organisms. Spec. Pap. Palaeont. 32, 237-263. -and cocks, l. r. m. 1974. A review of Silurian brachiopods from Gotland. Fossils Strata , 3, 1-56, pis. 1 11. benson, r. h. 1961. Ecology of ostracode assemblages. Pp. 56- 63. In moore, r. c. (ed.). Treatise on Invertebrate Paleontology , Part Q , Arthropoda 3. Geological Society of America and University of Kansas Press, New York and Lawrence, Kansas. Bergstrom, j. 1968. Some Ordovician and Silurian brachiopod assemblages. Lethaia , 1, 230-237. boucot, A. J. 1975. Evolution and extinction rate controls , xvi + 427 pp. Elsevier, Amsterdam. 1981. Principles of benthic marine paleoecology, xv + 463 pp. Academic Press. — brace, w. and DeMAR, r. 1958. Distribution of brachiopod shells by currents. J. sedim. Petrol. 28, 321-332. brenchley, p. j. and newall, G. 1970. Flume experiments on the orientation and transport of models and shell valves. Palaeogeogr. Palaeoclinuit. Palaeoecol. 7, 185-220. bromley, r. G. 1967. Marine phosphorites as depth indicators. Mar. Geol. 5, 503-509. brood, k. 1979. Bryozoans. In jaanusson, v., laufeld, s. and skoglund, r. (eds.). Lower Wenlock faunal and floral dynamics — Vattenfallet section, Gotland. Sver. geol. Unders. C762, 172-180. — 1984. Bryozoan ecology in the Silurian of Gotland. Spec. Pap. Palaeont. 32, 21 1 224, pis. 1 4. calef, c. e. and Hancock, n. j. 1974. Wenlock and Ludlow marine communities in Wales and the Welsh Borderland. Palaeontology , 17, 779-810. cave, r. and white, d. e. 1971. The exposures of Ludlow rocks and associated beds at Tites Point and near Newnham, Gloucestershire. Geol. J. 7, 239-254. — 1978. Stratigraphy of the Brookend (Vine Farm) borehole. Bull. Ludlow Res. Grp. 25, 44 45. cherns, L. 1979. The environmental significance of Lingula in the Ludlow Series of the Welsh Borderland and Wales. Lethaia , 12, 35-46. — 1980. Hardgrounds in the Lower Leintwardine Beds (Silurian) of the Welsh Borderland. Geol. Mag. 117, 31 1 326. cocks, l. r. m. 1970. Silurian brachiopods of the superfamily Plectambonitacea. Bull. Br. Mus. nat. Hist. (Geol.), 19, 139-203, pis. I 17. coppack, b. c. and white, d. e. 1974. (Log of Dean borehole). Inst. geol. Sci. Rep. 74/7, 6. CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 499 coward, m. p. and siddans, a. w. b. 1980. The tectonic evolution of the Welsh Caledonides. In Harris, a. L., Holland, c. H. and leake, b. E. (eds.). The Caledonides of the British Isles- reviewed. Spec. Pubis geol. Soc. Land. 8, 187 198. craig, G. y. 1967. Size-frequency distributions of living and dead populations of pelecypods from Bimini, Bahamas, B.W.I. J. Geol. 75, 34 45. — and hallam, a. 1963. Size-frequency and growth-ring analyses of Mytilus edulis and Cardium edule , and their palaeoecological significance. Palaeontology , 6, 731 750. — and jones, n. s. 1966. Marine benthos, substrate and palaeoecology. Ibid. 9, 30-38. curry, G. b. 1981. Variable pedicle morphology in a population of the Recent brachiopod Terebratulina septentrionalis. Lethaia , 14, 9-20. dorjes, j. 1971. Der Golf von Gaeta (Tyrrhenisches Meer) IV. Das Makrobenthos und seine kiistenparallele Zonierung. Senckenberg. marit. 3, 203-246. dunham, r. j. 1962. Classification of carbonate rocks according to depositional texture. In ham, w. e. (ed. ). Classification of carbonate rocks. Mem. Am. Ass. Petrol. Geol. 1, 108 121. earp, J. R. 1938. The higher Silurian rocks of the Kerry district, Montgomeryshire. Q. Jl geol. Soc. bond. 94, 125 160, pis. 12 and 13. — 1940. The geology of the south-western part of Clun Forest. Ibid. 96, I 11, pi. 1. elles, G. L. and slater, i. l. 1906. The highest Silurian rocks of the Ludlow District. Ibid. 62, 195-222, pi. 22. emery, k. o. 1968. Positions of empty pelecypod valves on the continental shelf. J. sedim. Petrol. 38, 1264 1269. emig, c. c. 1982. Terrier et position des Lingules (Brachiopodes, Inarticules). Bull. Soc. zool. Fr. 107, 185-194. — gall, j.-c., pajaud, d. and plaziat, j.-c. 1978. Reflexions critiques sur l’ecologie et la systematique des Lingules actuelles et fossiles. Geobios , 11, 573-609, pis. 1 and 2. fursich, f. t. and hurst, j. m. 1974. Environmental factors determining the distribution of brachiopods. Palaeontology , 17, 879-900. — 1981. Autecology of the Silurian brachiopod Sphaerirhynchia wilsoni (J. Sowerby, 1816). J. Paleont. 55, 805-809. greig, d. c., wright, J. e., hains, b. a. and mitchell, G. h. 1968. Geology of the country around Church Stretton, Craven Arms, Wenlock Edge and Brown Clee. Mem. geol. Surv. UK , xiii -E 1 379 pp. hallam, a. 1967. The interpretation of size-frequency distributions in niolluscan death assemblages. Palaeontology , 10, 25 42. hammond, L. s. 1983. Experimental studies of salinity tolerance, burrowing behaviour and pedicle regeneration in Lingula anatina (Brachiopoda, Inarticulata). J. Paleont. 57, 131 1 1316. Hancock, n. h., hurst, j. M. and fursich, f. t. 1974. Depths inhabited by Silurian brachiopod communities. J. geol. Soc. Lond. 130, 151 156. harms, J. c. 1975. Stratification produced by migrating bed forms. In Depositional environments as interpreted from primary sedimentary structures and stratification sequences. Soc. econ. Paleont. Miner., Short Course, 2, 45-61. Harrington, h. j. 1959. General description of Trilobita. Pp. 38-117. In moore, r. c. (ed.). Treatise on Invertebrate Paleontology, Part O , Arthropoda 1 . Geological Society of America and University of Kansas Press, New York and Lawrence, Kansas. havliCek, v. 1967. Brachiopoda of the order Strophomenidina in Czechoslovakia. Rozpr. ustred. Ust. geol. 33, I 235. and vanek, j. 1966. The biostratigraphy of the Ordovician of Bohemia. Sb. geol. Vecl. Praha, Rada P, 8, 7-69. hewitt, r. a. and watkins, R 1980. Cephalopod ecology across a late Silurian shelf tract. Neues Jb. Geol. Paldont. Abh. 160, 96-117. Holland, c. h. 1959. The Ludlovian and Downlonian rocks of the Knighton district, Radnorshire. Q. Jl geol. Soc. Lond. 114, 449-482, pi. 21. 1971. Some conspicuous participants in Palaeozoic symbiosis. Scient. Proc. R. Dubl. Soc., Ser. A, 4, 15-26, pi. 2. — 1980. Silurian series and stages: decisions concerning chronostratigraphy. Lethaia , 13, 238. — and lawson, j. d. 1963. Facies patterns in the Ludlovian of Wales and the Welsh Borderland. Lpool Manchr geol. J. 3, 269-288, pi. 14. — and walmsley, v. g. 1959. A revised classification of the Ludlovian succession at Ludlow. Nature, Lond. 184, 1037 1039. 500 PALAEONTOLOGY, VOLUME 31 Holland, c. h., lawson, j. d., and walmsley, v. g. 1963. The Silurian rocks of the Ludlow district. Bull. Br. Mus. nat. Hist. (Geol.), 8, 95 171, pis. 1-7. and white, d. e. 1980. Ludlow Stages. Lethaia, 13, 268. hurst, j. m. Wcnlock carbonate, level bottom, brachiopod-dominated communities from Wales and the Welsh Borderland. Palaeogeogr. Palaeoclimat. Palaeoecol. 17, 227-255. ingham, j. k. 1968. British and Swedish Ordovician species of Cybeloides (Trilobita). Scott. J. Geol. 4, 300- 316, pis. 1 and 2. jaanusson, v. 1971. Evolution of the brachiopod hinge. Smithson. Contr. Paleobiol. 3, 33-46. 1976. Faunal dynamics in the Middle Ordovician (Viruan) of Balto-Scandia. Pp. 301 326. In bassett, M. G. (ed . ). The Ordovician System: proceedings of a Palaeontological Association symposium , Birmingham , September 1974. University of Wales Press and National Museum of Wales, Cardiff, 696 pp. 1979. Stratigraphical and environmental background. Ecology and faunal dynamics. In jaanusson, v., laufeld, s. and skoglund, r. (eds.). Lower Wenlock faunal and floral dynamics— Vattenfallet section, Gotland. Sver. geol. Unders. C'762, 1 1 -38, 253-294. 1984. Ordovician benthic macrofaunal associations. In bruton, d. l. (ed.). Aspects of the Ordovician System. Palaeont. Contr. Univ. Oslo , 295, 127-139. jackson, j. b., goreau, t. f. and hartman, w. d. 1971. Recent brachiopod-coralline sponge communities and their paleoecological significance. Science , 173, 623-625. Johnson, m. e., 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. king, w. w. and lewis, w. j. 1912. The uppermost Silurian and Old Red Sandstone of south Staffordshire. Geol. Mag. 9, 437 443, 484-491. kirk, n. h. 1951. The Silurian and Downtonian rocks of the anticlinal disturbance of Breconshire and Radnorshire: Pont Faen to Presteigne. Abstr. Proc. geol. Soc. Lond. 1474, 12-1 A. larsson, K. 1979. Coiled serpulid polychaetes. In jaanusson, v., laufeld, s. and skoglund, r. (eds.). Lower Wenlock faunal and floral dynamics— Vattenfallet section, Gotland. Sver. geol. Unders. C762, 103. lawson, j. d. 1954. The Silurian succession at Gorsley (Herefordshire). Geol. Mag. 91, 227-237. 1955. The geology of the May Hill inlier. Q. Jl Geol. Soc. Lond. Ill, 85 116, pi. 6. 1960. The succession of shelly faunas in the British Ludlovian. C.r. Int. geol. Congr., 21st Session, Norden , 7, 114 125. 1973. Facies and faunal changes in the Ludlovian rocks of Aymestrey, Herefordshire. Geol. J. 8, 247-278. 1975. Ludlow benthonic assemblages. Palaeontology , 18, 509 525. -and whitaker, J. H. mcd. 1969. Correlation of the Leintwardine Beds. Geol. J. 6, 329-332. lever, j. 1958. Quantitative beach research. 1. The ‘left-right phenomenon’; sorting of lamellibranch valves on sandy beaches. Basteria, 22, 21-51. — and thijsson, R. 1968. Sorting phenomena during the transport of shell valves on sandy beaches studied with the use of artificial valves. Symp. zool. Soc. Lond. 22, 259-271. lockley, m. G. and antia, d. d. j. 1980. Anomalous occurrences of the Lower Palaeozoic brachiopod Schizocrania. Palaeontology , 23, 707-713. mccammon, h. m. and buchsbaum, r. 1968. Size and shape variations of three Recent brachiopods from the straits of Magellan. In schmitt, w. l. and llano, g. a. (eds.). Biology of Antarctic Seas 3. Antarctic Res. Ser. Washington, 11, 215-225. martinsson, a. 1962. Ostracodes of the family Beyrichiidae from the Silurian of Gotland. Bull. geol. Instn. Univ. Uppsala, 41, 1-369. [Also issued as Publ. pal. Inst. Univ. Uppsala, 41, 1-369.] mutvei, H. 1979. Cephalopods. In jaanusson, v., laufeld, s. and skoglund, r. (eds.). Lower Wenlock faunal and floral dynamics — Vattenfallet section, Gotland. Sver. geol. Unders. C762, 113- 115. noble, j. p. a. 1979. The relative significance of depth and substrate in Lower Paleozoic community distributions. Palaeogeogr. Palaeoclimat. Palaeoecol. 27, 167-178. osgood, r. g. 1970. Trace fossils of the Cincinnati area. Palaeontogr. am. 6, 281-444. palmer, t. j. 1982. Cambrian to Cretaceous changes in hardground communities. Lethaia, 15, 309 323. peel, J. s. and wangberg-eriksson, k. 1979. Gastropods. In jaanusson, v., laufeld, s. and skoglund, r. (eds.). Lower Wenlock faunal and floral dynamics— Vattenfallet section, Gotland. Sver. geol. Unders. C762, 105 108. phipps, c. B. and reeve, f. a. e. 1967. Stratigraphy and geological history of the Malvern, Abberley and Ledbury Hills. Geol. J. 5, 339-368. CHERNS: SILURIAN FAUNAL AND FACIES DYNAMICS 501 plaziat, j.-c., pajaud, d., emig, c. c. and gall, j.-c. 1978. Environments et distribution bathymetrique des Lingules actuelles; consequences pour les interpretations paleogeographiques. Bull. Soc. geol. Fr. 20, 309-314. potter, j. f. and price, j. h. 1965. Comparative sections through rocks of Ludlovian-Downtonian age in the Llandovery and Llandeilo districts. Proc. Geol. Ass. 76, 379- 402. reineck, h.-e. and singh, I. b. 1980. Depositional sedimentary environments (2nd edn.), x + 549 pp. Springer- Verlag, New York, Heidelberg, Berlin. rickwood, a. e. 1968. A contribution to the life history and biology of the brachiopod Pumilus antiquatus (Atkins). Trans. R. Soc. NZ (Zoo!.), 10, 163-182. rudwick, m. j. s. 1961. The anchorage of articulate brachiopods on soft substrata. Palaeontology, 4, 475-476. — 1965. Ecology and paleoecology. Pp. 199-214. In moore, r. c. (ed. ). Treatise on Invertebrate Paleontology , Part H , Brachiopoda. Geological Society of America and University of Kansas, New York and Lawrence, Kansas. 1970. Living and fossil brachiopods [Reprinted 1974], 199 pp. Hutchinson, London. sheehan, p. m. 1977. Ordovician and Silurian brachiopods from graptolitic shales and related deep-water argillaceous rocks. Lethaia , 10, 201-203. shergold, j. h. and shirley, j. 1968. The faunal-stratigraphy of the Ludlovian rocks between Craven Arms and Bourton, near Much Wenlock, Shropshire. Geol. J. 6, 119-138, pis. 14 and 15. siveter, david j. 1978. The Silurian. In bate, r. ll, robinson, e. and sheppard, l. m. (eds.). A stratigraphical index of British Ostracoda. Geol. Jl special issue , 8, 57-100, pis. 1-9. 1984. Habitats and modes of life of Silurian ostracodes. Spec. Pap. Palaeont. 32, 71 85, pis. 1, 2. siveter, derek j. 1983. Calymene lawsoni and allied species from the Silurian of Britain and their stratigraphical significance. Ibid. 30, 69-88, pis. 7-10. squirrell, H. c. and downing, R. a. 1969. Geology of the South Wales Coalfield. Part 1. The country around Newport (Mon.) (3rd edn.). Mem. geol. Surv. UK. xiii + 333 pp. — and tucker, e. v. 1960. The geology of the Woolhope inlier, Herefordshire. Q. Jl geol. Soc. Lond. 116, 139- 185, pi. 15. — and white, d. e. 1978. Stratigraphy of the Silurian and Old Red Sandstone of the Cennen Valley and adjacent areas, south-west Dyfed, Wales. Rep. Inst. geol. Sci. 78/6. Stanley, s. m. 1970. Relation of shell form to life habits of the Bivalvia (Mollusca). Mem. geol. Soc. Am. 125, 1-296. — 1972. Lunctional morphology and evolution of byssally-attached bivalve mollusks. J. Paleont. 46, 165- 212. straw, s. h. 1937. The higher Ludlovian rocks of the Builth District. Q. Jl geol. Soc. Lond. 93, 406 453, pis. 27-29. — 1953. The Silurian succession at Cwm Graig Ddu, Breconshire. Lpool Manchr geol. J. 1, 208-219. thayer, c. w. 1975. Size-frequency and population structure of brachiopods. Palaeogeogr. Palaeoclimat. Palaeoecol. 17, 139-148. — 1981. Ecology of living brachiopods. In broadhead, t. w. (ed.). Lophophorates. Univ. Tennessee , Dept. Geol. Sci., Stud. Geol. 5, 110- 126. — and steele-petrovic, h. m. 1975. Burrowing of the lingulid Glottidia pyramidata: its ecologic and paleoecologic significance. Lethaia, 8, 209-22 1 . trewin, n. h. and walsh, w. 1972. Transport, breakage and sorting of the bivalve Mactra corallina on Aberdeen beach, Scotland. Palaeogeogr. Palaeoclimat. Palaeoecol. 12, 193-204. tucker, e. v. 1964. The ecology of the brachiopod Dayia navicula (J. de C. Sowerby). Ann. Mag. nat. Hist., Ser. B, 7, 339-345. 1968. The atrypidine brachiopod Dayia navicula (J. de C. Sowerby). Palaeontology, 11, 612-626, pis. 118-121. turner, s. 1973. Silurian-Devonian thelodonts from the Welsh Borderland. ,/. geol. Soc. 129, 557 584. walmsley, v. G. 1959. The geology of the Usk inlier (Monmouthshire). Q. Jl geol. Soc. Lond. 114, 483-521, pi. 22. watkins, r. 1979. Benthic community organization in the Ludlow Series of the Welsh Borderland. Bull. Br. Mus. nat. Hist. (Geol.), 31, 175-280. — and aithie, c. j. 1980. Carbonate shelf environments and faunal communities in the Upper Bringewood Beds of the British Silurian. Palaeogeogr. Palaeoclimat. Palaeoecol. 29, 341-368. 502 PALAEONTOLOGY, VOLUME 31 watkins, R. and berry, w. b. n. 1977. Ecology of a Late Silurian fauna of graptolites and associated organisms. Lethaia, 10, 267-286. westbroek, p., NEiJNDORFF, f. and stel, j. h. 1975. Ecology and functional morphology of an uncinulid brachiopod from the Devonian of Spain. Palaeontology, 198, 367-375, pi. 51. whitaker, j. h. mcd. 1962. The geology of the area around Leintwardine, Herefordshire. Q. J I geol. Soc. Lond. 117, 319-351. white, d. e. and lawson, j. d. 1978. The stratigraphy of new sections in the Ludlow Series of the type area, Ludlow, Salop, England. Rep. Inst. geol. Sci. 78/30. williams, a., lockley, m. G. and hurst, j. m. 1981. Benthic palaeocommunities represented in the Lfairfach group and coeval Ordovician successions of Wales. Palaeontology, 24, 661-694. wood, e. M. r. 1900. The Lower Ludlow formation and its graptolite fauna. Q. Jl geol. Soc. Lond. 56, 415- 492, pis. 25 and 26. woodcock, n. h. 1976a. Ludlow Series slumps and turbidites and the form of the Montgomery Trough, Powys, Wales. Proc. Geol. Ass. 87, 169-182. - 19766. Structural style in slump sheets: Ludlow Series, Powys, Wales. J. geol. Soc. 132, 399 415. - 1984. Early Palaeozoic sedimentation and tectonics, Wales. Proc. Geol. Ass. 95, 323-335. worsley, d. and broadhurst, f. w. 1975. An environmental study of Silurian atrypid communities from southern Norway. Lethaia , 8, 271 286. yonge, c. m. 1962. On the primitive significance of the byssus in the Bivalvia and its effects in evolution. J. mar. biol. Ass. UK, 42, 113 125. ziegler, a. m. 1965. Silurian marine communities and their environmental significance. Nature, Lond. 207, 270 272. - 1970. Geosynclinal development of the British Isles during the Silurian Period. J. Geol. 78, 445-479. boucot, a. j. and sheldon, r. p. 1966. Silurian pcntameroid brachiopods preserved in position of growth. J. Paleont. 40, 1032-1036, pis. 121 and 122. — cocks, l. r. m. and bambach, r. k. 1968. The composition and structure of Lower Silurian marine communities. Lethaia, 1, 1 -27. LESLEY CHERNS Department of Earth Sciences University College of Swansea Typescript received 26 Lebruary 1987 Singleton Park Revised typescript received 21 July 1987 Swansea SA2 8PP, UK MORPHOLOGY AND PHYLOGENETIC SIGNIFICANCE OF THE ANGIOS PERM PLATANITES HEBRIDICUS FROM THE PALAEOCENE OF SCOTLAND by P. R. CRANE, S. R. MANCHESTER and D. L. DILCHER Abstract. Fossil platanoid leaves from the Palaeocene of Mull, north-west Scotland, are assigned to Platanites hebridicus Forbes. The leaves closely resemble those of extant Platanaceae, and differ only in being pinnately compound. Each leaf consists of a shallowly three-lobed terminal leaflet and two smaller asymmetric lateral leaflets. Reproductive structures associated with the fossil foliage are also similar to those of extant Platanaceae, and the only unequivocal differences are the ellipsoidal achene shape and the smaller number of achenes per fruiting head in the Palaeocene material. The combined information from leaves and reproductive structures establishes the "P. hebridicus plant’ as one of the most completely understood fossil Platanaceae. The recognition of pinnately compound leaves in this critical angiosperm group has important implications for understanding the early divergence of major clades within the dicotyledons. During the mid- Cretaceous, simple, palmate platanoid leaves and pinnately compound Sapindopsis leaves exhibit partially intergrading patterns of venation and cuticular structure, and this has been used to suggest a close phylogenetic relationship between these early primitive representatives of the extant dicotyledonous subclasses Hamamelidae and Rosidae. The occurrence of extinct Platanaceae with compound leaves adds to the similarities between platanoid and Sapindopsis foliage, strengthens the proposed close relationship between the Platanaceae and Rosidae, and highlights the need to clarify relationships within the mid-Cretaceous platanoid -Sapindopsis complex. The extant family Platanaceae includes a single genus, Platanus , which consists of approximately nine species of temperate to tropical dicotyledonous trees (Li 1957; Ernst 1963) divided between two subgenera (Leroy 1982). Subgenus Castaneophyllum contains a single species, P. kerrii (Gagnepain 1939), that has unlobed elliptical leaves and occurs today only in tropical south-east Asia (Buzek el al. 1967, 1976; Baas 1969; Kvacek 1970; Leroy 1982). P. kerrii is clearly distinct from all other species in the genus (Hsiao 1972, 1973). Subgenus Platanus contains approximately eight species with simple palmately lobed leaves, and has its centre of diversity in south-western North America, and Mexico. Two broadly distributed but disjunct species in subgenus Platanus are the familiar sycamore of eastern North America (P. occidentalism and the plane tree of the eastern Mediterranean (P. orientalis). Fossil leaves, inflorescences and infructescences very similar to those of extant Platanaceae first appear in the fossil record during the Albian (Lower Cretaceous) (Doyle and Hickey 1976; Hickey and Doyle 1977; Dilcher 1979; Crane et al. 1986; Upchurch and Wolfe 1987; Crane, in press a\ Friis et al. in press; Schwarzwalder and Dilcher, in press). Platanus- like plants are therefore known very early in the initial diversification of non-magnoliid (’higher’) dicotyledons, which are characterized by tricolpate or tricolpate-derived pollen (subclasses Asteridae, Caryophyllidae, Dilleniidae, Hamamelidae, Ranunculidae, Rosidae, sensu Takhtajan, 1980). Subsequently platana- ceous leaves are common constituents of middle to high latitude fossil floras in the late Cretaceous and Tertiary of the Northern Hemisphere (Crane 1987; Manchester 1986; Upchurch and Wolfe 1987). The variety of foliar morphology among fossil Platanaceae strongly suggests that extant Platanus reflects only a small proportion of the total diversity of the platanaceous clade, and this is supported by the morphological variety among platanoid inflorescences and infructescences [Palaeontology, Vol. 31, Part 2, 1988, pp. 503 517.| © The Palaeontological Association 504 PALAEONTOLOGY, VOLUME 31 recently reported from the Cretaceous and early Tertiary (Friis 1984, 1985b; Crane et al. 1986; Manchester 1986; Crane, in press a; Friis et al., in press). In this paper we describe unusual compound platanaceous leaves ( Platanites hebridicus Forbes) and associated reproductive structures from the Palaeocene of Mull, north-west Scotland. We suggest that these different isolated organs were produced by a single extinct species. The combined information establishes P. hebridicus as one of the more completely understood fossil Platanaceae and permits a detailed assessment of its similarities to Recent species. MATERIAL AND METHODS Plant fossils from the interbasaltic sediments of Mull were first described by Forbes (1851). They have since been the focus of several palaeobotanical studies, and additional plant fossil localities have been discovered in other parts of north-west Scotland (Gardner and Ettingshausen 1879 1882; Gardner 1883 1886, 1887; Edwards 1923; Seward and Holltum 1924; Johnson 1933, 1934, 1935, 1936, 1937; Seward 1939; Crane, in press b). The Mull flora is dominated by angiosperm leaves: fruits, seeds, and other reproductive structures are rare in the collections currently available. All of the specimens considered in this study are from the classic interbasaltic leaf beds’ (Argyll 1851; Gardner 1887) on the Ardtun peninsula near Bunessan in south-western Mull (National Grid Reference NM 377247). At least two different localities and two different stratigraphic levels on the Ardtun peninsula have yielded Platanites leaves (Gardner 1887), but the exact provenance of individual specimens is unknown. The maximum igneous activity in north-west Scotland is dated as early Palaeocene, between 66 58 million years before present (Curry et al. 1978). Palynological assemblages from Mull and other localities in this area have received considerable attention (Simpson 1937, 1961; Martin 1968; Phillips 1974; Srivastava 1975) and the most recent palynological assessments (Curry et al. 1978), which suggest an early Palaeocene age, are in broad agreement with the radiometric data. We have examined specimens from Ardtun in the British Geological Survey, Edinburgh and Keyworth; the Cockburn Museum, University of Edinburgh; the Royal Scottish Museum, Edinburgh; the Hunterian Museum, Glasgow; the British Museum (Natural History), London; the City Museum and Art Gallery, Glasgow. The most informative specimens, and all of the material cited and illustrated in this paper have either a ‘V’ prefix and are in the collections of the Department of Palaeontology, British Museum (Natural History), or a BGS prefix and are in the collections of the British Geological Survey, Keyworth, Nottinghamshire. Carbonaceous fragments from the staminate inflorescence were cleaned in hydrofluoric acid and macerated in concentrated nitric acid followed by ammonia to yield pollen for light and scanning electron microscopy. Terminology of leaf architectural features follows Hickey (1973). Annotations of the synonymy list follow the recommendations of Matthews (1973). SYSTEMATIC PALAEONTOLOGY Division magnoliophyta Cronquist, Takhtajan and Zimmermann, 1966 Class magnoliopsida Cronquist, Takhtajan and Zimmermann, 1966 Subclass hamamelidae Takhtajan, 1966 Family platanaceae Dumortier, 1829 FOLIAGE Genus platanites Forbes, 1851 Type species. Platanites hebridicus Forbes, 1851. Generic diagnosis. Leaves compound with a trilobed terminal leaflet and a pair of ovate, asymmetrical lateral leaflets. Terminal leaflet with three palinactinodromous primary veins; lateral leaflets with a single primary vein. Secondary venation of both terminal and lateral leaflets pinnate and craspedodromous with the veins ending in teeth. Tertiary venation percurrent, at right angles to the secondary veins. CRANE ET AL.\ PALAEOCENE PLATANOI D ANGIOSPERM 505 Remarks. No formal diagnosis was provided by Forbes and the specimens figured as P. hebridicus (Forbes 1851, pi. 3, fig. 5; pi. 4, fig. 1) are fragments of two terminal leaflets, and do not show the compound morphology of the complete leaf. However, there is no evidence of more than one taxon of Platanus-Uke leaves at the Ardtun locality, and all specimens in which the leaf rachis is complete have either attached lateral leaflets (text-figs, ia, b, f and 2B, c) or distinct leaflet scars (text-fig. id, h). This feature is the major character separating leaves of Platanites from those of extant Platanus , and it is therefore incorporated in the generic diagnosis. Our revised concept of Platanites will not accommodate simple platanoid leaves that differ in other respects from those of extant Platanus. Fossil wood assigned to this genus (Mathiesen 1932) must also be excluded. Several fossil angiosperm leaves illustrated in the literature, previously assigned to other genera, now fall within our circumscription of Platanites. These include some of the specimens previously assigned to Negundo fremontensis Berry (Berry 1931, pi. 11, figs. 13), N. decurrens Lesquereux (Knowlton 1930, pi. 45, fig. 10), and Platanus guillelmae Goppert (Knowlton 1930, pi. 33, fig. 2) from the uppermost Cretaceous or Palaeocene Denver Formation of Colorado, and Cissus marginata (Lesquereux) Brown (Brown 1962, pi. 53, fig. 4; pi. 54, figs. 3 and 4) from the Palaeo- cene, Fort Union Formation of Montana and the Middle Eocene of northwestern Wyoming (MacGinitie 1974). In the absence of a detailed investigation of the original material of these species, the necessary new combinations are not formally proposed in this paper. Platanites hebridicus Forbes, 1851 Text-figs. 1a h, 2a-c, 4a v * V 1851 Platanites hebridicus Forbes, p. 1 03, pi. 3, fig. 5; pi. 4, fig. I (also pi. 4, fig. 2, and possibly pi. 3, fig. 1, both listed as ‘affinities doubtful’). 1856 Platanites hebridicus Forbes; De La Harpe, p. 136. 1886 Platanites aceroides Gardner, p. 104. 1887 Platanites hebridicus Forbes; Gardner, pp. 289, 290, 296, pi. 13, figs. 4, 12 1 4r/ . 1924 Platanus hebridica (Forbes); Seward and Holltum, p. 83, fig. 14. 1937 Hamamelis suborbiculata Johnson, p. 317, pi. 20, fig. 4 (lateral leaflet). Specific diagnosis. Terminal leaflet broad, length to width ratio approximately 1:1. Apex acute, base typically broadly cuneate. Sinuses between lobes shallow. Lateral primary veins of terminal leaflet diverging alternately above the base of the lamina at acute angles. Secondary veins diverging from the primary veins at angles of 40-65° and terminating in well-developed teeth. Other teeth supplied by strong tertiary veins that arise abmedially from the supra-adjacent secondary vein. Lateral leaflets smaller than the terminal leaflet, subsessile and subopposite on the rachis; attached well above the base of the rachis and well below the terminal leaflet. Lamina of lateral leaflets prominently expanded along the side closest to the base of the leaf, the expansion sometimes forming a discrete lobe supplied by a strongly developed basal secondary vein. Teeth of both terminal and lateral leaflets simple; usually asymmetric with the upper margin concave and the lower margin convex. Lectotype. BGS GSM 76599 (Forbes 1851, pi. 3, fig. 5). Syntypes. BGS GSM 77352 (Forbes 1851, pi. 4, fig. 1), 77353 (Forbes 1851, pi. 4, fig. 2). Other material. V.2479, 24977, V. 25031. V.25034, V. 25035, V.25036, V. 25039 25041, V. 25061. V. 25064, V. 25065, V. 25068, V.25089, V.25206, V.25237. Locality. Ardtun peninsula, 3 km north of Bunessan, Mull, Scotland. National Grid Reference NM 377247. Stratigraphy. ‘Ardtun Leaf Beds’, probably early Palaeocene. Description. At least six specimens illustrate that the leaves of Platanites hebridicus are compound (text-figs. 1a, b, f and 2b, c) and this unusual feature was first noted by Seward and Edwards (unpublished manuscript BM(NH) ). The two lateral leaflets are attached sub-oppositely to the rachis, 2 to 6 cm below the terminal leaflet and 3-5 7-5 cm above the base of the rachis. The rachis may be up to about 10 cm long. 506 PALAEONTOLOGY, VOLUME 31 text-fig. 1 . Platcmites hebridicus Forbes, leaf morphology, a, compound leaf showing large terminal leaflet, long rachis, and a single lateral leaflet, V. 25040, x 0-35. b. compound leaf showing large terminal leaflet, and partially superimposed lobed lateral leaflet, V. 25031, x0-3. c, detail of intercostal venation of terminal leaflet in h, showing percurrent tertiary veins and quaternary venation, x 0-85. D, detail of rachis of terminal leaflet in h showing the subopposite scars of two lateral leaflets, x 1 . e, detail of base of terminal leaflet in h showing venation of basal margin; note strongly palinactinodromous primary veins, x 0-85. f, leaf rachis showing the base of a terminal leaflet and two complete lateral leaflets; note maximum development of lamina directed toward the leaf base, V. 25061, x 0-5. G, detail of lateral leaflet margin in f showing venation and marginal teeth, x I. H, terminal leaflet of a compound leaf; note two scars toward the base of the leaf rachis, V. 25039, x 0-35. CRANE ET AL.: PALAEOCENE PLATANOID ANGIOSPERM 507 Terminal leaflets are slightly asymmetric, broadly trilobed, and are about 18 24 cm long and 16 28 cm wide based on estimates from the most complete specimens. Johnson (1933) mentions a specimen 37 cm long. Often terminal leaflets are wider than they are long. They have palinactinodromous venation (text-figs. 1a, b, h and 2a), with the lateral primary veins diverging alternately from the midvein at angles of 40 65° ( text- fig. Ie). The lowermost primary vein may diverge very close to the base of the lamina (text-figs. 1a and 2a) or several millimetres above the base of the leaflet (text-fig. Ie). The points of divergence of the secondary veins are usually several millimetres apart. The lobes of the terminal leaflet are broad and sinuses shallow such that the distance from the leaflet base to the sinus is about two-thirds of the distance from the base to the apex. None of the specimens clearly show the leaf apex although the form of the lamina indicates that it is acute (text-fig. 2a). The base of the terminal leaflet is broadly cuneate, frequently slightly asymmetric, and forms an angle of approximately 75 1 40 A The lateral leaflets are more or less sessile or with a short petiolule up to 5 mm long. In some specimens they are detached, leaving prominent scars on the rachis (text-fig. Id, h). Isolated lateral leaflets (Forbes 1851, pi. 4, fig. 2; Johnson 1937, pi. 20, fig. 4; Phillips 1974, pi. 1, fig. 1) are easily recognized by then- asymmetry and pinnate venation (text-fig. 2b, c). They are broadly ovate but distinctly asymmetric with the maximum development of the lamina on the side closest to the base of the leaf. Occasionally the expanded portion of the lamina forms a weakly developed lobe supplied by a prominent secondary vein (text-fig. 2c). The sinus formed by the lobe is very shallow. Lateral leaflets are about 40 120 mm long and 35-90 mm wide. The apex of the leaflet is acute and the base obtuse to truncate. In both terminal and lateral leaflets there are four to eight pairs of secondary veins which arise alternately from the midvein at angles of about 40-65°. Secondary veins are craspedodromous, slightly admedially curved, and terminate in prominent teeth. Tertiary veins are percurrent, typically straight, occasionally branched (text-fig. 1c) and arise from the secondary veins at approximately 90° with intervals of 5-10 mm. Higher order venation is orthogonal with quaternary veins arising approximately at right angles and typically delimiting two rows of more or less isodiametric areolae between each pair of tertiary veins. Teeth are simple, obtuse or occasionally glandular, with a concave upper flank and convex lower flank (text-fig. 1g). Each tooth is supplied either directly by a secondary vein or by an abmedial branch from the supra-adjacent secondary vein. In both terminal and lateral leaflets, teeth toward the base of the lamina are supplied by strongly developed tertiary veins derived admedially from the lowermost secondary (text-fig. If). The vein is positioned medially in each tooth. Cuticular details have not been obtained from any of the specimens. ASSOCIATED REPRODUCTIVE STRUCTURES Infructescences and fruits Description. An isolated fruiting head from Mull was illustrated by Gardner (1887, pi. 13, fig. 12). Heads are 10 17 mm in diameter, but there are no specimens that show whether they were originally sessile along an infructescence axis or other details of how they were borne. Compressed heads show approximately forty curved styles protruding from the periphery (text-fig. 2d, e), and we estimate that each head probably contained approximately 200 300 achencs. Floral details are not preserved, and the number of carpels per flower cannot be determined. Isolated achenes are typically 6-7 mm long and consist of a slender elliptical body and an apical elongated, curved persistent style (text-fig. 2f). The body is about 4 mm long and 2 mm wide, and the style typically I 2 mm long. No hairs have been observed associated either with the infructescences or the dispersed achenes, although it is possible that they were present and not preserved. Material. V.25052, V.25057, V25058, V.62186. Staminate Inflorescences Description. Staminate inflorescences are more common in the Mull assemblage than infructescences (Gardner 1887, p. 290) and several specimens have been illustrated previously (Gardner 1887, pi. 13, figs. II, 13, 14, 14a, 15; Phillips 1974, pi. 1, fig. 3). They consist of spherical heads that are borne on short stalks up to 3 mm long, distributed along elongated inflorescence axes. The heads are typically 5 mm apart and the most complete specimen shows the positions of at least six (possibly nine) heads borne along a single inflorescence axis (text-fig. 3a). Heads are 6 1 1 mm in diameter and the surface consists of polygonal areas approximately 0-5 mm in diameter formed by the distally expanded connective of the stamens (text-fig. 2d). We estimate that each head contained about 300 stamens. Floral details, including the number and arrangement of stamens per flower, are unknown, although fragmentary staminate heads (text-fig. 3c) show short persistent PALAEONTOLOGY, VOLUME 31 text-fig. 2. Platanites hebridicus Forbes, leaf morphology and associated infructescences and fruits, a, terminal leaflet, V. 25036, x0-5. b, leaf rachis showing attached asymmetrical lateral leaflet, V.2479, x 1. c, leaf rachis showing asymmetrical lateral leaflet with two lobes separated by a shallow sinus, V. 25064, x 1. d, infructescence showing numerous projecting styles, V. 25052, x 2-5. e, infructescence showing numerous projecting styles, V.62186, x2-5. F, four dispersed achenes with curved elongated styles at the apex, V. 25058, x 4. CRANE ET AL.: PALAEOGENE PLATANOID ANGIOSPERM 509 text-fig. 3. Staminate inflorescences and pollen associated with Platanites hebridicus Forbes. A, staminate inflorescence showing a single staminate head and the attachment points of five (possibly eight) other staminate heads, V. 25054, x 2. b, staminate inflorescence showing three attached staminate heads, V.25051, x 2. c, detail of staminate head showing two attached anthers; note the distally expanded peltate connective, short filaments, and short persistent perianth parts, V.25051, xlO. d, staminate inflorescence with four attached heads, V. 25055, x 3. E, contents of a single pollen sac isolated from staminate head in d, x 100. f, light micrograph of pollen isolated from staminate head in d, x 750. G, SEM of several pollen grains isolated from staminate head in d, showing equatorial and polar views, x 1000. H, SEM of pollen grain isolated from staminate head in D, equatorial view showing colpus and microreticulate-rugulate tectum in mesocolpial areas, x 2500. perianth parts. Stamens were shed individually (Gardner 1887, pi. 13, fig. 14) and consist of a short filament, elongate anthers typically 0-8 mm long and a short capitate connective (text-fig. 3c). Pollen isolated from organically preserved staminate heads is prolate, 16-20 pm in polar length, tricolpate, and very finely microreticulate (text-fig. 3f h). Exine sculpture resembles that of the dispersed pollen species Tricolpites dubhensis S. K. Srivastava isolated from the Shiaba lignite on Mull which is approximately contemporaneous 510 PALAEONTOLOGY, VOLUME 31 text-fig. 4. Reconstructions of compound platanoid leaves. A, Platanites hebridicus Forbes; based on specimens illustrated in text-figs. 1 and 2. b, ‘ Cissus ’ marginata (Lesquereux) Brown; based on Brown (1962, pi. 53, fig. 4; pi. 54, fig. 4). with the Ardtun leaf beds (Simpson 1961; Srivastava 1975). Both the pollen isolated from the fossil staminate inflorescences and the holotype of T. dubhensis (Srivastava 1975, pi. 12, figs. 1 and 2) have muri that are approximately triangular in cross-section as in pollen of extant Platanus (e.g. Hesse 1978; Zavada and Dilcher 1986). The grains described here differ from T. dubhensis only in having a smaller polar length (16-20 gm vs. 20-30 pm). Material. V. 14850, V.25003, V.25051, V. 25053-25056, V.25060, V.25221. DISCUSSION We have no direct evidence of attachment that conclusively links Platanites hebridicus with the associated infructescences and staminate infructescences, and we therefore do not formally assign the leaves and reproductive structures to the same species. However, each dispersed organ can be assigned to the Platanaceae on independent evidence, and in all of the Mull material examined we found no other plant fossils referable to this family. P. hebridicus is one of the more common leaves in the Ardtun flora, and together with the absence of other platanaceous fossils this makes it likely that the fossil leaves, inflorescences, and infructescences were produced by a single species. If this is accepted, then the combined information from different organs permits a more useful comparison of the Mull species with other fossil and extant Platanaceae. Comparison with extant Platanaceae In all features of morphology and venation the terminal leaflets of Platanites hebridicus closely resemble the simple leaves of extant Platanus subgenus Platanus , such as those of P. occidentalis. Lateral leaflets do not occur in any extant Platanus species (Jaennicke 1899; Brown 1962; Depape and Brice 1966), but elaborately developed stipules may be present both in extant and fossil Platanus (Crane 1981; Leroy 1982; Schwarzwalder and Dilcher in press). These stipules have CRANE ET AL.\ PALAEOCENE PLATANOID ANGIOSPERM 511 been considered homologous to the basilaminar expansions seen in the leaves of some extant and fossil Platanus species (Ward 1888) but could also be interpreted as homologous to the leaflets of Platanites and similar taxa. The arrangement of staminate flowers and the morphology of stamens is typical of that seen in extant Platanus. The stamens are aggregated into more or less spherical heads, and have a characteristic short, domed apical extension of the connective (Boothroyd 1930). The tricolpate microreticulate pollen is also similar to that in extant Platanus. Pollen size (16-20 /mi) is close to the range typical of subgenus Castaneophyllum (16-18 /an) and smaller than is typical of subgenus Platanus (20-25 pm) (Ludlow-Wiechers and Nieto 1982; Manchester 1986; Zavada and Dilcher 1986). The arrangement of sessile staminate heads along the fossil inflorescences is also similar to that in extant Platanus. The number of staminate heads per inflorescence is intermediate between the one to five heads seen in subgenus Platanus and the numerous (> 20) heads seen in subgenus Castaneophyllum. The diameter of the fruiting heads is smaller than is typical of subgenus Platanus and more like that in subgenus Castaneophyllum. Each fruiting head is much less compact than in both extant subgenera. Details of floral structure are not preserved in the Mull material, but there is no evidence in the compressed heads that the fruits are clustered into discrete floral units. In some fossil species (Friis 1985 b, Manchester 1986; Friis el al. in press) there are prominent perianth parts and compressed fruiting heads clearly show discrete clusters of fruits. By comparison with extant Platanus , the fruits in the fossil material are interpreted as achenes. Fruit shape differs from that of all extant species in being narrowly ellipsoidal rather than obovoid. We cannot determine whether the apparent absence of hairs on the fossil fruits is original or the result of poor preservation. Abundant hairs typically develop from the base of the fruit wall in extant Platanus. However, well-preserved fossil material indicates that these hairs were absent in some extinct taxa (Manchester 1986). Comparison with other fossil Platanaceae The recognition that Platanites hebridicus has compound leaves clearly invalidates the early comparisons made by Heer (1856) with taxa with simple leaves ( Platanus aceroides Gbppert, P. guillelmae Goppert). However, although this is the first unequivocal demonstration of pinnately compound leaves in Platanaceae, specimens of five other platanoid species from the uppermost Cretaceous or early Tertiary of North America have been illustrated previously that resemble Platanites hebridicus in this respect. A single specimen of Platanus appendiculata Fesquereux from the early Eocene (MacGinitie, pers. comm.', Wolfe, pers. comm.) of the Sierra Nevada (Fesquereux 1878, pi. 3, fig. 3) shows two lateral laminae attached to the leaf rachis, but because these occur at the base of the leaf rachis or petiole, it is uncertain whether they are lateral leaflets or a pair of basal stipules. MacGinitie (1941) collected no further specimens that showed this feature and considered it to be of little systematic significance in his revision of the Sierra Nevada flora. Negundo fremontensis Berry (1931, pi. 11, figs. 1-3) from the Middle Eocene of the Wind River Basin, Wyoming is only known from isolated terminal and lateral leaflets but Berry’s reconstruction (1931, text-fig. 6) based on field association is closely similar to P. hebridicus. In particular the shape of the lateral and terminal leaflets is almost identical to that in P. hebridicus, and re- examination of the original material from Wyoming may demonstrate that lN\ fremontensis and P. hebridicus are conspecific. “N\ fremontensis was assigned to Aleurites (Euphorbiaceae) by MacGinitie (1974). The holotype, and only specimen, of N. decurrens Lesquereux (1889, p. 54), illustrated by Knowlton (1930, pi. 45, fig. 10; Museum of Comparative Zoology, Harvard, no. 1523) from the uppermost Cretaceous or earliest Palaeocene of the Denver Basin was regarded by Brown (1962) as conspecific with a specimen of Platanus guillelmae Goppert illustrated from the same flora (Knowlton 1930, pi. 33, fig. 2). The holotype shows part of a poorly preserved probable lateral leaflet adjacent to the rachis of a platanoid leaf. Brown (1962) placed both of these specimens and the single specimen assigned to the fossil taxon Winchellia triphylla Fesquereux (1893, pi. 8, fig. 1 ) 512 PALAEONTOLOGY, VOLUME 31 in the species Cissus marginata (Lesquereux) Brown. Brown illustrated three specimens of C. marginata from the Palaeocene Fort Union Formation of western North America that show the compound nature of this leaf (1962, pi. 53, fig. 4; pi. 54, figs. 3 and 4; see also Dorf 1942, pi. 17, fig. 4), and a similar specimen is reported from the Middle Eocene of north-western Wyoming by MacGinitie, who noted the similarities to leaves of extant Platanus (1974, pi. 14, fig. 1). Other compound platanoid leaves occur in the Palaeocene Ravenscrag Formation of southern Saskatchewan (Basinger pers. comm.). With the exception of ‘AT fremontensis (Berry 1931) all of these specimens differ from Platanites hebridicus in having a terminal leaflet with a more acute base that is frequently decurrent along the leaf rachis (text-fig. 4b). The specimens of ‘C. marginata ’ also have more symmetrical lateral leaflets than those of P. hebridicus. Although the reproductive structures of "C. marginata ’ are unknown, there may have been several species of Platanaceae with compound leaves during the early Tertiary. Although fossil Platanaceae are abundant in the mid-Cretaceous none of these taxa have so far been reconstructed in detail. Isolated mid-Cretaceous reproductive structures differ from those associated with P. hebridicus in having more numerous smaller heads borne along the infructescence axis, more prominent perianth parts, smaller pollen, and usually stamens with a more elongated apical extension of the connective (Dilcher 1979; Friis 19856; Crane et al. 1986; Friis, Crane and Pedersen, in press). The most completely reconstructed species of extinct Platanaceae is the plant which bore the leaves Macginitiea angustiloba (Lesquereux) Manchester (1986) from the Middle to late Eocene of Oregon. The leaves of M. angustiloba differ from those of P. hebridicus in being simple and having five to seven palmately arranged lobes. However, there are some similarities in the associated reproductive structures. Both species differ from extant Platanus in having ellipsoidal achenes, with a persistent elongated style, that apparently lacks a mass of dense hairs at the base. The ‘M. angustiloba plant’ and the "P. hebridicus plant’ differ in the aggregation of Macginitiea achenes ( Macginicarpa ) into clusters of five surrounded by prominent perianth parts, the tendency of Macginitiea stamens ( Macginistemon ) to adhere together in groups of five by hairs arising from an elongated apical extension of the connective (Manchester 1986) and the smaller pollen size in Macginistemon (11-16 /;m). In summary, P. hebridicus possesses a mosaic of characters that is unique among fossil and extant Platanaceae. Although some, probably derived, features, e.g., the shape of the expanded connective and pollen size, are indicative of a close relationship to extant Platanaceae, other characters such as ellipsoidal achenes, elongated inflorescences with several heads of flowers, may be generalized characters within the platanaceous clade. Evolution of the Platanaceae Fossil leaves with the architectural features of extant Platanaceae first appear in the mid-late Albian (Lower Cretaceous) and provide some of the earliest evidence of the dicotyledonous subclass Hamamelidae (Wolfe et al. 1975; Doyle and Hickey 1976; Hickey and Doyle 1977; Upchurch and Wolfe 1987; Crane, in press a). This early platanoid foliage exhibits considerable morphological variation and had previously been assigned to several extant genera in different families (e.g. Aralia, Sassafras , Sterculia , Lesquereux 1892; Berry 1902, 1903; Seward 1927; Schwarzwalder 1984, 1986; Schwarzwalder and Dilcher, in press). Fossil platanaceous wood is also known from the late Albian (Cedar Mountain Formation, Utah), although its structure is rather generalized and it does not show certain specialized features, such as simple perforation plates, that occur in the wood of the extant genus (Tidwell, pers. comm., and pers. observ.). In the late Cretaceous and early Tertiary platanoid foliage is diverse and widely distributed in fossil floras from the Northern Hemisphere (e.g. Ward 1888, 1890; Berry 1914; Depape and Brice 1966; MacGinitie 1969; Knappe and Ruffle 1975, Nemejc and Kvacek 1975; Walther 1985; Wolfe and Wehr 1987), and the abundant record of leaves is paralleled by numerous reports of fossil wood (e.g. Felix 1896; Prakash and Barghoorn 1961; Prakash et cd. 1971; Page 1968; Suss 1971, 1980; Brett 1972; Suss and Miiller-Stoll 1975, 1977; Wheeler et al. 1977; Mai and Walther 1978; Scott and Wheeler 1982; Manchester 1986). CRANE ET AL.: PALAEOCENE PLATANOID ANGIOSPERM 513 The fossil record of platanaceous reproductive structures is also extensive and is receiving increasing attention (Velenovsky 1889; Brown 1933; Krassilov 1973; Friis 1985a, 6; Crane et al. 1986; Knobloch and Mai 1986; Manchester 1986; Friis et al. in press). In the mid-Cretaceous, inflorescences superficially resembling those of extant Platanus kerrii are associated with early platanoid foliage and provide some information on the organization of flowers and inflorescences in the early representatives of the group (Lesquereux 1892; Hickey and Doyle 1977; Dilcher 1979). Further details on floral structure are provided by three-dimensionally preserved inflorescences and flowers from the mid-late Alhian of the Potomac Group, eastern North America that have been linked with platanoid foliage on the basis of field association and similarities in cuticular anatomy (Crane et al. 1986). Taken together, the information currently available from the mid- Cretaceous indicates that early platanoids resembled extant Platanaceae in having unisexual flowers clustered into globose heads, and were particularly like P. kerrii in possessing numerous sessile heads borne along a single inflorescence axis. These fossil plants however differed from the living taxa in the presence of five carpels per flower, the absence of hairs on the achenes, their well- developed perianth parts, and their small pollen. All of these features were apparently retained in some members of the platanaceous clade through the late Cretaceous and well into the early Tertiary (Friis 1984, 19856; Manchester 1986) and probably reflect significant differences in reproductive biology. In particular the size of in situ pollen from Cretaceous and some early Tertiary forms is smaller than is typical of wind pollinated plants (Wodehouse 1935; Crane 1986) and suggests insect pollination (Manchester 1986; Crane et al., 1986). The larger pollen and more reduced perianth of extant representatives, perhaps also Platanites hebridicus , may correlate with wind pollination (cf. Hesse 1978). During the mid-Cretaceous palinactinodromous palmately lobed platanoid leaves intergrade in leaf architecture, venation, and cuticular structure with pinnatifid or pinnately compound foliage assigned to Sapindopsis (Doyle and Hickey 1976; Hickey and Doyle 1977; Upchurch 1984; Crane, in press a). Furthermore, preliminary work on mid-Cretaceous reproductive structures suggests that the inflorescences, flowers and pollen of platanoid and Sapindopsis plants were similar (Crane et al. 1986). Sapindopsis foliage has been interpreted as early evidence of the subclass Rosidae and the morphological similarities between platanoid and Sapindopsis plants have been used to infer a sister-group relationship between this subclass and the Hamamelidae (Wolfe et al. 1975; Doyle and Hickey 1976; Hickey and Doyle 1977; Crane, in press a; see also Cronquist 1981). This hypothesis receives further support from the leaves described in this paper. P. hebridicus significantly expands the foliar diversity known to occur within the platanaceous clade, clearly demonstrates the occurrence of pinnately compound leaves in the Platanaceae, and raises the question as to whether compound or simple leaves are basic within the group. The apparent absence of pinnately compound platanoid foliage during the late Cretaceous suggests that the leaves of P. hebridicus are specialized within the platanaceous clade, although the possibility that the compound condition reflects retention of a primitive character cannot be excluded. Resolving this question will be critical to interpreting the early radiation of non-magnoliid dicotyledons and will involve clarification of relationships within the mid-Cretaceous platanoid -Sapindopsis complex. Acknowledgements. We thank P. Aspen, P. Brand, M. Crawley, A. Gunning, C. R. Hill, H. Ivimey-Cook, W. D. I. Rolfe, C. D. Waterston, and the Keeper of Palaeontology, British Museum (Natural History), for assistance in locating specimens and permission to examine the collections in their care. Clara Richardson drew the reconstructions of the Platanites leaves and R. Schwarzwalder prepared the pollen from fossil staminate inflorescences and provided helpful comments on the manuscript. We also thank P. K. Endress and E. M. Friis for many helpful discussions. This work was supported in part by National Science Foundation Grants BSR 8314592 to P. R. C„ 84-07841 to S. R. M., and DEB-79-10720 and BSR-85-16657 to D. L. D. 514 PALAEONTOLOGY, VOLUME 31 REFERENCES Argyll, duke of. 1851. On Tertiary leaf-beds in the Isle of Mull. Q. Jl geol. Soc. Lond. 7, 90-103. baas, p. 1969. Comparative anatomy of Platanus kerrii Gagnep. Bot. J. Linn. Soc. 62, 413-421. berry, e. w. 1902. Notes on Sassafras. Bot. Gaz. 34, 426-450. — 1903. Aralia in American paleobotany. Ibid. 36, 421-428. 1914. Notes on the geological history of Platanus. PI. World , 17, 1-8. — 1931. A flora of Green River age in the Wind River Basin of Wyoming. Prof. Pap. US geol. Surv. 165, 55-80. boothroyd, l. e. 1930. The morphology and anatomy of the inflorescence and flower of the Platanaceae. Am. J. Bot. 17, 678-693. brett, D. w. 1972. Fossil wood of Platanus from the British Eocene. Palaeontology , 15, 496-500. brown, r. w. 1933. Fossil plants from the Aspen Shale of southwestern Wyoming. Proc. US natn. Mus. 82 (12), 1-10. 1962. Paleocene flora of the Rocky Mountains and Great Plains. Prof. Pap. US geol. Surv. 375, 1 119. buzek, c., holy, f. and kvacek, z. 1967. Eine bemerkenswerte Art der Familie Platanaceae Lindl. (1836) in nordbohmischen Tertiar. Mber. dt. Akad. Wiss. Bert. 9, 203-215. — 1976. Tertiary flora from the Volcanogenic Series at Markvartice and Veselicko near Ceska Kamenice (Ceske stredohofi Mts.). Sb. geol. Ved, , Rada P , Paleont. 18, 69-132. crane, p. r. 1981. Studies on the flora of the Reading Beds (Upper Palaeocene). Ph.D. thesis (unpublished), University of Reading. 1986. Form and function in wind dispersed pollen. In blackmore, s. and ferguson, i. k. (eds.). Pollen and Spores: Form and Function , 179-202. Academic Press, London. — 1987. Vegetational consequences of the angiospcrm diversification. In friis, e. m., chaloner, w. g. and crane, p. R. (eds.). The origins of angiosperms and their biological consequences, 107-144. Cambridge University Press, Cambridge. — In press a. Paleobotanical evidence on the early radiation of nonmagnoliid dicotyledons. Plant Syst. Evol. In press b. Abelia-\\ke fruits from the Palaeogene of Scotland and North America. Tert. Res. — friis, E. m. and pedersen, k. r. 1986. Lower Cretaceous angiosperm flowers: fossil evidence on the early radiation of the dicotyledons. Science, Washington, 232, 852-854. cronquist, A. 1981. An integrated system of classification of flowering plants, 1262 pp. Columbia University Press, New York. — takhtajan, A. and zimmermann, w. 1966. On the higher taxa of Embryobionta. Taxon, 15, 129-134. CURRY, D., ADAMS, C. G., BOULTER, M. C., DILLEY, F. C., EAMES, F. E., FUNNELL, B. M. and WELLS, M. K. 1978. A correlation of Tertiary rocks in the British Isles. Spec. Rep. geol. Soc. Lond. 12, 1 72. de la harpe, p. 1856. Quelques mots sur fiore tertiaire de I’Angleterre. Bull. Soc. vaud. Sci. nat. 5, 123-143. depape, G. and brice, d. 1966. Platanes actuels et platanes fossiles. Revue gen. Bot. 73, 41-81. dilcher, d. l. 1979. Early angiosperm reproduction: an introductory report. Rev. Palaeobot. Palynol. 27, 291-328. dorf, e. 1942. Upper Cretaceous floras of the Rocky Mountain Region. Pubis Carnegie Instn, 508, 1 168. doyle, J. a. and hickey, l. j. 1976. Pollen and leaves from the mid-Cretaceous Potomac Group and their bearing on early angiosperm evolution. In beck, c. b. (ed.). Origin and early evolution of angiosperms , 139- 206. Columbia University Press, New York. dumortier, b. c. j. 1829. Analyse des families des p/antes, 104 pp. J. Casterman, Tournay. edwards, w. N. 1923. An Eocene microthyriaceous fungus from Mull, Scotland. Trans. Br. mycol. Soc. 8, 66 72. ernst, w. r. 1963. The genera of Hamamelidaceae and Platanaceae in the southeastern United States. J. Arnold Arbor. 44, 193-210. felix, j. 1896. Untersuchungen fiber fossile Holzer. V. Dt. geol. Ges. 48, 249-260. forbes, e, 1851. Note on the fossil leaves represented in plates II. III. IV. Q. Jl geol. Soc. Lond. 7, 103. friis, e. m. 1984. Platanaceous inflorescences from the late Cretaceous of Sweden and eastern North America. Abstracts of Contributed Papers and Poster Session, 14. Second International Organization of Paleobotany Conference, Edmonton, Alberta. CRANE ET AL.\ PALAEOCENE PLATANOID ANGIOSPERM 515 1 985a. Angiosperm fruits and seeds from the Middle Miocene of Jutland (Denmark). Biol. Skr. 24 (3), 1-165. 19856. Structure and function in late Cretaceous angiosperm flowers. Ibid. 25, 1 37. — crane, p. r. and pedersen, k. r. In press. Reproductive structures of Cretaceous Platanaceae. Ibid. gagnepain, F. 1939. Un genre nouveau de Butomacees et quelques especes nouvelle d’Indo-Chine. Bull. Soc. hot. Fr. 86, 300-303. Gardner, J. s. 1883 1886. A Monograph of the British Eocene Flora , //, Gymnospermae , 159 pp. Palaeonto- grapliical Society, London. 1887. On the leaf-beds and gravels of Ardtun, Carsaig, etc., in Mull. Q. Jl geol. Soc. Land. 43, 270 300. — and ettingshausen, c. von. 1879- 1882. A monograph of the British Eocene Flora , /, Filices , 86 pp. Palaeontographical Society, London. heer, o. 1856. Flora Tertiaria Helevetiae\ Die Tertiare Flora der Schweiz. Vol. 2, 110 pp. J. Wurster, Winterthur. hesse, m. 1978. Entwicklungsgeschichte und Ultrastruktur von Pollenkitt und Exine bei nahe verwandten entomophilen und anemophilcn Angiospcrmensippen: Ranunculaceae, Hamamelidaceae, Platanaceae und Fagaceae. Plant Syst. Evol. 130, 13 42. hickey, l. j. 1973. Classification of the architecture of dicotyledonous leaves. Am. ./. Bot. 60, 17-33. — and doyle, j. a. 1977. Early Cretaceous fossil evidence for angiosperm evolution. Bot. Rev. 43, 3-104. hsiao, j. y. 1972. Biochemical systematic and numerical taxonomic studies of the genus Platanus. Ph D. thesis (unpublished). University of Pennsylvania, Philadelphia. — 1973. A numerical taxonomic study of the genus Platanus based on morphological and phenolic characters. Am. J. Bot. 60, 678 684. jaennicke, F. 1899. Studien fiber die Gattung Platanus L. Nova Acta Abh. Kaiserl. Leap. -Carol. Deutschen Akad. Natur. 77, 113-227. Johnson, t. 1933. The occurrence of the genus Platanus in the Lough Neagh clays and other Tertiary deposits of the British Isles. Mem. Proc. Manchr. lit. phil. Soc. 77, 109 1 16. 1934. The leaf beds of Ardtun, Canna and Skye. Rep. Br. Ass. Admit Sci. 102, 388. — 1935. Quercus L. in the Tertiary Beds of Ireland and Scotland. Mem. Proc. Manchr. lit. phil. Soc. 79, 83-98. — 1936. A conifer new to the British Isles: Cunninghamia Squamata sp. n. Ibid. 80, 25-27. — 1937. Notes on the Tertiary flora of Scotland. Trans. Proc. bot. Soc. Eclinb. 32, 291 340. knappe, h. and ruffle, l. 1975. Beitrage zu den Platanaceen-Funden und einigen Hamamelidales der Oberkreide. Wiss. Z. Humboldt-Univ. Berk R. 24 (4), 487 492. knobloch, e. and mai, d. h. 1986. Monographic der Fruchte und Samen in der Kreide von Mitteleuropa. Rozpr. ustred. Ust. geol. 47, 1-219. knowlton, f. h. 1930. Flora of the Denver and associated formations of Colorado. Prof. Pap. US geol. Surv. 155, 1 142. krassilov, v. a. 1973. Upper Cretaceous staminate heads with pollen grains. Palaeontology , 16, 41 44. kvaCek, z. 1970. A new Platanus from the Bohemian Tertiary. Paldont. Abh. Abt. B. 3, 435 439. leroy, j. F. 1982. Origine et evolution du genre Platanus (Platanaceae). C. r. hebd. Seanc. Acad. Sci., Paris , 295, 251 254. lesquereux, l. 1878. Report on the fossil plants of the Auriferous Gravel deposits of the Sierra Nevada. Mem. Harv. Mus. Comp. Zook 6 (2), 1 62. — 1889. Specimens of fossil plants collected at Golden, Colorado, 1883, for the Museum of Comparative Zoology at Cambridge, Mass. Bulk Harv. Mus. Comp. Zook 16, 43-59. 1892. Flora of the Dakota Group. Monogr. US geol. Surv. 17, 1-400. 1893. The genus Winchellia. Am. Geol. 12, 209 213. li, H. L. 1957. The origin and history of the cultivated plane-trees. Morris Arbor. Bulk 8, 3 9, 26 31. ludlow-wiechers, b. and nieto, m. l. a. 1982. Catalogo Palinologico para la Flora de Veracruz. No. 9. Familia Platanaceae. Biotica , 7, 611 615. macginitie, h. d. 1941. A Middle Eocene flora from the central Sierra Nevada. Pubis Carnegie Instn , 534, 1-178. 1969. The Eocene Green River flora of northwestern Colorado and northeastern Utah. Univ. Calif. Pubis geol. Sci. 83, 1 203. 1974. An early Middle Eocene flora from the Yellowstone Absaroka volcanic province, northwestern Wind River Basin, Wyoming. Ibid. 108, I 103. 516 PALAEONTOLOGY, VOLUME 31 mai, d. h. and walther, h. 1978. Die Floren der Haselbacher Serie im Weisselster-Becken (Bezirk Leipzig, DDR) Abh. st. Mus. Miner. Geol. Dresd. 28, I -200. Manchester, s. R. 1986. Vegetative and reproductive morphology of an extinct plane tree (Platanaceae) from the Eocene of western North America. Bot. Gaz. 147, 200 226. martin, a. R. H. 1968. Aquilapollenites in the British Isles. Palaeontology , 11, 549 -553. mathiesen, F. J. 1932. Notes on some fossil plants from East Greenland. Meddr Gronland, 85, 1 62. Matthews, s. c. 1973. Notes on open nomenclature and synonymy lists. Palaeontology , 16, 713-719. nemejc, F. and kvacek, z. 1975. Senonian plant macrofossils from the region of Zliv and Hlubokd (near Ceske Budejovice) in South Bohemia , 83 pp., 24 pis. Universita Karlova, Prague. page, v. m. 1968. Angiosperm wood from the Upper Cretaceous of central California: Part II. Am. J. Bot. 55, 168-172. Phillips, L. 1974. Reworked Mesozoic spores in Tertiary leaf-beds on Mull, Scotland. Rev. Palaeobot. Palynol. 17, 221-232. prakash, u. and barghoorn, e. s. 1961. Miocene fossil woods from the Columbia basalts of central Washington. J. Arnold Arbor. 42, 165 195. brezinova, d. and buzek, c. 1971. Fossil woods from the Doupovske Flory and Ceske Stredohori Mountains in northern Bohemia. Palaeontographica Abt. B. 133, 103-138. schwarzw alder, r. n. 1984. Aspects of the evolution of the Platanaceae and its relation to the Hamamelidae. Am. J. Bot. 71(5, pt. 2), 111 (Abstract). 1986. Systematics and early evolution of the Platanaceae. Ph.D. thesis (unpublished), Indiana University. — and dilcher, d. l. In press. Systematics and early evolution of the Platanaceae. Ann. Mo. bot. Gdn. scott, r. a. and wheeler, e. a. 1982. Fossil woods from the Eocene Clarno Formation of Oregon. I.A.W.A. Bull. 3, 135-154. seward, a. c. 1927. The Cretaceous plant bearing rocks of western Greenland. Phil. Trans. R. Soc. B. 215, 57 175. 1939. The Western Isles through the mists of ages. Rep. Br. Ass. Admit Sci. 1, 11-29. and holltum, r. e. 1924. Tertiary plants from Mull. In bailey, e. b., clough, c. t., wright, w. b., richey, j. e. and wilson, g. v. (eds.). Tertiary and Post-Tertiary Geology of Mull, Loch Aline, and Oban. Mem. geol. Surv. UK , 67-90. HMSO, Edinburgh. simpson, J. b. 1937. Fossil pollen in Scottish Tertiary coals. Proc. R. Soc. Edinb. 56, 90-108. 1961. The Tertiary pollen flora of Mull and Ardnamurchan. Trans. R. Soc. Edinb. 64, 421 468. srivastava, s. k. 1975. Maastrichtian microspore assemblages from the interbasaltic lignites of Mull, Scotland. Palaeontographica Abt. B. 150, 125-156. suss, h. 1971. Uber das Vorkommen von Platanen und platanenahnlichen Holzresten in Miozan der Ungarischen V.P. Geologic, 20, 74-81. 1980. Ein Platenholz aus dem Untereozan der Greifswalder Oie, Platanoxylon cohenii (Schuster) comb, nov. SchrReihe geol. wiss. Berlin , 16, 401-416. -and muller-stoll, w. r. 1975. Zur Frage der Unterscheidung von Platanus- Arten nach dem Bau des Holzes. Feddes Reprium , 86, 57-70. 1977. Untersuchungen fiber fossile Platanenholzer. Ibid. 88, 1 62. takhtajan, a. 1966. Systema et phylogenia Magnoliophytorum , 61 1 pp. Akad. Nauk, Moscow. [In Russian.] - 1980. An outline of the classification of flowering plants. Bot. Rev. 46, 225-359. upchurch, G. r. 1984. Cuticle evolution in Early Cretaceous angiosperms from the Potomac group of Virginia and Maryland. Ann. Mo. bot. Gdn , 71, 522-550. -and wolfe, J. a. 1987. Mid-Cretaceous to Early Tertiary vegetation and climate: evidence from fossil leaves and woods. In friis, e. m., chaloner, w. g. and crane, p. r. (eds.). The origins of angiosperms and their biological consequences , 75-105. Cambridge University Press, Cambridge. velenovsky, j. 1889. Kvetena Ceskeno Cenomanu. Rozpr. math.-pfir. K. ceske Spot. Nauk , 7 (3.3), 1 75. walther, H. 1985. Das Vorkommen der Gattung Platanus L. im Tertiar des Weisselster-Beckens (Bezirk Leipzig, DDR) Hallesches Jb. Geowiss. 10, 9-19. ward, l. f. 1888. The paleontological history of the genus Platanus. Proc. US natn. Mus. 11, 39 42. 1890. Origin of the plane-trees. Am. Nat. 24, 797 810. wheeler, e., scott, r. a. and barghoorn, e. s. 1977. Fossil dicotyledonous woods from Yellowstone National Park. J. Arnold Arbor. 58, 280 302. wodehouse, r. p. 1935. Pollen grains, 574 pp. McGraw Hill, New York. wolfe, j. a. and wehr, w. c. 1987. Middle Eocene dicotyledonous plants from Republic, northeastern Washington. Bull. U.S. geol. Surv. 1597: 1 25. CRANE ET AL.: PALAEOCENE PL ATANOID ANGIOSPERM 517 doyle, J. a. and page, v. m. 1975. The bases of angiosperm phylogeny: Paleobotany. Ann. Mo. hot. Gdn , 62, 801-824. zavada, m. s. and dilcher, d. l. 1986. Comparative pollen morphology and its relationship to phylogeny of pollen in the Hamamelidae. Ibid. 73, 348-381. p. R. CRANE Department of Geology Field Museum of Natural History Roosevelt Road at Lake Shore Drive Chicago, Illinois 60605, USA Typescript received 6 March 1987 Revised typescript received 15 July 1987 S. R. MANCHESTER and D. L. DILCHER Departments of Biology and Geology Indiana University, Bloomington Indiana 47405, USA COLONY GROWTH PATTERN AND ASTOGENETIC GRADIENTS IN THE CRETACEOUS CHEILOSTOME BRYOZOAN HERPETOPORA by P. D. TAYLOR Abstract. The common Chalk anascan Herpetopora has runner-like encrusting colonies with uniserial branches. Following larval settlement, the ancestrula, described for the first time, budded two daughter zooids to initiate two first order colony branches which grew in opposite directions. Branches of higher orders were added by lateral zooidal budding from both sides of parent branches, usually at c. 80°. The ‘mature’ colony consisted of two conjugate sets of branches. Size/frequency distributions of zooidal length in H. laxata demonstrate the existence of two autozooidal polymorphs which differ in the length of their caudae. Independent astogenetic gradients of changing zooid size can be distinguished in each branch; in H. laxata these consist of an initial phase of progressively lengthening non-caudate polymorphs, followed by a threshold jump to a phase of caudate polymorphs, also of progressively increasing length. Each zooid in a colony could normally bud three potential daughters (one distal and two lateral). However, the frequency of buds actually formed declined with increasing branch order, and caudate autozooids generally budded more daughters than non-caudate autozooids in H. laxata. Intersections between branches had various possible outcomes; usually the growing branch terminated against the skeletal margin of the earlier branch, but sometimes growth deviated towards a pore window, probably by chemotropism, and occasionally branches were overgrown. Evidence of colony damage and repair includes ‘intramural’ budding, and normal and reverse polarity ‘extramural’ budding. Many colonies had complex histories involving mortality of zooids, fission, regrowth, and fusion. Functional interpretation of morphology suggests that growing colonies were proficient at exploring substratum space and seeking spatial refuges. They could withstand extensive damage and fragmentation, and had the capacity to repair damage and re-establish connections between the ramets formed by fragmentation. Among colonial organisms, bryozoans provide a good opportunity to study developmental patterns in fossil taxa (e.g. Anstey et al. 1976; Taylor and Furness 1978; Podell and Anstey 1979; Lidgard 1985). Development of the colony, termed astogeny to distinguish it from the ontogeny of a solitary organism or an individual zooid in a colony, is manifested in the sequence of budded zooids, their changing morphology during colony development and the way in which they are arranged in the growing colony. It is easier to unravel astogeny in encrusting, two-dimensional colonies than in more complex erect colonies in which parts formed during early growth are often hidden. Especially suitable for astogenetic study are encrusting colonies with runner-like uniserial branches. Flere the genealogy of the zooids and the temporal order of budding may be most obvious, as in the genus Herpetopora Lang, 1914. Herpetopora is an anascan cheilostome classified currently within the paraphyletic Suborder Malacostegina and Family Electridae (see Taylor 1987). Species can be extremely abundant in the Upper Cretaceous chalks of north-west Europe. Thomas and Larwood (1960) last revised Herpetopora and placed it in synonymy with Pyripora d’Orbigny. This synonymy was rejected by Voigt (1982) and is also not accepted here because Herpetopora zooids lack a pustulose cryptocyst and lateral pore chambers with septulae, have considerably narrower and often much longer caudae than Pyripora, and colonies never develop oligoserial branches. Thomas and Larwood described two similar species which occur commonly in the English Chalk: H. anglica Lang ranges from the late Turonian to early Campanian, whereas H. laxata (d’Orbigny), distinguished by the slightly broader zooidal opesia, is found mainly in the late Campanian and early Maastrichtian. As [Palaeontology, Vol. 31, Part 2, 1988, pp. 519-549, pis. 42-45| © The Palaeontological Association 520 PALAEONTOLOGY, VOLUME 31 text-fig. 1. Colony of Herpetopora laxata (d’Orbigny) encrusting a guard of Belemnitella. BGS Yc 2708, Campanian, Compton, Hampshire, x 1-4. Professor E. Voigt and Dr G. P. Larwood are together preparing a revision of Herpetopora , the systematics of the genus will not be further considered in this paper. Colonies of H. anglica and H. laxata can be found encrusting a variety of skeletal substrata, including bivalves, echinoid tests, and belemnite guards. H. anglica is particularly common on shells of Inoceramus where other bryozoans tend to be relatively rare. However, it is very easy to overlook colonies because of their narrow branches and poor visual contrast with the substratum. Colonies frequently cover large areas of substratum, but are always incomplete, either because their substratum is fragmented, or because not all of their zooids are preserved. Typical specimens lack zooids formed during early colony growth but preserve numerous disconnected chains of zooids which may have been derived from one or more colonies. Consequently, it is impossible to study astogeny by assembling a ‘growth series’ of colonies of increasing size, as is often done in growth studies of solitary animals and in some bryozoan species (Hakansson 1975). This paper aims to describe: the detailed growth pattern of Herpetopora colonies; their unusual astogenetic gradients; the hitherto unknown ancestrula and early astogeny of colonies; and morphological structures indicating extensive repair of damage in living colonies. These observations lead to a discussion of the factors controlling astogenetic gradients of zooid size, and aspects of colonial integration and functional morphology. MATERIAL AND METHODS The specimens used in this study are lodged in the collections of the British Museum (Natural History), abbreviated BM(NH), British Geological Survey, Keyworth (BGS), and Voigt Collection, Geologisch- Palaontologisches I nstitut und Museum, Universitat Hamburg (VH). A complete listing of the BM(NH) material is given by Thomas and Larwood (1960). Detailed analysis of astogenetic gradients was undertaken in a single colony of H. laxata (d'Orbigny): BGS Yc 2708; Campanian, zone of Gonioteuthis quadrata (presumably the ‘ Hagenowia Horizon’ at the base, C. J. Wood, pers. comm. 1986), ‘Southampton Waterworks: new pit’ (NGR SU 469236), Compton, Hampshire; collected by R. M. Brydone and presented by E. Brydone, 1943. This colony (text-fig. 1) encrusts a guard of the belemnite Belemnitella sp., and is exceptional in preserving the ancestrula and a large number of connected post-ancestrular zooids (though preservation of surface detail is poorer than in many other specimens). The principal growth axis of the colony is defined by the two first order branches which are orientated almost parallel to the length of the belemnite guard. The colony was drawn (text-fig. 2) using a drawing tube attached to a Wild M7 binocular microscope, rotating the belemnite guard about its long axis to enable inclusion of zooids around the full circumference; however, some peripheral zooids distant from the ancestrula and difficult to relate to the main mass of the colony were omitted. The resulting ‘map’ of the TAYLOR: GROWTH IN HERPETOPORA 521 text-fig. 2. Drawing of part of Herpetopora laxata colony BGS Yc 2708. The first order branches of the colony are almost vertically orientated, and the caudate autozooids have been stippled. 522 PALAEONTOLOGY, VOLUME 31 colony included over 300 zooids, each of which was given a serial number. An eyepiece micrometer graticule was used to measure lengths and other dimensions of the numbered zooids. Angles between branches of the colony were measured to the nearest 5° using an eyepiece protractor graticule; slight curvature of zooids and the consequent problem in defining exactly their long axes precluded a more precise determination. Observations of branch intersections and additional morphological features in BGS Yc 2708 and other specimens were initially made with the optical microscope but many specimens were then studied in more detail using an ISI 60A scanning electron microscope equipped to accommodate large, uncoated specimens (Taylor 1986c). It should be noted that all scanning electron micrographs in this paper are back-scattered electron images rather than the more conventional secondary electron images. ZOOIDAL POLYMORPHISM Autozooids The feeding zooids of a bryozoan colony are termed autozooids. In fossil bryozoans, autozooids can often be recognized by comparison with closely related Recent species. Though not necessarily the commonest polymorph, autozooids are usually the polymorph to occupy the largest area of the living surface of the colony. The autozooids of Herpetopora are identifiable by analogy with Recent Pyripora and are the most numerous polymorph. They have a pyriform outline shape, dilated distally in the region of the opesia (PI. 42, fig. 2). Unlike Pyripora , however, the opesia is not bordered by a pustulose cryptocyst. Closure plates occlude opesiae of some autozooids (text- fig. 3; PI. 45, fig. 6), and the occurrence of impressions of the operculum on the closure plates substantiates the autozooidal nature of these zooids. So-called 'regenerations’, indicating reparative budding (see p. 540), often occur as a concentric series of mural rims within the opesiae. The proximal part of the autozooid is a slender cauda, sometimes slightly curved and of very variable length; the extreme range of autozooidal length within colonies of Herpetopora is due almost entirely to variability in the length of the caudae. As discussed below (p. 536), the length of successive autozooids increases distally along each branch of the colony. Histograms (text-fig. 4) of autozooidal length within colonies of H. anglica and H. laxata reveal a previously overlooked difference between the two species. In H. anglica , the size distribution is positively skewed but near normal, with modal and mean values of autozooidal length between 0-8 and TO mm. Although autozooidal length appears to increase continuously along branches of this species (see p. 539), attaining a maximum recorded value of 6 mm in BM(NH) D.8213, branches with more than four zooidal generations are not often preserved and very long autozooids are comparatively rare. Autozooidal size distribution in H. laxata is distinctly bimodal. The modal value of the first peak (0-8 TO mm) corresponds with that of H. anglica. The second, smaller peak has a modal value of 2-0-2-4 mm and is positively skewed. Very few autozooids occur in the length range of T4-T6 mm between the two peaks. Bimodality implies autozooidal polymorphism in H. laxata and permits the distinction of two types of autozooids: «, ‘non-caudate autozooids’ less than T6 mm in length (PI. 42, fig. 2); 6, ‘caudate autozooids’ more than 1 -6 mm in length and with relatively long caudae (PI. 42, fig. 3). EXPLANATION OF PLATE 42 Figs. 1 4. Herpetopora laxata (d’Orbigny); except for 1, all are back-scattered SEM images. I and 3, BGS Yc 2708, Campanian (quadrata Zone), Compton, Hampshire. 1, proximal part of colony with narrow, runner-like branches encrusting a belemnite guard (circular white patches are abraded sheet-like bryozoans), x2-8. 3, caudate autozooid, x42. 2, BM(NH) D. 42375, branch (originating just beneath centre of right margin and growing WNW) consisting of two non-caudate autozooids followed by a caudate autozooid, Campanian (mucronata Zone), Webster’s Pit, Norwich, Norfolk, x22. 4, BM(NH) D. 42361, damaged zooid (with two lateral buds) followed by a caudate kenozooid (with incomplete lateral buds), Campanian (mucronata Zone), Thorpe St Andrew, Norfolk, x 32. PLATE 42 TAYLOR, Herpetopora 524 PALAEONTOLOGY, VOLUME 31 3rd order text-fig. 3. Diagram of a hypothetical colony of Herpetopora laxata showing arrangement of branches and other morphological features. Abbreviations: a, ancestrula; bi, branch intersection (type 1 ) entailing termination of the growing branch; cbz, caudally budded zooid; cz, caudate autozooid; ic, intramural cauda; ip. cauda growing into pore window (branch intersection type 3); kz, kenozooid (opcsiate); pz, partially formed zooid; zc, autozooid with closure; zr, autozooid with intramural reparative bud; zrc, autozooid with intramural reparative bud which has a closure. Branches of orders 1-3 are indicated. As both types of polymorph may be found with closure plates bearing opercular impressions of similar size, both are thought to have been autozooids. Other aspects of their skeletal morphology appear identical, although the distally dilated parts of caudate autozooids are often narrower. In addition, the frequency of closure plates is greater in caudate autozooids; in colony BGS Yc 2708, 24 of 54 (44 %) caudate autozooids have closure plates whereas only 43 of 276 (9 %) of non- caudate autozooids have them. The possibility that caudate autozooids are in reality cormidia formed of a proximal kenozooid and a distal autozooid is difficult to dismiss. However, there are no indications in abraded examples of the pore plate that would be expected between the two table 1 . Lengths (in mm) of non-caudate and caudate autozooids in Herpetopora laxata colony BGS Yc 2708. S.D. = standard deviation; C.V. = coefficient of variation. Non-caudate Caudate Mean 0-97 2-61 S.D. 0 198 0-989 C.V. 20-4 38-0 Range 0-27-1-47 1-83-8-28 Determinations 236 63 TAYLOR: GROWTH IN HERPETOPORA 525 zooecial length (mm) text-fig. 4. Frequency distributions of autozooidal length in two colonies of Herpetopora anglica and two of H. Uixata. zooids, and no junction is visible in external morphology. Table 1 gives mean values and ranges for non-caudate and caudate autozooids in BGS Yc 2708. The frequency distribution of autozooidal lengths (text-fig. 5) in the same colony of H. laxata has been broken down according to branch order number (see p. 529). First and second order branches growing across free substratum space are typically long and therefore contain a large number of caudate autozooids. It should be noted that the atypically small autozooids in the first order branches are those budded during early colony growth (see p. 529). Autozooidal polymorphism, excepting the distinction between brooding and non-brooding autozooids, has been recognized in relatively few cheilostome bryozoans (see Silen 1977 for a review of polymorphism in bryozoans). The most similar example to H. laxata occurs in the Albian ‘malacostegan’ Spinicharixa dimorpha Taylor. Colonies of this species have uniserial chains of caudate autozooids which budded non-caudate autozooids distolaterally to infill the areas of substratum between the caudate autozooids (Taylor 1986/5). As in Herpetopora , the caudate autozooids of S. dimorpha very often have closure plates. % frequency 526 PALAEONTOLOGY, VOLUME 31 zooecial length (mm) text-fig. 5. Frequency distributions of autozooidal length in branches of different orders in Herpetopora laxata colony BGS Yc 2708. EXPLANATION OF PLATE 43 Figs. 1-6. Herpetopora laxata (d’Orbigny), back-scattered SEM images. 1, BGS Yc 2708, colony origin showing the ancestrula (just right of centre) and early generations of budded autozooids, Campanian ( quadrata Zone), Compton, Hampshire, x 36. 2 and 3, VH 10300, Lower Maastrichtian, Hemmoor, W. Germany. 2, ancestrula and the two periancestrular zooids, x 55. 3, ancestrula showing closure plate bearing an opercular scar, x220. 4 6, BM(NH) D. 42361, Campanian ( mucronata Zone), Thorpe St Andrew, Norfolk. 4, caudate autozooid (with occluded intramural bud) which in addition to the usual distal and two lateral buds, has produced a proximolateral bud (origin arrowed) and two closely spaced caudal buds, x 32. 5, type 3 branch intersection in which a cauda has curved in growing towards the lateral pore window of a neighbouring zooid, x93. 6, type 2 branch intersection (overgrowth) followed by type 3 intersection in which the overgrowing cauda has deviated towards the left lateral pore window of the overgrown zooid, x 37. PLATE 43 TAYLOR, Herpetopora 528 PALAEONTOLOGY, VOLUME 31 Kenozooids The term kenozooid is generally applied to a range of cheilostome heterozooids (i.e. non-feeding zooids) lacking tentacles, functional guts, and opercula (cf. avicularia). Kenozooids typically have a comparatively simple skeletal morphology, are usually smaller than autozooids, and are often present in fewer numbers. Many kenozooids (text-fig. 3; PI. 44, figs. 3 and 5) in Herpetopora resemble similar polymorphs in closely related genera such as Pyripora, Pvriporopsis , and Conopeum. Like the autozooids, these kenozooids have a narrow cauda of variable length, but instead of having an ovoidal distal part, the kenozooid has a subtriangular distal outline, narrower than an autozooid and with a small circular to longitudinally elliptical opesia. Some of these ‘opesiate kenozooids’ have closure plates which differ from those of the autozooids in the absence of an opercular scar. Prolongations of variable length may extend from the two distolateral corners of the kenozooid. These are usually blind but on occasions connect with other zooids in the colony (PI. 44, fig. 3). The overall shape of the kenozooid can be highly irregular and asymmetrical, sometimes (but not always) in relationship to growth within a restricted space, e.g. where branches intersect. Opesiate kenozooids arise as lateral or distal buds from an autozooid (or possibly as fused buds from two autozooids), and do not normally produce buds of their own (i.e. they usually occur as terminal zooids within branches). Voigt (pers. comm. 1987) notes that kenozooids occur in abundance on substrata with numerous obstacles, e.g. echinoid tests with tubercles. In addition to opesiate kenozooids, two other types of kenozooids are present in Herpetopora. The outer parts of very long branches may comprise a continuous, stolon-like cauda which is difficult to subdivide into individual zooids. However, occasional dilations, smaller in width than an autozooid, signify the distal ends of a second type of kenozooid (PI. 42, fig. 4). These ‘caudate kenozooids’ lack an opesia and show no indications of having had one which was subsequently occluded by a closure plate. Their identification as kenozooids (rather than sections of autozooidal caudae) is supported by their similar length to contiguous autozooids, and the occurrence of paired lateral buds originating from their dilated distal ends. When a growing branch intersected and abutted the side of an existing branch (see p. 533), a third type of ‘kenozooid’ was formed. This is simply a length of cauda of a presumptive autozooid which was unable to develop into either an autozooid or an opesiate kenozooid because of the spatial restrictions imposed by branch abutment. ANCESTRULA AND EARLY GROWTH Ancestrula The ancestrula has not been previously described in Herpetopora and an initial search among the numerous colonies in the BM(NH) collections was unsuccessful. This may seem surprising in view of the abundance of Herpetopora in the Chalk. However, pre-mortem loss of the ancestrula and early parts of the colony may occur very commonly in runner-like bryozoan colonies which experience considerable partial mortality (i.e. death of individual zooids though not necessarily the entire colony). The high incidence of reparative structures in Herpetopora (see p. 540) confirms the widespead occurrence of pre-mortem colony damage. It is notable that the ancestrula was also undescribed until recently (Taylor 1986a) in the closely related living species Pyripora catenularia (Fleming). Two colonies of H. laxata are now known with their ancestrulae intact: BGS Yc 2708 (PI. 43, fig. 1) and VH 10300 (PI. 43, figs. 2 and 3). The ancestrulae are very small (Table 2), ovoidal in outline shape, and attain their maximum width about mid-length. Opesiae of both ancestrulae are occluded by closure plates, convex in VH 10300, but apparently flat (?compressed) in BGS Yc 2708. The junctions between the closure plates and the surrounding gymnocyst, and the ultrastructural details of the closure plates are poorly preserved. Both closure plates bear distal, crescent-shaped impressions of the operculum (PI. 43, fig. 3). No spines or open lateral pore windows are visible in either ancestrula. TAYLOR: GROWTH IN HERPETOPORA 529 table 2. Dimensions (in mm) of the ancestrula in two colonies of Herpetopora laxata. Height and width of the operculum are estimated from impressions on the closure plates. VH 10300 BGS Yc 2708 Length 0-22 0-20 Width 012 Oil Opesia length 017 014 Opesia width 009 006 Operculum height 002 0-02 Operculum width 003 004 The simple ancestrular morphology of H. laxata is similar to that of P. catenularia. The principal differences are that the ancestrula of the living species is about twice the size and has a more extensive proximal gymnocyst with the opesia occupying only the distal half of the frontal surface (Taylor 1 986a). Early growth The ancestrula produces two buds, one distal and one proximal (PI. 43, figs. 1 and 2). These periancestrular zooids initiated the two first order branches of the colony which grew in opposite directions parallel to the long axis of the ancestrula (text-fig. 2). In BGS Yc 2708, in which the two first order colony branches are orientated parallel to the encrusted belemnite guard (text-fig. 1; PI. 42, fig. 1), the distal branch grew towards the apex of the guard and the proximal branch towards the alveolus. The distal periancestrular zooid is larger than the proximal periancestrular zooid in both colonies, and whereas the distal buds have open opesiae, the proximal buds have closure plates. If the relative size of the periancestrular zooids corresponds to their order of budding, then the proximal bud would have formed before the distal periancestrular bud. Early budding patterns are known in three other pyriporids (Voigt 1982; Taylor 1986u): P. catenularia (Fleming), P. huckei Buge, and Pyriporopsis portlandensis Pohowsky. In the two latter species there is a distal and a proximal (or almost proximal) periancestrular bud, as in H. laxata. However, Pyripora catenularia differs in having two additional, lateral periancestrular buds. Colonies therefore possess four branches of the first order orientated at about 90° to one another. Like Herpetopora , the smaller size of the proximal periancestrular zooid in Pyriporopsis portlandensis may imply that it too was budded before the distal zooid. LATER COLONY GROWTH Budding loci and branching angles Most postancestrular zooids in Herpetopora have, in common with many cheilostomes (see Silen 1987), three potential sites of budding which can be termed budding loci. These are situated one distally, and two laterally to distolaterally on either side of the zooid (PI. 42, fig. 4; PI. 45, fig. 6). From the distal locus was budded a zooid orientated parallel to and extending the branch of the parent zooid. Zooids budded from each of the lateral loci were orientated at 70 1 10° to the parent zooid (see p. 530) and initiated new branches of an order one higher than that of the parent zooid. For example, whereas distally budded zooids in a first order branch contributed to that first order branch, each laterally budded zooid formed a new branch of the second order (text-fig. 3). The exact position of the lateral budding loci varies from opposite the mid-point of the opesia to between the mid-point and the distal end of the opesia. Often the two lateral loci are not 530 PALAEONTOLOGY, VOLUME 31 table 3. Branching angles (measured to the nearest 5°) between different branch orders in Herpetopora laxata colony BGS Yc 2708. Branch orders 1-2 2-3 3 4 4-5 5-6 Mean 83-7 79-7 79-6 81 -4 81 -3 S.D. 7-94 8-91 7-35 — — C.V. 9-5 11 -2 9-2 — — Range 70 105 50-100 60-90 75-90 75-90 Determinations 50 78 25 7 4 precisely opposite one another. Lateral budding loci that have not given rise to a bud may be visible as pore windows in the gymnocyst close to the level of the substratum. In some specimens the gymnocyst (here used for all inferred exterior walls, both frontal and vertical) is sufficiently transparent to show that the pore window leads to a parallel-sided canal which passes through the thick vertical wall and opens into the interior of the zooid. The canal is not dilated, as in P. portlandensis (Banta 1975), nor is there a pore plate of the kind present in the pore chambers of many other cheilostomes (see Banta 1969). Distal budding loci are similar to lateral loci except that at least some appear to possess an apparent pore plate which may be visible in abraded specimens as a transverse wall between the zooid and its distal bud (PI. 44, fig. 1). Daughter zooids are not usually formed from all three potential budding loci. One or both of the lateral loci very commonly fails to produce a daughter zooid, and occasional failure of the distal locus may also occur resulting in branch termination. The frequency of bud formation at lateral loci decreases markedly in branches of higher orders. For example, in BGS Yc 2708, lateral buds are about seven times more common in zooids belonging to first than third order branches (see p. 535). Angles between parent branches and laterally budded daughter branches are variable. The observed range in BGS Yc 2708 is 50-105°, with a mean value of about 80°. Differences in branching angle according to the orders of parental and daughter branches are slight and probably insignificant (Table 3). However, an early astogenetic decrease in branching angle is evident if angles between the first order branches and their second order daughters are plotted outwards from the ancestrula (text-fig. 6). Branching angle declines gradually from about 100° until the typical value of about 80° is reached after approximately nine zooidal generations. Attainment of this astogenetic repetition of branching angle is roughly coincidental with the major threshold in zooid length discussed below (p. 536). Zooids were occasionally budded from sites other than the distal and two lateral loci. A very EXPLANATION OF PLATE 44 Figs. I and 4. Herpetopora anglica Lang, back-scattered SEM images. 1, BM(NH) D.4189, right lateral bud and its reparative bud, both heavily abraded (the remains of a possible pore plate are visible at the distal end of the original bud), Coniacian or Santonian, Chatham, Kent, x49. 4, BM(NH) D. 45920, original (left) and reparative (right) buds partially superimposed, Santonian ( Uintacrinus Band), Margate, Kent, x 65. Figs. 2, 3, 5. H. laxata (d’Orbigny), back-scattered SEM images, BM(NH) D. 42361, Campanian (mucronata Zone), Thorpe St Andrew, Norfolk. 2, reverse polarity intramural opesiate kenozooid within a badly damaged non-caudate autozooid, x 53. 3, intramural opesiate kenozooid within non-caudate autozooid, and small opesiate kenozooid (bottom left) linked to an autozooid (centre right) by a cauda which over- grows the autozooid with the intramural bud, x 78. 5, complex arrangement of branches comprising kenozooids and autozooids, including examples with long caudae, closure plates and intramural buds, x 21. PLATE 44 TAYLOR, Herpetopora 532 PALAEONTOLOGY, VOLUME 31 IIO-i 100 - • • 90- ■ ■ ■ • 80 - 70- ■ • •• ■ • • ■ a a aaa a • • 5 i 1 r 10 — i 1 1 l 15 generation along 1st order branch text-fig. 6. Scatter diagram showing astogenetic gradient in angle of branching between first and second order branches according to zooidal generation along the first order branches. Dots are angles from the proximal first order branch, squares from the distal first order branch. Measurements made to the nearest 5° from Herpetopora laxata colony BGS Yc 2708. few zooids produced proximolateral daughters which originated from a budding locus opposite the proximal end of the opesia (PI. 43, fig. 4). These proximolaterally budded zooids are orientated at about 120° to the parent zooid and initiated branches which grew proximolaterally relative to the parent branch. More frequently buds arose from the caudal regions of parent zooids with long caudae (PI. 43, fig. 4). These caudally budded zooids are orientated at approximately 90° to their parent, two or more may originate from the same parent, and there is no apparent pattern in their distribution. The formation of caudal buds is unclear; open pore windows signifying possible additional budding loci have not been observed in the caudae of caudate zooids. It seems possible, therefore, that the formation of caudal buds may necessitate either skeletal resorption or damage in order to breach the cauda and form a site for zooidal budding. Unless reparative growth has occurred (see p. 540), the sequence of budding within a single branch is unequivocal and follows a proximal to distal direction. However, the relative timing of bud formation between branches is not easily inferred in view of possible between-branch variations in budding rates. One possible source of between-branch variation is retardation of lateral bud formation, suggested by the common occurrence of open lateral pore windows from which a bud has not yet formed in zooids with a fully formed distal bud. This would be conceivable in Herpetopora because the rarity of semi-formed zooids suggests that bud formation was probably episodic (similar to the intrazooidal budding defined by Lidgard 1985, but not necessarily involving a pore chamber). Observation of branch intersections can be used to estimate the extent of between-branch variations in budding rate. Because later zooids must abut or overgrow earlier-formed zooids (see p. 533), major differences in budding rate between branches will result in the common occurrence of intersections in which the abutting or overgrowing zooid has a lower generation number than the abutted or overgrown zooid. This is not the case in colony BGS Yc 2708; in only 7 of 59 (12 %) intersections does the abutting or overgrowing zooid have a lower generation number than the abutted or overgrown zooid. Therefore, appreciable differences in budding rates between branches seem to have been uncommon in Herpetopora , and any retardation of lateral bud growth relative to distal bud growth was probably minor. TAYLOR: GROWTH IN HERPETOPORA 533 text-fig. 7. Diagram of types of branch intersection observed in Herpetopora colonies. In type I intersec- tions, the growing branch terminates against the existing branch, immediately in the case of type 1 a but after partial overgrowth in the case of type \b. Complete overgrowth occurs in type 2 intersections. Type 3 intersections involve growth of a cauda towards an open pore window. Head-on collisions between growing branches result in type 4 inter- sections. type 4 Basic colony growth pattern Positions of budding loci and orientations of daughter zooids, and consequently colony branches, resulted in the following basic pattern of colony growth in Herpetopora. The two first order branches grew distally and proximally from the ancestrula, forming a ‘backbone’ to the colony. From both sides of the first order branches there diverged numerous second order branches in a rib-like arrangement at angles of about 80° to the first order branches (as in text-fig. 2). Third and higher order branches continued the pattern, branches of odd number orders orientated subparallel to the first order branches, and branches of even number orders subparallel to the second order branches (text-fig. 3). This conjugate pattern of branches, resembling the trichotomous branching pattern illustrated by Harper (1985) for a hypothetical modular organism, is disrupted in real colonies by variability in branching angles, slight curvature of branches, failure of buds, the presence of obstacles to branch growth, and irregularities in the shape of the substratum. Branch intersections Numerous examples of branches intersecting one another can be observed in colonies of Herpetopora. Branch intersections occurred routinely as growing branches collided with existing branches, or more rarely as two growing branches met. The availability of colony BGS Yc 2708 with a large number of connected zooids provides an opportunity to study the varying types and frequencies of branch intersections which occurred demonstrably within a single colony (and therefore a genetic entity). A total of fifty-nine branch intersections were observed in BGS Yc 2708, and qualitative observations were also made on several other colonies. Branch intersections can be classified (text-fig. 7) as follows: Type 1 (abutment). Here a zooid of a growing branch abutted the side of an existing branch. Generally the growing branch terminated soon after its first contact with the existing branch (type la), but sometimes it overgrew the mid-line summit of the existing branch before terminating (type 1 b). In both cases the distal end of the growing branch is sealed to the exterior. It is not known whether resorption of the gymnocystal skeleton of the earlier branch occurred to permit soft tissue 534 PALAEONTOLOGY, VOLUME 31 connection between the two branches. The zooid in the growing branch usually had insufficient space to develop into an autozooid and is represented by a short length of cauda. Occasionally, however, space was available for the abutting zooid to become an opesiate kenozooid or even an autozooid. Intersections of type 16 have only been observed in BGS Yc 2708 where the growing branch met the cauda or distal gymnocyst of a zooid on an existing branch; overgrowth of opesia does not seem to have occurred. In this colony, 61 % of intersections are of type 1 (51 % of type 1 a and 10 % of type 16). Herpetopora colonies frequently overgrew obstacles higher than their own branches, suggesting that type 1 intersections involved a mechanism of self-recognition to halt branch growth. Type 2 (overgrowth). Here the growing branch completely overgrew the existing branch and continued normal distal growth. This type of intersection in BGS Yc 2708 occurred only when a growing branch met parts of a zooid proximal or distal of the opesia, and never on meeting the opesia itself. Ten per cent of observed intersections in this colony are of type 2. Type 3 (pore location). The most interesting type of intersection entailed branch growth towards a lateral pore window of a zooid in an existing branch (PI. 43, fig. 5). As the second commonest type of intersection (25 % of observations), it is unlikely to have been a fortuitous occurrence. Furthermore, the growing branch often bent considerably in order to locate the pore window of the zooid in the existing branch. Branches growing away from or parallel to the existing branch began to bend towards the pore window at a distance of about 0-21 -0-23 mm from the window in well-preserved type 3 intersections in H. laxata colony BM(NH) D. 42361. In one example (PI. 43, fig. 6) from this colony, a growing branch evidently approached the right-hand side of an existing branch, overgrew the distal gymnocyst of one zooid, and then turned sharply towards the left lateral pore window of the next proximal zooid in the branch (i.e. a type 2 followed immediately by a type 3 intersection). Significantly, the right lateral pore window of this zooid, which could have been contacted first by the growing branch, is lacking. Examination of type 3 intersections with an optical microscope reveals the precise alignment of the growing branch and the pore window of the earlier branch which is visible through the slightly transparent gymnocyst. Abraded type 3 intersections also show this alignment and the lack of any skeletal barrier to potential soft tissue linkage between the two intersecting branches. Sometimes an opesiate kenozooid provides the connection between two branches and it may be difficult to decide whether this was formed by a type 3 intersection or by the fusion of separate buds originating from each of the two branches. Type 4 (collision). Rarely (4 % of observations in BGS Yc 2708) the distal growing tips of two branches collided. The probability of such ‘head-on’ collisions was undoubtedly small because of the narrowness of Herpetopora branches. Collisions occurred more often between branches constructing the dilated distal parts of zooids than branches constructing the narrow proximal caudae of zooids. The two colliding zooids are typically distorted in shape although they may have fully formed opesiae suggesting that they were able to function as autozooids. Intersection types 1-3 were all formed as the result of a growing branch meeting an existing branch. However, it is unclear what factors determined which of the three would occur. For example, there is no obvious correlation between type 3 intersections and the close proximity of a pore window; some growing branches abutted the sides of existing branches very close to open pore windows without contacting the window, whereas others had to deviate considerably in order to locate a pore window. Intuitively, angle of encounter would seem to be a possible determinant of intersection type, but again no clear correlation is apparent (cf. the effects of encounter angle on interspecific overgrowth between multiserial cheilostomes, see Jackson 1979u). Another factor may have been the condition of the zooids on the earlier branch; perhaps zooids with cuticles in poor condition, dead or without a polypide were overgrown rather than abutted, and pore linkage was dependent upon a healthy zooid in the earlier branch. As noted above (p. 532), intersecting zooids normally (88 % of observations in BGS Yc 2708) abutted, overgrew, or linked with pore windows of zooids of the same or lower generation number. This is to be expected if zooid generation number closely reflects the succession of zooidal budding. TAYLOR: GROWTH IN HERPETOPORA 535 All seven of the anomalous intersections in colony BGS Yc 2708 were type 1 intersections, i.e. branch abutments. Type 3 intersections entail deviation in branch growth towards lateral pore windows which is strongly suggestive of tropism. Ryland (1977) has reviewed tropisms in living bryozoans but makes no mention of any examples resembling the type 3 intersections of Herpetopora. Soft parts which could provide a tactile tropism are not known to protrude from the open pore windows of living cheilostomes, and the most probable source of the tropism would appear to be a chemical released from the pore window. The growing branch commenced its response to this inferred chemical when it reached a distance of about 0-2 mm from the pore window. By causing the fusion of parts of the same colony, type 3 intersections are a form of autosyndrome (Knight-Jones and Moyse 1961). Little comparative information has been published on branch intersections in runner-like bryozoans. Personal observations of the Recent species Pyripora catenalaria have shown that most intersections are of type 1 (abutment), but a few type 2 intersections (overgrowth) also occur. No examples have been seen of growth towards lateral pore windows in P. catenu/aria , but Marcus ( 1 949) observed ‘interzooecial bridges’ apparently linking lateral pore windows of zooids in adjacent branches of P. audens. Colonies of the runner-like cyclostomes Corynotrypa and Stomatopora exhibit predominantly abutment intersections, with occasional overgrowths (Gardiner and Taylor 1982; Carthew 1987). BUD FREQUENCY AND SUCCESS In common with other branching, runner-like bryozoans, the number of growing branches in a Herpetopora colony increased more rapidly than the space available at the perimeter of the expanding colony. Each zooid normally had the potential to bud three daughter zooids. Therefore, if all potential buds were to have formed, the number of zooids in successive generations would have increased in proportion to the cube of the generation number, whereas the area of substratum available for encrustation increased in proportion roughly to the square of generation number. As a result, the frequency of branch intersections and/or the density of zooids must increase in the younger, outer parts of colonies. However, most branch intersections (types 1, 3, and 4) resulted in the elimination of one growing branch (and all of the daughter branches which it may have subsequently formed), thus reducing the problem of crowding. Failure of one or more of the three potential zooidal budding loci to form an autozooid is a second factor which reduced crowding in Herpetopora colonies. This failure appears to correlate with the proximity of other zooids. To assess the effects of such crowding, frequency of bud formation and success (i.e. full development of the bud into an autozooid) have been studied in colony BGS Yc 2708 in relation to the branch order and polymorph type of the parental zooid. Conspicuous differences are seen in bud frequency according to branch order (text-fig. 8). Whereas zooids in first order branches gave rise to an average of 2-67 daughter autozooids, those in second order branches produced only 1-51 autozooids, third order branches 0-62 autozooids, and fourth order branches 0-50 autozooids. The decrease is proportionally greater for lateral buds than distal buds; very few zooids in third and fourth generation branches gave rise to lateral daughter autozooids. The pattern of diminishing bud success with increasing branch order may be explained by the geometry of the Herpetopora colony. The two first order branches grew in opposite directions across free substratum space and could bud zooids unimpeded by the presence of earlier autozooids. The second order branches grew parallel to one another and, although they also crossed free substratum space, their buds developed in the close proximity of adjacent second order branches (and the two first order branches). In contrast, third order branches converged during growth with other third order branches arising from adjacent second order parents. Therefore, a considerable degree of interference occurred between each third order branch and other branches of the same or lower orders. Similarly, fourth order branches were also greatly 536 PALAEONTOLOGY, VOLUME 31 number of zooids 30 71 102 32 text-fig. 8. Frequency of autozooidal buds from parent zooids of different branch orders in Herpetopora laxata colony BGS Yc 2708. Stippled area represents distally budded autozooids, unstippled laterally budded autozooids. branch order disrupted by convergent growth with other branches, and zooids in these branches were prevented from budding the maximum possible number of daughter zooids. There is a striking difference in the number of daughter buds produced by non-caudate and caudate autozooids. For lateral buds only, and summing the data across all zooid generations and branch orders, non-caudate autozooids produced an average of 0-43 buds, whereas caudate autozooids produced 1-54 buds. This difference may be related to the fact that the lateral budding loci of caudate autozooids are considerably more distant from their neighbours than the loci of the shorter non-caudate autozooids, and are therefore less prone to interference from other zooids. ASTOGENETIC GRADIENTS Gradients of changing zooid length have been quantified along branches of H. laxata colony BGS Yc 2708 for which the generation of each zooid (from the ancestrula) is known. These astogenetic gradients are more complex than is usual for bryozoans; individual branches exhibit a high degree of autonomy, and first and higher order branches demand separate treatment. First order branches The two first order branches are derived directly from the ancestrula and reveal astogenetic gradients of early colony development. Text-fig. 9 plots zooid length against generation along these branches. The astogenetic gradient of each branch has two phases: an initial phase (phase 1 ) of steadily increasing zooid length, separated by an abrupt increase in length from a later phase (phase 2) of fluctuating but generally increasing length. The slope of phase 1 is slightly steeper (019) in the distal branch than in the proximal branch (0T3), and the threshold between the phases occurs earlier in the distal branch (between zooid generations 5 and 6) than in the proximal branch (between generations 8 and 9). However, in both branches the zooid immediately preceding the threshold has a length of 1-23 mm. The first zooid after the threshold is 2-22 mm long in the distal TAYLOR: GROWTH IN HERPETOPORA 537 text-fig. 9. Astogenetic gradients of auto- zooidal length in the two first order branches of Herpetopora laxata colony BGS Yc 2708. The changeover from non-caudate to caudate autozooids occurs between generations 5 and 6 of the distal branch and generations 8 and 9 of the proximal branch. The abnormally long zooid of generation 15 in the proximal branch occurs in a damaged part of the colony and may be spurious. zooecial length mm generation branch and 2-31 mm long in the proximal branch. Zooidal length generally increases with generation in phase 2 but fluctuations are more pronounced than in phase 1 . The size-frequency distribution of autozooidal length in colonies of H. laxata (text-fig. 4) demonstrated the existence of two types of autozooidal polymorph: non-caudate autozooids with lengths less than 1-6 mm and caudate autozooids with lengths greater than 1-6 mm. The threshold along the first order branches between phases 1 and 2 clearly corresponds to a changeover from the budding of non-caudate to caudate autozooids. Higher order branches Thomas and Larwood (1960, p. 372) in their description of H. laxata remarked Tn any series of zooecia successive caudae are commonly increasingly longer’. This tendency of zooids to become progressively longer along each new branch is one of the most striking features of H. laxata colonies (PI. 42, fig. 2), and also occurs in H. anglica. Measurements of zooidal lengths in BGS Yc 2708 allows an analysis of the exact pattern of these astogenetic gradients in second and higher order branches. As there are no detectable differences between branches of different order, the data for all branch orders have been pooled. Branches usually start with a series of one to three non-caudate autozooids, followed by a series of caudate autozooids. Reversion to the budding of non-caudate autozooids has never been observed in colony BGS Yc 2708 or any other colony of H. laxata , but sometimes (7 % of branches in Yc 2708, all second order) new branches start with caudate autozooids instead of non-caudate autozooids. Pairwise comparison of zooidal length between non-caudate autozooids and their non- caudate distal daughters revealed that in seventy-three (99 %) of cases the daughter zooid was longer than its parent. A similar comparison for pairs of caudate autozooids revealed a more variable pattern with ten (63 %) examples of the daughter being longer than the parent, one (6 %) of it being the same size, and five (31 %) of it being shorter. Values of length for non-caudate and caudate autozooids according to their position along the branch are given in Table 4. The most proximal non-caudate autozooid averages 0-90 mm long, the next I TO mm, and the next T22 mm. The most proximal caudate autozooid averages 2-26 mm long, the next 2-62 mm, and the next 3-45 mm. Trends of increasing length within non-caudate and caudate series are depicted using size-frequency histograms (text-fig. 10). An astogenetic increase in length is very clear from the size distributions of non-caudate autozooids, but for caudate autozooids the trend seems to result from the occurrence of some unusually long autozooids in later generations which cause a rise in mean value. 538 PALAEONTOLOGY, VOLUME 31 table 4. Autozooidal length (in mm) in Herpetopora laxata colony BGS Yc 2708 according to position within branches. For non-caudate autozooids the value of position is relative to the branch origin, for caudate autozooids it is relative to the point of transition between astogenetic phases 1 and 2. Zooids from first order branches have been excluded. Non-caudate Caudate 1 2 3 1 2 3 Mean 0-90 1 10 1-22 2-26 2-62 3-45 S.D. 0 150 0 159 0-131 0-266 0-844 1-897 C.V. 16 6 14 5 10 7 11-8 32-2 54-9 Range 0-75 1 -47 0-90 1 -44 1 08 1 44 1 74 3-09 1-89 4-83 2-31 5-64 Determinations 157 74 11 30 11 3 zooecial length (mm) text-fig. 10. Size-frequency histograms showing astogenetic gradients of autozooidal length in non-caudate and caudate series of autozooids along branches of second and higher orders in Herpetopora laxata colony BGS Yc 2708. Arrows indicate mean length for each generation. TAYLOR: GROWTH IN HERPETOPORA 539 text-fig. 1 1 . Between-branch astogenetic gradient of autozooid length in Herpetopora laxata colony BGS Yc 2708. The points rep- resent the lengths of laterally budded daughter autozooids (i.e. first autozooids of second order branches) of parental autozooids located in the first order branches. Non-caudate autozooids are shown as dots, caudate autozooids as triangles. Two regression lines have been fitted to the data from the proximal first order branch; the line of higher slope includes all autozooids, whereas that of lower slope ex- cludes caudate autozooids. zooecial length mm Between-branch gradients The astogenetic gradients of individual branches have so far been treated separately as if each branch were entirely independent. However, the size of the first zooid of the second order branches bears a relationship to the position of the second order branches along the parental first order branch. This produces a between-branch astogenetic gradient (text-fig. 11). Length of the first zooid in the second order branches increases significantly in a distal direction away from the ancestrula, but values of the slope are low (0 017 along the distal first branch, and 0 053 along the proximal first order branch, but 0 031 if only non-caudate zooids are included in the linear regression). There is some indication of the slope levelling-out by about five zooid generations along the distal branch and ten along the proximal branch. Similar between-branch gradients are not obvious between branches of second and third, third and fourth orders, etc., though the comparatively short lengths of these higher order branches may hinder their detection. Summary Each separate branch in a colony of H. laxata exhibits an astogenetic gradient of zooidal length which consists of two phases: an early phase (phase I ) of non-caudate autozooids which steadily increase in length, followed by an abrupt rise in zooidal length into a later phase (phase 2) of caudate autozooids which also increase in mean length during branch growth. The threshold between phases 1 and 2 typically occurs after one to three generations of non-caudate autozooids have been budded, but a few branches omit phase 1 and bud caudate autozooids from the outset. Superimposed on these within-branch gradients there is a between-branch gradient, for first to second order branches at least, of increasing size of the first zooid in the branch depending on the position of origin of the branch distally along its parental branch. For other species of Herpetopora the form of the astogenetic gradients may be different. Size- frequency distributions (text-fig. 4) of zooid length in H. anglica reveal only one autozooid polymorph and hence exclude the possibility of a two phase astogenetic gradient. However, zooid length does increase distally along branches. Quantification of astogenetic gradients in sixteen branches, each of three to six generations in length, from four colonies of H. anglica gave the following results: forty-five instances in which the distal daughter zooid was longer than its parent; two in which it was the same length; and six in which it was shorter. Text-fig. 12 shows the 540 PALAEONTOLOGY, VOLUME 31 T 1 1 I | I I I I | I I I 1 1 1 2 3 4 5 6 7 8 9 1011 12131415 generation text-fig. 12. Astogenetic gradients of autozooidal length in long branches from two BM(NH) colonies of Herpetopora anglica. gradients in two exceptionally long branches of H. anglica. The overall trends seem to be of a sustained increase in zooidal length throughout astogeny but with considerable ‘noise’ superimposed. Astogenetic gradients of zooid size in most bryozoans are divisible into a primary zone of astogenetic change, generally brief, which is succeeded by a typically extensive primary zone of astogenetic repetition (Boardman and Cheetham 1973; Taylor and Furness 1978). Secondary zones of astogenetic change and repetition occur in some taxa. Flowever, application of this terminology to Herpetopora is difficult as there is little evidence that a stable zooidal size, indicating a zone of astogenetic repetition, is ever attained. Each branch (and the colony as a whole) may perhaps be in perpetual astogenetic change. REPARATIVE STRUCTURES Use of the SEM to study colonies of Herpetopora reveals abundant evidence of damage and its repair. Many specimens have zooids with extensively fractured skeletal walls. Sometimes all that remains of the zooids are short, disconnected lengths of caudae. Distinguishing pre-mortem from post-mortem damage is often difficult. Certain damage inflicted during the life of the colony can be recognized in the various types of reparative buds which occur. The frequency of reparative budding provides some indication of the extent of pre-mortem damage; however, this is a minimum estimate because more subtle repairs may not be detectable, and unrepaired damage cannot be distinguished from post-mortem damage. Intramural reparative budding Concentric, so-called regeneration rims are found within the opesiae of many Herpetopora zooids (PI. 43, figs. 5 and 6; PI. 44, fig. 5). These rims may occur singly or in multiples of up to at least three within each zooid. Use of the term ‘regeneration’ (Levinsen 1907 and most subsequent authors) to describe these structures in cheilostomes is unsuitable as they are probably not related to polypide regeneration, but are instead new zooids budded within the empty skeleton of an older damaged zooid (Banta 1969). As such they are the daughter buds of zooids adjacent to the damaged zooid and are presumed to have been budded through the open communication canals linking zooids. Banta (1969) termed these structures ‘reparative buds’ but the additional term ‘intramural’ TAYLOR: GROWTH IN HERPETOPORA 541 is here introduced in order to avoid confusion with the other type of reparative bud described below. Only the mural rims surrounding the opesia of intramural buds are normally visible, but fractured examples indicate that at least some of the cauda is also calcified. Obviously, the opesia of the intramural bud is smaller than that of the zooid within which it occurs. In examples of multiple intramural budding, the youngest (innermost) buds can have severely reduced opesial dimensions, and it seems possible that diminishing size restricted the number of times that intramural budding could occur within a host zooecium. However, even the smallest intramural buds may have closure plates bearing opercular impressions suggesting the autozooidal nature of the bud. Intramural buds with closure plates can be distinguished from normal zooids with closure plates by the occurrence of a narrow gap around the closure plate which separates it from the mural rim of the host zooecium. In all but one out of many observed examples the opercular scar is located distally relative to the host zooid, showing that the reparative bud had the same polarity as the host zooid. A single example of a reverse polarity intramural bud (cf. Jebram 1978, fig. 4 (2)) has been found (PI. 45, fig. 1) in which the bud is presumed to have originated from the zooid distal to the host zooid. Occasionally (PI. 44, fig. 3), intramural buds are opesiate kenozooids, including one example (PI. 44, fig. 2) of a reverse polarity intramural kenozooid occupying a severely damaged autozooid. Intramural buds appear to be more frequent in non-caudate autozooids of H. laxata than in caudate autozooids. This may correlate with the observation that caudate autozooids more often have closure plates (p. 524), the presence of which would prohibit intramural budding without first resorbing the closure plate. Related structures, recognized for the first time in Herpetopora , are lengths of cauda-like skeleton visible within autozooids (PI. 45, figs. 4 and 5). These ‘intramural caudae’ generally parallel the long axis of the host zooid, but can be oblique, can deviate laterally, and may even bifurcate. Intramural caudae could represent the proximal parts of autozooids, or alternatively they may be a type of non-opesiate kenozooid. Similar structures (‘tubules’) have been described by Marcus (1938, 1955) from two genera of living cheilostomes. Both intramural caudae and normal intramural buds potentially may provide a linkage between living zooids through the empty skeletons of dead zooids, thus maintaining physiological continuity of the colony despite mortality of intervening zooids. Extramural reparative budding Following destruction of entire zooids, or their very severe damage, ‘extramural buds’ were sometimes formed occupying the same site on the substratum as the damaged zooid. In reality, there is a continuum between intramural buds and extramural buds via original zooids which have suffered increasing levels of damage. Two types of extramural bud can be distinguished: buds having the same polarity as the original zooid and buds of opposite polarity. Buds of the same polarity (PI. 44, figs. 1 and 4) may have their long axes slightly oblique to the original zooid, and can include remnants of the original zooid within their walls. The reparative zooids originate from budding loci almost but not quite coincident with the loci which gave rise to the original zooid. Lengths of original and reparative zooids are approximately the same. It is likely that extramural reparative budding is of commoner occurrence than is obvious but that its recognition is precluded by total removal of the original zooid or complete overgrowth by the reparative zooid. Complete removal of one or more zooids from the middle of branches, or branch fracturing with removal of a portion of a zooidal cauda, sometimes resulted in the budding of reparative zooids from the proximal fractured end of the branch. These zooids grew in a proximal direction relative to branch growth direction (and that of the zooid they replaced). The reparative zooid and the parental zooid from which it originates have opposite polarities, and form a bipolar zooid pair (sensu Taylor 1986«). Comparatively few examples of bipolar zooid pairs have been identified in Herpetopora , possibly because branches of severely damaged colonies in which they should be most frequent are so badly fragmented that relationships between zooids are obscure. An example (PI. 45, figs. 2 and 3) involving a caudate distal zooid shows only a small remnant of the mural 542 PALAEONTOLOGY, VOLUME 31 rim of its proximal maternal zooid within which is situated the reparative bud of opposite polarity. Although the reparative bud appears to be short (? non-caudate autozooid), its exact point of origin is unclear, and its distally budded daughter is a caudate autozooid. A second, unfigured example of a bipolar pair in colony BM(NH) D. 42364 involves a distal zooid and a reparative zooid both of which are of similar size and are non-caudate autozooids. Very similar bipolar zooid pairs have been previously described (Taylor 1 986a) in colonies of the Recent cheilostome Pyripora catenularia , and also in the Palaeozoic cyclostome Corynotrypa (see Taylor 1985; Carthew 1987). SUMMARY OF COLONY DEVELOPMENT The sequence of development in colonies of Herpetopora can be summarized as follows: 1. Following settlement and metamorphosis of the larva on a firm substratum (e.g. an Inoceramus shell), the resultant ancestrula budded two daughter zooids, one distally and the other proximally, thereby initiating the two first order colony branches which grew in opposite directions. 2. These periancestrular zooids and subsequently budded zooids were normally capable of each producing three daughter zooids, one distally and two laterally on either side of the zooid. Distal daughter zooids extended the branch of the parent, whereas lateral daughter zooids initiated new branches of an order one higher than that of the parent zooid. The angle between branches averaged about 80° but varied between 50° and 105°, with higher angles occurring especially during early colony development. This pattern of branching resulted in a colony with two conjugate sets of parallel branches of odd and even number orders. However, variation in branching angle, branch curvature, and substratum irregularities typically obscures these relationships in actual specimens of Herpetopora (e.g. PI. 44, fig. 5). 3. The length of successive zooids along the two first order branches in H. laxata increased steadily for six to nine generations from the ancestrula. Thereafter, length rose abruptly and subsequently continued to increase but more erratically. Similar two-phase gradients of astogenetic increase in zooidal size also occurred in each new branch of H. laxata formed by lateral budding, but the later phase (phase 2) followed an early phase (phase 1) comprising only one to three generations of zooids and sometimes missing altogether. The zooids budded during the two phases were autozooidal polymorphs distinguished principally by their lengths, the ‘caudate autozooids’ of phase 2 having longer caudae than the ‘non-caudate autozooids’ of phase 1 . Caudate autozooids more often developed closure plates occluding their opesiae than did non-caudate autozooids. No autozooidal polymorphism is detectable in H. anglica, although astogenetic gradients of increasing zooidal length also occur along each branch. EXPLANATION OF PLATE 45 Figs. 1 and 6. Herpetopora anglica Lang, back-scattered SEM images. 1, BM(NH) D.4189, reverse polarity intramural autozooid (arrowed) with a closure plate bearing an opercular scar, Coniacian or Santonian, Chatham, Kent, x 27. 6, BM(NH) D. 29840, autozooids with open opesiae and opesiae occluded by closure plates, Turonian (planus Zone), Bridgewick Pit, Mailing, Sussex, x 37. Figs. 2-5. H. laxata (d'Orbigny), back-scattered SEM. 2-4, BM(NH) D. 42375, Campanian (mucronata Zone), Webster’s Pit, Norwich, Norfolk. 2, bipolar zooid pair involving an original caudate autozooid (left) which has an intramural bud with closure plate, and a reparative autozooid (right), x 23. 3, detail of the reparative autozooid of the bipolar pair showing the remains of the zooid which has been repaired, various segments of caudae external to and a possible oblique intramural kenozooid within the reparative zooid, x 87. 4, damaged autozooid with an apparently bifurcating segment of cauda within, x 61. 5, BGS Yc 2708, segment of cauda passing through a heavily worn autozooid, Campanian (quadrata Zone), Compton, Hampshire, x 72. PLATE 45 TAYLOR, Herpetopora 544 PALAEONTOLOGY, VOLUME 31 4. Lateral buds were possibly slightly retarded in their development relative to distal buds and often were not produced at all. Zooids located in branches of lower orders were more likely to bud all three potential daughter zooids, and caudate autozooids produced more buds on average than non-caudate autozooids in H. laxata. Caudal buds sometimes arose adventitiously from the caudae of zooids, and more rarely proximolateral buds were formed close to the opesiae of parental zooids. 5. Branches often intersected during colony growth. Most frequently the growing branch abutted the existing branch and ceased growth, but sometimes it overgrew the cauda of a zooid on the existing branch. Growing branches could also apparently respond to the presence of an open pore window of a nearby zooid and grow into it, probably chemotropically. Rarely, two growing branches collided and their terminal zooids became distorted. 6. Apart from autozooids, kenozooids were occasionally budded. These were either produced as lateral or distal buds, opesiate (but lacking an operculum) and with a subtriangular distal end, or produced as distal buds in long stolon-like branches, apparently non-opesiate with very long caudae and a slightly dilated distal end. A third type of kenozooid is the short portion of cauda remaining in a growing branch after abutment against the side of an existing branch. 7. Zooids and branches were often damaged during the lifetime of the colony and various reparative structures were formed. Particularly common was the intramural budding of zooids (usually autozooids) within the empty zooecia of dead autozooids. Intramural buds normally had the same polarity as their hosts and may have been derived as buds from the proximal neighbour of the host zooid. However, they occasionally had the opposite polarity. Severely damaged or entirely obliterated zooids were sometimes replaced by extramural reparative buds either of the same polarity or of opposite polarity to the zooids they replaced. 8. Partial colony mortality resulting from damage may often have caused the fission of colonies into spatially separated ramets (cf. Taylor 1985). However, through reparative budding and the ability of branches to grow towards open pore windows, colonies had the potential to reunite ramets. DISCUSSION Astogenetic gradients and their origin Herpetopora , especially H. laxata , has a more complex pattern of astogenetic gradients than is usual in a runner-like bryozoan. Typically (e.g. Stomatopora , see Taylor and Furness 1978), a primary zone of astogenetic change, during which there is a progressive increase in size of the first few generations of zooids, is followed by a primary zone of astogenetic repetition when zooid size is astogenetically constant (though ontogenetic and microenvironmental variations may occur). All zooids budded at the same time, regardless of which branch they belong to and their spatial location within that branch, are morphologically similar in the zone of astogenetic repetition. Therefore, the astogenetic gradient is colony-wide. P. catenularia and Pyriporopsis portlandensis also have a primary zone of astogenetic change followed by a primary zone of astogenetic repetition (Taylor 1986a). However, each new branch formed by lateral budding in these pyriporids initiates a brief secondary zone of astogenetic change through which zooid size increases (e.g. Pohowsky 1973). This is followed by a series of zooids, similar in size to those in the primary zone of astogenetic repetition, which form a secondary zone of astogenetic repetition (text-fig. 13). At any one time during colony growth, therefore, zooids may be developing which belong to primary or secondary zones of change or repetition. Astogenetic gradients within the colony are subcolony- wide, each branch in the colony constituting a subcolony. In Herpetopora , similar subcolony-wide astogenetic gradients occur (text-fig. 13) but there is little evidence that zones of astogenetic repetition are ever formed in either H. anglica or H. laxata. Branches of H. anglica exhibit a single phase of generally increasing zooidal length, whereas those of H. laxata have a two-phase gradient beginning with a series of progressively lengthening non-caudate autozooids (astogenetic phase 1) TAYLOR: GROWTH IN HERPETOPORA 545 l 1 1 1 1 1 1 1 1 2 3 4 5 6 7 8 generation text-fig. 13. Simplified astogenetic gradients present in branches of typical pyriporid cheilostomes, Herpetopora anglica , and H. laxata. which is followed by a series of caudate autozooids (phase 2) whose mean length increases with astogeny. The physiological basis of astogenetic gradients is completely unknown in living bryozoans (e.g. Jebram 1978), and has formed a subject for speculation in fossil bryozoans. Two principal theories have been proposed: 1. In the trophic theory (Dzik 1975, 1981) the astogenetic gradient is controlled simply by the food resources available to the bud at the time of its growth. A juvenile colony has relatively few zooids and little energy to supply the growing buds and it is assumed that the zooids formed will consequently be small. 2. In the morphogenetic theory (Anstey et al. 1976) the astogenetic gradient is the result of diffusion through the colony or subcolony of a morphogenetic substance which determines zooid size. A similar theory has been applied to graptolites (see Urbanek 1973) in which the siculozooid is the presumed source of the morphogen. The equivalent of the siculozooid in bryozoans is the ancestrula, and it or, in the case of subcolony-based gradients in trepostomes, the monarchic zooid, is thought to be the source of the morphogen (Anstey et at. 1976; Podell and Anstey 1979). Autoradiographic evidence (Best and Thorpe 1985) of translocation of labelled carbon- 14 towards the growing edge in living colonies of the cheilostome Membranipora is important in demonstrating the plausibility of morphogenetic control in bryozoans, although both the range of substances translocated and the mechanism of their translocation remain unknown in living bryozoans. The trophic theory cannot be applied to the subcolony-based astogenetic gradients of Herpetopora , unless the unlikely assumption is made that new branches depend for their growth entirely on their own food resources. A morphogenetic theory is more easily applicable to Herpetopora. However, there is an alternative to the diffusion theory in which zooid size (and other characteristics) is 546 PALAEONTOLOGY, VOLUME 31 determined by the zooid’s spatial position along a branch; zooid size may have been determined by the age of the branch at the time of zooid formation, older branches producing larger zooids. This alternative can be discounted in Herpetopora because extramural reparative buds (p. 541) are of a similar size to the zooids they replaced, despite having been formed later in branch growth. The source of the morphogens determining zooid size in each branch of a colony of Herpetopora was most likely to have been the parental zooid of the branch, through which the branch is connected via a canal-like interzooidal pore. In living cheilostomes strands of mesenchymal cells form the funicular system (see Bobin 1977) which links with the interzooidal pores and is thought to function as a system for the colonial transportation of metabolites (see Ryland 1979). Colonial integration Colonial organisms can be arranged in hypothetical series from poorly integrated to highly integrated. In poorly integrated colonies, the colony behaves in the manner of a simple aggregation of clonal solitary individuals, whereas in highly integrated colonies it behaves as a ‘superindividuaf whose constituent zooids can have no independent existence. Although this is a useful conceptual way of viewing colonial animals, integration is a rather nebulous concept; it is impossible to quantify integration in a consistent way because the various morphological features (e.g. presence of polymorphism, extrazooidal tissue, astogenetic variation, etc.) which have been used for estimating integration in the absence of any firm physiological data, cannot be combined into a single measure of integration along a linear scale. Integrational states for particular morphological parameters can, however, be estimated, and may potentially (as in Herpetopora) yield conflicting results. The runner-like colony form of Herpetopora is poorly integrated along the spectrum of colony forms found in bryozoans (see Boardman and Cheetham 1973; Jackson \919b). The absence of extrazooidal parts, of shared interior skeletal walls, or marked ontogenetic changes in zooid morphology, are also features of a poorly integrated bryozoan. Set against these are indicators of high integration in the several types of polymorphs and their often predictable positions, the subcolonial organization into branches with autonomous astogenetic gradients, the apparent chemotropic growth of branches towards open lateral pore windows, and the abundant reparative structures. The last two features are particularly noteworthy. Herpetopora colonies were clearly able to survive substantial damage and mortality of zooids which often led to fission of the colony into several spatially separated ramets (sensu Harper 1977), as in the Palaeozoic cyclostome Corynotrypa (Taylor 1985). However, repair of damage, and the ability to grow towards pore windows, presumably by chemotropism, gave the potential for re-establishing communication between ramets, and multiplying communications within ramets. What functional value might there have been in this ‘replumbing’? Herpetopora colonies do not seem to have possessed active heterozooids such as avicularia whose energy requirements must presumably be supplied by nutrients gained from autozooids. However, it was probably advantageous for previously disconnected parts of the colony to be re-linked in order that all feeding zooids could contribute energy to the formation of new buds in growing regions of the colony. Functional morphology Colony growth pattern in runner-like bryozoans determines how the zooids, and most importantly their feeding tentacle crowns, are distributed over the area of the substratum. In Herpetopora, astogenetic gradients of increasing zooid length along each branch meant that tentacle crowns of successive zooids were progressively more distantly spaced along the branch, and new branches were spaced at increasing intervals. The density of tentacle crowns would therefore have diminished outwards from the colony origin. Such a distribution is functionally advantageous if the colony origin is located on the optimal area of the substratum. As location of the colony origin depends upon the settlement behaviour of the larva, a highly selective pattern of larval settlement might be predicted in H. laxata. This prediction is not currently testable, however, in view of the scarcity of colonies preserving their origins. TAYLOR: GROWTH IN HERPETOPORA 547 Conversely, colonies with runner-like growth (equivalent to the ‘guerilla’ growth strategy identified in plants, Lovett-Doust 1981) have a high probability of locating spatial refuges where chances of mortality are diminished (Buss 1979). This is because these straggly colonies spread rapidly from the point of colony origin per zooid budded in comparison with compact, sheet-like colonies. Therefore, Herpetopora colonies should have been good at locating refuges on a patchy substratum. One important source of substratum patchiness may have been caused by the activities of grazers and other predators. Abundant evidence of colony damage often followed by repair supports the notion that predators had a significant impact, abiotic damage seeming a less likely option in the relatively tranquil habitats colonized by Herpetopora. While some parts of colonies may have been destroyed or severely damaged by predators, there is a high probability that others would have survived relatively unscathed because of the wide ‘dispersal’ of zooids across the substratum. The refuge-locating abilities of colonies would have been further enhanced if, as seems possible, branches with caudate zooids had high rates of linear growth. Coates and Jackson (1985) predicted that uniserial colonies should have relatively elongate zooids to maximize their ability to locate spatial refuges, and were able to support this prediction with data comparing zooid elongation in uniserial and multiserial encrusting bryozoans. Caudate zooids of Herpetopora have very narrow, tube-like caudae which may have grown rapidly. If so, the rates of linear extension of distal parts of branches would have been great, resulting in enhanced ability to explore areas of substratum distant from the colony origin. From the point of view of colony feeding, the disorderly arrangement of tentacle crowns in colonies of Herpetopora was probably of little importance. The runner-like colony form is normally associated with colonies whose zooids feed autonomously (Winston 1978) rather than co-operating in the formation of colony-wide or subcolony-wide feeding currents (e.g. Cook 1977). Furthermore, calculations suggest that overlap of tentacle crowns may not have occurred even between closely spaced zooids at branch intersections. There is a good correlation between the width of the zooid orifice and the diameter of the tentacle crown in cheilostomes (Winston 1981). Orifice width in Herpetopora can be measured from zooids with closure plates bearing impressions of the operculum. Operculum (and orifice) width is about 0 07 mm in H. laxata , giving an estimated tentacle crown diameter of 0T8 mm using the regression data of Winston. As this value is less than the width of the zooid (0-24 0-45 mm, fide Thomas and Larwood 1960), the extremely small tentacle crowns of Herpetopora probably did not even overlap the edges of the zooids. Boardman et at. (1983, fig. 4) give a reconstruction of a primitive cheilostome zooid of similar skeletal morphology to Herpetopora showing the very small size of the tentacle crown. There is considerable scope for further work on the functional morphology of Herpetopora and its relationship to the ecology of the substrata encrusted. Using the branching parameters already determined from fossil colonies, it would be particularly instructive to simulate model colonies (cf. Bell 1986) for the purpose of ascertaining: а, the exact distribution of zooid tentacle crowns across the substratum; б, the rate of change of zooid numbers during colony growth (assuming either constant linear growth rate or constant budding rate); c, the effects of obstacles and areas of partial mortality on growth pattern and ultimate size of the colony. Acknowledgements Impetus for this research was gained from the dissertation work of Andrew Butterworth (University of Bristol), and from the discovery of a relatively intact colony among BGS specimens loaned to Julian Hammond. John Bishop, Julian Hammond, Beth Okamura, and Professor E. Voigt kindly commented on the manuscript. 548 PALAEONTOLOGY, VOLUME 31 REFERENCES anstey, r. l., pachut, j. f. and PREZBiNDOWSKi, d. r. 1976. Morphogenetic gradients in Paleozoic bryozoan colonies. Paleobiology, 2, 131-146. banta, w. c. 1969. The body wall of cheilostome Bryozoa. II. Interzooidal communication organs. J. Morph. 129, 149-170. 1975. Origin and early evolution of cheilostome Bryozoa. Docum. Lab. Geol. Fac. Sci. Lyon , H.S. 3, 565-582. bell, a. d. 1986. The simulation of branching patterns in modular organisms. Phil. Trans. R. Soc. Load. B 313, 143-159. best, m. a. and thorpe, J. p. 1985. Autoradiographic study of feeding and the colonial transport of metabolites in the marine bryozoan Membranipora membranacea. Mar. Biol. Berlin , 84, 295 -300. boardman, r. s. and cheetham, a. h. 1973. Degrees of colony dominance in stenolaemate and gymnolaemate Bryozoa, 121-220. In boardman, r. s., cheetham, a. h. and Oliver, w. a. (eds.). Animal colonies. Dowden, Hutchinson and Ross, Stroudsburg. — et al. 1983. Bryozoa. In moore, r. c. and robison, r. a. (eds.). Treatise on invertebrate paleontology, part G (revised), volume 1, 625 pp. Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas. bobin, g. 1977. Interzooecial communications and the funicular system, 307-333. In woollacott, r. m. and zimmer, r. l. (eds.). Biology of bryozoans. Academic Press, New York. buss, L. w. 1979. Habitat selection, directional growth, and spatial refuges: why colonial animals have more hiding places, 459-497. In larwood, g. p. and rosen, b. r. (eds.). Biology and systematics of colonial organisms. Academic Press, London. carthew, r. 1987. The cyclostome bryozoan Corynotrypa from the Silurian of Gotland, Sweden, 57-64. In ross, J. R. p. (ed.). Bryozoa: past and present. Western Washington University, Bellingham. coates, a. g. and jackson, j. b. c. 1985. Morphological themes in the evolution of clonal and aclonal marine invertebrates, 67-106. In jackson, j. b. c., buss, l. w. and cook, r. e. (eds.). Population biology and evolution of clonal organisms. Yale University Press, New Haven. cook, p. l. 1977. Colony-wide water currents in living Bryozoa. Cah. Biol. mar. 18, 31 47. dzik, J. 1975. The origin and early phylogeny of the cheilostomatous Bryozoa. Acta paleont. pol. 20, 395- 423. - 1981. Evolutionary relationships of the early Palaeozoic ‘cyclostomatous’ Bryozoa. Palaeontology , 24, 827 861. Gardiner, a. r. and taylor, p. d. 1982. Computer modelling of branching growth in the bryozoan Stomatopora. Neues Jb. Geol. Palaont. Abh. 163, 389-416. hakansson, e. 1975. Population structure of colonial organisms. A palaeoecological study of some free-living Cretaceous bryozoans. Docum. Lab. Geol. Fac. Sci. Lyon, H.S. 3, 385-399. harper, j. l. 1977. Population biology of plants, 892 pp. Academic Press, London. — 1985. Modules, branches, and the capture of resources, 1-33. In jackson, j. b. c., buss, l. w. and cook, r. e. (eds.). Population biology and evolution of clonal organisms. Yale University Press, New Haven. jackson, j. b. c. 1979m Overgrowth competition between encrusting cheilostome ectoprocts in a Jamaican cryptic reef environment. J. Anim. Ecol. 48, 805-823. 1979b. Morphological strategies of sessile animals, 499-555. In larwood, g. p. and rosen, b. r. (eds.). Biology and systematics of colonial organisms. Academic Press, London. jebram, d. 1978. Preliminary studies on ‘abnormalities’ in bryozoans from the point of view of experimental morphology. Zool. Jb. Anat. Bd. 100, 245-275. knight-jones, e. w. and moyse, j. 1961. Intraspecific competition in sedentary marine animals. Symp. Soc. Exp. Biol. 15, 72-95. lang, w. d. 1914. On Herpetopora, a new genus containing three new species of Cretaceous cheilostome Polyzoa. Geol. Mag., ns, decade 6, 1, 5-8. levinsen, g. m. r. 1907. Sur la regeneration totale des Bryozoaires. Oyers. Kgl. Danske vidensk. selsk. Forhandl. 1907, 151 159. lidgard, s. 1985. Zooid and colony growth in encrusting cheilostome bryozoans. Palaeontology, 28, 255- 291. lovett-doust, l. 1981. Population dynamics and local specialization in a clonal perennial (Ranunculus repens). I. The dynamics of ramets in contrasting habitats. J. Ecol. 69, 743 755. marcus, E. 1938. Bryozoarios marinhos Brasileiros II. Bol.fac. Phil. Sci. e Letr., Univ. Sao Paulo, 4, 1 37. TAYLOR: GROWTH IN HERPETOPORA 549 — 1949. Some Bryozoa from the Brazilian Coast. Comun. zool. Mus. Hist. nat. Montev. 3 (53), 1 33. 1955. Notas sobre briozoos marinhos Brasileiros. Arqu. Mus. Nac. Rio dc Janeiro , 42, 273 324. podell, m. e. and anstey, r. l. 1979. The interrelationship of early colony development, monticules and branches in Palaeozoic bryozoans. Palaeontology , 22, 965 982. pohowsky, R. A. 1973. A Jurassic cheilostome from England, 447 461. In larwood, g. p. (ed . ). Living and fossil Bryozoa. Academic Press, London. ryland, j. s. 1977. Taxes and tropisms of bryozoans, 411 436. In woollacott, r. m. and zimmer, r. l. (eds.). Biology of bryozoans. Academic Press, New York. 1979. Structural and physiological aspects of coloniality in Bryozoa, 211 242. In larwood, g. p. and rosen, B. r. (eds.). Biology and systematic s of colonial organisms. Academic Press, London. silen, l. 1977. Polymorphism, 183 231. In woollacott, r. m. and zimmer, r. l. (eds.). Biology of bryozoans. Academic Press, New York. — 1987. Colony growth pattern in Electro pilosa (Linnaeus) and comparable encrusting bryozoans. Acta zool., Stockh. 68, 17 34. taylor, p. d. 1985. Carboniferous and Permian species of the cyclostome bryozoan Corynotrypa Bassler, 1911 and their clonal propagation. Bull. Br. Mus. nat. Hist. (Geol.), 38, 359 372. — 1986a. The ancestrula and early growth pattern in two primitive cheilostome bryozoans: Pyripora catenularia (Lleming) and Pyriporopsis portlandensis Pohowsky. J. nat. Hist. 20, 101 110. — 1986/l Charixa Lang and Spinicharixa gen. nov., cheilostome bryozoans from the Lower Cretaceous. Bull. Br. Mus. nat. Hist. (Geol.), 40, 197 222. — 1986c. Scanning electron microscopy of uncoated fossils. Palaeontology , 29, 685 690. — 1987. Skeletal morphology of malacostegan grade cheilostome Bryozoa, 269 -276. In ross, j. r. p. (cd. ). Bryozoa: past and present. Western Washington University, Bellingham. — and furness, r. w. 1978. Astogenetic and environmental variation of zooid size within colonies of Jurassic Stomatopora (Bryozoa, Cyclostomata). J. Paleont. 52, 1093 I 102. thomas, H. D. and larwood, G. P. 1960. The Cretaceous species of Pyripora d’Orbigny and Rhammatopora Lang. Palaeontology , 3, 370-386. urbanek, a. 1973. Organization and evolution of graptolite colonies, 441 514. In boardman, r. s., cheetham, a. H. and Oliver, w. A. (eds.). Animal colonies. Dowden, Hutchinson and Ross, Stroudsburg. voigt, e. 1982. Uber Pyripora huckei Buge (Bryoz. Cheilostomata) in Geschieben des Holsteiner Gesteins (Unt. Miozan). Der Geshiebesammler , 16, 49-56. winston, j. e. 1978. Polypide morphology and feeding behavior in marine ectoprocts. Bull. Mar. Sci. 28, 1 31. — 1981. Leeding behavior of modern bryozoans, I -21. In dutro, j. t. and boardman, r. s. (organisers). Lophophorates. Notes for a short course. University of Tennessee, Department of Geological Sciences, Studies in Geology, 5. Typescript received 12 March 1987 Revised typescript received 8 July 1987 P. D. TAYLOR Department of Palaeontology British Museum (Natural History) London SW7 5BD NEW CHAETETIFORM TREPOSTOME BRYOZOA FROM THE UPPER MISSISSIPPIAN OF THE WESTERN UNITED STATES by JUNE R. P. ROSS Abstract. Two new bryozoan (ectoproct) genera are part of the fauna of the Late Mississippian carbonate facies in Utah, Idaho, Wyoming, and Montana. Helenopora gen. nov. is widespread and abundant whereas Astralochoma gen. nov. is more restricted and sparse. Gross external features of the bryozoan colonies (size, colony form, and shape of zooecial tubules) resemble chaetetiform colonies. Internal characteristics of the two new genera show them to be bryozoans with distinctive budding patterns. Because of the distinctive features of these two genera, the new families Hclenoporidae and Astralochomidae are erected to accommodate them. Chaetetiform trepostomatous bryozoans (ectoprocts) with subspheroidal or hemispheroidal colonies are found in Late Mississipian (Chesterian) faunas of the carbonate shelf facies that extended along the edge of the craton from Utah to Idaho, Wyoming, and Montana (text-fig. 1a). One genus, Helenopora gen. nov., was generally common and widespread in these shelf faunas whereas the other genus, Astralochoma gen. nov., appears to be much less common and has been found thus far only in Utah and Wyoming. The bryozoan colonies appear to have been located on shallow, shoreward edges of carbonate buildups, particularly where currents moved sand and other detritus. Features of sedimentation suggest this western outer shelf and trough region probably had good circulation of marine waters. Southern Utah and south-eastern Nevada, the area where the bryozoans lived, was most likely a cul-de-sac between the Antler orogenic belt and the transcontinen- tal arch, at least for part of Chesterian time. At localities in Utah and Wyoming, Helenopora is associated with corals and brachiopods. It predominates in what has been called the Caninia ( Siphonophyllial ) Zone of the western interior of the United States. The Caninia Zone in the western United States (Sando et al. 1975) is identified as Western Interior North American Coral Zone V in the biostratigraphic sequence of Sando and Bamber (1979, 1984, 1985). Based on Foranrinifera, this zone is correlated (text-fig. 1b) with the middle and upper parts of the Chesterian Series of the midcontinent region of the United States and with the latest Visean (V3 ) and possibly early Namurian (Ei and E2) of Europe (Sando 1975; Sando and Bamber 1984, 1985)3. The extraordinary external similarity of Helenopora and Astralochoma in colony form, size, and tubular structure of the zooecia to Chaetetes has resulted in the bryozoan colonies being mistaken for Chaetetes, particularly in the field. The late Helen Duncan, US Geological Survey, recognized the homeomorphy and segregated for study some of chaetetiform bryozoans. The specimens de- scribed in this report are part of that material. Duncan (in Tooker and Roberts 1970) identified the occurrence of two genera, here named Helenopora and Astralochoma (Chaetetiform bryozoan n. gen. A and Chaetetiform bryozoan n. gen. B, respectively, in Duncan’s terminology; Tooker and Roberts 1970, table 1). The two genera are both present, but in different beds, in the Green Ravine Formation, Upper Mississippian, in the northern Oquirrh Mountains, Utah (region 2; text-fig. 1a). Sando (1975) discussed and illustrated the two chaetetiform trepostomes, still calling them Chaete- tiform bryozoan genus A and Chaetetiform bryozoan genus B, from the Amsden Formation, Salt River Range, Wyoming (region 7; text-fig. 1a). In that 1975 report, Sando described C. wyomingensis Sando, the first Chaetetes from the Mississippian of North America. This species of Chaetetes, from IPalaeontology, Vol. 31, Part 2, 1987, pp. 551-566, pis. 46-51. | © The Palaeontological Association 552 PALAEONTOLOGY, VOLUME 31 text-fig. 1. a, distribution of localities in regions 1 to 9. Shading on the map delineates areas of similar depositional environments: region 1 is situated on the Great Blue carbonate shelf and slope; region 2 lies on a thrust sheet in the Green Ravine Formation; regions 3 and 4 are situated in the clastic and carbonate margin and basin that lie between the inner clastic shelf and the carbonate shelf and slope; region 5 lies on the Aspen Range carbonate shelf edge; region 6 is on the Surrett Canyon outer carbonate shelf; and regions 7, 8, and 9 lie on the shallow water, carbonate, and fine-grained clastic cratonic shelf-margin and intracratonic basin, b, stratigraphic sections for regions 1 to 9 and correlation of the geological formations. Regions 1 to 9 are listed at the top of the figure. The diagonal shading indicates missing stratigraphic intervals which, except for region 2, represent non-depositional or erosional hiatuses. In region 2, the Green Ravine Formation is the lowest unit exposed on a thrust sheet. This text-figure was compiled from data included in: Lageson et al. (1979): Sando (1976); Sando and Bamber (1979, 1984, 1985); Sando et ai (1975); Sando et al. (1969); Sando et al. (1981); Skipp et al. (1979); Smith and Gilmour (1979); Tooker and Roberts (1963, 1970); and Welsh and Bissell (1979). EXPLANATION OF PLATE 46 Figs. 1-5. Helenopora duncanae gen. et sp. nov. Upper Mississippian. 1, external view of hemispheroidal colony, USNM 419783, x 1, Chainman Shale, Granite Mountain, Confusion Range, Utah (USGS 20547- PC). 2, weathered surface of colony shows longitudinal section, USNM 419784, x Doughnut Formation, near Mount Raymond, Wasatch Mountains, Utah (USGS 14496-PC). 3-5, tangential sections close to colony surface show distal structures within some zooecia, USNM 419785, x 50, Aspen Range Formation, Caribou County, Idaho (USGS 101-A). PLATE 46 ROSS, He/enopora 554 PALAEONTOLOGY, VOLUME 31 the Moffat Trail Limestone fauna, is present at the same locality as the homeomorph chaetetiform bryozoan genus B ( = Astralochoma). Thin sections of the bryozoan colonies rapidly dispel any consideration that these homeomorphs belong to the genus Chaetetes. As Sando (1975) noted, the laminate thin walls, the polymorphism of the zooecia (including mesozooecia), the presence of styles, and the lack of pseudosepta do not permit assignment to the chaetetids. Although the two new families Helenoporidae and Astralo- chomidae are distinctly trepostomes and are not assignable to some other bryozoan order, they differ from established trepostome suborders and families in the combination of a number of diagnostic characters, such as parallel colony growth form, zooecial wall microstructure and other features, and zooecial budding pattern, that are discussed in the family diagnoses. The Helenopori- dae includes certain features that characterize both the suborders Esthonioporoidea and Halloporo- idea (Astrova 1978). For example, the Esthonioporoidea have the primitive character of parallel budding of zooecia from the basal epitheca, thin granular-laminate zooecial wall microstructure, and acanthoforms, but lack mesozooecia and exilazooecia. The Halloporoidea do not display the parallel budding of the zooecia from the basal epitheca, but do have mesozooecia and acanthoforms. The Halloporoidea have a more clearly defined laminate microstructure of the zooecial walls which is lacking in both the Helenoporidae and Astralochomidae. The apparent restricted distribution of these two new trepostome families to a part of the western North American shelf margin and to a relatively thin zone in the upper Mississippian suggests they represent endemic genera with short stratigraphic ranges. On the other hand, because they superficially closely resemble chaetetids, it is possible they are geographically much more wide- spread, but have been misidentified as Chaetetes in hand specimens. Repository of material. National Museum of Natural History, Smithsonian Institution, Washington, DC (USNM) and United States Geological Survey, Paleontological Collections, Smithsonian Institution, Wash- ington, DC (USGS). SYSTEMATIC PALAEONTOLOGY Phylum bryozoa Class STENOLAEMATA Order trepostomata Family helenoporidae fam. nov. Type genus. Helenopora gen. nov. Derivation of name. Family name is derived from the genus Helenopora. Diagnosis. Colonies subspheroidal or hemispheroidal. Parallel type of colony with zooecia arranged more or less parallel to one another and rising from the basal epitheca at about right angles. Maculae present. Zooecia large and tubular with diaphragms thin, rare to common. Zooecial walls very thin and gently crenulate. Microcrystallites in the walls aligned in indistinct laminate pattern. Steep wall laminae almost parallel the zooecial tube. Mesozooecia present but not common. Acanthoforms common, large, and at zooecial wall junctions. Acanthoform wall structure EXPLANATION OF PLATE 47 Figs. 14. Helenopora duncanae gen. et sp. nov. Upper Mississippian. Holotype. USNM 165087. Moffat Trail Limestone Member, Amsden Formation, Moffat Trail, Salt River Range, Wyoming (USGS 22987-PC). 1, longitudinal section shows budding pattern of expanding colony, x 10. 2 and 4, longitudinal sections show thin granular walls, some of which are penetrated by acanthoforms, x 20 and x 50, respectively. 3, deep tangential section shows polygonal zooecia and acanthoforms at junctions of zooecial walls, x 20. PLATE 47 ROSS, Helenopora 556 PALAEONTOLOGY, VOLUME 31 distinctly laminate. Steeply sloping laminae of acanthoforms curve gently convexly in the central region. Zooecia bud from the floor of diaphragms. Remarks. Characters distinctive of Helenoporidae are: thin, indistinctly laminate walls; large zoo- ecia; large acanthoforms with distinctly laminate walls; unique zooecial budding pattern. These characters distinguish the family from all other trepostome families. The term acanthoform is used in preference to acanthostyle to identify a rod-like structure with laminate walls and an axial region. The term carries no inference about its function and no inference as to whether the structure was a solid rod or a hollow tube. Helenopora is the only genus presently known in this family. Occurrence. Late Mississippian of the western interior of the US (Utah, Idaho, Wyoming, and Montana) (text-fig. 1a, west of region 1 near western border of Utah, regions 2 9). Text-fig. 1b gives the stratigraphic units for the specific regions. Specific distribution data are given in the appendix. Genus helenopora gen. nov. Type species. Helenopora duncanae sp. nov. Derivation of name. The genus is named in honour of the late Helen M. Duncan who had an extensive and remarkable knowledge of bryozoans. Diagnosis. See family diagnosis. Additional features are: larger than normal zooecia in clusters that form indistinct maculae. Mesozooecia small, rare, and with polygonal outlines. Remarks. The distinctive budding pattern, colony form, large zooecial tubes with diaphragms, and large acanthoforms characterize this genus which is dissimilar to other trepostome genera. For example, Chondraulus Duncan has a thin laminate colony form, a granular microstructure in its zooecial walls, numerous acanthoforms, and no mesozooecia. H. duncanae sp. nov. is the only species presently known in the genus. Girty (in Mansfield 1927, pp. 68, 69) in a palaeontological report on fossil collections from Idaho referred to a bryozoan genus Anomalopora in species lists, but this genus and the species were never described or figured. In unpublished reports, specimens of Helenopora were informally referred to Anomalopora by several geologists including Helen Duncan. The sample USGS 101 A-PC, collected by Girty from Idaho, contains small, hemispheroidal colonies of Helenopora. Helenopora duncanae sp. nov. Plates 46-49 Type material. Holotype, USNM 165807, locality (20). Paratypes: USNM 419790, locality (21); USNM 419791, locality (18); USNM 419792, locality (19); USNM 419785, 419793, locality (13); USNM 419794, locality (11); USNM 419795, locality (2); USNM 419796, locality (22). Detailed stratigraphic and locality data are listed in the Appendix. Derivation of species name. Dedicated to the late Helen Duncan. EXPLANATION OF PLATE 48 Figs. 1-4. Helenopora duncanae gen. et sp. nov. Upper Mississippian. Holotype. USNM 165087. Moffat Trail Limestone Member, Amsden Formation, Moffat Trail, Salt River Range, Wyoming (USGS 22987-PC). 1 and 2, longitudinal sections show budding pattern of zooecia, x 20 and x 50, respectively. 3 and 4, tan- gential sections show variation in wall thickness and acanthoform diameter, x 50. PLATE 48 ROSS, Helenopora 558 PALAEONTOLOGY, VOLUME 31 Description. Subspheroidal or hemispheroidal colonies (PI. 46, figs. 1 and 2), sometimes laminate, ranging in diameter at the base of the colony from about 3 cm to about 9 cm and in height from 2 cm to 27 cm. Width across the distal part of a large colony may reach more than 25 cm. The colonies vary in appearance from large oversize globular buttons, sometimes with arched convex bases, to large spheroidal or hemispheroidal boulder-size masses. Polygonal to subpolygonal zooecial openings range in size across a colony surface and in different colonies from (0 09-0-46) x (0-05-0-38) mm. In maculae, zooecial openings range from (0-28-0-34) x (0-32-0-46) mm. Number of zooecial openings per square mm ranges from ten to fifteen. In tangential sections (PI. 48, figs. 3 and 4), the narrow, indistinctly laminate zooecial walls generally are 0.008 -0-026 mm in thickness, but they may reach 0-079-0 092 mm, particularly in maculae. The wider walls commonly enclose an acanthoform. In maculae, zooecial wall thickness averages 0-025 mm. Acanthoforms, located at the junctions of zooecial walls, have concentric laminate walls and clear axial regions. Acanthoform diameter is 0 05 0-09 mm. In maculae, the acanthoform diameter is generally 0 07-0-09 mm, however, there is a range from 0-05 to 0-09 mm. In longitudinal sections (PI. 4, figs. 1, 2, 4), thin, gently crenulate zooecial walls have steep wall laminae. Acanthoforms, where present, fill the zooecial walls and have steep sloping laminae in their walls that curve gently convexly in the central region. Zooecial tubes show a distinctive budding pattern with a new zooecial tube budding from the floor of a diaphragm (PI. 47, figs. 1 and 2; PI. 48, figs. 1 and 2; PI. 49, fig. 2). Diaphragms are thin, flat, and generally widely spaced. In some colonies from locality USGS 20253, diaphragms are more numerous indicating a variation in the abundance of diaphragms in different colonies. Mesozooecia, small polygonal tubes, are rare and scattered among the zooecia. There is no increase in numbers of mesozooecia in the poorly defined maculae which are clusters of larger than normal zooecia. Discussion. In some colonies, such as those in samples USNM 419785 (loc. USGS 101A-PC) (PI. 46, figs. 3-5; PI. 49, fig. 1), there appear to be distal structures in a few of the zooecia. The distal structures have an inner circular area and an outer area with radiating lines (between four to eight), like spokes of a wheel, which join the inner area to the zooecial walls. The inner area shows concentric features in some sections (PI. 49, fig. 1). Conti and Serpagli (1987) described a 'cap-like apparatus’ in some zooecia of Hallopora elegantula from the Upper Ordovician of Sardinia. The cap-like apparatus had two parts, an inner raised part which had six to eight porous radial ridges and a slightly depressed outer part of six to eight porous radial ridges. The ridges were more or less aligned from the inner to the outer parts. The structures in Helenopora duncanae are similar, but not identical to those in Hallopora elegantula. Some colonies from localities USGS 15291 and 20255 in Utah and USNM 419786 (loc. USGS 18521-PC) (PI. 49, fig. 5) and USNM 419788 (loc. USGS 18167) (PI. 49, fig. 6) in Idaho show short arrow-shaped spines along the sides of the zooecial walls. This species recolonized surfaces of colonies by means of overgrowths by parallel colony growth (Mannil 1961, fig. 1 a) (PI. 49, figs. 3 and 4), many of which would then extend as laminae further increasing the size of the colony. Seasonal growth may be present in the colonies, but it is not marked by regular thickening across a colony or by alignment of diaphragms across a colony. EXPLANATION OF PLATE 49 Figs. 1-6. Helenopora duncanae gen. et sp. nov. Upper Mississippian. 1, tangential section shows distal structure within zooecium, x 50. USNM 419785. Aspen Range Formation, Caribou County, Idaho (USGS 101-A-PC). 2 and 5, longitudinal sections. 2, shows budding from floor of zooecium; 5, shows spines projecting from zooecial walls, USNM 419786, x 20, Aspen Range Formation, Wells Canyon, Idaho (USGS 18521). 3 and 4, longitudinal sections show overgrowths, USNM 419787 and USNM 419788, respectively, x 20. 3, Doughnut Formation, near Mount Raymond, Utah (USGS 14496-PC); 4, probably in Surrett Canyon Formation, near Arco, Butte County, Idaho (USGS 18167-PC). 6, longitudinal section cuts tips of spines which appear as dark dots, USNM 419788, x 50, see stratigraphic and locality data for 4 above. PLATE 49 ROSS, Helenopora 560 PALAEONTOLOGY, VOLUME 31 Family astralochomidae fam. nov. Type genus. Astralochoma gen. nov. Derivation of name. Family name is derived from the star-studded arrangement of maculae on the colony surface. Diagnosis. Colonies subspheroidal or hemispheroidal. Parallel type of colony with zooecia arranged more or less parallel to one another and rising from the basal epitheca at about right angles. Maculae distinct with one or more clusters of mesozooecia in a central region and enclosed by very large zooecia. Zooecia large and tubular with diaphragms. Zooecial walls thin and granular. Microcrystallites in the walls form no distinct pattern. Mesozooecia numerous, partly surround zooecia but do not isolate zooecia on all sides. Acanthoforms numerous and small and at junctions of zooecia and mesozooecial walls. Acanthoform wall structure thin, indistinct. Zooecia bud by fission and bud from distal wall of preceding zooecium. Remarks. Characters distinctive of Astralochomidae are: thin, indistinct granular walls; large zoo- ecia; numerous small acanthoforms; maculae, large and distinct; zooecial budding by fission of preceding zooecium. These characters distinguish the family from all other trepostome families. Astralochoma is the only genus presently known in this family. Occurrence. Late Mississippian in age; western interior of the US (Utah and Wyoming) (text-fig. 1a, regions 1, 2, 7 and west of region 1 ). Specific distribution data are given in the appendix. Genus astralochoma gen. nov. Type species. Astralochoma helenae sp. nov. Derivation of name. The generic name describes the star-like appearance of the maculae. Diagnosis. See family diagnosis. Additional features are: diaphragms sparse in zooecial tubes. Zooecial walls gently crenulate. Mesozooecia small. Acanthoforms indistinct. Remarks. The distinctive maculae, colony form, large zooecial tubes with granular walls, numerous small acanthoforms, and zooecial budding by fission characterize this genus which is dissimilar to other trepostome genera. Astralochoma helenae sp. nov. Plates 50 and 51 Type material. Holotype, USNM 165086, locality (17). Paratypes: USNM 419797, locality (16); USNM 419798, 419799, locality (1). Detailed stratigraphic and locality data are listed in the Appendix. Description. Subspheroidal or hemispheroidal colonies (pi. 50, fig. 1), sometimes laminate, ranging in diameter at the base of the colony from about 4 cm to 10 cm and in height from 2 to 14 cm. Maculae (PI. 50, fig. 2) vary is size from circular areas 3-8 mm2 to elliptical areas which are larger 10-6 mm2 (axes 1 -4x2-4 mm). The colonies generally appear as large globular buttons or more massive boulder-like concretions. Rounded EXPLANATION OF PLATE 50 Figs 1 6. Astralochoma helenae gen. et sp. nov. Upper Mississippian. 1 and 2, external views of hemispheroidal colony and colony surface with zooecial openings, USNM 419789, x 1 and x 5, respectively. Great Blue Limestone, near Dry Canyon, Oquirrh Mountains, Utah (USGS 21146-PC). 3 and 4, tangential sections show zooecia, mesozooecia, and acanthoforms, holotype, USNM 165086, Moffat Trail Limestone Member, Amsden Formation, Covey Cutoff Trail, Salt River Range, Wyoming (USGS 6965-PC). 5 and 6, longitudinal sections show budding pattern, holotype, USNM 165086, x 10 and x 20 respectively, locality data same as for 3 and 4. PLATE 50 ROSS, Astralochoma 562 PALAEONTOLOGY, VOLUME 31 subpolygonal zooecial openings (PI. 50, figs. 3 and 4; PI. 51, fig. 1) range in size from 0T7 to 0-23 mm. In maculae, zooecial openings (PI. 51, figs 3 and 4) are larger than normal and range in size from 0-28 to 0-36 mm. The number of zooecia per square mm is fourteen to twenty and in maculae the number of zooecia per square mm. is nine to twelve. In tangential section, narrow granular walls, 0-01 5-0-025 mm thick, have small acanthoforms at the junctions of zooecia and mesozooecia (PI. 50, figs. 1 and 2). The wall thickness in maculae is 0 021 0 029 mm. Acantho- forms lie at each corner shared with adjacent zooecia or mesozooecia so that there are between four to six acanthoforms per zooecium. The acanthoforms are 0-017 0-025 mm in diameter and are smaller in the maculae, 0-017 or slightly smaller. The mesozooecia are triangular to rhomboidal and measure 0-06 0-1 0 mm. In maculae, the mesozooecia approximate to squares and measure 0100-0 125 mm. In longitudinal sections (PI. 50, figs. 5 and 6; PI. 51, fig. 2) thin, granular, and gently crenulate walls enclose the zooecia. Thin, flat diaphragms periodically cross the zooecia. Acanthoforms are not particularly distinctive unless the section passes through a junction between three walls of zooecia and mesozooecia. Remarks. Although the colony form is very similar to that of H. duncanae sp. nov., this species has very different characters of the zooecial walls, zooecial budding, structure of the maculae, and the structure and distribution of acanthoforms and mesozooecia. Acknowledgements. I thank Drs J. T. Dutro, Jr., Mackenzie Gordon, Jr., O. Karklins, W. J. Sando, and Mr H. Saunders, US Geological Survey, for their considerable assistance with information about the location of specimens and their stratigraphic occurrence. Dr C. A. Ross, Chevron USA, greatly assisted in the preparation of illustrations and in discussions on the palaeogeography. Support for publication from the Bureau for Faculty Research, Western Washington University, is gratefully acknowledged. REFERENCES astrova, g. G. 1978. The historical development, systematics and phylogeny of the Bryozoa. Trudi Paleont. Inst. Akad. Nauk SSSR , 169, 240 pp. [In Russian.] conti, s. and serpagli, e. 1987. Functional morphology of the cap-like apparatus in autozooids of a Palaeozoic trepostome bryozoan. Lethaia , 20, 1 -20. lageson, d. r., maughan, e. k. and sando, w. j. 1979. The Mississippian and Pennsylvanian (Carboniferous) Systems in the United States— Wyoming. Prof. Pap. US geol. Surv. 1 1 10-U, 38 pp. mannil, r. m. 1961. On the morphology of the hemispheric zoaria of the Trepostomata (Bryozoa). Trudi Inst. Geol. Akad. Nauk Est. SSR , 6, 113-140. [In Russian.] mansfield, g. R. 1927. Geography, geology, and mineral resources of part of southeastern Idaho with descriptions of Carboniferous and Triassic fossils by G. H. Girty. Prof. Pap. US geol. Surv. 152, 453 pp. sando, w. j. 1975. Coelenterata of the Amsden Formation (Mississippian and Pennsylvanian) of Wyoming. Ibid. 848-C, 31 pp. 1976. Mississippian history of the northern Rocky Mountains region. J. Res. US geol. Surv. 4, 317 338. — and bamber, e. w. 1979. Coral zonation of the Mississippian System of western North America. 9th Internat. Cong. Carb. Strut. Geol., Urbana, Illinois, 1979. Abstracts of Papers, p. 191. 1984. Coral zonation of the Mississippian System of western North America. In Sutherland, p. k. and manger, w. l. (eds.). Neuvieme Congres International de Stratigraphie et de Geologie du Carbonifere, Compte Rendu 2, Biostratigraphy, 289-300. Southern Illinois University Press, Carbondale and Edwardsville. — 1985. Coral zonation of the Mississippian System in the western interior province of North America. Prof. Pap. US geol. Surv. 1334, 61 pp. EXPLANATION OF PLATE 51 Figs. 1 4. Astralochoma helenae gen. et sp. nov. Upper Mississippian. Holotype. USNM 165086. Moffat Trail Limestone Member, Amsden Formation, Covey Cutoff Trail, Salt River Range, Wyoming (USGS 6965- PC). 1, 3, 4, tangential sections. 1, area between maculae with regular, subrounded-polygonal zooecia, x 50; 3 and 4, areas with maculae (aggregations of mesozooecia) and enlarged zooecia adjacent to maculae, x 50 and x 20, respectively. 2, longitudinal section shows thin granular and gently undulate walls and zooecia budding by fission, x 50. PLATE 51 ROSS, Astralochoma 564 PALAEONTOLOGY, VOLUME 31 sando, w. j., Gordon, m., jr. and dutro, J. t., jr. 1975. Stratigraphy and geologic history of the Amsden Formation (Mississippian and Pennsylvanian) of Wyoming. Prof. Pap. US geol. Surv. 848-A, 83 pp. — mamet, b. l. and dutro, J. T. jr., 1969. Carboniferous megafaunal zonation in the northern Cordillera of the United States. Ibid. 613-E, 29 pp. Sandberg, c. a. and gutschick, r. c. 1981. Stratigraphic and economic significance of Mississippian sequence at North Georgetown Canyon, Idaho. Bull. Am. Ass. petrol. Geol. 65, 1433-1443. skipp, b., sando, w. j. and hall, w. e. 1979. The Mississippian and Pennsylvanian (Carboniferous) Systems in the United States— Idaho. Prof. Pap. US geol. Surv. 1 1 10-AA, 42 pp. smith, d. L. and gilmour, e. h. 1979. The Mississippian and Pennsylvanian (Carboniferous) Systems in the United States— Montana. Ibid. 1110-X, 32 pp. tooker, e. w. and Roberts, r. j. 1963. Comparison of Oquirrh Formation sections in the northern and central Oquirrh Mountains, Utah. Ibid. 450-E, E32-E36. — 1970. Upper Paleozoic rocks in the Oquirrh Mountains and Bingham Mining District, Utah, with a section on biostratigraphy and correlation by Gordon, M., Jr. and H. M. Duncan. Ibid. 629-A, 76 pp. welsh, j. e. and bissell, h. j. 1979. The Mississippian and Pennsylvanian (Carboniferous) Systems in the United States— Utah. Ibid. 1 1 10-Y, 35 pp. Typescript received 3 1 March 1987 Revised typescript 1 December 1987 JUNE R. P. ROSS Department of Biology Western Washington University Bellingham, WA 98225, USA APPENDIX Locality data and occurrence of species The trepostome-bearing samples are from regions (numbered I to 9 on text-fig. 1 ) which during the late Mississippian were located on various parts of the western shelf and shelf margin of the North American craton. In these regions, different stratigraphic nomenclatures are applied to upper Mississippian strata and emphasize differences in their depositional histories and sedimentary environments (Text-fig. 1). Utah. Region 1. Locality (1). USGS collection 21 146-PC. Stockton 15-minute quadrangle, Tooele County. Near centre N| sec. 21, T. 5 S., R. 4 W. in fault block in Great Blue Limestone about 61 m (200 ft.) west of Lakes Killarney fault, 15-30m (50 1 00 ft.) in elevation below top of ridge and on its south slope, and ^-mile north-west of mouth of Dry Canyon. Upper part of Great Blue Limestone. Collected by M. Gordon, Jr. and E. W. Tooker in 1962. A. helenae. West of region 1. Locality (2). USGS collection 20547-PC. Confusion Range, near the western border of Utah. Granite Mountain, E| S| NW^ sec. 18, T. 14 S., R. 16 W. From 7 m (30 ft.) of beds with large Caninia specimens. On the western edge of the Mississippian Great Blue carbonate shelf where it became predominantly clastic and about at the same latitude as the previous collection. Chainman Shale. Collected by M. Gordon, Jr., H. Duncan, R. A. Lewandowski, and A. Rieke in 1961 . Helenopora duncanae and A. helenae. Region 2. Locality (3). USGS collection 16330-PC. Farnsworth Peak 7 ^-minute quadrangle, Tooele County. Green Ravine-Rogers Canyon measured section, Oquirrh Mountains (Tooker and Roberts 1970). On ridge south of Green Ravine below tramway tower, approximately on 5400-ft. contour in the NWJ SW^ sec. 6, T. 2 S., R. 3 W. Green Ravine Formation. Collected in limestone, 62 m (203 ft.) above base of unit 11 of type section, 327 m (1074ft.) above base of formation. Collected by E. W. Tooker and R. J. Roberts in 1956. A. helenae. Locality (4). USGS collection 17143-PC. Farnsworth Peak 7|-minute quadrangle, Tooele County. Green Ravine-Rogers Canyon measured section, Oquirrh Mountains (Tooker and Roberts 1970). Below 5400-foot contour, a few feet up north slope from bottom of Green Ravine, about one-eighth mile east of hill 5244' in the SE| SW| NW| sec. 6, T. 2 S., R. 3 W. Green Ravine Formation. Collected from interbedded limestone and argillaceous limestone, 78-6m (258ft.) from base of unit 11 of type section, 344m (1129ft.) above the base of the formation. Collected by M. Gordon, Jr. and R. J. Roberts in 1957. H. duncanae. Locality (5). USGS collection 20253-PC. Farnsworth Peak 7j-minute quadrangle, Tooele County. Green Ravine-Rogers Canyon measured section, north end of Oquirrh Mountains (Tooker and Roberts 1970). North slope of Green Ravine at 5320-foot contour in SW| SE4 NW| sec. 6, T. 2 S., R. 3 W. Green Ravine ROSS: MISSISSIPPI AN TREPOSTOME BRYOZOA 565 Formation. Collected from limestone about 49m (160ft.) above base of unit 11 of type section, 314m (1031ft.) above base of formation. Collected by H. Duncan, E. W. Tooker, and R. A. Lewandowski in 1961. H. duncanae. Locality (6). USGS collection 20255-PC. Farnsworth Peak 74-minute quadrangle, Tooele County. Green Ravine-Rogers canyon measured section, Oquirrh Mountains (Tooker and Roberts 1970). North of Green Ravine near middle of west line of SE{ NWj sec. 6, T. 2 S., R. 3 W., at 5400-foot contour. Green Ravine Formation. Collected from interbedded fissile limestone and argillaceous limestone, 26-5 m (87 ft.) above base of unit 12 of type section, 393m (1289 ft.) above base of formation. Collected by H. M. Duncan, E. W. Tooker, and R. A. Lewandowski in 1961. H. duncanae. Locality (7). USGS collection 20256-PC. Farnsworth Peak 74-minute quadrangle, Tooele County. Green Ravine-Rogers Canyon measured section, Oquirrh Mountains (Tooker and Roberts 1970). On ridge north of Green Ravine in the NE| SW,| NW| sec. 6, T. 2 S., R. 3 W., at 5320-foot contour. Green Ravine Formation. Collected from argillaceous limestone, 91 m (298ft.) above base of unit I I of type section, 356m (1 169 ft.) above base of formation. Collected by H. M. Duncan, E. W. Tooker, and R. A. Lewandowski in 1961. H. duncanae. Locality (8). USGS collection 20257-PC and USGS collection 21131 -PC. Farnsworth Peak 7^-minute quadrangle, Tooele County. Green Ravine-Rogers Canyon measured section, Oquirrh Mountains (Tooker and Roberts 1970). On ridge north of Green Ravine, same as locality (4). Green Ravine Formation. Collected by H. M. Duncan, M. Gordon, Jr., E. W. Tooker, and R. A. Lewandowski in 1961. H. duncanae. Locality (9). USGS collection 21130-PC. Farnsworth Peak 74-minute quadrangle, Tooele County. Green Ravine-Rogers Canyon measured section, Oquirrh Mountains (Tooker and Roberts 1970). Summit of ridge north of Green Ravine, about one-eighth mile east of knoll 5244'. Same general locality as USGS collection 20257-PC. Green Ravine Formation. Collected from 2-4 m (8 ft.) of dark grey shaly limestone immediately underlying base of limestone bed, 60m (198 ft.) above base of unit 1 1 of type section, 326 m (1069 ft.) above base of formation. Collected by H. M. Duncan, E. W. Tooker, and R. A. Lewandowski in 1961. H. duncanae. Region 3. Locality (10). USGS collection 14496-PC. Morgan 15-minute quadrangle. About 2 miles east- north-east of Morgan, near Mount Raymond, Wasatch Mountains. North slope of east trending ridge which is the next ridge south of the one on which the type Morgan section was measured (Tooker and Roberts 1970). NWi sec. 29, T. 4 N., R. 3 E. Upper part of Doughnut Formation. Collected by M. Gordon, Jr., E. Yochelson, and M. Crittenden in 1953. H. duncanae. Region 4. Locality (11). USGS collection 15177-PC. Mount Pisgah 74-minute quadrangle, Cache County. T. 10 N., R. 1 W., SWi Limestone 23 m (75 ft.) stratigraphically below cairn at top of Dry Lake section hill. Northern facies of the Great Blue Limestone where it starts to grade laterally into the inner shelf Doughnut Formation. Great Blue Limestone. Collected by M. Gordon, Jr., G. Sohn, and P. Knopf in 1954. H. duncanae. Locality (12). USGS collection 15291-PC. Mount Pigsah 74-minute quadrangle, Cache County. T. 10 N., R. 1 W., SWi. Bryozoan from float 9 m (30 ft.) stratigraphically below station 18 in Dry Lake section. Great Blue Limestone. In its northern facies where it starts to grade laterally into the inner shelf Doughnut Forma- tion. Collected by M. Gordon, Jr. H. duncanae. Idaho. Region 5. Locality (13). USGS collection 101A-PC. Crow Creek 15-minute quadrangle, Caribou County. Sec. 10, T. 10 S., R. 45 E. Shelf margin Aspen Range Formation. Probably from thick-bedded limestone (Girty in Mansfield 1927). Collected by G. H. Girty in 1911. H. duncanae. Locality (14). USGS collection 18521-PC. Crow Creek 15-minute quadrangle. Caribou County. Wells Canyon, north side along tributary creek about in NE sec. 9, T. 10 S., R. 45 E. and above stream to west. About 15 to 20 m (50 to 60 ft.) east of easterly point in first switchback of road and about 6 m (20 ft.) below road. Shelf margin Aspen Range Formation. Probably from thick-bedded limestone. Collected by W. D. Keller and J. S. Williams in 1935. H. duncanae. Region 6. Locality (15). USGS collection 18167-PC. Near Arco, Butte County, Idaho. North-east and uphill about one-half mile on Main Street. The road went north-east to a small pass and the outcrops were on the south-west side of road. At 6 m (20 ft.) above lowest outcrop, there was a bed with many ‘ Cliaetetes '. Outer edge of the thick carbonate wedge which formed on the Mississippian cratonic shelf margin. Collected by J. S. Williams in 1953. This collection is probably from the Surrett Canyon Formation. II. duncanae. Wyoming. Region 7. The Mississippian shallow water, mixed clastic-carbonate shelf area of south-western Wyoming forms region 7 and includes five collections from the Moffat Trail Limestone Member of upper part of the Amsden Formation, Salt River Range, Lincoln County, Wyoming (Sando et al. 1975): Locality (16). USGS collection 6957-PC. Bear Creek, Salt River Range. NE| SW{ SW4 sec. 27, T. 33 N., R. 117 W. Collected by J. S. Williams in 1931. A. helenae. 566 PALAEONTOLOGY, VOLUME 31 Locality (17). USGS collection 6965A-PC. Covey Cutoff Trail, Salt River Range. NW| NE^ sec. 27, T. 34 N., R. 1 17 W. Limestone 84 m (277 ft.) above base of Amsden Formation. Collected by J. S. Williams in 1931. A. helenae. Locality (18). USGS collection 17907-PC. Haystack Peak section. Salt River Range. Centre sec. 19, T. 34 N., R. 1 17 W. 61 -64m (199-209 ft.) above base of Amsden Formation. Collected by W. .1. Sando and J. T. Dutro, Jr. in 1958. H. duncanae. Locality (19). USGS collection 22986-PC. Moffat Trail, Salt River Range. NW| NE' sec. 3, T. 33 N„ R. 117 W. Limestone 64m (210ft.) above base of Amsden Formation. Collected by J. T. Dutro, Jr. and Mario Suarez in 1966. H. duncanae. Locality (20). USGS collection 22987-PC. Locality same as Locality (19) but 66m (217ft.) above base of Amsden Formation. H. duncanae. Region 8. Locality (21). USGS collection 16209-PC. Hoback Canyon, Teton County, sec. 2, T. 38 N., R. 1 15 W. Float from limestone 23-28 m (76 91 ft.) above top of Darwin Member, Moffat Trail Limestone Member. Along the Mississippian edge of the western Wyoming shallow water, clastic-rich shelf where the Amsden Formation may be subdivided into several additional members. Collected by M. Gordon, Jr., and others in 1955. H. duncanae. Montana. Region 9. Locality (22). USGS collection 17517-PC. Tostin 15-minute quadrangle, Broadwater County. Lombard section. SW^ SE^ sec. 7, T. 4 N., R. 3 E. 73 m (240 ft.) above base of Big Snowy Formation. About 0-25 mile west of railroad station at the town of Lombard, Montana. In the late Mississippian, region 9 in south-central Montana contained a broad channel which marked the northern limit of the Wyoming shallow water shelf and which also served as a connection to the Williston evaporite basin to the east. Collected by J. T. Dutro, Jr. and W. J. Sando. H. duncanae. 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. Dianne Edwards, Department of Plant Science, University College, P.O. Box 78, Cardiff CF1 1XL, who will supply detailed instructions for authors on request (these are published in Palaeontology 1985, 28, pp. 793-800). 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 1988 is £37-50 (U.S. $66) for Institutional Members, and £ 1 8-50 (U.S. $32) 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 1 19 are still in print and are available (post free) together with those listed below: 20. (for 1977): Fossil Priapulid Worms, by s. conway morris. 155 pp., 99 text-figs., 30 plates. Price £16 (U.S. $24). 21. (for 1978): Devonian Ammonoids from the Appalachians and their bearing on International Zonation and Correlation, by m. r. house. 70 pp., 12 text-figs.. 10 plates. Price £12 (U.S. $18). 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. $38). 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. $45). 24. (for 1980): Dinoflagellate Cysts and Acritarchs from the Eocene of Southern England, by j. p. bujak, c. downie, g. l. eaton and g. l. williams. 100 pp., 24 text-figs., 22 plates. Price £15 (U.S. $2-3). 25. (for 1980): Stereom Microstructure of the Echinoid Test, bv a. b. smith. 81 pp., 20 text-figs., 23 plates. Price £15 (U.S. $23). 26. (for 1981): The Fine Structure of Graptolite Periderm, bv p. R. crowther. 1 19 pp., 37 text-figs., 20 plates. Price £25 (U.S. $38). 27. (for 1981): Late Devonian Acritarchs from the Carnarvon Basin, Western Australia, by g. playford and r. s. dring. 78p/7., 10 text-figs., 19 plates. Price £1 5 (U.S. $23). 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. $38). 29. (for 1982): Fossil Cichlid Fish of Africa, by j. a. h. van couvering. 103 pp., 35 text-figs., 10 plates. Price £30 (U.S. $45). 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. $60). 31. (for 1984): Systematic palaeontology and stratigraphic distribution of ammonite faunas of the French Coniacian, by w. j. Kennedy. 160 pp., 42 text-figs., 33 plates. Price £25 (U.S. $38). 32. (for 1984): Autecology of Silurian organisms. Edited by m. g. bassett and j. d. lawson. 295 pp., 75 text-figs., 13 plates. Price £40 (U.S. $6oU 33. (for 1985): Evolutionary Case Histories from the Fossil Record. Edited by j. c. w. cope and p. w. skelton. 202pp., 80 text- figs., 4 plates. Price £30 (U.S. $45). 34. (for 1985): Review of the upper Silurian and lower Devonian articulate brachiopods of Podolia, by o. l. NIKIFOROVA, T. l. modzalevskaya and m. g. bassett. 66 pp., 6 text-figs., 16 plates. Price £10 (U.S. $15). 35. (for 1986): Studies in palaeobotany and palynology in honour of N. F. Hughes. Edited by d. i. batten and d. e. g. BRIGGS. 178 pp., 29 plates. Price £30 (U.S. $50). 36. (for 1986): Campanian and Maastrichtian ammonites from northern Aquitaine, France, by w. j. Kennedy. 145 pp., 43 text-figs., 23 plates. Price £20 (U.S. $35). 37. (for 1987): Biology and revised systematics of some late Mesozoic stromatoporoids, by rachel wood. 89 pp., 31 text- figs., 7 plates. Price £20 (U.S. $35). 38. (for 1987): Taxonomy, evolution, and biostratigraphy of late Triassic-early Jurassic calcareous nannofossils, by p. r. bown. 1 18 pp., 19 text-figs., 15 plates. Price £30 (U.S. $50). 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. $12). Other Publications 1982. Atlas of the Burgess Shale. Edited by s. conway morris. 31 pp., 24 plates. Price £20 (U.S. $30). 1985. Atlas of Invertebrate Macrofossils. Edited by j. w. Murray. Published by Longman in collaboration with the Palaeontological Association, xiii + 241 pp. Price £13-95. Available in the USA from Halsted Press at U.S. $24.95. © The Palaeontological Association, 1988 Palaeontology VOLUME 31 • PART 2 CONTENTS Disarticulated bivalve shells as substrates for encrustation by the bryozoan Cribrilina puncturata in the Plio-Pleistocene Red Crag of eastern England J. D. d. bishop 237 Fish trails in the Upper Carboniferous of south-west England R. HIGGS 255 The oldest freshwater decapod crustacean, from the Triassic of Arizona G. L. miller and s. R. ash 273 Changes in life orientation during the ontogeny of some heteromorph ammonoids t. okamoto 281 Middle Jurassic ammonites of Tibet and the age of the Lower Spiti Shales G. E. G. WESTERMANN and WANG YI-GANG 295 Cretaceous wood-boring bivalves from Western Antarctica with a review of the Mesozoic Pholadidae S. R. A. KELLY 341 Early Cretaceous acolumellate semitectate pollen from Egypt j. H. J. penny 373 Fortipecten takahashii, a reclining pectinid from the Pliocene of north Japan H. HAYAMI and I. HOSODA 419 A Late Triassic cynodont from the American south-west S. G. LUCAS and w. OAKES 445 Faunal and facies dynamics in the Upper Silurian of the Anglo-Welsh Basin l. cherns 451 Morphology and phylogenetic significance of the angiosperm Platanites hebridicus from the Palaeocene of Scotland P. R. CRANE, S. R. MANCHESTER and D. L. DILCHER 503 Colony growth pattern of astogenetic gradients in the Cretaceous cheilo- stome bryozoan Herpetopora P. D. TAYLOR 519 New chaetetiform trepostome Bryozoa from the Upper Mississippian of the western United States J. R. P. ROSS 551 Printed in Great Britain at the University Printing House, Oxford by David Stanford, Printer to the University IN 0031-0239 0 1 I « 5 § : | 5 \JV_DC^/ > V0P 2 v > xgunsgx 5 ^ > 2 CO 2 CO “2 CO *■• 2 saiavaan libraries Smithsonian institution NouniiiSNi nvinoshiisns saiavaan l CO 3 __ CO — CO INSTITUTION NOIlfUUSNl'JNVINOSHlllNS S3 I d VB 3 IT LIBRARI ES^ SMITHSONIAN^ INSTITUTION ^ N — “ > 2 r* 2: r~_2 » > CO jjj | | ^ saiavaan libraries SMiTHS0NiAN~iNST!TUTi0Na,N0iinuiSNi”NviN0SHiii/\is saiavaan L Jfe. I ^ 1 | J * 1 sW 1 ^SK!p 1 1 |. 5* ^ V5>v > ' s > • oo '•*' 2 2o -** 2 W 2 CO INSTITUTION NOlinilJLSNI NVINOSHillNS SBlHVasn LIBRARIES SMITHSONIAN INSTITUTION N CO > CO ~ CO ~ CO CO o 2 O X£!vo?*i22 Z O =J 2 ' _J 2 ^ 2 saiHvnan libraries Smithsonian institution NouniiiSNi nvinoshiiws samvaan i ___ z c ^ Z r- ^ .z F= m Xgft pcj2 ^ 'yW$ m ^ rn rn co — to £ co X 5 co institution NouniiiSNi NViNOSHiiyMS S3iavaan libraries Smithsonian institution n 2 CO 2 CO 2 v... Co 2 CO O 2 _ . > xguussx 2 ^ > S3 1 ava a n_ li b rar s es^smithsonian institution NouniiiSNi nvinoshiiwsws3 i ava a n2i CO ^ CO “ CO INSTITUTION NOSiniilSN! NVINOSHIISNS S3IBVBan LIBRARIES SMITHSONIAN INSTITUTION N 2 f“ \- 2 r- ■ 2 •” 2 H* 3 CO TO 33 > PI ^ XjVASt^X m ___ SaiHVHan LIBRARIES SMITHSONIAN~INSTITUTION%OlinitJLSNI~NVINOSHllWS S3iavnan L 2 w ^ Z \ OT 2 W 2 , #jwi w INSTITUTION NOIIfllllSNI NVIN0SH1IINS^S3 I B VH 3 ll^ll B RAR I ES^SMITHSONIAN INSTITUTION N1 CO = > CO ” CO = CO cc X 5 'w' > ^ > ''40^ S CO 2 CO * 2 to ^ Z CO ES SMITHSONIAN INSTITUTION NOlinillSNI NVIN0SH11WS S3iava8ll LIBRARIES SMITHSONIAI' CO Z = . CO X to q: v%. o 5 NKjsftsy - m A ;Nl“JNVIN0SHIISAIS2S3 I avy an^LI BRAR I ES^MiTHSONIAN-'lNSTITUTION^NOlinillSNlAlVINOSHlIlM “ > z r- z ir- Z ■ r- v 03 33 > 70 m * m ^ m to — co _ to ”C0 ES SMITHSONIAN INSTITUTION NOlinillSNI NVINOSHISINS SBiaVaail LIBRARIES SMITHSONIAI Z X to Z CO Z % § ^ CO •VAS^iX g " S X^osvO^ > ■ INI_NV1N0SH1ISNSWS3 S avaan^LIB RAR I ES^SMSTBSONIAN^SNSTITUTION ^NQIiniiiSNI^NVINOSHIIW — . (O “ in — ■ in — '■» to to o Z o O x^^sjx “ x^sx o 2 —I Z _j z — 1 2 ES SMITHSONIAN INSTITUTION NOlinillSNI NVINQSHiliAIS S3IHVH3n LIBRARIES SMITHSQNIA z E 2 r- . z r= z ^ - ™ X\T'^2;v7^x O . O XvKVaT**-. — o rm > V4T' 2 v > Xfoms<£X 5 ^ 2 to 2 to * z — « *■ z ■ to ES SMITHSONIAN INSTITUTION NOlinillSNI NVIN0SH11NS S3iaVH0l1 LIBRARIES SMITHSONIAI to , to — .... (/) ~ to TO 39 ;ni„nvinoshims S3 lava an libraries Smithsonian institution Noiiniiism nvinoshhw , “ i I #&, § ! /0^ m jg m g' ^ m | m ES SMITHS0NIAN""lNSTITUTI0NWN0linillSNl”NVIN0SHlllrtlSLh'aSV^// ;Ni_NviN0SHims saiavaan libraries Smithsonian institution NoiiniusNi nvinoshiiin — -n- CO *= m —» m ~ k 3 %